摘要:

physicochemical topics, thereby encouraging scientists of different disciplines to contribute their varied viewpoints to a coiiinion theme. A recent Discussion is :- The Royal Soci- of Chemistry- No.75 lntraamolecwlar Kinetics No. 75 in the series, this publication is the result of a general discussion held at the University of Warwick in April 1983. Contents: The Spiers Meniorlal Lecture; Vibrational Redistribution within Excited Electronic States of Polyatomic Molecules Inrraniolecular R e h u t i o n o f 1. vcited States lsomerization of Intcrnal~ncrgy-selected Ions Kinetics of Ion-Molecule Collision Coinple\es in the Gas Phase, E\periinent and Theory lntrainolccular Decay 01' Soinc Open-shell Pulya t o niic Ca lions On tlic Theory u i Iiitrdniolccul~r I n e r g y Transfer Pulsed Laser Preparation and Ouaniuin Superposition Statc Evolution in ReguLtr and Irregular Systems A Ouantuiii-iiicclianical Internal-collision Model for State-sclcctcd Uniinolccular Decoiiiposilio n The Correspondence Principle and Intramolecular Dynamics lntrainoleculdr Dcphasiiig.t'icusecond Evolution of Wavepacket States in a Molecule with Int erinediate-casc level Struct urc Energy Conversion in van der Waals C'u~~iplc\c\ ol s-Tetrarine and Argon Tim-dependent Processes in Polyatuinic Molecules During and After Intense Intrarcd Irradiation Energy Distributions in tlic (.N(X'L+) bragnient froiii tlie Infrared Multiplepholun Dissociation ol' CI. ICN. A Coinparison between 1:xperiiiiental Results and the Predictions ot Statistical Theories of ChFO + Product Energy Partitioning in the Decoiii- position of State-selectively Excited HOON and IIOOD Low-power Inl-rarcd Laser I'hoiolysis o f Tetramethy ldioxetan Uniinolecular Reactions lnduccd by Vibrational Overtone Excitation Uniiiiolecular Decomposition of t-Butylhydro- peroxide by Direct Excitation of the 6-0 0-11 Stretching Overtone I'icosecond-jet Spectroscopy and Photoclieinistry.Energy Redistribution and its Iiiipact'on Coherence, Isoincrization, Ihssociatiun and Solvalioii knergy Redistribution in Large Molecules. Duect St ud y o f In1 rainolucular Rehxa lion in the Gas Phase with Picosecond Gating Rotation-dependent Intrainolecuhr I'r~)cessc.sofSO:(A'A.) in a Superwnic Jct Role of Rotation-Vibration Interaction in Vibrational Keh\ation. Energy Kcdistribution in k,xcitcd Singlet I~'ornialdc1iyde Sub-lhppler.Spectroscopy of Benrcnc in the "('liaiinel-lliree" Region Intraiiiulccular 1:lectronic Kclau~tion and I'liotois~)iiieruati[)n Processes in tlie lsuhted Azabenrene Molecules Pyridinc, Pyrazinc and I'yriiiiidinc Softcover 434pp 0 85186 658 1 Price f25.00 ($48.00) Rest of the World f26.00 RSC Members f 16.25 Faraday Discussions of the Chemical Society 7< lnrruniolei u h r Kincrit I Faraday Symposia are usually held annually and are confined to more specialiscd topics than Discussions, with particular reference to recent rapidly developing lines of rescuch. A recent Symposium is :- No.l?The Hydrophobic Interadion No. 17 in the series, this publication is the result of a symposium on The Hydrophobic Interaction held at the Uiiiversity of Reading in December 1982.Contents: Hydrophobic Interdctionr a llistaric.11 Per spect ivr llydrupliobic Ilydration Geometric Kelaution in Water. Its Role in Precise Vapour-pressure Measureiiients of the SolubilkdtiorI of Benzene by Aqueous Sodiuiii Octylsulphate Solutions Nuclear Magnetic Resonance R e b u t i o n Investigation of Tetrahydrofuran and Methyl Iodide Clathrdtes Infrared and Nuckar Magnetic Kcwnance Studies Pertaining to the (age Model t o r Solutions oS Acetone in Water Irothernial Transport Properties in Solutions o f S y mmet r ica I Tet ra-alk y hmnioniuiii Bromides Thermodynamics of Cavity I'oriiiaiion in Water. A Molecular Dynamics Study Molecular Librations and Solvent Oricnt- ational Correlations in Hydrophobic Phenomena Monte Carlo Computer Siniulation Study of the Hydrophobic Effect.Potential ot Mean Force for ECfir)gaq at 25 and SOv C Hydroplicibic Moments and Protein Structure Application 01' the Kirkwood-Buff Theory to the t'roblcin 01 Hydrophobic Interactions Ihentangleinent of Ilydrophubic and IFlcctrosi~tic Contributions t o the I.ilni Pressures O i Ionic Surfactants llydrophobir. Intcracliuns in Dilute Su lut io ns u t 1'0 1 y (vin y I a Ico lio I) ('onioriii;tiionaI and 1:unc.i ional I'ropertics of tiaeiiwglobin in Water+Alcohol Mixtures. Dependence o f Bull. Electrostatic and tlydrupliohic I n t c r x t i o n s upon ptl and KCI concentrations Softcover 24Opp 0 85186 668 9 Price f36.50 ($70.00) Rest of the World f38.50 RSC Members f 23.75 ORDERING RSC Members should send their orders to: The Royal Society of Chemistry.The Membership Officer. 30 Russell Square, Non-RSC Members The Royal Society of Chemistry, Distribution Centre, Blackhorse Road, L London WC1 B 5DT. Letchworth, Herts SO6 IHN, England. Faradaj Symposia of the Chemical Society hGi 17 I hc HI drophohr' Inrcrm rron 1 9 X ? (viii)physicochemical topics, thereby encouraging scientists of different disciplines to contribute their varied viewpoints to a coiiinion theme. A recent Discussion is :- The Royal Soci- of Chemistry- No.75 lntraamolecwlar Kinetics No. 75 in the series, this publication is the result of a general discussion held at the University of Warwick in April 1983. Contents: The Spiers Meniorlal Lecture; Vibrational Redistribution within Excited Electronic States of Polyatomic Molecules Inrraniolecular R e h u t i o n o f 1.vcited States lsomerization of Intcrnal~ncrgy-selected Ions Kinetics of Ion-Molecule Collision Coinple\es in the Gas Phase, E\periinent and Theory lntrainolccular Decay 01' Soinc Open-shell Pulya t o niic Ca lions On tlic Theory u i Iiitrdniolccul~r I n e r g y Transfer Pulsed Laser Preparation and Ouaniuin Superposition Statc Evolution in ReguLtr and Irregular Systems A Ouantuiii-iiicclianical Internal-collision Model for State-sclcctcd Uniinolccular Decoiiiposilio n The Correspondence Principle and Intramolecular Dynamics lntrainoleculdr Dcphasiiig. t'icusecond Evolution of Wavepacket States in a Molecule with Int erinediate-casc level Struct urc Energy Conversion in van der Waals C'u~~iplc\c\ ol s-Tetrarine and Argon Tim-dependent Processes in Polyatuinic Molecules During and After Intense Intrarcd Irradiation Energy Distributions in tlic (.N(X'L+) bragnient froiii tlie Infrared Multiplepholun Dissociation ol' CI.ICN. A Coinparison between 1:xperiiiiental Results and the Predictions ot Statistical Theories of ChFO + Product Energy Partitioning in the Decoiii- position of State-selectively Excited HOON and IIOOD Low-power Inl-rarcd Laser I'hoiolysis o f Tetramethy ldioxetan Uniinolecular Reactions lnduccd by Vibrational Overtone Excitation Uniiiiolecular Decomposition of t-Butylhydro- peroxide by Direct Excitation of the 6-0 0-11 Stretching Overtone I'icosecond-jet Spectroscopy and Photoclieinistry. Energy Redistribution and its Iiiipact'on Coherence, Isoincrization, Ihssociatiun and Solvalioii knergy Redistribution in Large Molecules.Duect St ud y o f In1 rainolucular Rehxa lion in the Gas Phase with Picosecond Gating Rotation-dependent Intrainolecuhr I'r~)cessc.sofSO:(A'A.) in a Superwnic Jct Role of Rotation-Vibration Interaction in Vibrational Keh\ation. Energy Kcdistribution in k,xcitcd Singlet I~'ornialdc1iyde Sub-lhppler. Spectroscopy of Benrcnc in the "('liaiinel-lliree" Region Intraiiiulccular 1:lectronic Kclau~tion and I'liotois~)iiieruati[)n Processes in tlie lsuhted Azabenrene Molecules Pyridinc, Pyrazinc and I'yriiiiidinc Softcover 434pp 0 85186 658 1 Price f25.00 ($48.00) Rest of the World f26.00 RSC Members f 16.25 Faraday Discussions of the Chemical Society 7< lnrruniolei u h r Kincrit I Faraday Symposia are usually held annually and are confined to more specialiscd topics than Discussions, with particular reference to recent rapidly developing lines of rescuch.A recent Symposium is :- No.l?The Hydrophobic Interadion No. 17 in the series, this publication is the result of a symposium on The Hydrophobic Interaction held at the Uiiiversity of Reading in December 1982. Contents: Hydrophobic Interdctionr a llistaric.11 Per spect ivr llydrupliobic Ilydration Geometric Kelaution in Water. Its Role in Precise Vapour-pressure Measureiiients of the SolubilkdtiorI of Benzene by Aqueous Sodiuiii Octylsulphate Solutions Nuclear Magnetic Resonance R e b u t i o n Investigation of Tetrahydrofuran and Methyl Iodide Clathrdtes Infrared and Nuckar Magnetic Kcwnance Studies Pertaining to the (age Model t o r Solutions oS Acetone in Water Irothernial Transport Properties in Solutions o f S y mmet r ica I Tet ra-alk y hmnioniuiii Bromides Thermodynamics of Cavity I'oriiiaiion in Water.A Molecular Dynamics Study Molecular Librations and Solvent Oricnt- ational Correlations in Hydrophobic Phenomena Monte Carlo Computer Siniulation Study of the Hydrophobic Effect. Potential ot Mean Force for ECfir)gaq at 25 and SOv C Hydroplicibic Moments and Protein Structure Application 01' the Kirkwood-Buff Theory to the t'roblcin 01 Hydrophobic Interactions Ihentangleinent of Ilydrophubic and IFlcctrosi~tic Contributions t o the I.ilni Pressures O i Ionic Surfactants llydrophobir. Intcracliuns in Dilute Su lut io ns u t 1'0 1 y (vin y I a Ico lio I) ('onioriii;tiionaI and 1:unc.i ional I'ropertics of tiaeiiwglobin in Water+Alcohol Mixtures. Dependence o f Bull. Electrostatic and tlydrupliohic I n t c r x t i o n s upon ptl and KCI concentrations Softcover 24Opp 0 85186 668 9 Price f36.50 ($70.00) Rest of the World f38.50 RSC Members f 23.75 ORDERING RSC Members should send their orders to: The Royal Society of Chemistry. The Membership Officer. 30 Russell Square, Non-RSC Members The Royal Society of Chemistry, Distribution Centre, Blackhorse Road, L London WC1 B 5DT. Letchworth, Herts SO6 IHN, England. Faradaj Symposia of the Chemical Society hGi 17 I hc HI drophohr' Inrcrm rron 1 9 X ? (viii)

ISSN:0300-9599    

DOI:10.1039/F198480FX005

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

AUTHOR INDEX xxxv Sabbatini, L., 1029 Sacco, A., 2669 Sanders, J. V., 571 Sangster, D. F., 291 1 Sarka, K., 521 Sasahira, A., 473 Sasse, W. H. F., 571 Satchell, P. W., 2395 Sato, K., 841 Sato, Y., 341 Savino, V., 759 Sayers, C. M., 1217 Schiller, R. L., 1257 Schmidt, J., 1 Schmidt, P. P., 2017 Schneider, H., 3275, 3285 Schulz, R. A., 489, 1323 Scott, J. M. W., 739, 1651, 2287, Scott, S. K., 3409 Segall, R. L., 2609 Sehested, K., 2929, 2969 Seidl, V., 1367 Sem, P., 297 Serratosa, J. M., 2225 Seyama, H., 237 Seyedmonir, S. R., 87, 2269 Shanahan, M. E. R., 37 Sheppard, A., 2999 Sherwood, P. M. A., 135, 2099, Shindo, Y., 879, 2199 Shiotani, H., 2145 Shizuka, H., 383, 341 Siekhaus, W. J., 61 Sircar, S., 1101, 2489 Smart, R. St C., 2957, 2609 Smith, I. M., 3021 Smith, R., 3233 Snow, R.L., 3463 Solar, S., 2929 Solar, W., 2929 Solymosi, F., 1841 Soma, M., 237 Soupart, J-B., 3209 Sourisseau, C., 3257 Spink, J. A., 3469 Spoto, G., 1875, 1891 Spotswood, T. M., 3147 Staricco, E. H., 2631 Stassinopoulou, K., 3095 Stedman, D. H., 285 Stout, D. R., 3481 Strohbusch, F., 1757 Strumolo, D., 1479 Struve, P., 813, 2167 Styring, M. G., 3051 Subramanian, R., 2405 2881, 3359 2549, 2867 Sundar, H. G. K., 3491 Sutcliffe, L. H., 669, 3021 Sutton, H. C., 2301 Sutton, L. E., 635 Suzuki, H., 803 Suzuki, T., 1925, 3157 Symons, M. C. R., 423, 1005, Szamosi, J., 1645 Szczepaniak, W., 2935 Takagi, Y., 1925 Takahashi, K., 803 Takahashi, N., 629 Takanaka, J., 941 Takao, S., 993 Takasaki, S., 803 Takegami, H., 1221 Tam, S-C., 2255 Tamamushi, R., 2751 Tamaru, K., 29, 1567, 1595 Tamilarasan, R., 2405 Tanabe, S., 803 Tanaka, K., 2563,2981 Tanaka, T., 119 Taniewska-Osinska, S., 1409 Tascon, J.M. D., 1089 Teo, H. H., 981, 1787 Tetenyi, P., 3037 Thomas, J. K., 1163 Thompson, L., 1673 Thomson, M., 1867 Thomson, S. J., 1689 Tiddy, G. J. T., 789, 3339 Tittarelli, P., 2209 Tominaga, T., 941 Tomkinson, J., 225 Tonelli, C., 1605 Toprakcioglu, C., 13,413 Tran, T., 1867 Trasatti, S., 913 Tripathi, A. D., 1517 Tronc, E., 2619 Troncoso, G., 2127 Truscott, T. G., 2293 Tsurusaki, T., 879 Tuck, J. J., 309 Turner, P. S., 2609 Tusk, M., 1757 Tvarbikova, Z., 2639 Tyrrell, H. J. V., 1279 Ueki, Y.. 341 Ueno, A., 803 Unno, H., 1059 Valencia, E., 2127 van de Ven, T. G. M., 2677 van Ommen, J. G., 2479 van Truong, N., 3275, 3285 Vargas, I., 1947 2767, 2803, 21 1, 1999 Vedrine, J.C., 1017 Veith, J., 2313 Velasco, J. R., 3429 Vesala, A., 2439 Vickerman, J. C., 1903 Vincent, B., 2599 Vinek, H., 1239 Vink, H., 507, 1297 Waghorne, W. E., 1267 Wagley, D. P., 47 Walker, R. W., 435, 3187, 3195, Wallington, T. J., 2737 Wang, G-W., 1039 Watkins, P. E., 2323 Watkiss, P. J., 1279 Watt, R. A. C., 489 Webb, G., 1689 Webster, B. C., 255, 267 Weiner, E. R., 1491 Wells, C. F., 2155. 2445 Wells, J. D., 1233 Whang, B. C. Y., 291 1 Whittle, E., 2323 Wichterlova, B., 2639 Wiesner, S., 3021 Wilhelmy, D. M., 563 Williams E. H., 3147 Williams, P. A., 403 Williams, R. J. P., 2255 Wokaun, A., 1305 Wolff, T., 2969 Wood, S. W., 3419 Woolf, L. A., 549, 1287 Wright, C. J., 1217 Wu, D. C., 1795 Wiirflinger, A., 3221 Wyn-Jones, E., 1915 Yamabe, M., 1059 Yamamoto, S., 941 Yamashita, H., 1435 Yamauchi, H., 2033 Yamazaki, A., 3245 Yariv, S., 1705 Yasumori, I., 841 Yeates, S.G., 1787 Yide, X., 969, 3103 Ylikoski, J., 2439 Yokokawa, T., 473 Yoneda, N., 879 Yonezawa, T., 1435 Yoshida, S., 119, 1435 Zambonin, P. G., 1029 Zanderighi, L., 1605 Zecchina. A., 2209, 2723, 1875, Zipelli, C., 1777 Zundel, G., 553 348 1, 2827 1891AUTHOR INDEX xxxv Sabbatini, L., 1029 Sacco, A., 2669 Sanders, J. V., 571 Sangster, D. F., 291 1 Sarka, K., 521 Sasahira, A., 473 Sasse, W. H. F., 571 Satchell, P. W., 2395 Sato, K., 841 Sato, Y., 341 Savino, V., 759 Sayers, C. M., 1217 Schiller, R. L., 1257 Schmidt, J., 1 Schmidt, P. P., 2017 Schneider, H., 3275, 3285 Schulz, R. A., 489, 1323 Scott, J. M. W., 739, 1651, 2287, Scott, S.K., 3409 Segall, R. L., 2609 Sehested, K., 2929, 2969 Seidl, V., 1367 Sem, P., 297 Serratosa, J. M., 2225 Seyama, H., 237 Seyedmonir, S. R., 87, 2269 Shanahan, M. E. R., 37 Sheppard, A., 2999 Sherwood, P. M. A., 135, 2099, Shindo, Y., 879, 2199 Shiotani, H., 2145 Shizuka, H., 383, 341 Siekhaus, W. J., 61 Sircar, S., 1101, 2489 Smart, R. St C., 2957, 2609 Smith, I. M., 3021 Smith, R., 3233 Snow, R. L., 3463 Solar, S., 2929 Solar, W., 2929 Solymosi, F., 1841 Soma, M., 237 Soupart, J-B., 3209 Sourisseau, C., 3257 Spink, J. A., 3469 Spoto, G., 1875, 1891 Spotswood, T. M., 3147 Staricco, E. H., 2631 Stassinopoulou, K., 3095 Stedman, D. H., 285 Stout, D. R., 3481 Strohbusch, F., 1757 Strumolo, D., 1479 Struve, P., 813, 2167 Styring, M. G., 3051 Subramanian, R., 2405 2881, 3359 2549, 2867 Sundar, H.G. K., 3491 Sutcliffe, L. H., 669, 3021 Sutton, H. C., 2301 Sutton, L. E., 635 Suzuki, H., 803 Suzuki, T., 1925, 3157 Symons, M. C. R., 423, 1005, Szamosi, J., 1645 Szczepaniak, W., 2935 Takagi, Y., 1925 Takahashi, K., 803 Takahashi, N., 629 Takanaka, J., 941 Takao, S., 993 Takasaki, S., 803 Takegami, H., 1221 Tam, S-C., 2255 Tamamushi, R., 2751 Tamaru, K., 29, 1567, 1595 Tamilarasan, R., 2405 Tanabe, S., 803 Tanaka, K., 2563,2981 Tanaka, T., 119 Taniewska-Osinska, S., 1409 Tascon, J. M. D., 1089 Teo, H. H., 981, 1787 Tetenyi, P., 3037 Thomas, J. K., 1163 Thompson, L., 1673 Thomson, M., 1867 Thomson, S. J., 1689 Tiddy, G. J. T., 789, 3339 Tittarelli, P., 2209 Tominaga, T., 941 Tomkinson, J., 225 Tonelli, C., 1605 Toprakcioglu, C., 13,413 Tran, T., 1867 Trasatti, S., 913 Tripathi, A.D., 1517 Tronc, E., 2619 Troncoso, G., 2127 Truscott, T. G., 2293 Tsurusaki, T., 879 Tuck, J. J., 309 Turner, P. S., 2609 Tusk, M., 1757 Tvarbikova, Z., 2639 Tyrrell, H. J. V., 1279 Ueki, Y.. 341 Ueno, A., 803 Unno, H., 1059 Valencia, E., 2127 van de Ven, T. G. M., 2677 van Ommen, J. G., 2479 van Truong, N., 3275, 3285 Vargas, I., 1947 2767, 2803, 21 1, 1999 Vedrine, J. C., 1017 Veith, J., 2313 Velasco, J. R., 3429 Vesala, A., 2439 Vickerman, J. C., 1903 Vincent, B., 2599 Vinek, H., 1239 Vink, H., 507, 1297 Waghorne, W. E., 1267 Wagley, D. P., 47 Walker, R. W., 435, 3187, 3195, Wallington, T. J., 2737 Wang, G-W., 1039 Watkins, P. E., 2323 Watkiss, P. J., 1279 Watt, R. A. C., 489 Webb, G., 1689 Webster, B. C., 255, 267 Weiner, E. R., 1491 Wells, C. F., 2155. 2445 Wells, J. D., 1233 Whang, B. C. Y., 291 1 Whittle, E., 2323 Wichterlova, B., 2639 Wiesner, S., 3021 Wilhelmy, D. M., 563 Williams E. H., 3147 Williams, P. A., 403 Williams, R. J. P., 2255 Wokaun, A., 1305 Wolff, T., 2969 Wood, S. W., 3419 Woolf, L. A., 549, 1287 Wright, C. J., 1217 Wu, D. C., 1795 Wiirflinger, A., 3221 Wyn-Jones, E., 1915 Yamabe, M., 1059 Yamamoto, S., 941 Yamashita, H., 1435 Yamauchi, H., 2033 Yamazaki, A., 3245 Yariv, S., 1705 Yasumori, I., 841 Yeates, S. G., 1787 Yide, X., 969, 3103 Ylikoski, J., 2439 Yokokawa, T., 473 Yoneda, N., 879 Yonezawa, T., 1435 Yoshida, S., 119, 1435 Zambonin, P. G., 1029 Zanderighi, L., 1605 Zecchina. A., 2209, 2723, 1875, Zipelli, C., 1777 Zundel, G., 553 348 1, 2827 1891

ISSN:0300-9599    

DOI:10.1039/F198480BX007

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

JOURNAL OF THE CHEMICAL SOCIETY FARADAY TRANSACTIONS, PARTS 1 AND I 1 The Journalof the Chemical Society is published in six sections, of which five are termed Transactions; these are distinguished by their subject matter, as follows: Dalton Transactions (Inorganic Chemistry). All aspects of the chemistry of inorganic and organometallic compounds; including bioinorganic chemistry and solid-state inorganic chemistry; of their structures, properties, and reactions, including kinetics and mechanisms; new or improved experimental techniques and syntheses. Furaday Transactions I (Physical Chemistry). Radiation chemistry, gas-phase kinetics, electrochemistry (other than preparative), surface and interfacial chemistry, heterogeneous catalysis, physical properties of polymers and their solutions, and kinetics of polymerization, etc.Faraday Transactions II (Chemical Physics). Theoretical chemistry, especially valence and quantum theory, statistical mechanics, intermolecular forces, relaxation phenomena, spectroscopic studies (including i.r., e.s.r., n.m.r., and kinetic spec- troscopy, etc.) leading to assignments of quantum states, and fundamental theory. Studies of impurities in solid systems. Perkin Transactions I (Organic Chemistry). All aspects of synthetic and natural product organic, organometallic and bio-organic chemistry, including aliphatic, alicyclic, and aromatic systems (carbocyclic and heterocyclic). Perkin Transactions I1 (Physical Organic Chemistry). Kinetic and mechanistic studies of organic, organometallic and bio-organic reactions. The description and application of physicochemical, spectroscopic, and theoretical procedures to organic chemistry, including structure-activity relationships. Physical aspects of bio-organic chemistry and of organic compounds, including polymers and biopolymers.Authors are requested to indicate, at the time they submit a typescript, the journal for which it is intended. Should this seem unsuitable, the Editor will inform the author. The sixth section of the Journal of the chemical Society is Chemical Communications, which is intended as a forum for preliminary accounts of original and significant work, in any area of chemistry that is likely to prove of wide general appeal or exceptional specialist interest. Such preliminary reports should be followed up eventually by full papers in other journals (e.g.the five Transactions) providing detailed accounts of the work. NOTES I t has always been the policy of the Faraday Transactions that brevity should not be a factor influencing accep!ability for publication. In addition however to full papers both sections carry at the end of each issue a section headed ‘Notes’, which are short self-contained accounts of experimental observations, results, or theory that will not require enlargement into ‘full’ papers. The Notes section is not used for preliminary communications. The layout of a Note is the same as that of a paper. Short summaries are required. The procedure for submission, administration, refereeing, editing and publication of Notes is the same as for full papers.However, Notes are published more quickly than papers since their brevity facilitates processing at all stages. The Editors endeavour to meet authors’ wishes as to whether an article is a full paper or a Note, but since there is no sharp dividing line between the one and the other, either in terms of length or character of content, the right is retained to transfer overlong Notes to the full papers section. As a guide a Note should not exceed 1500 words or word-equivalents.NOMENCLATURE AND SYMBOLISM Units and Symbols. The Symbols Committee of The Royal Society, of which The Royal Society of Chemistry is a participating member, has produced a set of recommendations in a pamphlet ‘Quantities, Units, and Symbols’ (1975) (copies of this pamphlet and further details can be obtained from the Manager, Journals, The Royal Society of Chemistry, Burlington House, London W 1 V OBN).These recommendations are applied by The Royal Society of Chemistry in all its publications. Their basis is the ‘ Systeme International d’Unites’ (SI). A more detailed treatment of units and symbols with specific application to chemistry is given in the IUPAC Manual of Symbols and Terminology for Physicochernical Quantities and Units (Pergamon, Oxford, 1979). Nomenclature. For many years the Society has actively encouraged the use of standard IUPAC nomenclature and symbolism in its publications as an aid to the accurate and unambiguous communication of chemical information between authors and readers. In order to encourage authors to use IUPAC nomenclature rules when drafting papers, attention is drawn to the following publications in which both the rules themselves and guidance on their use are given: Nomenclature of Organic Chemistry, Sections A , B, @, D, E, F, and H (Pergamon, Oxford, 1979 edn). Nomenclature of Inorganic Chemistry (Butterworths, London, 197 I , now published by Pergamon).Biochemical Nomenclature and Related Documents (The Biochemical Society, London, 1978). A complete listing of all IUPAC nomenclature publications appears in the Index issues of J. Chem. Soc., Faraday Transactions. It is recommended that where there are no IUPAC rules for the naming of particular compounds or authors find difficulty in applying the existing rules, they should seek the advice of the Society’s editorial staff.( ii)THE FARADAY DIVISION OF THE ROYAL SOCIETY OF CHEMISTRY GENERAL DISCUSSION NO. 77 Interfacial Kinetics in Solution University of Hull, 9-11 April 1984 This Discussion will focus attention on reactions involving liquid-gas, liquid-liquid and liquid-solid interfaces (but it will not include electrode kinetics as such). The subject encompasses processes of fundamental, industrial and environmental importance and includes such topics as the rate of dissolution of reactive gases, kinetics at liquid membranes, metal and solvent extraction, Marangoni effects, heterogeneous catalysis and photocatalysis in solution, and the kinetics of dissolution of minerals and drugs. The aim of the meeting is to bring together workers in these diverse fields to highlight the complementary nature of the problems encountered and of the results obtained, and to disseminate ideas concerning new and effective experimental techniques and novel theoretical approaches. The programme and application form may be obtained from: Mrs Y.A. Fish, The Royal Society of Chemistry, Burlington House, London W1V OBN THE FARADAY DIVISION OF THE ROYAL SOCIETY OF CHEMISTRY GENERAL DISCUSSION NO. 78 Radicals in Condensed Phases University of Leicester, 4-6 September 1984 Organising Committee Professor M. C. R. Symons (Chairman) Dr G. 6. Buxton Dr T. A. Claxton Dr K. A. McLauchlan Professor Lord Tedder Dr R. L. Willson The discussion will be primarily concerned with the structure and reactions of radicals in liquids and solids. It is designed to bring together theoretical work on structure, environmental effects and reactivity with spectroscopic and mechanistic studies directly concerned with radicals.Fundamental aspects will be stressed, and particular attention will be given to new developments including measurement at short time intervals, special solvent effects, and the effects of external fields. A special area for inclusion will be electron gain and loss processes including trapped and solvated electrons, electrochemical reactions, and specific electron capture and electron loss in low-temperature systems. Photochemical charge-transfer processes will also be included. 1 The preliminary programme may be obtained from: Mrs Y. A. Fish, The Royal Society of Chemistry, Burlington House, London W1 V OBN ITHE FARADAY DIVISION OF THE ROYAL SOCIETY OF CHEMISTRY I SYMPOSIUM NO.19 University of Cambridge, 1-3 April 1985 The object of the meeting will be to discuss all aspects of the developing subject of polymeric liquid crystals. The hope is to bring together scientists from the fields of conventional polymer science and monomeric liquid crystals who are active in this field. The discussion is aimed at understanding the following facets: (a) The chemical characteristics that give rise to polymer liquid crystalline behaviour. (b) The nature of the high local anisotropy of these systems and their structural organisation at the molecular, micron and macroscopic levels. (c) The physical properties and their industrial exploitation, with particular reference to the influence of external force fields such as flow, electric and magnetic fields.(d) The inter-relations of polymer liquid crystals with small-molecule mesophases, conventional flexible polymers and biopolymers which exhibit liquid-crystalline behaviour. Contributions are invited for consideration by the Organising Committee. A title and 300- word abstract should be submitted as soon as convenient and not later than 31 May 1984 to: I Professor 8. R. Jennings, Electro-optics Group, Department of Physics, Brunei l University, Uxbridge UB8 3PH. Molecular Electronic Structure Calculations- Methods and ~ Applications University of Cambridge, 12-1 3 December 1984 N.B. Please note change of date Molecular electronic structure calculations have now developed into a powerful predictive tool and are necessary in several different fields to aid the understanding and interpretation of experimental observations.The meeting will review the current state of this rapidly developing discipline and will bring together experts on some of the most advanced methods and their applications. The meeting will provide an opportunity for discussion and comparison of the various techniques currently in use. It will therefore not only be a valuable forum for discussion among research workers in the field, but should also show the non-specialist what theoretical calculations can be expected to achieve now and in the near future. The preliminary programme may be obtained from : Mrs Y. A. Fish, The Royal Society of Chemistry, Burlington House, London W1 V OBN FARADAY DIVISION OF THE ROYAL SOCIETY OF CHEMISTRY GENERAL DISCUSSION NO.79 Polymer Liquid Crystals~~ ~~ FARADAY DIVISION INFORMAL AND GROUP MEETINGS Division London Symposium : The Selective Use and Disposal of Energy in Elementary Processes (Including the Tilden Lecture by I. W. M. Smith) To be held at the Scientific Societies Lecture Theatre, London on 9 February 1984 Further information from Mrs Y. A. Fish, The Royal Society of Chemistry, Burlington House, London W1 V OBN Neutron Scattering Group Quantum Molecular Motion in Crystals and Intercalates and on Surfaces To be held at the University of Nottingham on 26-28 March 1984 Further information from Professor S. Clough, Department of Physics, University of Nottingham, Nottingham NG7 2RD Statistical Mechanics and Thermodynamics Group Thermodynamics of Mixed Polymer Systems To be held at the University of Sheffield on 4-5 April 1984 Further information from Dr G.C. Maitland, Department of Chemical Engineering, Imperial College, London SW7 2BY Division Annual Congress: Electronic Processes in Thin Films and Novel Conductors To be held at the University of Exeter on 16-1 9 April 1984 Further information from: Dr J. F. Gibson, The Royal Society of Chemistry, Burlington House, London W1V OBN Electrochemistry Group Engineering Aspects of Electrochemistry To be held at Loughborough University of Technology on 17 April 1984 Further information from Dr N. A. Hampson, Department of Chemistry, Loughborough University of Technology, Loughborough LE11 3TU Industrial Physical Chemistry Group The Metal-Polymer Interface To be held at Girton College, Cambridge on 10-1 2 July 1984 Further information from Dr T.G. Ryan, ICI New Science Group, PO Box 11, The Heath, Runcorn, Cheshire WA7 4QE Gas Kinetics Group 8th International Symposium on Gas Kinetics To be held at the University of Nottingham on 16-20 July 1984 Further information from Prof. J P Simons, Department of Chemistry, University of Nottingham, University Park, Nottlngham NG7 2RD Industrial Physical Chemistry Group with the Food Chemistry Group Cell Adhesion to Solid Surfaces To be held at Girton College, Cambridge on 11 -1 3 September 1984 Further information from Dr I . D. Robb, Unilever Research, Port Sunlight, Bebington, Wirral, Merseyside L63 3JW Division with the Gas Kinetics Group Autumn Meeting: Combustion Chemistry in the Gas Phase To be held at the University of Hull, 18-20 September 1984 Further information from Dr R.W. Walker, Department of Chemistry, The University, Hull HU6 7RXDivision jointly with the Deutsche Bunsen Gesellschaft fur Ph ysikalische Chemie, Societe de Chimie Physique and Associazione ltaliana di Chimica Fisica Laser Studies in Reaction Kinetics To be held at the Evangelische Akademie, Tutzing, West Germany on 24-27 September 1984 Further information from: Prof. J. Troe, lnstitut fur Physikalische Chemie, Tammannstr 6, Gottingen, West Germany Colloid and interface Science Group with the Society for Chemical Industry Foaming, Aeration and Dynamic Surface Tension To be held at Imperial College, London on 24-25 September 1984 Further information from: Prof. A. I. Bailey, Department of Chemical Engineering, Imperial College, London SW7 2BY ~~ ~~~~~ ~ ~~~ ~ ~ Electrochemistry Group with the Statistical Mechanics and Thermodynamics Group The Electrical Double Layer To be held at the University of Southampton on 25-26 September 1984 Further information from Dr A. J. 6. Cutler, Research Division, CERL, Kelvin Avenue, Leatherhead, Surrey KT22 7SE ~ ~ ~~~~~ Division Annual Congress: Solid State Chemistry To be held at the University of St Andrews on 26-28 March 1985 Further information from Professor P. A. H. Wyatt, Department of Chemistry, University of St Andrews, St Andrews KY16 9ST ( vi)

ISSN:0300-9599    

DOI:10.1039/F198480FP015

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. I, 1984, 80, 215-283 Temperature-programmed Desorption of Methanol, Ethanol, Propan- l-ol and Propan-2-01 on Silica-Magnesia Mixed Oxides BY HEINRICH NOLLER AND GABRIELE RITTER* Institut fur Physikalische Chemie, Technische Universitat Wien, Getreidemarkt 9, A- 1060 Vienna, Austria Received 27th August, 1982 The adsorption of methanol, ethanol, propan- l-ol and propan-2-01 on silica, silica-magnesia mixed oxides and magnesia has been studied by means of temperature-programmed desorption (t.p.d.). Besides the desorption of the alcohols themselves, CH,O, CO and H, have been found to desorb upon adsorption of methanol, the corresonding olefins upon adsorption of the other alcohols and also propan-2-one upon adsorption of propan-2-01. The desorption behaviour and alcoholconversion are discussed in terms ofthe EPA (electron-pair acceptor)-EPD (electron-pair donor) concept.The t.p.d. peaks of the alcohols have been assigned to two adsorption structures with different interaction strengths. The desorption maxima of the products have been found at higher temperatures than those of the alcohols, the difference being at least 50 "C. CO, H, or the olefins were found with all oxides, whereas formaldehyde was only observed with oxides of lower base strength and propan-2-one with oxides of higher base strength. The interaction with and decomposition on oxide surfaces of lower alcohols has been extensively studied by means of i.r. spectroscopy and kinetic experiments. Four different adsorption structures (depicted in the Discussion section) were suggested by Tench et aZ.l on the basis of i.r.studies and found to be of different bond strengths. Besides i.r. spectroscopy, the t.p.d. technique is a powerful tool for studying bond strengths in adsorption structures. T.p.d. studies with methanol on magnesia were carried out by Foyt and White,2 who compared it with the methanol/alumina system where bonding of methoxyl to the surface was found to be stronger. The interaction strength between the adsorptive and the surface in different adsorption structures is governed by the electron-donor and electron-acceptor capacities of both the adsorptive and the surface. The subject of the present work is to study the variation in the strength of the interaction of a series of alcohols with the same oxide (mixed oxide) and similarly of one alcohol with several mixed oxides of varying composition.In order to note any regularities it was decided to vary the composition of the mixed oxide systematically. Mixtures of silica and magnesia were chosen for this purpose. T.p.d. was considered to be the most powerful tool for studying the interaction strength. EXPERIMENTAL MATERIALS The silica-magnesia mixed oxides covered the range from 10 to 90 mol % magnesia. They are labelled with roman figures from I (10 mol % magnesia) to IX (90 mol % magnesia). The oxides were prepared by stirring Aerosil (Fluka) and magnesium hydroxide (Fluka purum p.a.) in water at 90 OC for 17 h, followed by centrifugation and drying to constant weight at 120 O C . 275276 T.P.D.STUDY OF ALCOHOLS ON SILICA-MAGNESIA Acidity values were determined according to Benesi3. in the following way: three to five drops of a solution of 4-dimethylaminoazobenzene (pK, = 3.3) in dry benzene were added to 1 g of calcined material (650 "C) in dry benzene. Subsequently, 0.5 mol dm-3 n-butylamine in benzene was added until the indicator changed colour. The consumption (mmol per g of sample or mmol per m2 of sample) of butylamine indicated the number of acid centres (pK, = 3.3). B.E.T. surface areas are listed in table 1. Table 1. Acidity (H,, < 3.3) and B.E.T. surface areas of silica- magnesia mixed oxides B.E.T. surface acidity magnesia area ( H , > 3.3) sample (mol %) /m2 g-l /mmol m-2 silica I I1 I11 IV V VI VII VIII IX magnesia 0 10 20 30 40 50 60 70 80 90 100 330 190 240 3 20 340 190 150 130 143 143 81 - 0.0085 0.01 15 0.0131 0.01 53 0.0027 0.01 17 0.0 148 0.0075 aO.O 1 16 a Colour change only poorly pronounced.Debye-Scherrer diagrams taken from oxides calcined at 650 "C (3 h) showed forsterite as well as magnesia in the range 60-90 mol % magnesia, whereas another solid phase, probably talc, was detected in addition to silica in the range 10-50 mol %. Methanol (Merck, Uvasol), ethanol (Merck, dehydrated p.a.), propan- 1-01 and propan-2-01 (Merck, Uvasol) were used without further purification. APPARATUS T.p.d. was carried out in vucuo (ca. 1 Pa) with a heating rate of 10 OC min-l in the range between 30 and 750 "C. The reactor was a quartz glass tube (10 mm in diameter) which was connected to a vacuum pump and to a quadrupole mass spectrometer for detecting the desor- bing species (m/e 1-100).The mass spectrometer and the t.p.d. furnace were coupled to a pro- cess computer IBM S/7. For further details, see ref. (5) and (6). PROCEDURE The oxide was calcined in the t.p.d. reactor in uucuo at 650 "C for 45 min and cooled to room temperature. An amount of alcohol, such that the initial pressure was 2000-2700 Pa, was injected through a septum. After 15 min the reactor was evacuated again for 45-60 min, this time at room temperature, before the t.p.d. was started. The amount of sample (24-100 mg) was adjusted in such a way as to obtain a total surface area of ca. 800 m2 after calcination, except for magnesia, the total surface area of which was ca. 560 m2.RESULTS In all cases, the desorption of products and of alcohol was observed. The desorbed alcohols and products were identified by comparison with mass fragmentation patterns obtained with the QMG 31 1 quadrupole mass spectrometer, which was also277 H. NOLLER AND G . RITTER Table 2. Desorption of the alcohols from silica-magnesia in the range 30-750 "C (1 0 "C min-l), temperatures of t.p.d. maxima alcohol SiO, I IIIU VII" IX MgO CH30H 88 140 236 - 250 245 - 220 239 - 205 I 216 -200 CH3CH,0H 89 100 CH,CH,CH,OH 90 100 CH,CHOH 87 80 CH3 117 130 119 100 126 122 I03 129 I l l - 260 105 128 96 - 240 230-280 250-270 141 250-300 120 - 270 I10 - 290 92 - 200 a High-temperature tailing for each peak. Table 3. Desorption from silica-magnesia of the products of alcohol conversion in the range 30-750 O C (10 OC min-l), temperatures of t.p.d.maxima products formed from (alcohol) SiO, I I11 VII IX MgO CO + €3, (M) > 700" 650 648 517 510 512 ethene (E) 577 b300, 575 395, 560 360 359 357 propene (P-1) 577 336, 560 337, 531 381 393 394 propene (P-2) 490 269,480 256,464 316 317 303 > 700 642 608 476 476 480 324 317 303 - - - - - - CH2O (M) CH, (El - - - CH3COCH3 (P-2) a Beginning of desorption around 600 OC; beginning of desorption. E, ethanol ; M, methanol ; P- 1, propan- 1-01 ; P-2, propan-2-01. used for detection in the t.p.d. experiments. Desorption maxima of the alcohols and products are listed in tables 2 and 3. Representative t.p.d. spectra are shown in fig. 1 (a)-(d). DISCUSSION COORDINATION CHEMICAL APPROACH TO CATALYSIS; EPA AND EPD STRENGTHS OF THE SURFACES AND THE ALCOHOLS The coordination chemical approach to catalysis7? will be used in this discussion of the results.Its main feature is that all interactions are considered to take place between an electron-pair donor (EPD, synonymous with a Lewis base) and an electron-pair acceptor (EPA, synonymous with a Lewis acid). The EPA sites on the surface of our oxides are all cationic sites, namely Mg and Si ions and the protons of hydroxyl groups. The EPD sites are all anionic sites, in our case oxygen. For the alcohols, the strongest EPA function is localized in the proton of the hydroxyl group, which interacts with surface EPD sites. (It seems convenient to use the term site when referring to the catalyst, the term function when referring0 2 00 400 600 800 0 200 4 00 600 800 temperature /" C Fig.1. T.p.d. spectra obtained after adsorption of alcohols on (a) silica, (b) + (6) sample IX, ( c ) sample 111. 1: methanol [(a) m/e 31, (b) m/e 321; 2, ethanol (m/e 31); 3, propan-1-01 [(a) m/e 3 1, (b) m/e 291; 4, propan-2-01 (m/e 45); 5, propan-1-01 (m/e 29); 6, propene (m/e 41); 7, water (m/e 17).H. NOLLER AND G. RITTER 279 to the reactant.) Other groups or atoms of the alcohols also exhibit EPD or EPA functions, but to a much lesser extent so they will not be taken into account here. The behaviour of the system (with respect to adsorption, desorption and catalysis) is governed, to a great extent, by the strength of these interactions, which in turn depend upon the strengths of the EPD and EPA species involved in each interaction.Therefore it is important to have an idea of these EPA and EPD strengths. A mixing rule was proposed by Noller et aL9 which stated that the acid and base strengths of mixed oxides are always between the acid and base strengths of the constituent oxides. Acid (base) strength rises monotonically with increasing fraction of the acidic (basic) component. The same tendency is expected on the basis of Sanderson electronegativity.1° Hence a monotonic variation of site strength is expected from silica to magnesia, decreasing strength for EPA sites, including hydroxyl groups, and increasing strength for EPD sites. This was indeed found in i.r. studies of the adsorption of acetonitrilell and of acetone12 on silica, magnesia and mixed oxides.Silica always turned out to have the most acidic hydroxyl groups. An important detail was that the EPA strength of the hydroxyl groups was in all cases lower than that of cationic sites (here Mg and Si sites). Acidity measurements by the method of Benesi shown in table 1 do not appear to favour the idea of a monotonic variation of EPA strength. Acidity was found to be highest for sample VII, followed by samples 111, I and IX, whereas silica and magnesia do not show acidity (H, < 3.3). Note, however, that acidity as determined in this way is the number of acid sites within the range of strength determined by an indicator. Noller and Parera13 and Vinek14 suggested that silica does not exhibit cations, which they ascribed to the perfect shielding of Si sites by oxygen.The hydroxyl groups of silica on the other hand are not acidic enough to be detected with an indicator of pK, 3.3, as their pK, is as high as 7 according to Hair and Hertl.15 Magnesia on the other hand is a solid base rather than a solid acid, with pK, > 3.3. The EPA strength of the alcohols (in terms of acceptor numbers given by Gutrnannl6 and Mayer and Gutmannl' and listed in table 4) decreases from methanol to propan-2-01 while their EPD strength (in terms of Gutmann's donor numbers) slightly rises in this order. Table 4. Donor numbers (DN) and acceptor numbers (AN) of water and alcohols according to Gutmann DN AN water methanol ethanol propan- 1-01 propan-2-01 18.0 54.8 19.0 41.7 20.0 37.9 37.3 33.3 - - DESORPTION OF ALCOHOLS As the t.p.d.spectra of alcohols [fig. 1 (a)-(d)] are rather complicated, we turn first to simpler systems. In the t.p.d. spectra of acetonitrile, diethyl ether and pyridine on silica,l1. l8 only one relatively narrow peak was found with its maximum below 100 "C and desorption was complete between 150 and 200 OC. This behaviour was assigned to adsorption on hydroxyl groups exclusively, which led to weak interaction. On mixtures with alumina or magnesia, and on magnesia alone, the same peak, but with280 T.P.D. STUDY OF ALCOHOLS ON SILICA-MAGNESIA pronounced high-temperature tailing, or several peaks appeared. Addition of lower- valent oxides to silica had obviously created strong adsorption sites, which in our opinion should be identified with cations exposed in the surface as some of the oxygen is now missing.13 With this relatively simple view in mind, it was surprising when studying alcohols to find that not only with the mixtures and magnesia but also with silica, several peaks appeared (one at ca. 90 OC, a second between 216 and 245 OC and a third one, which is not given in table 2 and may be classified as high-temperature tailing).There must be alcoholic species on silica comparable to those on mixtures, but with so high an interaction strength as to be retained up to at least 500 O C (see the results of the desorption of the products of alcohol conversion, table 3). With samples I and IX and with magnesia one peak with a shoulder at higher temperature was observed, with samples I11 and VII one peak with strong high-temperature tailing was found.Obviously, we are dealing with more than one type of interaction in each case (several interaction structures). The interaction of alcohols with oxides has already been investigated by means of i.r. spectroscopy.'? 1 9 9 2o Investigating the system methanollmagnesia Tench et a1.l proposed four adsorption structures (fig. 2), which appear suitable for discussing our results. S1 is for alcohol molecules adsorbed in a second layer and interacting so weakly with the surface as to desorb on evacuation even at room temperature. S2 describes interaction of the alcoholic oxygen with a Lewis acid or EPA site of the surface, accompanied by the interaction of the hydrogen of the alcoholic hydroxyl group with a surface Lewis base or EPD site.S3 is the result of dissociative adsorption where the alkoxy group is bound to a surface EPA site and the proton to a surface EPD site (surface oxygen), thus giving a surface hydroxyl group. R \ R I H I s2 53 54 Fig. 2. Adsorption structures S2, S3 and S4 of alcohols on oxides. 0, EPA site; 0, EPD site, RCH,, CH,CH,, CH,CH,CH, or CH,CHCH,. In structures S2 and S3 we have EPA,/EPD,, as well as EPAJEPD,, interactions (A, adsorbate, S, surface), i.e. EPA and EPD functions of both the adsorbate and the surface are assumed to be involved. S2 and S3 are considered to be responsible for the desorption of the alcohols themselves. The desorption of alcohol from S3 must be preceded by the recombination of the alkoxy species with the proton, while no such recombination will be required for the desorption of alcohol from S2.Therefore we suggest that alcohol desorption observed at lower temperatures may be associated with desorption from S2. Alcohol desorption in the higher-temperature range (> ca. 200 OC : second peak, shoulder, high-temperature tailing) may be assigned to desorption from S3, after recombination. S4 corresponds to a second type of alkoxy group on the surface where the alkoxy group is within the surface layer rather than on it. S4 is expected to be the most tightly bound surface species, the desorption of which may result in the appearance of products rather than alcohol.H. NOLLER AND G. RITTER 28 1 There are also indications of alkoxy species formed upon the adsorption of ethanol and propan-2-01 on magnesia.lg? 2o We believe that this system can also be used for silica and silica-magnesia mixed oxides.The next question is whether the EPA sites for S2 are hydroxyl groups or other cationic sites, namely Mg or even Si sites. This question is easiest to answer for silica. The first peak, assigned to S2, appears at approximately the same temperature as the peaks for the systems silica/acetonitrile,ll silica/diethylether and silica/pyridine.ls As these latter peaks can only be assigned to adsorption on hydroxyl groups, we assign the peak that with alcohols appears at the lowest temperature to adsorption on hydroxyl groups as well, possibly accompanied by adsorption on oxygen sites (interacting with the proton of the alcoholic hydroxyl group). The second peak or the shoulder or the high-temperature tailing is assigned to S3, for which an Mg or Si site is assumed to be the EPA site.As this interaction is considerably stronger than that with surface hydroxyl groups, the alcoholic hydroxyl groups dissociate. An Mg or Si site is easily accessible on all samples except silica, because of the above-mentioned shielding of Si by oxygen. We assign the second peak observed with silica to a surface species (fig. 3) which is formed by reaction of alcohol with strained Si-0-Si bridges, as discussed by Borello et aL2’ for silica/methanol. Note that such a structure is impossible for acetonitrile, diethylether and pyridine for stoichiometric reasons, which is why only one peak, appearing below 100 OC, is observed with these molecules on silica.Fig. 3. Adsorption of methanol on silica leading to surface methylation reaction. We now discuss the variation of desorption temperature as a function of the oxides and alcohols. Taking into account that there are at least two interactions between the alcohols and the surface in structures S2 and S3 enables us to interpret several details. Usually more emphasis is given to acidic rather than basic surface sites. However, much evidence on the role of basic sites can be obtained from our results. We will first compare silica and magnesia. Except for the second peak of propan-2-01, desorption temperatures are higher with magnesia. As magnesia is more basic than silica, this must be attributed to the EPAA-EPDS interaction being the decisive factor. A further interesting phenomenon is the difference of the desorption temperatures of the four alcohols on the same oxide.When both EPA, and EPD, are very weak sites, interaction will be weak, even though the strengths of EPA, and EPDA are notable. This is so with silica, whose hydroxyl groups and oxygen sites are relatively weak EPA and EPD sites, respectively. Thus, the first peak of the four alcohols appears at nearly the same temperature. However, with magnesia, which has strong EPD and moderately strong EPA sites, a difference of ca. 50 OC is observed. Again the EPA,-EPD, interaction appears to be decisive. An increase in the EPA strength of the alcohols leads to stronger retention on the surface. Propan-2-01, which is the weakest EPA, is desorbed at lower temperature than the other alcohols, not only from silica and magnesia but also from samples I and IX.With samples 111 and VII, only one peak with high-temperature tailing was observed, appearing at roughly the same temperature for all four alcohols. Possibly on these oxides acid and base strengths are more balanced than on the other oxides.282 T.P.D. STUDY OF ALCOHOLS ON SILICA-MAGNESIA As the increasing EPA strength of the alcohol is accompanied by decreasing EPD strength, a compensation effect may be operative which reduces the differences between the desorption temperatures. FORMATION OF PRODUCTS Product formation cannot be unambiguously assigned to one of the two surface alkoxy groups S3 or S4, but it seems likely that S4 will play a significant role. Products are also formed over silica, whose catalytic activity is known to be very low.Indeed, the products do not appear until a temperature of ca. 500 OC is reached, even with propan-2-01 which is the most reactive alcohol in this series. This shows that the interaction with silica must be extremely strong, and we must now ask how is it possible that in spite of so strong an interaction the catalytic activity of silica is so poor? It has been pointed out for many years that for catalysis an interaction of medium strength is most favourable. Too weak an interaction cannot have a notable effect, but also too strong an interaction is of no use as the retention of reactants and products is too strong. This concept is known as the denomination volcano curves.22 A further idea about the low catalytic activity of silica will now be put forward.There is no doubt that the EPA strength of silica is extremely high. Its EPD strength, on the other hand, is extremely low. Provided both EPA and EPD sites are indispensable for bringing about catalysis, as has been pointed out by several 23 the EPD strength of silica could be too low. METHANOL CO and H, were the main products from methanol. Furthermore, CH,O was found to desorb from silica and samples I and 111. In agreement with ref. (24) we suggest a two-step base-catalysed reaction. The temperature of the maxima of desorption from the samples of higher base strength (samples VII and IX and magnesia) was at least 130 OC below that observed on silica and samples I and 111: CH30H f CH,O + H, CH,O +CO+H,. CH,O was only found to desorb from oxides with lower basicity at temperatures (silica > 700 OC; I, 642 OC; 111, 608 "C) below those of the desorption maxima of CO and H, (table 3).We would suppose that, on oxides with sufficient base strength, CH,O is immediately transformed into CO and H,. On silica and samples I and 111, the formation of CH,O occurs at higher temperatures and CH,O may be partly desorbed before it is completely transformed to CO and H,. ETHANOL, PROPAN- 1 -0L AND PROPAN-2-OL The main product was the corresponding olefin from these alcohols. Water formed by dehydration did not desorb simultaneously with the olefin. In some cases, on samples VII and IX and magnesia, broad water peaks were found at temperatures higher than those of the maxima of the olefin desorption peaks.Often only a rise of the base line of water occurred in the high-temperature range. This is a further clear indication of active sites being blocked by dehydration products, as has often been discussed in the literature. On samples I and 111, two olefin desorption peaks were found. The first one (at lower temperatures) was associated with alkoxy species adsorbed on the mixed phase while the second one could be due to alkoxy species on excess silica. With respect to the temperature of the desorption maxima (table 3), the oxides mayH. NOLLER AND G . RITTER 283 be divided into three groups: group A, silica; group B, samples I and 111; group C, samples VII and IX and magnesia. Dehydration is well known to be best catalysed by acid sites.The question is whether the olefin is desorbed once it is formed, i.e. if the temperature of its desorption is the temperature of its formation, or if it is formed below the temperature of its desorption. If the olefin were desorbed on formation, the olefin peaks in group B should be found at lower temperatures than those in group C . On the other hand, if the olefin, after its formation, were retained on the oxide (surface EPA interacting with CC double bond) the temperature of the desorption maxima would be determined by the strength of the interaction between the olefin and the surface EPA. In this case the olefin peaks in group B should be found at higher temperatures than those in group C . The temperature of the desorption maximum of propene in group B is at least 40 "C lower than in group C, showing that the temperature of desorption is determined by reaction and not retention of olefin.A second finding agrees with this: propene formed from propan-2-01 is desorbed at temperatures at least 65 OC lower than those for propene formed from propan- 1-01. Nevertheless, ethene produced from ethanol was found at higher temperatures in group €3. We suggest that ethene, in contrast to propene, does not desorb immediately after formation, at least on oxides which exhibit enough sites of sufficient acid strength. Possibly retention of ethene on the surface is favoured because of the smaller size of ethene. On oxides of group B, products occur from ethanol and propan-2-01 which are typical results of base-catalysed reactions : CH, and propan-2-one. Propene and propan-2-one show desorption maxima at practically the same temperature (sample VII : propan-2-one 324 "C and propene 3 16 "C ; sample IX : 3 1 7 "C ; magnesia : 303 "C).We assume that propan-2-one and propene are evolved from the same adsorption sites. A. J. Tench, D. Giles and J. F. J. Kibblewhite, Trans. Faraday Soc., 1971, 67, 854. D. C. Foyt and J. M. White, J . Catal., 1977, 47, 260. H. A. Benesi, J. Am. Chem. Soc., 1956,78, 5490. H. A. Benesi, J. Am. Chem. Soc., 1957, 61, 970. G. Hatzl, Diplomarbeit (TU Wien, 1978). J. Latzel and G. Kaes, React. Catal. Lett., 1978, 9, 183. H. Noller and W. Kladnig, Catal. Rev., 1976, 13, 149. a H. Noller, Acta Chim. Acad. Sci. Hung., 1982, 109, 429. H. Vinek, H. Noller, M. Ebel and K. Schwarz, J. Chem. SOC., Faraday Trans. I , 1977, 73, 743. lo R. T. Sanderson, Chemical Bonh and Bond Energy (Academic Press, New York, 1976). G. Ritter, H. Noller and J. A. Lercher, J . Chem. SOC., Faraduy Trans. I , 1982, 78, 2239. J. A. Lercher and H. Noller, J. Catal., 1982, 77, 152. l3 H. Noller and J. M. Parera, J. Res. Inst. Catal., Hokkaido Univ., 1981, 29, 95. l4 H. Vinek, Z . Phys. Chem., N . F., 1980, 120, 119. l5 M. L. Hair and W. Hertl, J . Phys. Chem., 1970, 74, 91. l6 V. Gutmann, The Donor-Acceptor Approach to Molecular Interactions (Plenum Press, New York, l7 U. Mayer, V. Gutmann and W. Gerger, Monatsh. Chem., 1975, 106, 1235. la J. Latzel and G. Ritter, unpublished work. H. Miyata, M. Wakamiya and Y. Kubokawa, J. Catal., 1974, 34, 117. 2o N. Takezawa, C. Hanamaki and H. Kobayashi, J. Catal., 1975,38, 101. 21 E. Borello, A. Zecchina and C. Morterra, J. Phys. Chem., 1967, 71, 2938. 22 J. M. Thomas, Introduction to the Principles of Heterogeneous Catalysis (Academic Press, London, 23 H. Pines and J. Menassen, Adv. Catal., 1966, 16,49. 24 J. Kijenski, B. Zielinski, R. Zadrozny and S. Malinowski, J . Res. Inst. Catal., Hokkaido Univ., 1979, 1978). 1967), p. 314. 27, 145. (PAPER 2/ 1494)

ISSN:0300-9599    

DOI:10.1039/F19848000275

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

J . Chem. SOC., Faraday Trans. 1, 1984, 80, 285-295 Quantum-yield Studies of Group V Hydride Chemiluminescent Reactions BY MARK E. FRASER, DONALD H. STEDMAN*~ AND THOMAS M. DUNN Department of Chemistry, The University of Michigan, Ann Arbor, Michigan 48109, U.S.A. Received 12th January, 1983 Quantum yields (photons per hydride molecule introduced) are reported in the wavelength region 275-825 nm for the chemiluminescent reactions of the Group V hydrides phosphine, arsine and stibine with ozone in oxygen, ozone in nitrogen and oxygen atoms in oxygen under excess oxidant conditions. Typical values are shown below system typical quantum yield PH,, ASH,, SbH,+O,/O, 1 x lo-,, 3 x 1 x PH,, ASH, + 03/N2 PH,, ASH,, SbH, + O/O, 3 x 10-3,i x 10-3 o.i5,2 x 10-2, 5 x 10-3 Under some conditions the quantum yield for the PH, + O/O, system approaches unity.In the chemiluminescent reaction between SbH, and O/O, a film is deposited on the flow tube walls which has been observed to enhance the intensity of the ultraviolet SbO band systems. It is attributed to energy transfer from O,(A 3X) formed by recombination of oxygen atoms on the film. A separate ultraviolet emission system is shown to be markedly enhanced by the absence of oxygen in the chemiluminescent reaction between phosphine and ozone. This continuum is spectroscopically and kinetically distinct from the visible continua observed from the reaction of PH3 with 0, and 0 atoms. Chemiluminescence from the reaction of Group V compounds with air or oxygen has been known for centuries,' yet the nature of the light emitters remains unknown and only a few quantum-yield investigations have been performed on these systems.The classical chemiluminescent reaction of phosphorus vapour with air had been estimated to have a minimum yield of 1 quantum of visible light per 2000 P, molecules;2 however, a recent quantitative study obtained a value of 1 quantum per 130 P, molecule^.^ In an oxygen deficient system, Edelstein et aZ., measured a photon yield of ca. lo-,. The chemiluminescent reaction between phosphine and nitrous oxide (initiated by a fast electrical discharge) has been found to exhibit a low photon yield (2 x lo-, at 666 T ~ r r ) . ~ Oxygen atoms are known to produce intense chemiluminescence upon reaction with phosphorus and phosphine6 at reduced pressures yet the quantum yields of these emissions have never been measured.We have recently reported that reaction with ozone produces chemiluminescence from phosphine, arsine and stibine.' In this paper we present spectra and quantum yields for the chemiluminescent reactions of the Group V hydrides phosphine, arsine and stibine with ozone and oxygen atoms. t Present address: Department of Chemistry, The University of Denver, Denver, Colorado 80208, U.S.A. 285286 CHEMILUMINESCENT REACTIONS OF GROUP V HYDRIDES EXPERIMENTAL The spectral and quantum-yield studies were performed in a 25 mm diameter Pyrex flow tube equipped with a movable sample inlet port and an arm for the addition of reagent gases. A spectrometer unit composed of a cooled EM1 9659QB photomultiplier tube, a 0.25 m McPherson monochromator, a Keithley electrometer, a 1 kV power supply and a strip chart recorder was used to perform the experiments.Medium-resolution studies were performed using a Bausch and Lomb spectrometer (0.1 8, resolution) with hour long exposures with a slit width of 60 pm. The film used was SA-1 for the ultraviolet region and 103af for the visible. Survey spectra were taken with the spectrometer viewing the downstream end of the Pyrex flow tube through a quartz window. All quantum-yield measurements and associated spectra were made with the spectrometer mounted in a fixed position viewing through the side of the flow tube. The inlet port was moved so spectra were obtained from several points on the decay of the chemiluminescent emissions.The flow tube was coated with phosphoric acid for these studies in order to minimize atom recombination. Generation of the reagent gases and addition to the flow system have been described previ~usly.~ The spectrometer was calibrated for absolute photon flux by comparison of an observed spectrum from the reaction NO + 0 (generated by titrating active nitrogen with measured flows of NO) with the standard spectrum reported by Fontijn et ~ 1 . ~ The calibration was extended to below 450 nm by observing the spectrum of a standard quartz halogen lamp as a measured temperature of 2500 K. The total photon flux, It, from the observed emissions was calculated from where L is the distance downstream from the inlet port at the measured emission intensity, ImeaS(A1 L), A is the cross-sectional area of the flow tube and S is the system sensitivity factor determined by the absolute calibration.The quantum yield, 4, in units of photons per hydride molecule input, was then where F is the measured flow rate of the hydride in molecule sP. All of the studies performed were carried out at a total pressure of 2 Torr* under conditions of excess oxidant. An examination of the inherent uncertainties of these measurements leads to an average uncertainty of f 20% for the reported quantum yields. For the O,/O, systems where the emissions are very short-lived the uncertainty is larger. Several of the systems studied here have emission spectra which continue into the infrared, which is past the upper wavelength limit of sensitivity of the detector.Thus, the quantum yields given for these systems are lower limits. Linear extrapolations of the absolute intensities indicate that the fractions of the estimated total intensity lying to the red of 825 nm are 7% and 14% for PH, + O/O, and PH, + 0,, respectively, and insignificant for the arsine chemiluminescent reactions. For the stibine chemiluminescent reactions undetected infrared emission may amount to 50% of the total intensity. CD = I J F (2) RESULTS REACTIONS WITH OZONE The hydrides PH,, ASH, and SbH, were reacted with 0,/02 (< 3%). Fig. 1 shows the spectrum for the chemiluminesence together with spectra from other chemilumi- nescent reactions of phosphine. The spectrum is characterized by a continuum peaking at 600 nm with a weak system to shorter wavelengths (this system has been found to be dependent upon reaction and viewing conditions and was not observed previously7).A spectrum of the chemiluminescence with arsine has been reported previo~sly.~~ @ It is characterized by discrete As0 p ( A 2C -+ X "n) emission in the 295-345 nm region * 1 Torr = (101 325/760) Pa.M. E. FRASER, D . H. STEDMAN AND T. M. D U " 287 I I I I wavelength/nm 200 4 00 600 80 0 Fig. 1. Uncorrected photoelectron spectra of the chemiluminescence generated by mixing phosphine with an excess of the oxidants shown at a total pressure of 2 Torr. The apparent dip in intensity between 440 and 460 nm and the feature at ca. 720 nm are spectrometer artifacts. The dashed line shows the experimentally determined response of the spectrometer as a percentage of the peak response. The spectra shown here were optimized for intensity and were not used in the quantum-yield studies.with an apparent continuum starting at 360 nm and peaking at 450 nm with intensity continuing into the infrared. Light emission from the reaction with stibine is characterized only by discrete SbO bands (B 2&4 2111/2, 3/2 + X 2111/2, starting at 340 nm and continuing into the infrared. The hydrides PH, and ASH, were reacted with O,/N, (< 3%). Substitution of nitrogen for oxygen in the SbH, + 0,/02 system has previously been reported to have little effect on the observed emission.' The spectral components of the ASH, + 03/N2 emission are the same as those observed with 0,/02 except that the ultraviolet (u.v-) intensity (295-345 nm) is enhanced relative to the visible continuum emission.Fig. 1 shows a spectrum of the chemiluminescence from PH,+O,/N,. The discrete288 CHEMILUMINESCENT REACTIONS OF GROUP V HYDRIDES emission bands between 240 and 270 nm are from PO y emission ( A 2E -+ X "n); OH emission is observed at 308 nm and PO p emission ( B 2C -+ X TI) at 326 nm. Relative to the observations with O,/O,, the ultraviolet emission (280-500 nm) is now greatly enhanced with respect to the visible continuum. Using the apparatus described, the spatial decay of the chemiluminescence was measured. For PH, + O,/N, the ultraviolet system showed faster decay than the visible continuum. Using the observed decays and the measured spectral distributions of the emissions, the quantum yields were calculated using eqn (1) and (2).The results are shown in tables 1 and 2. Table 1. Quantum yields for Group V hydrides + 03/0, ~~ flow / 10'' molecule s-l quantum yield/photons (hydride molecule)-1 hydride hydride ozone system 1 system 2 total PH3 1.33 1.33 1.33 1.77 2.28 3.01 ASH, 1.90 1.90 1.90 4.12 4.12 4.12 SbH, 1.41 1.41 1.41 4.59 4.59 4.59 66.2 34.4 14.1 66.2 66.2 66.2 69.8 58.5 38.3 69.8 58.5 38.3 70.4 38.6 15.9 70.4 38.6 15.9 7.3 x 10-5 5.7 x 10-5 2.2 x 10-5 7.0 x 10-5 5.9 x 10-5 3.2 x 10-5 1.4 x 10-3 1.5 x 10-3 5.5 x 10-4 1.4 x 10-3 1.6 x 10-3 1.4 x 10-3 2.3 x 10-4 1.6 x 10-4 9.4 x 10-5 2.6 x 10-4 1.8 x 10-4 1.0 x 10-4 6.9 x 10-5 9.2 x 10-5 1.2 x 10-4 1.2 x 10-4 1.2 x 10-4 6.7 x 10-5 1.4 x 10-3 1.5 x 10-3 5.5 x 10-4 1.4 x 10-3 1.6 x 10-3 1.4 x 10-3 3 .o ~ 10-4 2.2 x 10-4 1.1 x 10-4 3.4 x 10-4 2 . 4 ~ 10-4 1.4 x 10-4 6.9 x 10-5 9.2 x 10-5 1.2 x 10-4 1.2 x 10-4 1.2 x 10-4 6.7 x 10-5 PH, system 2 = 275-825 nm; SbH, system 2 = 325-825 nm; ASH, system 1 = 295-345 nm; ASH, system 2 = 375-775 nm. REACTIONS WITH OXYGEN ATOMS The hydrides PH,, ASH, and SbH, were reacted with O/O, (24%). With a clean system, the reaction of low flows of phosphine with oxygen atoms produced little luminescence. At higher PH, flows a small amount of red phosphorus was deposited at the sample inlet but not on the flow-tube walls. In the presence of this deposit the chemiluminescence intensity from low and high PH, flows increased markedly, the red phosphorus facilitating the initial reaction between 0 and PH,, which has been noted to be slow.6 A spectrum of the emission from PH,+0/0, is shown in fig.1. The only discrete emissions are from OH at 308 nm and singlet oxygen at 760 nm (these emissions were found to be present to some extent in the absence of added hydride and thus were subtracted when the quantum-yield determinations were performed). Under both low and medium resolution, the visible emission from 350 to > 800 nm appears only as a featureless continuum.M. E. FRASER, D . H. STEDMAN AND T. M. DUNN 289 Table 2. Quantum yields for Group V hydrides+O,/N, flow / 10'' molecule s-l quantum yield/photons (hydride molecule)-' hydride hydride ozone system 1 system 2 total PH3 0.55 0.79 1.05 0.58 0.58 1.05 ASH, 1.55 1.55 1.55 0.61 0.98 2.08 75 75 75 44.9 14.3 12 64.1 37.8 21.1 64.1 64.1 64.1 3.6 x 10-4 4.8 x 10-4 5.2 x 10-4 4.2 x 10-4 3.5 x 10-4 2.5 x 10-4 6.1 x 10-4 4.3 x 10-4 3.1 x 10-4 4.4 x 10-4 6.5 x 10-4 5.5 x 10-4 2.6 x 10-3 2.7 x 10-3 2.8 x 10-3 2.5 x 10-3 1.8 x 10-3 1.4 x 10-3 4.3 x 10-4 3.4 x 10-4 2.5 x 10-4 1.4 x 10-4 4.ox 10-4 4.6 x 10-4 3.0 x 10-3 3.2 x 10-3 3.3 x 10-3 2.9 x 10-3 2.2 x 10-3 1.6 x 10-3 1.0 x 10-3 7.8 x 10-4 5.6 x 10-4 5.9 x 10-4 1.0 x 10-3 1.0 x 10-3 PH, system 1 = 275-475 nm and system 2 = 475-825 nm; ASH, system 1 = 295-345 nm; ASH, system 2 = 375-775 nm.Also shown in fig. 1 is a spectrum of the intense chemiluminescence from the reaction of an equimolar mixture of oxygen atoms and nitrogen atoms with phosphine. We have observed the intensity from the PH,+O/N reaction to be independent of red phosphorus deposits. It has been proposedlO that this effect arises from the efficiency of nitrogen atoms as initiators of the reaction.The spectrum of the emission from AsH,+0/0, has been reported previ~usly.~ Intense As0 y bands (B 2C -+ X ") in the 240-270 nm region, weak As0 bands and an intense continuum with onset at 350 nm are observed. Analysis at higher resolution shows the As0 y emission to arise primarily from the lowest vibrational state of the excited state and that the rotational structure indicates a thermalized rotational temperature. We have also observed at medium resolution that the visible emission exhibits evidence of structure superimposed on the underlying continuum. Whether or not this structure is related to the nature of the continuum emitter is unknown.The spectrum of the chemiluminescence from the reaction of stibine with oxygen atoms is shown in fig. 2, which illustrates the observed intensity distribution shift as a function of stibine flow. At low flows the glow appears violet-white while at higher flows the glow becomes orange-white. The emission is characterized by ( B 2C,A 3/2) SbO bands, as has been observed with OJO, reagent in addition to previously unobserved (D 2Z and C 2A3/2, 5/2 + X 2111/2, 3/2) SbO bands in the ultraviolet and a continuum starting at 500 nm and continuing into the infrared. The reaction deposits a light-coloured film on the flow-tube walls which has been observed to affect the nature of the emission. With the sample probe moved upstream away from the deposit and the stibine flow adjusted such that the emission appeared orange, moving the inlet port downstream into the area of the deposit changed the colour of the emission to violet-white.Viewed end-on with both emissions visible, a radial distribution is observed with the ultraviolet occurring primarily near the walls of the flow tube and the visible near the centre. Under medium resolution, no evidence of underlying continuum is observed to 3/2 -+ X290 CHEMILUMINESCENT REACTIONS OF GROUP V HYDRXDES I I I i I I 1 200 400 600 8 0 wavelength/nm Fig. 2. SbH, + 0/02 chemiluminescence spectrum. Uncorrected photoelectron spectra of the chemiluminescence from (a) 1.65 x lo1' molecule s-l (high flow) and (b) 6.3 x 10l6 molecule s-l (low flow) of SbH, reacting with 5.62 x 1OI8 molecule s-' of 0 atoms in 0, at total pressures of 2 Torr.Table 3. Quantum yields for Group V hydrides+O/O, flow /lo1' molecule s-' quantum yield/photons (hydride molecule)-' hydride hydride 0/02 system 1 system 2 total PH3 0.55 1.41 2.51 0.55 ASH, 0.49 0.83 1.42 2.19 2.65 0.49 0.83 1.42 2.19 2.65 SbH, 0.63 1.19 1.65 1.65 56.2 56.2 56.2 84.2 56.2 56.2 56.2 56.2 56.2 84.2 84.2 84.2 84.2 84.2 56.2 56.2 56.2 56.2 0.27 0.15 0.10 0.72 1.1 x 10-2 1.3 x 2.7 x lop2 1.5 x 2.1 x 10-2 6.6 x 2.1 x 10-2 2.0 x 10-2 2.3 x 7.1 x 10-3 4.1 x 10-3 3.2 x 10-3 2.9 x 10-3 3.0 x 10-3 0.27 0.15 0.10 0.72 1.1 x 10-2 1.3 x 2.7 x 1.5 x 2.1 x 10-2 6.6 x 2.1 x 10-2 2.0 x 10-2 2.3 x lo-, 7.1 x 10-3 6.9 x 10-3 4.5 x 10-3 3.7 x 10-3 6.2 x 10-3a PH, system 2 = 325-825 nm; ASH, system = 325-775 nm; SbH, system 1 = 275-475 nm; SbH, system 2 = 475-825 nm.a Repeated point after film build-up.M. E. FRASER, D. H. STEDMAN AND T. M. D U " 29 1 I I 1 I 1.0 2 .o SbH, flow/lO" molecule s-l Fig. 3. Quantum yields for the visible system, (475-825 nm), and the ultraviolet system, A (275-475 nm), plotted as a function of stibine flow. The open symbols are values obtained after noticeable surface film deposition. wavelengths < 550 nm. A structured continuum is observed to longer wavelengths with some of this structure attributable to SbO. The reaction SbH, + O/N produces dim bluish-white chemiluminescence. The spectrum of this emission showed some bands but none corresponding to the SbO emission observed in the SbH, + O/O, system. It was subsequently determined that these bands were due to the reaction SbH,+N/N,, which exhibits violet-white chemiluminescence.This reaction produces a black deposit on the flow-tube walls which proved extremely difficult to remove as it was resistant to both aqua regia and concentrated alkali. Spectra of this emission showed intense neutral antimony atomic lines from 200 nm continuing into the visible and molecular emission in the 440-480 nm region. Closely spaced rotational levels are observed under medium resolution for the molecular emission with vibrational spacing similar to that in ground-state Sb,, but the emission corresponds to no known transition reported for Sb,. Atomic emission from reaction with active nitrogen has been observed previously for the Group V hydrides.ll Using the observed decays and eqn (1) and (2), quantum yields for these systems were calculated and are shown in table 3.The data for stibine are also plotted in fig. 3. Also shown in fig. 3 are values obtained after several experiments when film deposits were noticeable. PHOSPHORUS CHEMILUMINESCENT SPECTRA With the exception of the spectrum of the chemiluminescence from excess P,+O, reported by VanZee and Khan,12 there are no recently published spectra of low-pressure chemiluminescent reactions of phosphorus. Fig. 4 shows spectra of the chemilumi- nescence from the reaction of P, vapour with excess 0,02 and 0, at reduced pressure. No solid residue was observed as a product of any of these reactions. The reaction292 CHEMILUMINESCENT REACTIONS OF GROUP V HYDRIDES I I I I wavelength/nm 2 00 400 600 800 Fig.4. Uncorrected photoelectron spectra of the chemiluminescence generated by mixing phosphorus vapour with the indicated oxidants. The O/O, and 0, spectra were taken at 2 Torr while the ozone spectra were taken at 1.2 Torr. The feature at 720 nm in all the spectra is a spectrometer artifact. P, + O/O, produces a continuum spectroscopically indistinguishable, but at signifi- cantly higher intensity, from that observed for P,+O,. This latter reaction also produces PO /? emission not observed by VanZee and Khan', under oxygen-deficient conditions. The chemiluminescent reactions of phosphorus with ozone have not been examined previously. Fig. 4 shows the spectrum of the reaction P,+O3/O,. PO /? emission appears superimposed on an apparent continuum starting at ca.270 nm and continuing into the infrared. In the PH, + 0, system, we observed enhancement of the ultraviolet system in the absence of oxygen. This is also observed for P,+O, as the ultraviolet emission increases relative to the shoulder at 600 nm. PO y emission is observed distinctly in the absence of oxygen from both P, and PH, + 0,.M. E. FRASER, D . H. STEDMAN AND T. M. DUNN 293 DISCUSSION PO y emission on the reactions of P, and PH, with O,/N, and As0 y emission in the reaction of ASH, with O/O, probably arise from P+O,-+PO*+Os (3) As+ 0 -+ AsO* (4) which are sufficiently exothermic (- 488.7 and - 48 1.8 kJ mol-l, respectively) to populate the ground vibrational levels of the excited states. Davies and Thrush6 observed ultraviolet continuum emission (down to 280 nm) and PO p emission in the PH, + 0 system and in excess PH, with ozone added downstream.They also noted that the ultraviolet emission demonstrated kinetic behaviour distinct from that of the visible continuum. Our observations of the PH, chemiluminescent reactions are consistent with those of Davies and Thrush.6 The ultraviolet emission appears to be weak from PH,+O/N and strong from PH,+O,. At increased PH, flows, the relative intensity of the ultraviolet emission from reaction with 0 atoms increases. We have also noted that ultraviolet emission in the PH, + 0, reaction exhibits kinetic behaviour dissimilar from that of the continuum. We suggest that the ultraviolet emission observed here is identical to that noted previously.6 The nature of the ultraviolet emitter is unknown but HOP0 can be discounted as the ultraviolet emission occurs strongly in systems with no source of H atoms (fig.4). We have also studied the emission from excess PH, reacting with 0 and observed intense PO /? bands and HPO bands, in addition to the continuum emission. Under these conditions the visible emission exhibits a short decay while the PO p emission decays slowly, extending far down the tube. These observations support the proposal that PO j? emission occurs via energy transfer.6 The spectrum of the /? system of PO was recorded spectrographically at a resolution of ca. 35000 and showed some unusual features. The p system is the B 2Z -+ X transition with 224 em-l spin-orbit splitting of the ,lIr term giving rise to the 2113/2 state and the lower 2111,2 state.This yields the characteristic doublet spacing of all PO band systems terminating with the ground ,II term. The spectrum obtained showed the 0,O and 1,l pairs of bands (each a spin-orbit pair) with a spacing of ca. 70 cm-l, but no trace of any higher-sequence bands was observed, presumably because of lack of PO ulation in the levels with v > 1 in the ,I: state. There is also a strong feature halfway between the 0,O and 1,l features. The vibrational levels of both the ,l'I and ,I: states are well known, as are various sequence bands from 1 , 1 through to 7,7, from previous spectra.6 The features at 3250.2 and 3274.5 A cannot be assigned to any known sequences of the /3 system and have the appearance of very sharp line(s) in the P,, and PI, branches of the origin sub-bands.In fact, the features coincide almost exactly with P,, (26.5) and P,, (27.5) lines but the resolution of the spectrum allows a precision of only ca. 1 in J. Construction of the complete rotational diagram for the 2Z -+ ,IIr transition shows that both P,,(J) and P,,(J) have the same upper-state J value and spin component so that a genuine case of a selective upper-state rotational excitation with a non-statistical population is possible only if the same lower-state J value is observed in both the P,, and P,, branches. For the O,/O, systems, the results show that for all hydrides the quantum yields are relatively small, 10-3-10-4. For the O,/N, systems, the quantum yields are in the range lo-,.The data show that the ultraviolet emission from PH, can account for ca. 4 x (ca. 36 cm-l) to lower energy of the 0,O pair of sub-bands, i.e. almost exactly294 CHEMILUMINESCENT REACTIONS OF GROUP V HY DRIDES up to 16% of the total photon flux, which is approximately twice the estimates made for the O,/O, system. The data in table 2 show, in general, that for phosphine the ultraviolet (275-475 nm)/visible (475-825 nm) ratio increases with increasing ozone flow. These differences, combined with the observed kinetic dissimilarity, indicate that the ultraviolet and the visible continua arise from different emitters or different states of the same emitter, via different precursors. The AsH,+O,/N, system shows enhanced ultraviolet emission relative to that from O,/O,, with the U.V.now accounting for ca. 60% of the total photon flux. This confirms previous qualitative spectroscopic observations . The data in table 3 show that the quantum yield for PH,+O/O, decreases with increasing oxygen-atom concentration, where it approaches unity. The high quantum yield for this reaction indicates that production of light is the major channel in the reaction mechanism and that the excited species, once produced, is not efficiently quenched in 2 Torr of 0,. Thus the lower quantum yields in other phosphorus systems imply less efficient production of the excited species. Table 4. Typical values of the quantum yield system typical quantum yield PH,, ASH,, SbH,+O,/O, 1 x lo-,, 3 x 1 x PH,, ASH, + O,/N, PH,, ASH,, SbH, + O/O, 3 x 10-3, 1 x 10-3 o.i5,2 x io-2,5 x 10-3 We recorded a spectrum using the same apparatus for the chemiluminescent reaction between phosphorus vapour and air at atmospheric pressure and found the intensity distributions of the visible continuum to be identical to those in fig.1 and 4 for PH, + 0, P, + O/O, and P, + 0,. If the proposed excitation mechanism PO + 0 is analogous to NO + 0 then a red shift in the peak of the emission should have been observed at higher pressure, as has been seen for NO+0.l3 The failure to observe a red shift, combined with the high quantum yield for these systems, indicates that quenching and vibrational relaxation of the excited state are minimal relative to the radiative pathway. The data plotted in fig. 3 show that the intensity of the ultraviolet (275-475 nm) system decreases more rapidly than that of the visible (475-825 nm) system with increasing stibine flow.Thus, at higher stibine flows the visible orange emission dominates. Fig. 3 shows that the visible quantum yield is unaffected by film deposits but that the ultraviolet quantum yield has increased by a factor of four. The phenomenon responsible is energy transfer to ground-state SbO molecules from excited molecular oxygen formed by enhanced oxygen-atom recombination on the surface film SbO(X ,II) + O,(A ,X) + SbO* + O,(X "). ( 5 ) Recombination of oxygen atoms on activated nickel surfaces has recently been observed to produce O,(A ,X),14 but energy transfer to other molecular species has not been reported previously. Table 4 summarizes the results of these studies. Typical quantum yields for each of the systems studied are given. For the oxidant systems, the order of quantum yields is PH, > ASH, > SbH, and the observed order of quantum yields for the oxidantsM. E. FRASER, D. H. STEDMAN AND T. M. D U " 295 is O/O, > OJN, > O,/O,. The absence of oxygen increases the quantum yield from both phosphine and arsine. The majority of the increase for arsine arises from the inrease in the ultraviolet As0 band emission. E. N. Harvey, A History of Luminescence (American Philosophical Society, Philadelphia, 1957). V. A. Nevrovskii and L. B. Soroka, Zh. Fiz. Khim., 1977, 51, 376. S. A. Edelstein, D. J. Eckstrom, B. E. Perry and S. W. Benson, J. Chem. Phys., 1974, 61, 4932. D. G. Harris, Ph.D. Thesis (Cornell University, 1980). P. B. Davies and B. A. Thrush, Proc. R. SOC. London, Set. A , 1968,302, 245. M. E. Fraser and D. H. Stedman, J. Chem. SOC., Faraday Trans. I , 1983, 79, 527. A. Fontijn, C. B. Meyer and H. I. Schiff, J. Chem. Phys., 1964, 40, 64. M. E. Fraser, D. H. Stedman and M. J. Henderson, Anal. Chem., 1982, 54, 1200. * J. Bowen and E. G. Pells, J. Chem. SOC., 1927, 1096. lo M. A. A. Clyne and M. C. Heaven, Chem. Phys., 1981,58, 145. l1 A. P. D'Silva, G. W. Rice and V. A. Fassel, Appl. Spectrosc., 1980, 34, 578. l 3 K. H. Becker, W. Groth and D. Thran, Chem. Phys. Lett., 1972, 15, 215. l4 R. D. Kenner and E. A. Ogryzlo, Inf. J . Chem. Kinef., 1980, 12, 501. R. J. VanZee and A. U. Khan, J. Phys. Chem., 1976,80,2240. (PAPER 3/045)

ISSN:0300-9599    

DOI:10.1039/F19848000285

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

J . Chem. Soc., Faraday Trans. 1, 1984’80, 297-308 CaC1, + KC1 + NaCl Molten-salt Mixtures Experimental and Estimated Enthalpies of Mixing BY PATRICK SEM, GERARD HATEM, JEAN-PIERRE BROS AND MARCELLE GAUNE-ESCARD* Universi te de Provence, Laboratoire de Dynamique et Thermophysique des Fluides, Rue Henri Poincare, 13397 Marseille Cedex 13, France Received 31st January, 1983 The CaC1, + KC1 + NaCl molten-salt mixture is a candidate system for thermal-energy storage. Estimates of the amount of energy which can be stored by such a system requires knowledge of the enthalpies of formation of the liquid mixtures. We have therefore performed direct measurements of the enthalpy of mixing at 1083 K for the three limiting binary systems and the ternary mixture. However, estimates of the ternary enthalpy of mixing were made from those referring to the binary limiting systems using either Kolher’s equation or the results derived from the surrounded-ion model, which was extended in the present work to ternary asymmetrical systems of the kind AX, + BX + CX.In the course of a general s t ~ d y l - ~ of the thermodynamics of molten-salt mixtures, undertaken because of their potential use for thermal-energy storage, we were led to estimate and to measure the enthalpies of formation of several systems. In the present paper we report the calculated and experimental values of the enthalpies of mixing at 1083 K of the CaCl,+KCl+NaCl ternary mixtures as well as those referring to the three limiting binary systems. EXPERIMENTAL METHODS All the enthalpies of mixing reported in this paper were measured using a high-temperature Calvet microcalorimeter. The principle of this calorimeter and its use have been the subject of several Only the main features of our apparatus will be recalled here: it can be operated up to 1300 K and the calorimetric block and the three shields were made of pure alumina.The two thermopiles, each one made up of Pt/Pt+ 10% Rh thermocouples, were connected in opposition. Each cylindrical calorimetric cell was 80 mm high and 16 mm in diameter. The whole assembly was located within a vertical cylindrical furnace heated with four resistors (two on the side wall, one on the bottom and one in the lid). Peripheral equipment consisted of a temperature regulator for the furnace, an automatic system (Apple 11) for acquisition and treatment of calorimetric data, a double potentiometric recorder (Sefram type Servorac BPD) for the measurement of temperatures and thermal effects and a thermostatted gas-tight charging device.The following points must be attended to in order to obtain reliable enthalpy-of-mixing data at high temperatures. (i) If the equation A + B+AB symbolizes the mixing reaction under investigation it is obvious that the most accurate value of the enthalpy of mixing will be obtained when the components A and B on the one hand and the mixture AB on the other are in the same reference state (liquid here) and at the same experimental temperature TE. In this way it is possible to avoid the correction term arising from the extensively used drop method,s which in some cases can exceed the actual enthalpy of mixing.This condition can be fulfilled for several experimental 297298 CaC1,KCI + NaCl MOLTEN-SALT MIXTURES device^.^. In the present investigation we used one very similar to that proposed by 0stv0ld.l~ One salt (or a binary mixture of two salts) was contained in a crucible 50 mm hgh and 9 mm diameter, the second salt (or the third component) was in a quartz ampoule (6 mm diameter) with a breakable tip. The crucible and the ampoule were located within the calorimeter at an experimental temperature of 1083 K. The mixing process was performed by breaking the tip of the ampoule. (ii) In order to obtain a homogeneous final product it was necessary to include a stirring system to this mixing device.Therefore the ampoule was connected to a long vertical quartz tube (t) which could be moved vertically, thus ensuring the homogeneity of the mixture. (iii) The calorimeter must be calibrated under the experimental conditions. This condition was achieved by dropping National Bureau of Standards a-alumina reference samples during the experimental runs. Insertion of this material was guided by the tube (t). Knowledge of the molar enthalpy increment of alumina between room temperature & and the experimental temperature TE allowed us to obtain the calorimeter constant. A gas-tight and thermostatted charging device12 allowed the calibration samples to be dropped without any perturbation of the gas atmosphere. Several blank runs (crucible and ampoule filled with the same salt) showed that the thermal effect from breaking the ampoule was negligible.MATERIALS The purity of the salts and their handling play an important role in the reproducibility of the experimental results. Sodium and potassium chlorides were suprapur reagents from Merck of 99.9980 and 99.9984% purity, respectively; they were dried under vaccum at 573 K for eight days. Particular attention was paid to the dehydration of calcium chloride (suprapur reagent from Merck of 99.9985%. The salt was gradually heated to its melting point under a flow of gaseous chlorine13 and then chlorine was allowed to bubble into the melt for 1 h. The excess gas was removed using a flow of pure and dry argon and the calcium chloride then solidified. All further handling of the dehydrated salts was performed in a dry-box. The argon used during the dehydratation procedure and for the calorimetric experiments was Argon-U supplied by Air Liquide.The chlorine gas was purchased from Merck. ACCURACY In isoperibol calorimetry the accuracy of the results depends on the operational temperature and on the number of the experimental steps required to reach the final result. The investigation of a ternary mixture is more difficult and therefore less precise than that of a binary system since two experimental determinations are necessary, namely the enthalpies associated with the reactions A + B+AB and AB + C-+ABC. This difficulty increases with temperature. A general study of the precision of calorimetric measurements of the enthalpy of mixing14 has allowed us to estimate the uncertainty of the present experimental results using the operating conditions given above as ca. 9%.RESULTS All calorimetric experiments were performed at 1083 K, a temperature at which the binary and ternary mixtures were single-phase liquids over the whole concentration range. The calorimetric results obtained for the systems NaCl+ KCl, CaCI, + KCl and CaCl, + NaCl are reported in table 1, together with those already published by Hersh and Kleppa14 and by Ostvoldlo in order to enable a critical comparison of all these data to be made. NaCl + KCI SYSTEM This mixture has weak thermicity. Our experimental results are compared with those obtained by Hersh and Kleppa14 at the same temperature (see table 1). They fit the equation AH/J mol = -xNa( 1 - xNa) (1921 + 53xNa)P.SEM, G. HATEM, J-P. BROS AND M. GAUNE-ESCARD 299 Table 1. Experimental molar enthalpies of mixing for liquid NaCl + KCl, CaCl, + KCl and CaCl + NaCl systems XNaCl AHIJ mol-1 AHIJ mo1-I AHIJ mo1-I XNaCl XNaCl XNaCl AHIJ rno1-I AH/J mo1-l XNaCl XCaC12 AH/kJ mo1-I 0.1 13 0.303 0.463 0.673 - 427 0.851" - 287" - 189 -418 - 500 0.096' 0.259 0.499 0.702' - 3.68' - 7.57 - 8.97 - 8.75' 0.101' - 1.64b 0.300' - 3.49' 0.5 18 - 3.65 0.751 - 2.40 NaCl + KCl 0.1 47" 0.1 48" 0.350 0.399" - 436 - 517" 0.495" 0.499" - 541" - 550a 0.700 0.701" - 400 - 472a 0.853" 0.892 - 262" - 264" - 292" - 161 CaCl, + KCl 0.099' 0.199' -3.74' -6.53' 0.307' 0.351 0.500 0.600' 0.750 0.751 -8.58' -8.33 - 9.37 8.41' - 6.25 - 5.93 CaC1, + NaCl 0.201' 0.249 0.406' 0.482 0.596' 0.699' 0.905' -2.88' - 1.98 - 3.85' - 3.65 - 3.54' - 2.92' - 1.1 1' 0.171 0.418" - 47ga 0.501 - 506 0.763 - 269 -381 0.200 0.427 0.604' 0.798' -6.59 -9.21 -8.16' - 5.23' 0.250 0.498' 0.749 - 3.14 - 3.78' - 2.69 0.242 - 339 0.447" -541" 0.543 0.800 - 483 -318 0.250 0.499' 0.696' 0.90 1 ' - 7.32 -9.17 - 7.12' - 2.85' 0.250 0.500 0.770' - 3.39 - 4.09 - 2.45' 0.300 - 446 0.448" 530a 0.58 1 - 494 0.850" - 252" a Ref. (14); ref. (10). while Hersh and Kleppa obtained a similar expression AH/J mol-l = - xNa( 1 - xNa) (2038 + 296~,,). At the equimolar composition the difference between both sets of results is ca. 50 J mo1-l. The limiting partial enthalpies obtained in the present work are, respectively, A@&(K)/J mol-l = - 1921 ARg(Na)/J mo1-1 = - 1974 and will be used in the rest of the paper.300 CaC1,KCl + NaCl MOLTEN-SALT MIXTURES CaCl, + KC1 SYSTEM 0stvold's results,lo shown in table 1, were represented by the analytical expression AH/kJ mo1-I = -xCa (1 -xCa) (43.72 - 1 3 .6 3 ~ ~ ~ ) . As they were very close to our experimental values at 1083 K, they were used to represent the enthalpy of mixing using the surrounded-ion model. First developed for binary systems with ions of the same valency,15 this model was later extended to systems with ions of different valency16 (CaCl, + KCl, for instance). The enthalpy of mixing was found to be described by the equation AH/kJ mol-1 = -(1 +x,,)x~,(l -xEa) (21.57+8.50~~,) where xEa = 2 xca/(l +xc,) is the equivalent ionic fraction as defined by F1~1r1and.l~ A@,,,,/kJ mo1-l = -43.14 AHg(ca)/kJ mol-l = -30.07.The limiting partial enthalpies are, respectively, CaCl, + NaCl SYSTEM This system has already been investigated by Ostvold,lo who fitted his experimental enthalpies of mixing (see table 1) to the equation AH/kJ mol-1 = -xca(l -xCa) (18.76-6.70). Our results at 1083 K, reported table 1, are in good agreement, and therefore we used both sets of values to represent the enthalpy of mixing using the surrounded-ion model. l6 The following expression was obtained AH/kJ mol-l = - (1 + xCa) xEa( 1 - xEa) (9.50 + 2.88~:~) which led to the limiting partial molar enthalpies AH/,",,,,,/kJ mol-1 = - 19.00 A ~ / g a a C , , / k J mol-1 = - 12.38. CaCl, + NaCl + KCl SYSTEM This system was investigated at 1083 K. The ternary mixture was obtained either by adding NaCl to a binary CaCl,+KCl mixture or by adding KCl to a binary CaCl, + NaCl mixture.The ternary mole fractions were such that the same composition could be obtained from both procedures, which provided a test of the internal consistency of the measurements; on the Gibbs composition triangle they were located either on constant x,/x,, quasi-binary sections (with xK/xCa = 1, 1/3 and 3) or on constant xNa/xCa quasi-binary sections (with xNa/xCa = 1, 1/3 and 3). The reactions of formation of the ABC ternary mixture can be written as nAA+nBB*(nAA+n,B) (nA A +n, B) +n, C --+ (nA A + nB B + n, C) with the corresponding enthalpies of formation AhAB and Ah. The ternary molar enthalpy of mixing A H is thus obtained from the experimentalP.SEM, G. HATEM, J-P. BROS AND M. GAUNE-ESCARD 30 1 Table 2. Experimental enthalpies of formation of ternary mixtures obtained by mixing liquid NaCl with binary KCI + CaCI, melts at 1083 K. Ah *K + nCa nK + nNa -k %a AH XNaCl / kJ / mol / l 0-3 mol /kJ mol-l 0.200 0.200 0.428 0.692 0.144 0.154 0.332 0.500 0.502 0.601 0.197 0.199 0.201 0.20 1 0.427 0.428 xK/xCa = 1/3, AHK(Ca) = -6.19 kJ mol-1 -8.58 6.516 8.143 - 6.79 5.358 6.697 - 10.57 3.706 6.483 - 15.52 2.873 9.330 xK/xCa = 1, = -9.08 kJ mol-l - 2.49 9.487 1 1.066 - 3.06 8.618 10.182 - 4.00 6.541 9.773 - 2.52 3.969 7.937 - 3.65 4.566 9.143 - 5.85 5.398 13.505 xK/xCa = 3, AH,(,,) = -7.49 kJ mo1-I -0.81 1 1.896 14.823 - 0.30 8.517 10.620 - 0.2 1 8.684 10.863 - 0.88 13.259 16.588 -0.15 6.162 10.760 - 0.25 6.091 10.656 - 6.00 - 5.96 - 5.16 - 3.57 - 8.01 - 7.99 - 6.49 -4.86 - 4.93 - 4.06 - 6.07 -6.03 - 6.00 - 6.04 -4.31 - 4.3 1 enthalpy Ah and from the enthalpy of mixing of the AB binary system, which was measured previously : Ah nA +nB +n, AH = + (1 + x c ) AHAB.Tables 2 and 3 report our experimental results at xK/xCa and xNa/xCa = 1,1/3 and 3, respectively. We also indicate in these tables the binary enthalpies of mixing of KCI + CaCI, and NaCl + CACl,, at the same binary mole fraction ratio, which were used in the calculation of the ternary enthalpy of mixing. ESTIMATION OF ENTHALPY OF MIXING Many relationships have been proposed to estimate the excess thermodynamic functions of multicomponent systems from those referring to the limiting binary mixtures. These relations are either empirica11s-20 or obtained from models based on energetic c~nsiderations.~~-~~ In the present work we used both kinds of procedure to estimate the ternary enthalpy of mixing.Thus we use Kohler’s equationla where the enthalpy of formation AH of the ternary liquid mixture ABC is a weighted function of the binary enthalpies of mixing AHij along the quasi-binary sections corresponding to constant x i / x j molar fraction ratio. We also estimated this ternary enthalpy of mixing from the surrounded-ion model, which was extended to the kind of system investigated in the present work. Indeed the formulation of the enthalpy 11 FAR 1302 CaC1,KCl-t NaCl MOLTEN-SALT MIXTURES Table 3. Experimental enthalpies of formation of ternary mixtures obtained by mixing liquid KC1 with binary NaCl+ CaC1, melts at 1083 K Ah *Na+%a nNa +nK +nCa AH XKCl kJ / mol / 1 0-3 mol /kJ mol-l 0.078 0.199 0.199 0.199 0.199 0.428 0.428 0.692 0.692 0.138 0.142 0.329 0.329 0.450 0.450 0.454 0.599 0.599 0.739 0.077 0.077 0.201 0.201 0.429 0.428 0.692 0.691 0.692 xNa/xCa = 1/3, AHNa(ca) = -2.56 kJ mol-1 - 11.74 6.632 7.195 - 22.25 5.073 6.338 -31.48 6.033 7.541 - 27.03 5.403 6.750 - 25.83 5.372 6.709 - 60.24 5.849 10.225 - 39.28 3.48 1 6.095 - 36.36 1.939 6.306 - 34.97 1.974 6.425 xNa/xCa = 1, AHNa(Ca) = -3.80 kJ mo1-I - 18.22 7.683 8.915 - 20.79 7.90 1 9.216 - 20.98 3.759 5.610 - 25.53 4.476 6.680 -31.53 4.018 7.469 -32.17 4.1 12 7.486 -34.17 4.055 7.438 - 44.36 4.177 10.444 -46.29 4.563 1 1.400 -31.90 2.399 9.207 XNa/XCa = 39 AHNa(Ca) = - - 8.74 13.692 - 8.58 12.880 - 14.14 7.320 - 14.00 7.2 10 - 18.23 4.796 - 18.59 4.910 - 18.38 3.189 - 16.59 2.974 - 15.64 2.567 .3.19 kJ mol-1 14.835 13.953 9.157 9.018 8.398 8.591 10.367 9.65 1 8.337 - 4.00 - 5.56 - 6.23 - 6.06 - 5.91 - 7.36 - 7.91 -6.56 - 6.23 - 5.32 - 5.52 - 6.29 - 6.37 - 6.32 - 6.23 - 6.67 - 5.77 - 5.59 - 4.46 -3.54 - 3.56 -4.10 -4.11 - 4.00 - 3.99 - 2.75 - 2.70 - 2.86 of mixing of a ternary molten-salt mixture has been previously establishedz3 for salts where the ions having the same valency.For the AX + BX + CX system, for one mole, the expression is AH = X,XB[XB AH&) + (1 - Xg) AR~(A)] where xi and AH., are the ionic ternary fractions and the limiting partial enthalpies of the component i in the i+j binary system, respectively. this equation was extended to As already done for ternary reciprocalP.SEM, G. HATEM, J-P. BROS AND M. GAUNE-ESCARD 303 ternary additive mixtures containing ions of different valencies. The reaction of formation of such a system AX,+BX+CX can be symbolized as: n A AX, + nB BX + nc CX+(nA AX, + nB BX + nc c x ) with the corresponding molar enthalpy increment AH. For divalent salts, L~msden,~ introduced the concept of ' equivalent salt' : one mole of the AX, real salt is assimilated to two moles of the A,,,X (equivalent) fictitious salt, which assumes that the fictitious ion occupies a single cationic site in a quasi-lattice model of the melt. Following this assumption, the previous reaction of mixing can be written: 2n~Ao.,X+n~ BX + ~ c C X + ( ~ ~ A A O . ~ X + nBBX + ncC,) which denotes the formation of a ternary mixture from the three symmetrical salts Ao.5X, BX and CX.The enthalpy of formation AH* is thus derived from eqn (1) AH* = xl;xg[~;AHz:(~)+(l -x;)AH~*(A,,.~)] + X~X;[X: AHg;) + (1 - x:) AEz&)] + x: xl;[xZ ARz;o.5) + (1 -a A~z:(c)l where 2nA - 2XA -- x); = 2nA+nB+n, 1 +xA are the equivalent ionic fractions defined by Farland'' and the ARGY are, as previously, the limiting partial enthalpies of component i in the i+j binary system: AHzo:(B), for instance, is the limiting partial enthalpy of the Ao.5X salt in the A,.,X + BX binary system. It can easily be shown that AR'o:(B) = +ART(B) and APg&5, = ARg(A) ARE;, = A@$)(,, and ARS& = ARSB) = A E z A ) and ARTAcc, = ;AH&). The relation between the ternary molar enthalpies AH and AH* is also immediate AH*(~~A +- nB + n,) - A H = - AH*(l +XA).nA + bB + n, These equalities, introduced in eqn (2), yield the enthalpy of formation A H of the AX, + BX + CX ternary liquid mixture - AH= ( l + x A ) x : x ~ ( x ~ ~ AH'(^) + (1 - xg) ARE(^,) 11-2304 CaC1,KCI + NaCL MOLTEN-SALT MIXTURES XNa 0.5 1 O P - 2 - I - 0 E - Y 3 - 4 -6 Fig. 1. Ternary molar enthalpies of mixing along the xK/xCa = 1/3 section. *, Experimental; (-) calculated using Kohler’s equation ; (- - -) calculated using surrounded-ion model. Fig. 2. Ternary molar enthalpies of mixing along the xK/xCa = 1 section. *, Experimental; (-) calculated using Kohler’s equation ; (- - -) calculated using surrounded-ion model.P. SEM, G. HATEM, J-P. BROS AND M.GAUNE-ESCARD Na 305 Fig. 3. Ternary molar enthalpies of mixing along the xK/xCa = 3 section. *, Experimental; (-) calculated using Kohler’s equation ; (- - -) calculated using surrounded-ion model. XK 0.5 1 ____. - I I I I I - 2 I I * I - 0 E - -4 Y 5 - 6 - 8 Fig. 4. Ternary molar enthalpies of mixing along the xNa/xCa = 1 /3 section. *, Experimental; (-) calculated using Kohler’s equation ; (- - -) calculated using surrounded-ion model.306 0 - 2 - I - g - 4 v .Y 5 -6 - 8 CaC1,KCl + NaCl MOLTEN-SALT MIXTURES XK I 0 . 5 1 / / / / / I / / Fig. 5. Ternary molar enthalpies of mixing along the xNa/xCa = 1 section. *, Experimental; (-) calculated using Kohler’s equation ; (- - -) calculated using surrounded-ion model. XK 0.5 1 I I /’ / / / / Fig. 6. Ternary molar enthalpies of mixing along the xNa/xCa = 3 section.*, Experimental; (-) calculated using Kohler’s equation ; (- - -) calculated using surrounded-ion model. This equation was applied to the CaCl,+NaCl+KCl ternary mixture (A = Ca, B -= Na, C = K). Using the values given above for the binary limiting partial enthalpies, we obtained for the ternary molar enthalpy of mixing (in kJ mol-l) AH = (1 + xCa) xGa[ - 9.50 x;Ga - 12.38 (1 - x$,)] + (1 + xCJ xSa xi;[ - 1.92 xi; - 1.97 (1 - x;)] +( 1 +x,,) X$ xEJ - 30.07 x&- 21.57 (1 -x&)].P. SEM, G. HATEM, J-P. BROS AND M. GAUNE-ESCARD 307 We show in fig. 1-6 the enthalpies calculated in this way for comparison with the experimental values reported in tables 2 and 3. On the same figures we also show the enthalpies estimated from Kohler’s equation.Note that in most cases better agreement was obtained between estimated and experimental values when using the equation derived from the surrounded-ion model. Therefore, we used this equation to evaluate the iso-enthalpy curves over the whole ternary composition range, shown in fig. 7. Whichever relationship is selected for the calculation, the differences between the estimated and measured enthalpies of mixing were small and of the order of magnitude of the experimental uncertainty. So, for most ionic ternary mixtures of this kind, an a priori calculation of the enthalpy of mixing should be sufficient to obtain a correct estimate of the thermal energy which can be stored by such system,s. A / \ Na C1 K C I Fig. 7. Ternary iso-enthalpy curves (kJ mol-I) calculated from the surrounded-ion model.1 J-P. Bros and M. Gaune-Escard, Rev. Int. Hautes Temp. Refract., 1978, 15, 99. 3 J. L. Bouju, J-P. Bros, R. Doyen, M. Gaune-Escard, J. Pantaloni and R. Santini, Proc. Utilisation J-P. Bros and M, Gaune-Escard, Rev. Phys. Appl., 1979,14, 107. rationnelle de l’dnergie (DGRST, Grenoble, 198 1). M. Gaune-Escard and J-P. Bros, to be published. E. Calvet and H. Prat, Microcalorimdtrie (Masson, Pans, 1955). 0. J. Kleppa, J. Phys. Chem., 1960,64, 1937. 0. Kubaschewski and C. B. Alcock, Metallurgical Thermochemistry (Pergamon Press, Oxford, 5th edn, 1979). 0. J. Kleppa and L. S. Hersh, J. Chem Phys., 1961,34, 351; J. L. Holm and 0. J. Kleppa, J. Chem. Phys., 1968,49, 2425; M. E. Melnichak and 0. J. Kleppa, J. Chem. Phys., 1970,52, 1790. ’ M. Gaune-Escard, Thesis (UniversitC de Marseille, 1972). lo T. Ostvold, Thesis (University of Trondheim, 1971). 11 P. Gaune, Y. Fouque and M. Gaune-Escard, J. Chim. Phys., 1981,78,621. l2 C. Girard, Thesis (UniversitC de Marseille, 1972). l4 L. S. Hersh and 0. J. Kleppa, J. Chem. Phys., 1965,42, 1309. l5 M. Gaune-Escard, J. C. Mathieu, P. DesrC and Y. Doucet, J. Chim. Phys., 1972, 9, 1390; 1972, 9, l6 G. Hatem and M. Gaune-Escard, J. Chim. Phys., 1977, 74, 754; 1980, 77, 925. l7 T. F~lrland, Discuss. Faraday Soc., 1961, 32, 122. l9 G. W. Toop, Trans. Metall. Soc. AIME, 1965, 233, 850. 2o E. Bonnier and R. Caboz, C. R. Acad. Sci., 1960, 250, 527. D. L. MaricIe and D. N. Hume, J. Electrochem. SOC., 1960, 107, 354. 1397; 1973, 11-12, 1666. F. Kohler, Monatsh. Chem., 1960, 91, 738.308 CaC1,KCl + NaCl MOLTEN-SALT MIXTURES *l J. H. Hildebrand and R. L. Scoott, The Solubility of Nonelectrolytes (Reinhold, New York, 1950). 22 M. L. Saboungi and M. Blander, J. Chem. Phys., 1975,63, 212. 23 M. Game-Escard, J . Chim. Phys., 1974,9, 1167; 1974,9, 1175. 24 G. Hatem, B. de Gasquet and M. Game-Escard, J. Chem. Thermodyn., 1979, 11, 927. 25 J. Lumsden, Thermodynamics of Molten Salt Mixtures (Academic Press, New York, 1966). (PAPER 3/141)

ISSN:0300-9599    

DOI:10.1039/F19848000297

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

J . Chem. SOC., Faraday Trans. 1, 1984,80, 309-324 Quasi-elastic Neutron-scattering Studies of the Dynamics of Intercalated Molecules in Charge-deficient Layer Silicates Part. 1 .-Temperature Dependence of the Scattering from Water in Ca2+-exchanged Montmorillonite BY JONATHAN J. TUCK, PETER L. HALL? AND MICHAEL H. B. HAYES Department of Chemistry, University of Birmingham, P.O. Box 363, Birmingham B15 2TT AND D. KEITH ROSS* Department of Physics, University of Birmingham, P.O. Box 363, Birmingham B15 2TT AND CHRISTIANE POINSIGNON Institut Laue-Langevin, 156X Centre de Tri, 38042 Grenoble Cedex, France Received 17th February, 1983 Quasi-elastic neutron scattering (QENS) measurements on a Ca2+-exchanged montmorillonite containing two layers of adsorbed interlayer water have been made at three different temperatures.The data were analysed using a multi-dimensional non-linear least-squares minimisation method which separated the intensity into an elastic and a quasi-elastic part and fitted a broadening to the quasi-elastic part. The resulting broadenings were then fitted to two alternative jump-diffusion models which differed in the shape of the jump-length distributions assumed. The resulting jump correlation times (the mean time between jumps) were found to be model dependent, but for both models the temperature dependence of the correlation times gave a straight line on an Arrhenius plot, with the same activation energy, namely 1 1 .O 2 1 kJ mol-I. However, because of the variation in the mean jump length with temperature, the effective diffusion coefficients did not show Arrhenius behaviour.A detailed analysis of the correctly internormalised elastic and quasi-elastic intensities is presented. This suggests that both the water in the first hydration shell of the exchangeable interlayer cations and the protons of the hydroxyl groups within the aluminosilicate lattice have a small Debye-Waller factor. For the non-hydration shell water the Debye-Waller factor was found to be comparable with that for bulk water and to increase markedly with temperature in the range of the experiment. Water directly coordinated to the exchangeable cation was shown to be immobile on the time scale of the neutron measurements, but the apparent intensity of this ‘bound’ water fraction was found to decrease with increasing temperature.The elastic incoherent structure factors (EISF) were derived from the quasi-elastic intensities to avoid the complications due to coherent effects in the elastic intensities. The form of these functions suggested that the volume of the restricted diffusion (ca. 300 A at 300 K) increased with increasing temperature. To complete the explanation of the observed peaks the coherent components in the total intensity were attributed to small-angle scattering caused by inhomo- geneities in the clay layer and to the (001) reflection. The above description shows that water in the two-layer hydrate of Ca2+-exchanged montmorillonite can be separated into relatively ‘ free ’ and ‘ bound’ components, the former exhibiting relatively rapid, spatially restricted motions. t Present address: Schlumberger Cambridge Research, P.O.Box 153, Cambridge CB2 3BE. 309310 NEUTRON-SCATTERING STUDIES IN LAYER SILICATES Intercalations of water and of positive ions into charge-deficient layered alumino- silicates such as montmorillonite and vermiculite provide not only two-dimensional electrolyte systems of fundmental interest but also models for the complex natural clay-water systems that are of immense practical importance. Information regarding the structure and dynamics of the hydrated interlayer cations and .water molecules has been obtained by a variety of physical methods such as X-ray and neutron diffraction, infrared spectroscopy, n.m.r., e.s.r., dielectric relaxation and QENS (quasi-elastic neutron ~cattering).l-~ All of these methods have contributed to an improved awareness of the nature of the water and ions between the silicate layers.The present paper is the first of a series describing the knowledge that has been accumulated in our group from an extensive programme of QENS measurements on water-clay and organo-clay systems. In QENS, low-energy neutrons scattered by materials are analysed according to scattering angle or, more precisely, momentum transfer TiQ and energy transfer h. For a rigid solid material neutrons are either scattered elastically or with the exchange of one or more quanta of energy with the vibrational modes of the solid. In a system where diffusive motions occur, however, neutrons scattered elastically by mobile nuclei are subjected to a distribution of small Doppler energy shifts which produce an energy broadening of the elastic peak.This is the so-called quasi-elastic peak, the shape of which is often well represented by a Lorentzian. From the variation of the width of this Lorentzian with the scattering variable, Q, one can obtain information regarding the microscopic motions involved. Thus, for example, for simple Fickian diffusion, AE (the width in energy of the quasi-elastic peak) increases as Q2, with gradient proportional to the macroscopic diffusion coefficient .4 The quasi-elastic scattering exhibits different and more complex behaviour when the diffusive motions are spatially re~tricted.~ Here the scattering has a purely elastic component superimposed on the quasi-elastic peak. The relative intensities of the elastic and broadened (quasi-elastic) components vary with Q, the scattering being entirely elastic at Q = 0.The normalized relative intensity of elastic scattering is known as the elastic incoherent structure factor (EISF).5 We have found that both restricted and unrestricted diffusion can be identified in neutron scattering from clay-water systems. The first study of adsorbed water in clays using QENS was by White and co- who investigated Li+- and Na+-exchanged montmorillonites and vermi- culites which had been equilibrated to constant mass uptake at various water-vapour partial pressures. Their results suggested that the water diffusion coefficients were exponentially related to the reciprocal of the water-layer thickness. The macroscopic thermodynamic model used to describe these results, which employed the Kelvin equation, assumed that the adsorbed water formed a concave, curved meniscus between the silicate sheets.These experiments indicated that the adsorbed water was liquid-like rather than ice-like but that the diffusion coefficients measured for three and fewer water layers were markedly lower than the value for bulk water. The slower diffusion was attributed to the influence of the aluminosilicate surfaces. There is some doubt about the accuracy of the resulting diffusion coefficients particularly at the lower water contents, due to the relatively low energy resolution of the experiments (ca. 250 peV) and the data-analysis methods employed. It has also been pointed out9 that, as capillary action is not regarded as a principal means of swelling, the notion of a curved meniscus is unreasonable.More recently, higher energy resolution QENS measurements have been made using two instruments at the Institut Laue-Langevin (ILL), Grenoble, France. OneJ. J. TUCK, P. L. HALL, M. H. B. HAYES, D. K. ROSS AND c. POINSIGNON 31 1 of these, the IN5 multichopper time-of-flight instrument, gives an energy resolution down to ca. 10 peV while the other, the INlO 'back-scattering' spectrometer, has an energy resolution of ca. 1 peV. The general conclusions from the work published to date have been recently reviewed.l0? l1 Cebula et a1.,12 using the IN5 instrument, made measurements on Li+-exchanged montmorillonite containing one, two and three layers of adsorbed water.They analysed their data using a model that assumed both translational and rotational diffusion. For example, for the two-layer hydrate they obtained D, = 7.0 x m2 s-l and zR = 15 ps. Their model implies that for Li+-exchanged montmorillonite all the adsorbed water appeared to have the same mobility. In contrast, the studies of Hall and c o ~ o r k e r s ~ ~ - ~ ~ on divalent cation-exchanged clays have shown that water coordinated into the first hydration shell of the divalent ions is immobile on the time scale of the neutron measurements ( s) and that the quasi-elastic scattering is due entirely to the non-coordinated water. This conclusion arose from measurements made using the IN5 time-of-flight spectrometer which showed that there was more elastic scattering than could be attributed to either the immobile clay lattice or to an EISF component (which must decrease to zero intensity at sufficiently high Q).Using the fitting method described below, it was found that the amount of extra elastic scattering could be correlated with the amount of the adsorbed water that was in the first hydration shell of the exchangeable cations. In addition, the observation of an EISF showed that the fastest diffusive motion observed was either rotational or spatially restricted translational diffusion. Recently measurements by Conard and coworkerslG~ l7 have suggested that for Li+-exchanged hectorite in a low hydration state the reorientational motions of the water molecules about the cations are observable on the neutron time scale.Correlation times of ca. 3 x and ca. 1.5 x 10-lo shave been reported for this system at room temperture, the former being attributed to the rotations of the water proton about the cation-oxygen axis and the latter to the complete rotations of the cation hydration shell. These data are not necessarily in conflict with the measurement reported here, as the total water content of their samples was considerably less than the present samples and as hydrated lithium ions are known to behave in a different fashion. From IN 1 Omeasurements at low Q on a Ca2+-exchanged montmorillonite containing two layers of adsorbed water, Hall and coworkers13-15 obtained a translational diffusion coefficient of 3.4 x 10-lo m2 s-l. This corresponds to broadenings at least four times smaller than those measured on the lower-resolution IN5 instrument.It was thus concluded that unrestricted translational diffusion is not the dominant contribution to the broadening as measured on INS, at least for the case of divalent-exchanged montmorillonites. Note that the broadening observed on INlO at low Q (< 5 peV) would appear as elastic scattering when measured at 20peV resolution on IN5. On the other hand, at higher Q the intensity of this component falls corresponding to the increase in the quasi-elastic component seen on IN5 and therefore does not interfere with the process of fitting a simple Lorentzian-plus-elastic component to the IN5 peaks, A rigorous justification of this separation of the effect seen on the different instruments can only be made, however, by carrying out identical fitting procedures on simulated data, as is described in a later paper.In the present paper the results of a QENS experiment on a Ca2+-exchanged montmorillonite at three different temperatures are presented, together with details of the experimental method and data-analysis techniques used. From the fitted parameters an Arrhenius activation energy for the observed motion is derived and -312 NEUTRON-SCATTERING STUDIES IN LAYER SILICATES the temperature dependence of the mean-square vibrational amplitude ( u2) is also presented and discussed. In addition, the absolute intensities of both the elastic and quasi-elastic scattering are discussed in terms of a Debye-Waller factor, an EISF and superimposed coherent contributions from small-angle scattering due to inhomogeneity in the specimen and from the (001) reflection.Two subsequent papers will report on further QENS measurements of the motions of water adsorbed on montmorillonite and vermiculite samples containing a variety of particle sizes, exchangeable cations and equilibrated at different water-vapour partial pressures. Possible models for the QENS due to the spatially restricted diffusion have been discussed previously.l09 l1 However, these can only be compared qualitatively with the data presented in the first three papers of this series as the models refer to perfect crystals and do not in fact predict simple Lorentzian broadening. In the fourth paper of the present series these models are extended to provide quantitative comparisons with experiment by generating simulated data in which specific models are averaged over the measured platelet orientation distribution functions for the sedimented samples1* and then convoluted with the resolution function of the spectrometer. The resulting simulated data are then fitted using the same model as was used in analysing the experimental data.The extent of agreement between the two sets of derived parameters reveals the adequacy of the model. An important advantage of this rather indirect method is that it enables a comparison between different models to be made relatively easily. THEORY The theory of neutron scattering and its application to the study of clay minerals has been presented el~ewhere.~? lo, l1 The procedure involves the scattering of mono- chromatic neutrons by materials and the subsequent determination of the scattered intensity as a function of energy (or time-of-flight) and scattering angle. The angle of scattering from the incident beam, 8, determines the momentum transfer between the neutron and the scattering particle, which is given by fiQ = hk' - hk,.Scattering from water is dominated by the scattering cross-section for hydrogen, which is over 98 % incoherent. The incoherent double-differential cross-section per unit final energy (E') and unit solid angle (a), d2a/dE'dS2, is related to the incoherent scattering function Sinc(Q, w). Van Hovel9 showed that this function may be obtained from the self-correlation function for the diffusive motions, Gs(r, t), via a double Fourier transform in space and time.The latter function is defined in the classical limit as the probability of finding a particle at a position r at a time t when the same particle was initially located at the origin. Thus In principle, from the measurement of Sinc(Q,w) it should be possible to obtain Gs(r, t ) by direct transformation, but in practice too few points in (Q, a) space are accessible to make this method reliable. Instead, it is usual to propose a model for Gs(r, t ) for which Sinc(Q, w ) is calculated. The experimental data can then be reduced to Sinc(Q, w) and acomparison between the theory and experiment made. Alternatively, as is done here, the actual form of the appropriate double-differential cross-section can be calculated from the model and all necessary corrections may be made to thisJ.J. TUCK, P. L. HALL, M. H. B. HAYES, D. K. ROSS AND c . POINSIGNON 313 cross-section before a least-squares refinement is carried out between the experimental points and the parameters of the model. This latter approach has the advantage that the proper statistical weight can be applied to each experimental point.20 The problem then arises as to the choice of model and hence to the form of Sinc(Q, o) to be used in the fitting process. If, in the direction defined by Q, the diffusing particle is confined to a restricted space defined by impermeable boundaries separated by a distance a, or, in the case of a rotational motion, is confined to a circular or a spherical surface of radius a, then Gs(r, t ) will remain finite at all times.Thus Gs(r, t ) can be divided into a term decaying to zero with time and a time-independent term. The latter leads to a delta function in Sinc(Q,w) which now has the form where L(Q, o) is a broadened component arising from the time decay of G&, t ) . The form of L(Q,w) is dependent on the details of the motion but it is commonly Lorentzian or the sum of several Lorentzians; i.e. where Ai(Q*a) is a model-dependent amplitude and X,(Q*a) are the model-dependent widths of the Lorentzians. The amplitude A,(Q*a) is the EISF which may be regarded as the Fourier transformation of G&, 00). As Gs(r, t ) is always normalised in space, it follows that at Q = 0 for a restricted motion the QENS will be totally elastic. The intensity of the elastic peak will decrease with increasing Q, there being a complementary increase in the quasi-elastic intensity.A computer program (QUELDA) which uses the simple form of eqn (3) (elastic plus one Lorentzian) to analyse ' time-of-flight' data has been developed at Birmingham.20 The parameters derived from QUELDA will be used to describe the experimental results presented in this paper. Although they do not have a direct correspondence with a microscopic model of the system they do allow general comments to be made and model-independent parameters such as activation energies to be derived. EXPERIMENTAL The Na+-exchanged form of a montmorillonite from Crook County, Wyoming,21 was prepared by steeping the clay for 10 days in 1 mol dm-3 NaCl (with frequent replacement of electrolyte).Excess electrolyte was then removed by repeated washing and centrifugation with distilled water. The 0.2-2 pm equivalent-spherical-diameter fraction of the clay was prepared using sedimentation and centrifugation techniques. Self-supporting films of diameter 90 mm were prepared from clay suspensions by sedimentation and suction through nominally 0.01 pm cellulose nitrate Sartorius* membrane filters. Sufficient quantities of 1 mol dm-3 CaC1, were then passed through the sedimented gels to ensure homoionic specimens. This was followed by sufficient quantities of distilled water to remove excess electrolyte. After drying in air at 385 K the clay film was equilibrated at a constant water-vapour partial pressure, p / p o , of 0.76 (above saturated NaCl solutions at 295 K).Six films were than stacked together and sealed in an air-tight cell having stainless-steel windows of thickness 0.125 mm. The sample thickness was such as to transmit 92% of the neutron beam. The individual clay platelets exhibit a marked degree of preferred orientation in samples prepared in this way. Thus Hall et d.'* have shown by neutron-diffraction methods that the distribution of platelet normals in these samples can be approximately described by Gaussians of full widths at half maximum height in the range 35-50°. * Sartorius Ltd, PF 142 Gottingen, Germany.314 NEUTRON-SCATTERING STUDIES IN LAYER SILICATES Quasi-elastic neutron-scattering spectra for the sample were obtained using the IN5 multichopper time-of-flight spectrometer at the ILL, Grenoble, France.On this instrument a monochromatic neutron pulse is scattered by the sample into a set of detectors which are positioned on Debye-Scherrer rings located ca. 3.98 m from the sample position. The resolution function of the instrument, i.e. the time spread of the incident neutron pulse, was measured using a 2 mm thick vanadium sheet, and the transmission factors of sample and vanadium were determined from the transmitted beam intensities. Both sample and vanadium reference were orientated at 4 5 O to the incident beam, such that the momentum transfer lay parallel to the plane of the sample for neutrons scattered at 90°. Spectra were recorded at three sample temperatures (300, 338 and 368 K), controlled within +2 K, and at an incident neutron wavelength of 10 A.At this wavelength, and at a chopper speed of 15000 r.p.m., the energy resolution of the instrument is 20 peV (f.w.h.m.). For each angle the actual measured spectra were fitted using the program QUELDA.~~ The delta function and the Lorentzian in o, which represent the elastic and quasi-elastic contributions to the total scattering, respectively, are convoluted with the instrument resolution (taken from the actual vanadium data corrected for attenuation effects and the Debye-Waller factor) and are then corrected to a direct time-of-flight profile. A sloping, linear term is included to represent the total inelastic and sample-independent backgrounds in the region of the quasi-elastic peak. All corrections, including the filtering or self-shielding corrections of both vanadium and sample, are built into the model, which is then fitted to the data by a multi-dimensional non-linear least-squares minimisation method incorporating six adjustable parameters.These were as follows: the width, AE (= hAo), of the Lorentzian quasi-elastic broadening, the relative fraction of quasi-elastic to total (elastic and quasi-elastic) scattering in the peak, an intensity normalising factor, the slope and intercept of the ramp background, and a ‘shift’ parameter, which allows for any slight displacement between sample and vanadium spectra (e.g. due to slight differences in sample positioning). A normalised x2 value was calculated for the data at each scattering angle over the N fitted time channels and used to assess the goodness of fit.Fig. 1 shows a typical fit for a value of Q2 = 0.751 A-2. The fitted Lorentzian has an f.w.h.m. AE of 57.2 peV, compared with a resolution function of 20 peV, while the relative fraction of quasi-elastic to total scattering is 0.65 and x 2 / N is 1.15. The quality of the fit is shown by the weighted difference plot, which shows no systematic deviations. Some 95 % of all detectors gave x 2 / N values below 1.5, showing that the data were almost perfectly described by the simplified model used. There was no dependence of the value x 2 / N on temperature or Q2. Thus there would appear to be no justification in fitting any of the more complex models directly to the data. The number of channels over which the fitting was done had to be large enough to include most of the quasi-elastic scattering so that the sloping background was adequately defined. However, fitting over too large a number of channels considerably increased the number of iterations required to find a fit because the contributions to x 2 / N were dominated by the background.In practice, fits to each detector over 150 channels required a few seconds on a Cray 1s computer. Note that the experimental data have not been corrected for multiple-scattering effects. Since the sample is only an 8% scatterer, these effects will only be significant at low Q. However, even at low Q the data are well fitted by the simplified model. It is therefore much simpler to add multiple-scattering contributions to data simulated from a model. As mentioned above, the comparison of experimental and simulated data will be discussed in a later paper.RESULTS The full widths at half height, AE, of the Lorentzians fitted to the quasi-elastic part of the measured peaks are plotted in fig. 2 as a function of Q2 for the three sample temperatures. As each point is obtained from an independent fit at that Q value, the smoothness of the curve is an indication of the stability of the fitting procedure. The absence of systematic errors in the fitting process for small and relatively narrow quasielastic components has already been demonstrated using simulated data.20 All the curves bend over towards a constant AE value at high Q2. This behaviour is to be expected in a diffusing system in thermal equilibrium.22 It can be explained in termsJ.J. TUCK, P. L. HALL, M. H. B. HAYES, D. K. ROSS AND C. POINSIGNON 315 230 360 380 400 L20 L40 4 60 L80 time of flight (channels) Fig. 1. Quasi-elastic time-of-flight spectrum for Ca2+-exchanged montmorillonite two-layer hydrate at 300 K: (------- ) total least-squares fit; (-) fitted quasi-elastic and background scattering. The weighted difference plot (magnified scale) shows no significant systematic deviation. Q2 = 0.75 A2, AE = 57.3 peV. 0 Fig. 2. Quasi-elastic broadening A E as a function of Q2 for Ca2+-exchanged montmorillonite two-layer hydrate: +, 300; 0, 338 and x , 368 K. Solid lines are least-squares fits to the Gaussian jump model, A E = (2?i/z) [l -exp (-Q2(<r2)/2].316 NEUTRON-SCATTERING STUDIES IN LAYER SILICATES of either jump-diffusion models or more continuous models, such as those based on the Langevin equation.22 Confining the discussion to the former approach, it can be shown that Lorentzian broadening is always obtained by neglecting the jump time in comparison with the residence time z and that, in this limit, the various models differ only in the jump-length distribution Thus, if a Gaussian jump-length distribution in an unbounded medium is assumed, we 25 AE = - [ 1 - exp (- Q2 ( r 2 ) / 2 ) ] z where ( r 2 ) is the mean-square jump length.As shown in fig. 2, this expression gives a good fit to the data. The resulting parameters are recorded in table 1. The data show a slight increase in ( r 2 ) with temperature. The mean residence time decreases strongly Table l.a~ Mean jump lengths, (r2);, inter-jump residence times, z, effective diffusion coefficients, Deff, and mean-square displacement ( u2) for water at three different temperatures in Ca2+-exchanged montmorillonite @/p0 = 0.76) and bulk Ca2+- 300&2 1.429 1.004 1.83 2.73 11.3 12.4 0.13 exchanged 338f2 0.823 0.555 1.90 2.79 21.7 23.1 0.29 montmoril- 368+2 0.658 0.480 1.94 3.05 28.6 32.4 0.50 lonite water3@ 295&4 0.205 0.136 0.99 1.44 23.7 25.5 0.29 a Values calculated for jump length distributions of (A) g a ~ s s i a n ~ ~ and (€3) Singwi-Sjolander26 Calculated values of Eact from residence times: 11.5 kJ mol-1 (Gaussian model), models; 10.5 kJ mol-l (Singwi-Sjolander model). with temperature and gives a straight line on a In z against inverse temperature plot, indicating Arrhenius behaviour with an activation energy of 11.5 kJ mol-l. The Sing~i-Sjolander~~ model has also been fitted to obtain an indication of the sensitivity of the data to the shape of the jump-length distribution. Here the effective jump-length distribution is23 This model gives rather different values of z and ( r 2 ) but similar apparent diffusion coefficients ( ( r 2 ) / 6 z ) and activation energies (10.5 kJ mol-l).The apparent diffusion coefficients and the activation energy would thus appear to be independent of the model. The Q dependence of the normalised intensities of both the elastic and quasi-elastic components of the peak is also of considerable interest. The total intensity for Li+-exchanged vermiculite, the only dry clay for which time was available in this experiment, is shown on a In I against Q2 plot in fig. 3.The intensity was purely elastic, the lattice hydroxyls being immobile on the time scale of the neutron measurement. Separate experiments on wet and dry clays using D11 at the ILL have shown that the small-Q part of the normalised intensity is1.0- 0 . 8 % 0.6- 4-4 .- c 3 4 0 . 4 - m W > 0.3 0.2 - 317 - 0 0 0 0 0 o o o o o 0 0 0 0 0 O O 0 - - I 1 Fig. 3. Plots of In I against Q z of the total scattering intensity for a dry Li+-exchanged vermiculite. independent of ion and is very similar for vermiculite and montmorillonite. The first point to be made is that both wet and dry samples show the same rise at small Q and we can therefore conclude that this small-angle scattering component is due to inhomogeneity of the clay layers themselves.Secondly, the slope of the intensity at higher Q in the dry clays is not significantly different from zero on the present Q2 scale, and hence the lattice hydroxyls are tightly bound. Fig. 4 shows a plot of In I against Q2 of the total intensity (elastic plus quasi-elastic) after the subtraction of the intensity arising from the lattice hydroxyls. This subtraction was done by plotting the uncorrected total intensity (In I against Q2) and fitting a straight line to the high-Q portion. By extrapolating this line to Q = 0 the total incoherent intensity was calculated. The incoherent intensity due to the lattice hydroxyls was then calculated using the published chemical constitution21 and the known mass of water absorbed/unit mass of montmorillonite.In calculating the scattering from the clay layer, the incoherent scattering cross-sections have been used and these are dominated by the lattice protons. There would be a small extra coherent diffuse scattering at Q values above the (001) peak due to isomorphous substitution within the unit cell. Because this, if significant, must be flat (see fig. 3) we can estimate it from the Laue expression 4nc(l -c) n((a,)-(a2))2 where ( a ) is the coherent scattering length, n is the number of octahedral sites per unit cell, c is the fraction of octahedral sites occupied by A1 and subscripts 1 and 2 refer to A1 and Mg, respectively. The resulting scattering would be ca. 0.3 b and can be safely ignored in comparison with the total incoherent cross-section of ca.790 b.* This incoherent scattering was then subtracted from the measured total intensity and the results replotted as in fig. 4. From this figure it can be seen that the total intensity shows a sharp peak that can be attributed to the (001) reflection from the hydrated clay. There is an unfortunate gap in the data here due to the failure of two detector banks but the peak is clearly quite well defined. For the centre of the peak the Q vector lies at an angle of 72S0 to the sample normal, but our earlier measurementsls show that the orientation distribution function of the platelet normals extends this far and hence the observed Bragg intensity is to be expected. The corresponding peak is absent from the dry clay and would be weak or absent from a D20 equilibrated clay, since the size of the (001) structure factor depends upon the large phase contrast introduced by the negative scattering length of the hydrogen in the water layer.For higher-order basal reflections this phase-contrast effect largely cancels out, and indeed no other reflections are apparent in the present data. Apart from these two effects it will be * 1 b = m2.318 0.1 0.1 h m 44 c 3 .- NEUTRON-SCATTERING STUDIES IN LAYER SILICATES - + + + T + . + A - + 0 0 0 0 0.3 2 W Ls c I 0.01 - X x x Fig. 4. Plot of In I against Q2 of the total scattering intensity for Ca2+-exchanged montmorillonite after removal of the lattice hydroxyl contribution: +, 300; 0, 338 and x , 368 K. seen that the total incoherent scattering intensity decreases smoothly with increasing Q2 as would be expected for incoherent scattering with a Debye-Waller factor due to the thermal vibrations of the hydrogen atoms of the adsorbed water. Mean-square displacements ( u2) were obtained from the slopes of the straight lines.The resulting values, which are an average over all the interlayer water, are shown as a function of temperature in fig. 5. As can be seen, ( u 2 ) increases strongly with temperature. Fig. 6 shows the plot of In I against Q2 of the quasi-elastic scattering (a) and the elastic scattering (b) for three sample temperatures after subtraction of the lattice hydrogen contribution. Again the smoothness of the quasi-elastic intensity curve and of the elastic intensity curve beyond the (001) peak demonstrates the stability of the fits that are made quite independently at each Q value.This, combined with the tests on the fitting procedure carried out using simulated data,20 gives considerable confidence in the correctness of the division into elastic and quasi-elastic parts. The straight lines through the high-Q portions of the quasi-elastic data have been drawn with the same slopes as those of the total intensity for the interlayer water shown inJ. J . TUCK, P. L. HALL, M. H. B. HAYES, D. K. ROSS AND C. POINSIGNON 319 0.4 N 2 *3 0.3- v 0.2 0.1 0.0 - _ _ _ _ _ _ _ _ _ _ _ _ _ - - _ _ _ - - - - - - 0 X / - - I 1 I fig. 4. In contrast, horizontal lines were drawn through the more widely scattered high-Q points of the elastic intensity. From this it can be concluded that the hydration-shell water (elastic scattering) has a small or zero Debye-Waller factor whereas the non-hydration water (quasi-elastic scattering) has approximately the same Debye- Waller factor as ascribed to the total adsorbed water.Such is to be expected from the dominance of the quasi-elastic component. The values of ( u2) for the non-hydration shell water are given in table 1 and plotted in fig. 5. Note that in the high-temperature limit of the harmonic approximation, ( u2) should extrapolate through the origin. The steeper gradient here implies a reduction in the intermolecular potential with increasing temperature. The EISF can now be plotted as in fig. 7 by correcting the quasi-elastic intensity for the Debye-Waller factor and plotting as a function of Q2, where IQ,(Q) is the corrected quasi-elastic intensity and IQE(Qmax) is the limiting value of IQ,(Q) at high Q.This assumes that the EISF has decreased to zero in the present Q range. On some models this would not be so, owing to sharp features in G(r, 00). However these are more likely to be a characteristic of the models than of a realistic distribution of the sizes of spaces between the hydrated atoms. The plot shows that the EISF becomes narrower in Q as the temperature increases, suggesting that the volume available for the diffusion increases with increasing temperature. Also, the total amount of bound water may be estimated by extrapolating the elastic and quasi-elastic intensities back to Q = 0 as in fig. 4. The resulting values are given in table 1.It will be seen that the Lorentzian broadened (i.e. ‘free’ water) fraction increases with temperature to values above 0.82, which is the calculated fraction assuming that the Ca2+ ions have six-fold coordination. This indicates that either the effective coordination number of the hydrated Ca2+ ions decreases with increasing temperature or that a fraction of the coordinated water becomes more mobile at the higher temperatures.320 NEUTRON-SCATTERING STUDIES IN LAYER SILICATES I X I O I + I t ' I 0 1 + I 0 1 X I + I I I I I I I I 0 1 ' x I I 0 I I I + 0 I X + + 0 + 0 r I I x I l l 1 1 1 1 1 I I 0009 7c? c" - c c 0 0 9 9 0 - 0 0 0 0 0 0 (silun 'q.re) 1 u~J. J. TUCK, P. 1,. HALL, M. H. B. HAYES, D. K. ROSS AND C. POINSIGNON 1 .o ia. 0 . 5 - 0 -0.1 32 1 + + 0 x + + X 0 0 0 + + + + + + 0 " O x x + X * - * ...* x - " X Z X ' x o -.+ 5 I I DISCUSSION The results presented in this paper contain the most detailed analysis of quasi-elastic neutron-scattering measurement of clay-water systems so far reported. The quality of the data demonstrates the ability of IN5, in combination with the data-analysis techniques used, to define accurately the bound and unbound water fractions. Similar separations into ' free' and 'bound' water components have been made from QENS data for water in cement pastes,26 lyotropic liquid crystals2' and aqueous solutions of A1C13.28 In studies of other heterogeneous systems, where no separation into two components was made, the effective diffusion coefficients were found to be smaller than for bulk water.Examples of such behaviour include water on silica as well as the earlier lower-resolution studies of water in Li+- and Na2+-exchanged In many respects water adsorbed by montmorillonite resembles water in concentrated ionic solutions, although in the former case the proximity of the negatively charged aluminosilicate surfaces provides an additional constraint to its m~bility.~ Recent work by Enderby and c o ~ o r k e r s ~ ~ - ~ ~ using neutron-diffraction difference methods and quasi-elastic scattering has elucidated details concerning the hydration structure and dynamics of aqueous solutions of various salts. For aqueous solutions of CaC1, at concentrations similar to those found in the interlamellar spaces of Ca2+-exchanged montmorillonite having two layers of water,32 the coordination number was found to be 6-7. By use of high-resolution QENS rneas~rements~~ they distinguished between (a) rapid exchange and (b) slow exchange of hydration-shell molecules with the bulk liquid as compared with the neutron observation time.They concluded that water molecules reside in the hydration shell of Li+ for only 10-l' s, whereas for the divalent ions Ni2+ and Mg2+ the lifetime in the hydration shell is considerably longer. These conclusions are consistent with the difference observed between Li+-exchanged montmorillonite12 and hectoritel6? l7 and our present results for Ca2+-exchanged montmorilloni te. The use here of unbounded diffusion models in fitting the observed broadening curves, where there is also a clearly defined EISF indicating spatially restricted motion, may appear to be a contradiction. Their use is justified in that both models give good empirical fits to the data from which activation energies and apparent diffusion coefficients may be derived.A more logically satisfactory model, incorporating the and in boehmite322 NEUTRON-SCATTERING STUDIES IN LAYER SILICATES influence of the aluminosilicate surfaces in restricting the environment for diffusion, will be presented in a subsequent paper. The apparent diffusion coefficient obtained at 300 K (see table 1) is some 50% smaller than that of bulk water at a similar temperat~re,~~ while the mean residence time between jumps (ca. 1 0 - l ' ~ at room temperature) is several times longer than values obtained for bulk water.The activation energy of 1 I & 1 kJ mol-1 for the jump diffusion of the non-hydration- shell water is comparable to values given by Ricci et ai.35 for bulk water and by Schon and we is^^^ for water in hydrated vermiculites. It is, however, much smaller than the 20-40 kJ mol-1 given by Fripiat and c o ~ o r k e r s ~ ~ ~ ~ ~ for the activation energy of rotational motions of water hydrogens about the metal-oxygen bonds in the hydration spheres. From the observed correlation times for these motions measured by Fripiat and others using n.m.r. techniques37 it would not be expected that any quasi-elastic broadening would be observable at the energy resolution of the present data. This conforms with our observations that the hydration-shell water is immobile on the time scale of the neutron measurements.Of the scattering from the adsorbed water at 300 K, a fraction 0.86 is quasi-elastic whereas the calculated proportion of non-hydration-shell water, assuming six-fold coordination, is 0.82. The correlation deteriorates at higher temperatures: at 338 and 368 K the observed values of the quasi-elastic fraction were 0.90 and 0.93, respectively. There are two possible reasons for the decrease in the apparent proportion of bound water at higher temperatures. First, the correlation time for rotational motion of hydration-shell water may decrease enough for the consequent broadening to become sufficiently large such that the least-squares fitting method attributes some of its intensity to the larger non-hydration shell broadening.If this is the case then two Lorentzians plus the elastic peak should be fitted. However, the goodness of the x2 fits show that, at the energy resolution of the present data, such detail cannot readily be distinguished. Secondly, the structure of the hydration shell may begin to break down and therefore the calculation of the proportion of hydration-shell water may no longer be valid. The Debye-Waller plot for water in Ca2+-exchanged montmorillonite yields a mean- square vibrational amplitude ( u2) = 0.13 * 0.01 A2 at 300 K (table 1 and fig. 5), which is in reasonble agreement with the value of 0.16 A2 obtained by Cebula et a1.12 for water in the two-layer hydrate of Li+-exchanged montmorillonite. At room temperature Hall et a1.,39 using a similar technique to that reported here, obtained ( u 2 ) = 0.29 A2 for bulk water at 295 K.Franks et aL40 reported a value of 0.34 A2 at an unspecified temperature, presumably room temperature. Similar values have also been reported by Page.34 On the other hand, Blankenhagen41 has reported a room-temperature value of 0.12 A2. This author, however, also found an increase in the Debye-Waller factor with temperature which is similar to but not as pronounced as in the present data. One might conclude that the more rapid increase in ( u2) observed in the clay-water system is due to the faster breakdown of the rather less structured hydrogen bonding as compared with bulk water. A further interesting feature of the QENS data for water in Ca2+-exchanged montmorillonite is the tendency of the broadening curves to turn over, i.e.to deviate below values linear in Q2, at lower Q values than for bulk water. This tendency has previously been noted for water in Li+-exchanged montmorillonite,12 suggesting that it is a feature of the behaviour of the montmorillonite-water system rather than of the specific exchange cation. The effect has also been noted in other aqueous ~ystexns.~~~ 4 2 Interpretation of this phenomenon depends on the model chosen and has been discussed elsewhere in terms of either an oscillatory phase in the motion or as being due to a reduction in the residence time as compared with bulkJ . J . TUCK, P. L. HALL, M. H. B. HAYES, D. K. ROSS AND c. POINSIGNON 323 ~ a t e r . ~ ~ ~ 43 We have preferred to use models in which the time during which the jumps occur, z,, is such tht z, 4 z, where z is the mean residence time.As pointed out elsewhere, these models predict Lorentzian-shaped quasi-elastic broadening with Q dependence determined by the jump-length distribution.l0? 239 24 For such models a more rapid deviation from Fickian behaviour merely indicates larger average jump lengths in the adsorbed water in comparison with bulk water, due perhaps to the less ordered arrangement in the absorbed phase or to the presence of molecular-sized vacancies. This is indeed indicated by QUELDA fits to data for bulk water (from IN5 at 8.5 A),39 which give mean jump lengths approximately half as large as for water on montmorillonite (see table 1). Note also that the mean jump length for the latter increases with temperature.The present experiment has accounted quantitatively for all the observed scattering and has confirmed that the hydration-shell water for Ca2+ is immobile on the time scale of the neutron measurements (i.e. any motions are slower than 10-lo s). The more mobile non-hydration-shell water undergoes jump diffusion having an activation energy similar to bulk water with an apparent diffusion coefficient approximately half that of bulk water. The influence of the aluminosilicate surface and of the hydrated cations on the properties of the non-hydration-shell water is mainly to enforce spatial restrictions on the motion of the latter. In general the data agree with the early work of White and coworkers6-s which suggested that the adsorbed water is more like bulk water than ice, a conclusion that emerges more clearly once the near-elastic scattering due to the hydration shell and the elastic structure factor have been removed.The value of the Debye-Waller factor for the non-hydration-shell water is also similar to that for bulk water. The difference in the reported temperature dependence may well reflect the greater accuracy of the present data. Note also that the evidence for the presence of the EISF (i.e. of the restricted diffusion) is now unambiguous, since after the separation of both the coherent scattering due to inhomogeneities at small angles, and to the (001) Bragg reflection, the elastic and quasi-elastic intensities plotted in fig. 6 show complementary deviations at low Q. The breadth in Q of the EISF decreases with increasing temperature, showing an increase in the volume available for diffusion.At 300 K the linear dimension of the available space corresponds to ca. 7 A. We thank Dr D. J. Picton for help in implementing QUELDA on the Cray 1s computer; the ILL, Grenoble, for the provision of the neutron facilities and the S.E.R.C. for provision of computational facilities. J. J. T. and P. L. H. thank the S.E.R.C. for financial support during the course of this work. We are also endebted to Drs G. Sposito and R. Prost for making available a preprint of ref. (3). Advanced Chemical Methods for Soiland Clay Minerals Research, ed. J. W. Stucki and W. L. Banwart (Reidel, Dordrecht, Holland, 1980). Advanced Techniques for Clay Minerals Analysis, ed.J. J. Fripiat (Elsevier, Amsterdam, 198 1). G. Sposito and R. Prost, Chem. Rev., 1982, 82, 553. T. Springer, Quasi-elastic Neutron Scattering for the Investigation of Diffusion Motions in Solid and Liquids (Springer-Verlag, Berlin, 1972). A. J. Dianoux, F. Volino and H. Hervet, Mof. Phys., 1975, 30, 1181. R. W. Hunter, G. C. Stirling and J. W. White, Nature (Phys. Sci.), 1971, 230, 92. 'I S. Olejnik, G. C. Stirling and J. W. White, Spec. Discuss. Faraday Soc., 1970, 1, 194. S. Olejnik and J. W. White, Nature (Phys. Sci.), 1972, 236, 15. P. F. Low, Soil Sci. Soc. Am. J., 1976, 40, 500. lo D. K. Ross and P. L. Hall, in Advanced Chemical Methods for Soil and Clay Minerals Research, ed. J. W. Stucki and W. L. Banwart (Riedel, Dordrecht, Holland), p. 93. P. L.Hall, in Advanced Techniques for Clay Minerals Analysis, ed. J. J. Fripiat (Elsevier, Amsterdam, 1981), p. 51.324 NEUTRON-SCATTERING STUDIES IN LAYER SILICATES l 2 D. J. Cebula, R. K. Thomas and J. W. White, Clays Clay Miner., 1981, 29, 241. l3 P. L. Hall, D. K. Ross, J. J. Tuck and M. H. B. Hayes, Proc. IAEA Symp. Neutron Inelastic Scattering l4 P. L. Hall, D. K. Ross, J. J. Tuck and M. H. B. Hayes, Proc. Int. Clay ConJ (Oxford, 1978), p. 121. l5 J. J. Tuck, P. L. Hall, D. K. Ross and M. H. B. Hayes, in Water at Interfaces, Institut Laue-Langevin Workshop, ed. C. Touret-Poinsignon and P. Timmins, 1981, Report 81T055S, p. 38. l6 H. Estrade-Szwarckoff, J. Conard, C. Poisignon and A. J. Dianoux, in Water at Interfaces, Institut hue-Langevin Workshop, ed. C. Touret-Poinsignon and T. Timmins, 1981, Report 81T055S, p. 17. J. Conard, H. Estrade-Szwarckoff, A. J. Dianoux and C. Poinsignon, J. Phys. (Paris), submitted for publication. l8 P. L. Hall, R. Harrison, M. H. B. Hayes, D. K. Ross and J. J. Tuck, J. Chem. Soc., Faraday Trans. I , 1983, 79, 1687. l9 L. Van Hove, Phys. Rev., 1954,95, 249. P. L. Hall, D. K. Ross and I. S. Anderson, Nucl. Instrum. Methods, 1979, 159, 347. 21 Data Handbook for Clay Materials and Other Non-Metallic Minerals, ed. L. Van Olphen and J. J. Fripiat (Pergamon, Oxford, 1979). 22 P. A. Egelstaff, An Introduction to the Liquid State (Academic Press, London, 1967). 23 D. J. Winfield and D. K. Ross, Mol. Phys., 1972, 24, 753. 24 P. L. Hall and D. K. Ross, Mol. Phys., 1981, 42, 673. 25 K. S. Singwi and A. Sjolander, Phys. Rev., 1960, 119, 863. 26 D. H. C. Hams, C. G. Windsor and D. C. Lawrence, Mag. Concr. Res., 1974,26,65. 27 J . B. Hayter, A. M. Hecht, J. W. White and G. J. T. Tiddy, Faraday Discuss. Chem. Soc., 1974, 57, 28 P. Martel and B. M. Powell, Solid State Commun., 1981, 39, 107. p g P. G. Hall, A. J. Leadbetter, A. Pidduck and C. J. Wright, Proc. IAEA Symp. Neutron Inelastic Scattering (Vienna, 1977), (IAEA Vienna, 1978), vol. 2, p. 51 1. 30 J. D. F. Ramsay, S. R. Daish and C. J. Wright, Faraday Discuss. Chem. Soc., 1978, 61, 65. 3 l J. E. Enderby and G. W. Neilson, Rep. Progr. Phys., 1981 44, 593. 32 N. A. Hewish, G. W. Neilson and J. E. Enderby, Nature (London), 1982, 2W, 138. 33 N. A. Hewish, J. E. Enderby and W. S. Howells, Phys. Rev. Lett., 1982, 48, 756. 34 D. I. Page, in Water: a Comprehensive Treatise, ed. F. Franks (Plenum Press, New York, 1972), 35 F. D. Ricci, M. A. Ricci and D. Rocca, Phys. Lett., 1974, 48A, 289. 36 G. Schon and A. Weiss, Z. Naturforsch., Teil B, 1973, 140. 37 J. J. Fripiat, in Advanced Chemical Methods for Soil and Clay Minerals Research, ed. J . W . Stucki 38 J. Hougardy, W. E. E. Stone and J. J. Fripat, J . Chem. Phys., 1976, 64, 3840. 39 P. L. Hall, G. J. Churchman, B. K. G. Theng and J. B. Hayter, to be published. 40 F. Franks, J. Ravenhill, P. A. Egelstaff and D. I. Page, Proc. R. Soc. London, Ser. A, 1970, 319, 189. 41 P. Von Blanckenhagen, Phys. Chem., 1972, 76, 891. 42 R. K. Thomas, Progr. Solid State Chem., 1982, 14, 1. 43 G. J. Safford and P. S. Leung, Ber. Bunsenges. Phys. Chem., 1971, 75, 366. (Vienna, 1977), (IAEA, Vienna, 1978), vol. 1, p. 617. 130. p. 333. and W. L. Banwart (Riedel, Dordrecht, 1980), p. 245. (PAPER 3/248)

ISSN:0300-9599    

DOI:10.1039/F19848000309

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

J. Chem. SOC., Faraday Trans. 1, 1984,80, 325-340 Influence of Adsorbed Water on the Dielectric Response of a Ceramic Material BY TERRY RAMDEEN, LEONARD A. DISSADO AND ROBERT M. HILL* The Dielectrics Group, Chelsea College, University of London, Pulton Place, London SW6 5PR Received 18th March, 1983 The dielectric response of a porous ceramic has been measured in the frequency range 10-2-105 Hz and in the temperature range 290-900 K when both dry and exposed to water. These measurements have allowed the frequency dependence of the susceptibility and the temperature dependence of its magnitude and characteristic frequency to be determined. In all cases an anomalous low-frequency dispersion has been observed and substantial differences have been found between the response of the sample when dry and when containing adsorbed water.The quantitative dependence upon water content has been determined, and a tentative explanation of the behaviour is presented in terms of imperfect transport within an adsorbate system. The applicability of this model to other types of dielectrically active system which contain adsorbed species is outlined. The dielectric properties of heterogeneous systems cannot always be described by the sum of those of the independent components. This is particularly true when one component adopts a network structure which is unrealisable in the bulk state, such as is typical of the behaviour of water in such hydrophilic solids as zeolites' and P-alumina2 and in materials of biological interest such as haemoglobin3y and protein^.^ In these systems it is the ability of the water sub-system to transport electrical charge that is of importance.In the work described here advantage has been taken of the known ability of a catalytic material to adsorb water in the form of surface adsorbate structures. In this way it has been possible to study the quantitative effect of such water structures on electrical transport with the knowledge that the adsorbed water does not alter the physical structure of the host material. The catalyst used in these experiments was comprised of a porous ceramic containing a dispersion of nickel particles and was of the type that has been used to catalyse the chemical reactions of natural oil at high temperatures.6 The modification of its electrical properties by adsorbed water is therefore of some importance with respect to the catalytic activity, but this aspect is not of prime importance in the investigation reported here.Instead we regard the catalyst as a host material which, under water-adsorbing conditions, exhibits a complex electrical response similar to that observed in a wide range of hydrophilic materials, and we analyse the behaviour in terms of the known properties of a hydrogen-bonded adsorbate. Because the dielectric response of a material at one frequency can usually be shifted to a lower frequency by an appropriate reduction in temperature, frequency-dependent measurements over the range 10-2-105 Hz have been carried out for a number of fixed temperatures between 290 and 900 K. In addition quantitative values of the water content were obtained for temperatures between 290 and 373 K by monitoring the 325326 DIELECTRIC RESPONSE OF A CERAMIC MATERIAL sample weight before and after dielectric measurement.Unfortunately the detailed composition and porosity of the sample are not available and hence an estimate of the surface coverage could not be made. The dielectric response of a system is described by the complex permittivity E(W) at the frequency o: (1) which can be related to the complex dielectric susceptibility, ~(o), by way of the effective infinite frequency permittivity E , : (2) These data are presented here in the form of master curves which have been constructed from double logarithmic spectral response plots by translation of individual sets of data obtained under different experimental conditions (temperature and/or water content) to form a single master response curve.In this investigation it was observed that all the data could not be used in this manner as there were significant differences in the spectral shape, particularly as a function of water content. For this reason a set of master curves has been constructed. The master curve, or normalisation, technique has been described in detail elsewhere7 and results in a significant improvement in the definition of the observed response as well as presenting the experimental information in a compact and concise manner. When the master curve can be achieved the experimental data obtained obeyed the equation E(O) = d(w) - iE”(o) ~ ( o ) = &’(a) - id’(o) - E, = x’(o) - ix”(co). x(o) a x’(0,) F(w/w,) (3) where the form of the complex function F(x) is given by the master curve.The scales of the axes are chosen to apply to F(x) for given values of the variables, and the variations of the scaling frequency and amplitude parameters w, and ~ ’ ( c o , ) can be determined, relative to the chosen values, by marking the shift of the datum point on the master plot. In this way the absolute values of the characteristic frequency w, and the magnitude of the response ~ ’ ( w , ) can be determined. In all cases the principal feature observed in the dielectric spectral response F(w/o,) was an anomalous low-frequency dispersion8. which can be characterised in the following manner. For frequencies less than the characteristic frequency (a,) ~ ’ ( w ) a ~”(o) a o - p 0 < p < 1 (4) and for frequencies greater than the characteristic frequency ~ ’ ( o ) ax”(o) a on-l 0 < n < 1 ( 5 ) although in some cases a loss peak89l0 was superimposed on the general response, which is shown schematically in fig.1 . This behaviour, termed the low-frequency dispersion (LFD) by Jonscher,** l1 has been shown to be a bulk response of materials containing channels of mobile ions such as the Hollandites12 and is known to be strongly influenced by the presence of water.8 Indeed in some materials such as biopolymers its observation has been completely attributed to the presence of water.5 If the frequency-dependent loss component, for o < o,, is expressed as a frequency- dependent conductivity, a(o), then a(o) = oy’(o) cc X’(0,) og o l - p (6) from which it is clear that as the index p approaches unity the process can be easily confused with that of a d.c.conduction. The two can, however, be distinguished byT. RAMDEEN, L. A. DISSADO AND R. M. HILL 327 QC log w Fig. 1. Schematic diagram of the anomalous low-frequency susceptibility response as a function of frequency showing two power-law regions and the characteristic frequency 0,. a simultaneous measurement of the real component ~’(o), as has been done in this investigation. Some evidence exists for the observation of the characteristics of the low-frequency dispersion process in a number of water-absorbing and equivalent systems. The proportionality between the real and imaginary parts of the susceptibility of eqn (4) has been observed, in terms of permittivity, in a mixed-valence salt [K,(MnO,),], with a value ofp of ca.0.8.13 Large low-frequency capacitance dispersions have been found in bovine serum albumin5? l4 and in haemoglobin., Unfortunately no loss component was reported in these results and hence it is not possible to be certain that the process observed was not one of interfacial polarisation. However, the thin haemoglobin layers examined by Hasted et aL4 give clear indications of the presence of LFD as the parallel dispersion of the real and imaginary parts of the permittivity, eqn (4), was observed. A fractional power law, with frequency, of the a.c. conductivity [eqn (6)] has been found in perylene and in a perylene-chloranil complex,15 and in these cases the values ofp were ca.0.9 and 0.6, respectively, a strong indication of the LFD dielectric process and not a true d.c. conductivity. A d.c. conductivity is commonly attributed to Na-B-aluminals and zeolites,17 and in the latter case it has become the norm to determine a ‘d.c.’ conductivity from results at the lowest frequency available. This is quite an arbitrary procedure and cannot give a true d.c. value, since it is clear from expression (6) that the observed response does not contribute to such a property because the a.c. conductivity approaches zero at zero frequency, albeit very slowly. The frequency dependence of a(o) and the large magnitude of the dispersion observed without a plateau occurring in ~’(o), however, indicates that the response should be regarded as originating in an imperfect charge-transport process.A useful analogy here is with the electrical conductivity in structurally disordered systems, where a similar anomalous frequency dependence is expected for a correlated transport of charge driven through a finite range of disorder.lg A similar structural situation can be expected to arise in materials where mobile ions or ionisable molecules do not form a structurally regular matrix, either as a result of partial occupancy of binding sites or incompatibility of328 DIELECTRIC RESPONSE OF A CERAMIC MATERIAL the binding-site separations and interionic interaction^.'^ In the present materials such a matrix may arise either from the presence of clusters of mobile ions or from the weakly connected islands formed by water adsorbate monolayers.It is possible20 that at frequencies below those used here the entities contributing to LFD will generate a true d.c. conduction; however, this will be as a result of long-time processes and will possess different characteristics to those measured here.18 At frequencies greater than o, on the other hand a different frequency dependence, eqn (9, is observed. Since o, is the characteristic rate for the LFD transport process with ~'(o,) dependent on the number of charge displacements and their effective chargel6, l9 this region of response corresponds to polarisation of those species which when ionised produce the transported charge displacements. It is therefore not surprising that the crossover frequency o, exhibits a well defined temperature dependence,8 and that it has proved impossible to separate the two frequency regions into observable independent processes by varying the temperature, water content,8 or in some cases the electric field.21 It is thus clear that the low-frequency imperfect transport process should be regarded as the long-time limit of the high-frequency local polarisation process, with polarisation going over continuously to ionic transport at 0,.The observed response has therefore been treated as a composite process with a single temperature dependence in both o, and its strength ~'(o,). EXPERIMENTAL The porous catalyst used contained nickel particles of ca. 0.02 ,um diameter22* 23 dispersed in a ceramic with potash present to promote the removal of carbon deposits.A sample of the ceramic was prepared with polished and roughly parallel faces onto which evaporated aluminium electrodes were deposited. Because the thickness of the sample was relatively poorly defined in comparison with the accuracy of the electrical measurements the results are reported in terms of the capacitance, C, and the a.c. loss, G/w, where G is the a.c. conductance and w the frequency of measurement. C and G/w correspond, respectively, to the real and imaginary parts of the complex permittivity, ~ ( w ) . All measurements were carried out either in a furnace, for the higher-temperature range, or in a cryostat. The dielectric measurements were made using a Solartron frequency-response analyser which was connected on line to a PDP-11 computer.Complete control of the former, including the selection of the frequency, output voltage level, input sensitivity and the number of cycles of integration per measurement, was made by software instruction^.^^ The machine automatically produced output plots of log (capacitance) and log (loss) as functions of log (frequency), as well as print-out of the same information to four significant digits. RESULTS A number of samples have been investigated and it has been found that the characteristic frequency and the spectral response are strongly dependent on both the temperature and water content. At moderate temperatures, generally < 200 O C , the water content is highly significant and hence the samples tend to be sensitive to their premeasurement history.For this reason one particular sample, sample I, was taken through a carefully controlled experimental cycle in order that the exact sample conditions should be known. Details of the behaviour for particular experimental conditions were then reinvestigated using this and other samples. Given a known starting condition the observed behaviour was reproducible from run to run and from sample to sample, indicating that the behaviour observed was the true response of the system.T. RAMDEEN, L. A. DISSADO AND R. M. HILL 329 SAMPLE I The sample was allowed to come to equilibrium at 20 OC in a desiccator and the frequency response was measured 2, 3 and 18 days after the initial desiccation. The sample was then removed and placed in a cryostat, under atmospheric conditions, and heated to 100 "C.Measurement runs were made after the lapse of +, 4,8 and 124 h. In each case the weight of the sample was measured before and after each frequency run. The frequency run itself took ca. 2 h but the weighing, which was carried out on a laboratory balance, took only 5 min. The second series of measurements with this sample took place in a furnace and the measurements were made at 50 OC steps between 150 and 600 OC as the sample was heated and then cooled. Time was given at each temperature for the sample to reach equilibrium but no weight measurements were made in this series in order to obviate exposure to atmospheric humidity. On the cooling cycle additional measure- ments were made at 100, 65 and 21 OC. The effect of controlled quantities of adsorbed water was investigated in the third series of measurements.Following saturation by exposure to excessive humidity at room temperature the sample was dried out by raising the temperature. Measurements were made at various temperatures with the sample weight recorded before and after the measuring run. This sequence culminated in a series of measurements taken at increasingly longer times with the sample held at 100 OC in the cryostat. Finally the cryostat was evacuated at this temperature for 60 h before the last measurement of this series was made. Fig. 2 presents an Arrhenius plot of the characteristic frequency for the three sets of measurements and gives a guide to the behaviour of the sample. With the specimen held initially in the desiccator the characteristic frequency was ca.103-104 Hz, and tended to decrease with desiccation time. On heating to 100°C the characteristic frequency fell to 10-1 Hz, and holding the sample at this temperature further reduced the characteristic frequency. On heating a further small reduction was observed initially, followed by a non-Arrhenius increase, and after a temperature of 350 O C had been reached Arrhenius behaviour with an activation energy of 1.77 eV was observed. At the highest temperature to which the sample was taken, 600 OC, the characteristic frequency could not be measured as it was in excess of los Hz, and this data point has not been plotted in the diagram. On cooling, the high-activation energy region was limited to temperatures in excess of 450 O C , below which an activation energy of 1.16 eV was observed down to 200 OC.A minimum in u), occurred in the region of 100 OC and the initial, room-temperature, value was recovered. The third series of measurements indicates that as the sample is dried the characteristic frequency decreases. Note that the lowest-frequency value observed, after evacuation, lies on the extrapolation of the 1.16 eV Arrhenius plot which was obtained during the cooling part of the second series. The spectral responses obtained from sample I are presented in fig. 3 in the form of master response curves. Response type A, fig. 3(u), was observed at room temperature after desiccation and during ageing at 100 OC in air. Response type B, fig. 3(b), was observed initially, after heating to 100 O C , and in the region of 100 O C after the heating cycle and saturation.Type A differs from type B in having a smaller gradient in the lower-frequency part of the spectrum and in the absence of a gradient change, in this frequency region, which is apparent in the type B response. This latter feature is also missing in the type C response which was observed on heating from 150 O C and on cooling down to 250 OC. Over the whole of this range a small loss peak at a frequency of about two orders of magnitude greater than w, can be observed330 6 4 1 s 2 W 3" c( 0" 0 -2 1 DIELECTRIC RESPONSE OF A CERAMIC MATERIAL TI"C 600 400 200 100 1 1 '01 8 1 1 I I 1 - 2 3 103 KIT Fig. 2. Arrhenius plot of the characteristic frequency observed in sample I of the porous ceramic catalyst.The arrows indicate the order in which the measurements were made, as do the run numbers beside each experimental point. Note that run no. 47 was carried out not only at 100 O C but after 60 h in vacuum. Where simple Arrhenius behaviour can be seen the activation energies have been given. Extrapolation of the higher-temperature region plots gives a pre-exponential frequency of ca. 1014 Hz. in the normalised response of fig. 3(c). At the highest temperatures examined there is an apparent tendency to saturation of the real component of the response at the lowest frequencies measured and a concomitant rise in the loss component. The characteristics of sample I are listed in table 1. SAMPLE I1 The temperature region from 20 to 100 "C was reinvestigated using a second sample and temperature steps of 20 "C.At each temperature the sample was allowed to come Fig. 3. Master spectral response characteristics for the experimental runs contained in fig. 2 and listed in table 1. (a) Type A response, (b) type B response and (c) type C response. The numbered points at the bottom of each plot are the datum shift points, from which the normalised plot has been constructed, for the experimental runs. The plots are scaled for run numbers (a) 3, (6) 1 and (c) 17. In (c) the curves through the experimental points are given by the complex susceptibility function of eqn (7) with p = 0.91 and n = 0.75.T. RAMDEEN, L. A. DISSADO AND R. M. HILL 33 1 7 - - t4 5 - u, h \ - 3 Y M 3 - - Q - ld E I - G - OO 3 -1 t 0 - 3 -1 1 3 5 7 9 11 log (o/Hz) Fig.3. For legend see opposite.332 DIELECTRIC RESPONSE OF A CERAMIC MATERIAL Table 1. Characteristic frequency, water content and type of response for sample I as a function of temperature ~ response type run no. T/K %/HZ water content (wt %) 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 40 41 43 44 45 46 47 289 289 289 373 373 373 373 423 473 523 573 623 673 723 773 823 873 823 773 723 673 623 573 523 473 423 373 338 294 323 333 353 363 373 373 373 4 x 103 104 1.1 x 103 2 x 10-1 1.7 x 10-l 1 . 4 ~ 4 x 10-2 2 x 10-2 2.5 x 10-l 1.3 1.4 x 10' 8.5 x 10' 8 x lo2 6 x lo3 7 x 104 2.4 x 105 2 x 105 4~ 104 1.2 x 104 1.8 x 103 - 1.5 x lo6 3.8 x lo2 7.7 x 10' 4.0 6.9 x 10-' 5.3 x 10-1 1.4 x 10' 2.7 x 103 1.2 x 10' 1.3 1.2 x 10-1 1.5 x 10-1 1.8 x 10-l 2.9 x 10-l 3.3 x 10-3 1.89-1.87 1.97-2.10 1.18-1.93 0.99-1.85 0.6 1-0.62 0.3 7-0.53 0.53 - - - - - - - - - - - - - - - - - - - - - - 1.6-2.2 1.47-1.55 0.98- 1.12 0.98 0.56-0.98 0.42-0.56 0 B B A B A A B C C C C C C C C C C C C C C C C B B B B B C C C C C C B - to equilibrium when mounted in the unevacuated cryostat and the frequency response was measured.No weight measurements were carried out during this series of measurements. The spectral response observed was of type B thoughout and the Arrhenius plot of the characteristic frequencies observed is given in fig. 4. SAMPLE 111 The non-activated behaviour in the heating cycle was reinvestigated using a third sample in the temperature range 120-500 O C . The spectral response observed was of type C , as observed orginally, and the characteristic frequency obtained is shown in fig.5 in an Arrhenius plot. In general terms the observed behaviour was identical to that found earlier.T. RAMDEEN, L. A. DISSADO AND R. M. HILL 333 / / O / O / O o/o / 2 . 5 3.0 3.5 103 K I T Fig. 4. Arrhenius plot of the characteristic frequency in the temperature range 20-100 O C for a second sample (11) of the ceramic catalyst. 6 5 8 4 1 $ - 3 v 00 2 1.5 2. 0 2.5 103 K I T Fig. 5. Arrhenius plot of the characteristic frequency in the temperature range 120-500 O C for the third sample (111) of the ceramic catalyst. Fig. 2-5, together with the fractional water content measured in the temperature range between 20 and 100 O C , contain all the information that has been obtained about the system investigated.In general terms it can be seen from fig. 2, 4 and 5 that the characteristic frequency is activated but that the presence of water in the sample increases this frequency. It is believed that the extrapolation shown in fig. 2 by the dashed line is a correct description of the behaviour of a totally dry sample, and this allows an estimation of the effect of the water content in a quantitative manner. 12 FAR 1334 DIELECTRIC RESPONSE OF A CERAMIC MATERIAL DISCUSSION In this section the form of the observed response and its dependence on water content and temperature is analysed in greater detail and a tentative explanation of the behaviour proposed. SPECTRAL RESPONSE The spectral responses observed in this investigation have been presented in fig.3 in classified form. The type B response is characterised by a slow transition through the characteristic frequency cross-over with a significant overshoot in the real part of the permittivity. The type A response, on the other hand, gives a very weak transition over a broad frequency range although the asymptotic gradients of the low-frequency dispersion for both types A and B are almost identical with p = 0.8 and n = 0.7. Because in both cases the deviation from the typical response shown in fig. 1 is limited to the cross-over region we consider that both the A and B spectral forms are indicative of an incompletely realised adsorbate system giving a weak effective distribution in o, about its mean value.Such a distribution can be estimated to be no more than a factor of one hundred to one thousand wide. The type C response is much more of the form indicated in fig. 1 and has been fitted by the theoretical spectral shape function proposed by Dissado and HilP9 F(o/o,) = (1 + io/o,)l-n zF, { 1 - n, 1 + p ; 2 - n ; (1 + io/oJ1} (7) where &{, ; ;} is the gaussian hypergeometric function,25 using the values of p = 0.91 and n = 0.75. The discrepancy between the theoretical curves in fig. 3(c) and the normalised experimental data at the highest frequencies is due to the presence of the weak loss process at the normalised frequency o, x 100, which has not been taken into account in the calculated function. The discrepancy at the lowest frequencies has already been noted and may be an artefact of measurement or possibly the onset of a weak d.c.conductivity. TEMPERATURE DEPENDENCE The characteristic frequency o, obtained for temperatures above 373 K has a different temperature dependence on the heating and cooling runs, as shown in fig. 2, although the form of response is effectively unchanged. Since the value of o, observed at 373 K at the start of this sequence, no. 7, is greater than its value when evacuated at this temperature, no. 47, it is likely that the sample initially has a residue of water content. It therefore appears that it is the presence of this water which is removed on heating that causes the observed difference in the rate process governing 0,. Note that the sample was measured in a sealed furnace during the heating/cooling cycle and hence the water removed on heating was retained as a humid atmosphere which could be readsorbed on cooling.The activated behaviour on cooling should therefore be attributed to a ‘dry’ sample in a humid atmosphere, and evidence for the formation of an equilibrium adsorbate is revealed by the deviation from an activated rate constant below temperatures of 500 K, fig. 2. The non-Arrhenius behaviour of w, on heating can be presented in a more revealing manner by plotting log (0,) as a function of the linear absolute temperature, as shown in fig. 6. Included in this plot is not only the information contained in fig. 2 but also the results of the measurements made on sample 111, fig. 5. Both sets of data exhibit straight lines with the same gradient, and hence 0, = exp(CT) (8) with C = 4.1 x K-l.T.RAMDEEN, L. A. DISSADO AND R. M. HILL 335 5 3 n N 3” v Do 4 -1 I X ” / x / n ’ I I 0 I I I I I I I 300 500 700 T/K Fig. 6. A plot of log(o,) as a function of linear temperature using the data presented in fig. 2 and 5. 0, Sample I; x , sample 111. It has been shown26 that the behaviour of expression (8) follows from a model of relaxation in which the optimum rate contains elements of both activation and tunnelling through the activation energy reduced potential barrier. Expression (8) re- sults when the discrete vibrational quantum states are smoothed out into a continuum. This requires either a large particle mass or a high barrier for a local quantum- mechanical tunnelling. It is not necessary, however, to assume a single particle tunnelling in order to derive this form of relaxation, which is obtained whenever a continuum of relaxation routes exist, each comprising a number of steps of equal statistical weight and increasing activation energy.It is possible from this behaviour to obtain the product of the particle mass and tunnelling distance at zero temperature and also the maximum barrier height. However, as the product cannot be separated out without further information a choice of particle mass must be made on the basis of hysical plausibility. The choice of an electron gives a tunnelling distance of ca. 25 x in the case considered here, whereas a proton gives a distance of ca.1 A. WATER CONTENT All results for which the water content was known were determined on sample I following the heating/cooling cycle.It could therefore be assumed that the effect of water adsorbate was to modify the ‘dry’ rate process as determined during the cooling run. This supposition was confirmed by recovery of a value of o, lying on the activation plot following evacuation at 100 “C for several hours (no. 47). The unusual behaviour of o, in the region 20-100 O C , fig. 2 and 4, can therefore be understood 12-2336 DIELECTRIC RESPONSE OF A CERAMIC MATERIAL to be the result of temperature-dependent changes in the water content, W, competing with an underlying activated process. The dependence upon water content can be determined by extracting the activated temperature dependence from 0,. It is convenient to use the ‘dry’ activation energy, A, which has a value of 1.16 eV, to convert all the measurements of o,( T, W) to an effective value o,( T,, W) at a standard temperature, T,, chosen to be 373 K, with o,(T,, W) = w,(T, W) exp - --- [X 31.(9) In constructing eqn (9) it has been assumed that the effect of water content can be represented by a factor fT W) multiplying the underlying ‘dry’ activated process,14 giving o,(T, W ) at an arbitrary temperature T as where with A = 1.16 eV and vo determined from the high-temperature cooling curve, the value of which should be of the order of a vibrational frequency, 1013 Hz, for a truly dry sample. Values of o,( T,, W) have been plotted as a function of Win fig. 7, where the water content has been expressed as the percentage fraction of the sample dry weight, determined from the evacuated sample.The weight error bars in these diagrams represent the change in weight of the sample over the spectral measuring period. For weights < 1% [fig. 7(a)] there is an essentially linear relationship between the logarithm of cue( T,, W) and the weight; i.e. F(W)=exp(AW), W < 1%. (12) With the weight in percentage units, as here, the constant A has the value 5.34. The figure, however, shows that at higher water content the linear relationship breaks down. Normalising the results to that of the evacuated (dry) sample at 373 K allows the plot of fig. 7(b) to be constructed in which log {In [o,(T,, W)/o,(T,)]) is plotted as a function of W. In this case a linear plot can be seen in the region above 1 %, and hence we have that with B = 1.68.These two functions can be combined into the form fTW CK exp [exp (BW)1, 1 % < W = 2% (13) (14) fTW) = exp [XWexp (BW)] where 2 is 0.9, as determined from the intercept of the extrapolated linear section. By writing the form offTW) becomes N = exp (BW (15) with which is immediately recognisable as the fraction y of the change in entropy required to disorder or unbind an ordered structure of N particles, or vice versa. The value of N will be proportional to the number of occupied adsorbate sites per one hundred y = Z / B = 0.54I I I I I ,-. / I / / / / / / // -0 0 0 1 2 H20 (wt %) (b )-d I--- / 0 /$ e-+ Fig. 7. Water-content dependence of the relative characteristic frequency scaled to 100 "C. (a) Log [o,(T,, w)] as a function of wt % water; (6) log (In [w,(T,,w)/w,(T,)]} as a function of wt % water.338 DIELECTRIC RESPONSE OF A CERAMIC MATERIAL available sites, and its form, expression (15), results from a fixed availability probability raised to the power of the number of attempted adsorptions.It is therefore clear that the effect of the water adsorbate in this temperature range is modifying the existing ‘dry’ relaxation process by increasing the number of routes available through the relaxation transition a feature that can be represented by an activation entropy26 given by k-l ln[f(W)] in expression (16). Although an exponential dependence such as expression (12) has been previously obtained in a similar form of response in hydrated proteins14 at low water content, it is believed that this is the first time that the super-exponential form of eqn (14) has been seen, and its significance recognised.INTERPRETATION First some comment is required as to the physical nature of the low-frequency dispersion process itself. This is characterised by a weakly frequency-dependent susceptibility, con-1, at high frequencies which is followed by a strongly frequency- dependent behaviour, o - p , at frequencies less than a critical frequency o, [eqn (4) and (5)]. The sample polarisability, which is proportional to the susceptibility, becomes enormous at low frequencies (typically six decades’ increase) and therefore corresponds to a charge separation over micrometre ranges at least. In contrast the high-frequency region with its near constant polarisability indicates that the charge separation is restricted to small regions in which the charges are highly bound.Under these circumstances the characteristic frequency o, is the rate at which charges are separated to a distance at which they are only weakly bound, that is the ionisation- recombination rate of nearly free charges. For these reasons a modification has been proposed to the model of dielectric relaxation put forward earlier2s-30 in which an effective ionic charge displacement is transported in a structurally disordered system composed of an irregular array of ionisable cluster^.^^ This is the basis of eqn (7), in which n measures the degree of correlation within the cluster and p the perfection of transport of the effective charge displacement by means of the irregular array.This latter process can be regarded as the transport of a non-classically diffusing ionic charge packet,ls and an alternative definition of p can be given in terms of its average localisation (correlation) length as determined by the array structure.lg In the present case adsorbed molecules and/or ions on the surface of the catalyst will fulfill the requirements of such a model, forming weakly connected islands or clusters whose internal structure is only partly ordered,22* 23 being determined by the competition of adsorbate sites and intermolecular forces. Although it is possible to envisage the transport of the cluster as a whole about the catalyst surface it is more likely that clusters will partially dissociate and associate causing the effective transport of a charge displacement as is required for the derivation of eqn (7).The equilibrium description of such a system will be that of a fluctuating cluster array3O? 31 rather than a static disorder. Since the values ofp and n remain almost identical despite the changes in mechanism of the cluster ionisation rate o,, the structure of the clusters and array responsible for the dielectric response must be controlled by the catalyst surface structure, with n 0.7 indicating ca. 30% irregularity of intracluster structure, and p N 0.8 ca. 20% reduction in screening of a transported charge displacement from the array of ionisable clusters.32 The effect of water content in the temperature range 20-100 O C is to increase the nominally dry recombination/ionisation rate co, by providing a large number of extra routes for the process, equivalent to the number of ways of binding or unbinding a number of molecules in a characteristic structural arrangement. The similarity of the index y , describing the fractional regularity of the water structure, and the frequencyT.RAMDEEN, L. A. DISSADO AND R. M. HILL 339 power index n, defining the intracluster regularity, indicates that in this temperature range water forms a major proportion of the surface clusters generating the response. This contention is supported by its ease of removal by heating, which indicates that the stability of its state in the system is mainly determined by the intermolecular interactions of water molecules. The activation energy for a,, however, is identical to that observed at high temperatures on the cooling run.This indicates that the underlying charge-transport mechanism must be the same under both conditions, a contention that is reinforced by noting that the number of transporting ions, which is proportional to ~'(o,), is essentially constant throughout. Therefore it must be concluded that clusters of adsorbed ionisable entities exist which can contribute an LFD type of transport at high temperatures, and about which water can be easily adsorbed either transiently from a humid atmosphere at high temperatures or permanently at low temperatures, facilitating the probability of cluster ionisation. Although it is possible that nickel particles may fulfil this role, it is more likely that cation and hydroxyl ions of the potash are responsible, with the activation energy being that required to desorb a cation from a solvated neutral cluster.On the heating curve shown in fig. 2 a mechanism occurs whose rate constant w, is less than that observed at the same temperature on the cooling run. In this series of measurements therefore the activated mechanism must be somehow suppressed. Since the temperature characteristic of this mechanism, fig. 6, can be extrapolated to include the evacuated result (no. 47) it seems likely that the sample includes only water that is not easily removable. The temperature dependence in this region shows that a tunnelling charge is responsible for transport with a maximum activation energy of 1.77 eV. In this case it appears that in the absence of water to solvate the alkali ions of the potash, electrons are transferred from hydroxyl ions to neighbouring clusters over a range of ca.25 A. It may be possible that the nickel particles play a role here, transiently converting some hydroxyl ions to water molecules and thus freeing the cations to take part in the activated process on cooling. Although the processes described here are those appropriate to a fractional monolayer coverage of the catalyst surfaces, no evidence for a bulk response of water has been observed in this frequency range in the data reported, except perhaps for the weak loss peak. When the sample contained large quantities of water, however, evidence was obtained for a bulk d.c. conductivity in parallel with the behaviour described here.In addition electrode effects could be discerned and the response was found to be time dependent. As a result these data have not been reported. We believe that it is reasonable to assume that observations that have been reported refer to a partially occupied monolayer system. CONCLUSIONS The anomalous low-frequency dispersion or near d.c. conductivity observed in heterogeneous systems containing mobile ions has been observed to be substantially enhanced in transport rate by adsorbed water molecules. Evidence has been presented consistent with the view that in a porous ceramic catalyst the water forms a cluster about the binding site with a structure whose form is close to that revealed by the correlation index of the high-frequency region of the dielectric response.These results are in agreement with the proposed interpretation of the LFD response as an imperfect transport of charge displacement in a weakly connected array of low-dimensional clusters. In addition the main body of data has been shown to agree with the response function determined theoretically from such a model. The unusual feature of the response mechanism shifting to lower frequencies with increasing temperature has been shown to be the result of a reduced water content340 DIELECTRIC RESPONSE OF A CERAMIC MATERIAL which successfully reduces the statistical probability of transport in competition with an increased thermal probability. By relating the statistical weight to an activation entropy it has been shown that this takes the form of a fractional amount of the classical entropy of dissociation of a group of elements.It is considered that both the process of transport discussed here and the technique which has been used to determine the nature of the transport process are of wider applicability than the particular material investigated. It has been pointed out that many materials of chemical and biological interest have been observed to exhibit some of the characteristics of the low-frequency dispersion process. A full characterisation, however, has not been able to be made because it had not been realised that it is essential to examine the frequency dependence of both the real and imaginary components of the dielectric response. A dispersion in the loss component, or in the conductivity, which is not associated with a dispersion in the real, capacitive, component is highly indicative of a d.c.conductivity in the sample. The evidence is that this is not the common behaviour observed in a wide class of materials which can be generally associated with the formation of a low-dimensional transport process. In this case the anomalous dispersion process observed here appears to be the common be haviour . We acknowledge the helpful criticism of a referee. J. J. Fripiat, A. Jelli, G. Poncelet and J. Andre, J. Phys. Chem., 1965, 69, 2185 J. Jensen and M. Kleitz, Solid State Protonic Conductors, Vol. I. Fuel Cells and Sensors (Odense University Press, Odense, 1982); D. P. Almond and A. R. West, J. Phys. (Paris), 1981, 42, (26-187. D. D. Eley, N. C. Lockhart and C. N. Richardson, J. Chem. Soc., Faraday Trans. 1, 1979,75, 323. J. B. Hasted, H. M. Millany and D. Rosen, J. Chem. SOC., Faraday Trans. 2, 1981, 77, 2289. S. Bone, J. Eden, P. R. C. Gascoyne and R. Pethig, J. Chem. Soc., Faraday Trans. I , 1981,77, 1729. A. Ovenston and J. R. Walls, J . Chem. SOC., Faraday Trans. I , 1983, 79, 1073. R. M. Hill, Nature (London), 1978, 267, 96. A. K. Jonscher, Philos. Mag., Sect. B, 1978, 38, 587. M. Shahidi, J. B. Hasted and A. K. Jonscher, Nature (London), 1975, 258, 595. R. M. Hill, J. Mater. Sci., 1981, 16, 1 18. A. K. Jonscher, J. Mater. Sci., 1981, 16, 2037. l2 A. K. Jonscher, K. L. Deori, J. M. Reau and J. Moali, J. Muter. Sci., 1979, 14, 1308. l3 D. R. Rosseinsky, J. A. Stephen and J. S. Tonge, J. Chem. Soc., Faraday Trans. I , 1981,77, 1719. l4 J. Eden, P. R. C. Gascoyne and R. Pethig, J. Chem. SOC., Faraday Trans. I , 1980,76,426. l5 S. Bone and R. Pethig, J. Chem. Soc., Faraday Trans, I , 1978, 74, 720. l6 P. C. Bruce, A. R. West and D. P. Almond, Solid Stale Ionics, 1982, 7, 57. l7 G. Jones and M. Davies, J. Chem. SOC., Faradav Trans. I , 1975, 71, 1791; G. Jones, J. Phvs. Chem. Solids, 1976, 37, 887. l 8 B. Shapiro and E. Abrahams, Phys. Rez;., 1981, B24,4889. l9 L. A. Dissado and R. M. Hill, J. Chem. Soc., Furaduy Trans. 2, 1984,80, 291. 2o J. Bernasconyi, H. U. Beyeler and S. Strassler, Phys. Rev. Lett., 1979, 42, 819. 21 A. K. Jonscher, F. Meca and H. M. Millany, J. Phys. C , 1979, 12, L293. 22 D. R. Sandstrom, Phys. Bull., 1982, 33, 277. 23 E. C. Marques, D. R. Sandstrom, F. W. Lytle and R. B. Gregor, J. Chem. Phys., 1982, 77, 24 J. Pugh, Phys. Bull., 1978, 29, 469. 25 L. J. Slater, Generalised Hypergeometric Functions (Oxford University Press, Oxford, 1966). 26 R. M. Hill and L. A. Dissado, J. Phvs. C , 1982, 15, 5171. 027. 27 M. S. Shablakh, L. A. Dissado and R. M. Hill, J. Chem. SOC., Faraday Trans. 2, 1983,79, 369. 28 L. A. Dissado and R. M. Hill, Nature (London), 1979, 279, 685. 29 L. A. Dissado, Phys. Scr., 1982, T1, 110. 3n L. A. Dissado and R. M. Hill, Proc. R. SOC. London, Ser. A , 1983, to be published. 31 R. M. Hill, L. A. Dissado and R. Jackson, J. Phys. C , 1981, 14, 3915. 32 A. K. Jonscher, Nature (London), 1975, 253, 5495. (PAPER 3/431)

ISSN:0300-9599    

DOI:10.1039/F19848000325

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

J . Chem. SOC., Faraday Trans. 1, 1984,80, 341-357 The 2pP-3dn Interaction in Aromatic Manes Fluorescence from the l(2pn, 3dn) Intramolecular Charge-transfer State BY HARUO SHIZUKA,* YOSHIHIRO SATO AND YUTAKA UEKI Department of Chemistry, Gunma University, Kiryu, Gunma 376, Japan AND MITSUO ISHIKAWA AND MAKOTO KUMADA Department of Synthetic Chemistry, Faculty of Engineering, Kyoto University, Kyoto 606, Japan Received 29th March, 1983 The 2pn*-3dn interaction in the excited state of aromatic silanes has been studied by means of absorption and emission spectroscopy. Broad and structureless fluorescence spectra of phenyldisilanes and naphthyldisilanes with large Stokes shifts have been observed and they have been attributed to the emissions from the intramolecular charge-transfer (c.t.) states with large dipole moments. Evidence that the c.t. emission originates from the l(2pn, 3dz) state produced by the 2pn* (aromatic ring) 3d7t (Si-Si bond) intramolecular charge transfer is given by the effect of steric twisting on the emission. It is shown that the fast formation (d 1 ns) of the c.t. state from the locally excited state I(n, n*) ‘B, (or ‘Lb) of phenyldisilanes takes place, followed by rapid decay (d 1 ns) of the intersystem crossing I(2pn, 3dn) + 3(7t, x * ) . However, no c. t. emission has been observed for aromatic monosilanes and polycyclic aromatic disilanes (aromatic rings 2 3) even in fluid polar solvents. The 2pn* -+ 3dn intramolecular c.t. mechanism is discussed in comparison with that of the twisted intramolecular charge-transfer (t.i.c.t.) state.Over the past thirty years silicon chemistry has been extensively studied, and a number of papers and reviews have been The photochemical reactions of silicon compounds have been widely in~estigated.~-ll However, relatively little attention has been paid to mechanistic work until recently.l29 l3 The electronic interaction between 2pn and 3d orbitals is currently an interesting aspect of chemistry. It is known that the n-d interaction in aromatic silanes is shown in u.v.,’~ 14-18 n.m.r.19 and e.s.r.20 spectra and dipole-moment 22 However, the interaction in the ground state is not so large. Athough the U.V. spectra of phenylsilanes have been interpreted in terms of excited-state interactions of the ‘n*-d’ type, no clear evidence of a 2pn*-3dn interaction of aromatic silanes has yet been found.15 In previous papersl29 l3 we found that the intramolecular charge-transfer (c.t .) fluorescence of aromatic disilanes has a broad and structureless band with a large Stokes shift. The c.t. state is closely related to the photochemical reactions of aromatic disilanes.l3? 2 3 9 24 At that time, no assignment of the c.t. state [l(2pn, 3dn) or ‘(2pn, a*)] was made.13 We now demonstrate that the c.t. fluorescence originates from the ‘(2~72, 3dn) state produced by 2pn* + 3dn intramolecular charge transfer. Evidence for the 2pn*-3dn interaction in aromatic disilanes is shown in the present work. The mechanism of the l(2pn, 3dn) c.t. state formation leading to the longer-wavelength emission is also discussed in comparison with that of the twisted intramolecular charge-transfer (t.i.c.t.) state.34 1342 INTRAMOLECULAR CHARGE TRANSFER IN AROMATIC SILANES EXPERIMENTAL MATERIALS The samples of pentamethylphenyldisilane ( 1),25 tris(trimethylsily1)mesitysilane (5),26 tris- (trimethylsily1)phenysilane (4),,' 1-naphthylpentamethyldisilane (7),,, 2-naphthylpenta- methyldisilane (8),24 1-naphthyltrimethylsilane (9),* and 2-naphthyltrimethylsilane were synthesised by the usual methods. Other samples were synthesised as follows (the solvents being purified by the usual methods30). 2,5-Dimethylphenylpentamethyldisilane (2) : In a 500 cm3 three-necked flask fitted with a stirrer, a dropping funnel and a reflux condenser was placed 2,5-dimethylphenylmagnesium chloride prepared from 2.1 g (0.086 mol) of magnesium and 12 g (0.086 mol) of 2,5- dimethylchlorobenzene in 150 cm3 of THF.To this was added 10 g (0.06 mol) of chloropenta- methyldisilane dissolved in 20cm3 of THF. The mixture was refluxed for 2 h and then hydrolysed with dilute hydrochloric acid. The organic layer was separated, washed with water and dried over potassium carbonate. Distillation under reduced pressure gave 6.8 g (49% yield) of 2,5-dimethlyphenylpentamethyldisilane, b.p. 75 OC (2 mm); lH n.m.r. 6 (ppm): 0.06 (9H, s, Me,Si), 0.36 (6H, s, Me,Si), 2.30 (3H, s, 0-CH,), 2.35 (3H, s, rn-CH,), 7.03 (2H, m, ring protons), 7.17 (lH, s, ring proton). Calculated analyses for C,,H,,Si,: C, 66.02; H, 10.23%. Found: C, 65.86; H, 10.48%. Mesitylpentamethyldisilane (3): To a mesityl-lithium reagent prepared from 10 g (0.050 mol) of mesitylbromide and 1.5 g of lithium metal in 100 cm3 of ether was added 4 g (0.024 mol) of chloropentamethyldisilane at room temperature. The mixture was refluxed for 5 h and then hydrolysed with water.The organic layer was separated, washed with water and dried over potassium carbonate. Distillation under reduced pressure gave 4.2 g of a light-yellow liquid boiling over a range of 60-75 OC (2 mm). Pure mesitylpentamethyldisilane was isolated by preparative vapour-phase chromatography. lH n.m.r. 6 (ppm): 0.06 (9H, s, Me,Si), 0.45 (6H, s, Me,Si), 2.22 (6H, s, 0-CH,), 2.33 (3H, s,p-CH,), 6.67 (2H, s, ring protons). Calculated analysis for C,,H,,Si,: c, 67.11; H, 10.46%. Found: C, 67.03; H, 10.21%. 9-(Pentamethyldisilany1)phenanthrene (11) : To a solution of 9-lithiophenanthrene prepared from 15 g (0.06 mol) of 9-bromophenanthrene and 36 cm3 (0.06 mol) of butyl-lithium hexane solution in 100 cm3 of a 1 : 1 mixture of ether and THF at -60 OC was added 9.7 g (0.06 mol) of chloropentamethyldisilane in 15 cm3 of THF.The mixture was stirred for 1 h at room temperature and then hydrolysed with water. The organic layer was separated, washed with water and dried over potassium carbonate. The solvent was evaporated and the residue was distilled under reduced pressure using a short column to give 7.9 g (43 % yield) of a colourless liquid, b.p. 165-167 OC (1 mm); 'H n.m.r. 6 (pprn): 0.10 (9H, s, Me3%), 0.56 (6H, s, Me,Si), 7.4-8.1 and 8.5-8.8 (9H, m, ring protons). Calculated analysis for C,,H,,Si,; C, 73.95; H, 7.84%.Found: C, 73.77; H, 7.71%. 9-(Pentamethyldisilany1)anthracene (12) : In a 300 cm3 flask was placed 13.0 g (0.05 mol) of 9-bromoanthracene dissolved in a mixed solvent consisting of 80 cm3 of ether and 100 cm3 of THF. To this was added 31 cm3 (0.05 mol) of butyl-lithium hexane solution over a period of 15 min at -60 to -40 "C. The mixture was stirred for 1 h at -20-0 OC. To this was added 8.3 g (0.05 mol) of chloropentamethyldisilane in 10 cm3 of THF at the same temperature. The mixture was then warmed up to room temperature, stirred for 2 h and hydrolysed with water. The organic layer was washed with water and dried over potassium carbonate. The solvent was evaporated and the residue of the flask was distilled under reduced pressure using a short column to give 7.5 g (49% yield) of pale yellow crystals, b.p.165-168 OC (1 mm); m.p. 87 OC (after recrystallisation from ethanol). 'H n.m.r. 6 (ppm in C,D,): 0.16 (9 H, s, Me,Si), 0.75 (6H, s, Me,Si), 7.3-7.4, 7.8-7.9 and 8.2-8.6 (9H, m, ring protons). Calculated analysis for C,,H,,Si, : C, 73.95; H, 7.84%. Found: C, 73.81; H, 7.59%. 1-(Pentamethyldisilany1)pyrene (13): In a 500 cm3 three-necked flask was placed 11.3 g (0.040 mol) of 4-bromopyrene dissolved in a 1 : 1 mixture of ether and THF. To this was added 28 cm3 (0.042 mol) of butyl-lithium hexane solution over a period of 20 min at - 60 to - 50 "C. The mixture was stirred for 1 h at - 50 to - 30 OC, and then warmed up to room temperature. Chloropentamethyldisilane (7.0 g, 0.042 mol) was added slowly to the solution of lithiopyrene, using ice to cool to solution.The mixture was stirred for 4 h at room temperature andH. SHIZUKA et al. 343 hydrolysed with water. The organic layer was washed with water and dried over potassium carbonate. The solvent was evaporated and the resulting crystals (19 g, 90% yield) was chromatographed to give light-yellow crystals of 1-(pentamethyldisilanyl)pyrene, m.p. 102-103 "C (recrystallization from ethanol); lH n.m.r. S (ppm): 0.12 (9H, s, Me,Si), 0.65 (6H, s, Me,Si), 7.8-8.3 (9H, m, ring protons). Found: C, 75.69; H, 7.08%. The compounds used are as follows: @Me,SiMe3 1 Calculated analysis for C,,H,,Si,: C, 75.83; H, 7.27%. Me Me 3 Me Me .Me 'Me 4 5 6 SiMe2SiMe3 SiMe3 ~ S i M e 2 S i M e 3 a 13 APPARATUS AND PROCEDURES All samples were thoroughly degassed by freeze-pumpthaw cycles on a high-vacuum line.The absorption and emission spectra were recorded with Hitachi 200 and 139 spectrophotometers and a Hitachi MPF 2A fluorimeter, respectively. Spectral corrections for emissions were made. The fluorescence quantum yields were measured by comparison with a quinine bisulphate 0.05 mol dm-, H,SO, solution (aF = 0.54).31-32 The fluorescence and phosphorescence quantum yields at 77 K were determined from the fluorescence quantum yield of toluene (aF.= 0.29)33 in MP glass at 77 K. The fluorescence response functions were recorded using a Hitachi nanosecond time-resolved fluorimeter (pulse width 1 1 ns) and the convolution method was applied.34 The phosphorescence lifetimes were measured using a Hitachi MPF 2A fluorimeter or a transient memory (Kawasaki MR-SOE) with a photomultiplier (1 P28).Low-temperature experiments were carried out using a cryostat (Oxford DN 704). RESULTS AND DISCUSSION ABSORPTION AND EMISSION SPECTRA OF AROMATIC SILANES Some examples of the solvent shifts of the absorption and fluorescence spectra for mol dm-3; 8: mol dm-3) at 300 K are shown in fig. 1 and 2, respectively. The first phenyldisilane (1 : ca. 8 x mol dm-3) and naphthyldisilanes (7:4 x344 INTRAMOLECULAR CHARGE TRANSFER IN AROMATIC SILANES wavelength, h/nm 500 400 300 0.5 ' " ~ 3yl J I' n - 20 25 30 35 40 45 wavenumber, F/ 1 O3 cm-' Fig. 1. Absorption and fluorescence spectra of 1 in various solvents at 300 K. CH: cyclohexane; THF : tetrahydrofuran; EtOH : ethanol; PN: propionitrile; AN: acetonitrile.absorption band of 1 with vibrational structures at 38.5 x cm-l corresponds to the lB,, +- lAlg (lLb + 'A) transition at 39.3 x lo3 cm-135 in benzene36 and the second band (43.3 x lo3 cm-l)maycorrespond to the lB1, + lAlB (lLa +- lA) transition (49.0 x lo3 in benzene. The solvent shifts for the former band were scarcely observed and only slightly for the latter band. Dual fluorescences for 1 were observed: one is the normal fluorescence corresponding to the 11?2u + lA,, radiative transition in benzene and the other the intramolecular c.t. fluorescence having a broad and structureless band at longer wavelengths. Slight solvent shifts for the former emission were observed, whereas the peak of the longer wavelength emission was shifted to the red depending upon solvent polarity.Similar results were obtained for 7 and 8, as shown in fig. 2. The absorption spectra of 7 and 8 in non-polar solvents are similar to that of naphthalene. The first [31.8, x lo3 cm-l (7); 31.7 x lo3 cm-l (8)] and second [34.8 x lo3 cm-l (7); 36.8 x lo3 cm-l (S)] absorption bands correspond to lLb +- lA (33.2 x lo3 cm-1)35 and lL, + lA (36.4 x lo3 ~ m - 9 ~ ~ transitions in na~hthalene,~~ respectively. Solvent shifts were also observed. There was very little solvent shift on the normal emission bands (lLb -+ lA) at ca. 30 x lo3 cm-l, but in constrast the broad and structureless emissionH. SHIZUKA et al. wavelength, X/nm 345 500 400 300 n wavenumber, v/103 em-' (a) ( b 1 Fig.2. Absorption and fluorescence spectra of 7(a) and 8(b) in various solvents at 300 K. DCE: 1,2-dichloroethane, band was red shifted in polar solvents. The solvent shifts of the longer-wavelength emission of 1, 7 and 8 indicate that their emitting states are very polar. The broad and structureless emission is not due to an excimer emission judging from the concentration effect on the spectral change. The excitation spectra at both fluorescence band maxima were comparable with those of the absorption bands at wavelengths > 260 nm. The broad and structureless spectra at longer wavelengths are therefore attributed to the emissions from intramolecular charge-transfer (c. t.) states of the compounds. The fact that there is a tendency for the broad and structureless fluorescence of the a-isomer (7) to be more prominent than that of the 8-isomer (8) supports this assignment, It is known that in general the intramolecular c.t.character in the excited state of a-substituted naphthalenes is predominant compared with that of 8-substituted naphthalene~.~'-~O The anomalous Stokes shifts in fig. 1 and 2 can be ascribed to the change in solute and solvent interaction during the lifetime of the excited state of the aromatic solute molecule. Similar dual emissions were observed for 2 and 4, as described later, but not for monosilyl derivatives (6, 9 and 10) and polycyclic aromatic compounds (11, 12 and 13: aromatic rings 2 3) even in polar solvents. The lack of c.t. emission from the monosilyl compounds suggests that the Si-Si bond might be the electron acceptor. For 3 and 5, dual emission were observed in both polar and non-polar solvents at room temperature, but not in MP and EPA rigid matrices at 77 K, as discussed later.The experimental data for the absorption and emission spectra of aromatic silanes in cyclohexane at 300 K are listed in table 1.346 INTRAMOLECULAR CHARGE TRANSFER IN AROMATIC SILANES Table 1. Absorption (2%) and fluorescence band (2%) maxima, fluorescence quantum yields (Qfm) and lifetimes (rfm) of aromatic silanes in cyclohexane at 300 Ka 1 2 3 4 5 6 7 8 9 10 11 12 13 23 1 234 243 24 1 247 260d 287 272 282. 279 373 350. , 301 ' 5 1.1, x 104 1.0, x 104 1.3, x 104 1.3, x 104 1.4, x 104 9.9, x 103 7.3 x 103 5.1 x 103 8.4, x 103 5.1, x 104 3.0 x 10, 6.6, x lo3 1.6, x lo4 28Y 29gC 285.3c 285. gc 285. 3c - 328. , 334 329 330., 367.6 434.7 388. 1 . R x 10-3 1 .4 x 10-3 - 0.05, 0.15 0.25 0.24 0.09, 0.72, 0.48, < 1 < 1 < 1 < 1 <1 8.0 68., 52., 49.6 73.5 12., 343 a Data in MP (methylcyclohexane: isopentane = 3 : 1) were the same as those in cyclohexane. For details, see text. The c.t. fluorescences at 340 nm were observed appreciably for 1 and slightly for 2-5. Corresponding to the 'B, (l&) band. ESTIMATION OF THE DIPOLE MOMENT OF THE INTRAMOLECULAR C.T. STATE In addition to the solvents shown in fig. 1 and 2, several other solvents were used when recording the fluorescence spectra. These data are listed in table 2. Values of Ap (= pct --p& where p,, and pg denote the dipole moments of the excited c.t. state and the ground state, respectively, can be estimated from the Lippert-Mataga equation:*'* 42 where VZ and VZ are the wavenumbers of the peaks of the absorption band and of the corresponding c.t.emission band, respectively, D and n are the dielectric constant and refractive index of the solvent, respectively, and a is the Onsager radius. The observed values of vs, V% and the Stokes shift Avct are listed in table 2, together with the values of F(D, n). A plot of Avct as a function of F(D, n) gives a straight line, as shown in fig. 3. Eqn (1) is approximately satisfied in the present work. The following experimental equations in units of cm-l are obtained from fig. 3: 1 : 7: Av,, = 3419+20137F(D, n) (r = 0.935) 8: Av,, = 4355+ 12 138F(D, n) Avct = 9087 + 7747F(D, n) (r = 0.768) ( r = 0.942) (3) (4)Table 2.Absorption (vZ), normal fluorescence (v&) and c.t. fluorescence (vZ) maxima and Stokes shifts (AvJ of 1, 7 and 8 in various solvents at 300 Ka CH ether THF DCE BN EtOH PN AN glycerol DMSO 0 0.167 0.210 0.221 0.275 0.289 0.290 0.306 0.263 0.264 38.5 38.5 38.5 38.5 38.5 38.5 38.5 38.5 - 34 29.4 - 34 28.0 - 34 28.4 - 34 27.0 - 34 27.9 - 34 28.0 - 34 27.6 - 34 26.2 9.1 10.5 10.1 11.5 10.6 10.5 10.9 12.3 9 3 2 31.8, 29.6, 22.7, 9.1, 31.6, 29.5, 24.1 7.5, * 31.7, 29.4, 22.4 9.3, 31.6, 29.5 23.9, 7.7, 2 31.8, 29.6, 22.4, 9.3, 31.6, 29.6, 23.6, 8.0, % 31.7, 29.8, 22.4, 9.2, 31.6, 29.6, 23.5 8.1, 31.7 30.4 - - 31.6, 29.9, - - - - - - m 31.8, 30.4 25.0 6.85 31.8, 29.6, 24.8, 6.9, 31.6, 29.5 25.0 6.6, 31.7, 30.4 23.3, 8.4, 31.6, 29.4, 24.5, 7.0, R 31.7, 29.4, 23.0, 8.7, 31.6, 29.4, 25.0 6.65 - - - - 31.7, 28.0, 23.5 8.7, a For details, see text.CH: Cyclohexane; ether: diethyl ether; THF: tetrahydrofuran; DCE: 1,2-dichloroethane; BN: n-butylonitrile; EtOH: ethanol; PN: n-propionitrile; AN: acetonitrile; DMSO: dimethylsulphoxide. In units of lo3 cm-l. w P 4348 INTRAMOLECULAR CHARGE TRANSFER IN AROMATIC SILANES 14 12 t c E 7 c Ei mE 10 : 5 \ 0 10 5 f 0 0.1 0.2 0.3 0.4 F (D, n ) Fig. 3. Plots of Apct as a function of F(D, n). (a) 1, (b) 7 and ( c ) 8. For details, see text. where r is a correlation coefficient. From eqn (2)-(4), the Ap values for 1, 7 and 8 can be evaluated to be ca. 4.4, 11.3 and 8.8 D, respectively, on the assumption that the Onsager radii for phenyldisilane (1) and naphthyldisilanes (7 and 8) are 3 and 4 A, respectively. The dipole moments in the ground state of aromatic silanes seem to be small (ca.0.4, D) because of + I and - M 22 Thus, the values of pet for 1, 7 and 8 are ca. 4.8, 11.7 and 9.2 D, respectively, on the assumption that pg x 0.4 D. We therefore conclude that the fluorescent state at the longer-wavelength emission has a strongly polar structure with complete charge separation. ASSIGNMENT OF THE INTERMOLECULAR C.T. STATE The mechanism for the intramolecular c. t. emission at longer wavelengths is described in this section. There are two possible explanations for the c.t. emission mechanism. One is the 2pn* -+o* c.t. mechanism: the longer-wavelength emission may be due to intramolecular charge transfer from the aromatic n electron system into a vacant Q* orbital of the Si-Si bond.On consideration of the likely molecular conformations arising from steric interactions, this 2pn* -+ o* c. t. interpretation may explain adequately the differences between 1 - and 2-naphthyldisilanes and the lack of c.t. fluorescence in the monosilane derivatives. If this is so, the intramolecular c.t. state is the l(2pn7 o*) state produced by the 2pn* + Q* c.t. The other is the 2pn* -+ 3dn c.t. mechanism: the broad and structureless emission may originate from an intra- molecular c.t. state resulting from charge transfer from the 2pn* orbital of the aromatic ring to the vacant 3dn orbital of the Si-Si bond. The assignment of the intramolecular c.t. state has been made experimentally. Samples 1-5 were chosen in order that molecular conformations arising from steric interactions might affect the intramolecular c.t.emissions, which would give information about the intramolecular c.t. state.H. SHIZUKA et aZ. 349 (a 1 (b 1 wavelength, h/nm 500 LOO 300 500 LOO 300 I ' 1 I 1 . 1 . I 1 x I0 1.0 - ' 0.5- h .z 0 2 X I 0 x 10 fluo. 20 30 LO 20 30 LO Fig. 4. (a) Absorption and fluorescence spectra of 1, 2 and 3 in acetonitrile at 300 K and (b) absorption (in EPA at 300 K) and emission (in EPA glass at 77 K) spectra of 1, 2 and 3. The emission spectra of 1 and 2 in EPA glass at 77 K consist of the normal (f.m.) and charge-transfer (c. t.) fluorescence spectra and the normal phosphorescence (phos). For details, see text. wavenumber, T/l O3 cm-' The absorption and emission spectra of the phenyldisilanes 1, 2 and 3 (ca.lov4 mol dm-3) in acetonitrile (AN) and EPA (ether + isopentane + alcohol, 5 : 5 : 2 ) are shown in fig. 4. In AN at 300 K, dual fluorescence was observed for 1 and 2: one is the normal fluorescence corresponding to the lB,, + lAlg radiative transi- tion in benzene and the other the intramolecular c.t. emission having a broad and structureless band with a large Stokes shift (12.3 x lo3 cm-l for 1 and 9.7 x lo3 cm-l for 2). The fluorescence intensities of the c.t. emissions relative to those of the corresponding normal emissions are in the order 1 > 2 >> 3, and in contrast the intensities for the normal emissions are 1 < 2 < 3, as can be seen in fig. 4(a). Similarly, dual fluorescence was observed for 1 and 2 even in an EPA rigid matrix at 77 K, but not for 3 [fig. 4(b)].From the total emission spectra, the quantum yields for the normal (Qfm) and c.t. (met) emissions and the phosphorescence (ap) in an EPA rigid matrix at 77 K following 263 nm excitation were determined to be 0.09, 0.09, and 0.35 for 1; 0.17, 0.03 and 0.42 for 2 ; and 0.29,, 0 and 0.62, for 3, respectively. The fluorescence quantum yields for 1 and 2 were very small (ca. at room temperature and the lifetimes of dual emissions were very short (< 1 ns). The values of mfm and met increased markedly with decreasing temperature. This may be because the photochemical reaction occurs to a considerable extent at room temperature; e.g. the reaction quantum yield for 1 in MP at 300 K is very large (0.86).23 The experimental data can be explained by the following mechanism. If the intramolecular c.t. fluorescence of phenyldisilanes originated from the '(2pn, a*) state, the intensity of the c.t. emissions for 3 or 5 would be large compared with those for 1,2 and 4, since for 3 and 5 the intramolecular charge transfer from the 2pn* orbital to a vacant Q* orbital of the Si-Si bond would be much more efficient than for 1,3 50 INTRAMOLECULAR CHARGE TRANSFER IN AROMATIC SILANES 2 and 4 because of the out-of-plane molecular structure (the n orbital being able to overlap with the CT* orbital) arising from steric interactions, as shown in fig. 5(b). However, the experimental results argue against this mechanism. An in-plane or in-plane-like structure is more favourable for intramolecular c.t. interactions between z and 3dn systems, as can be seen in fig. 5(a). There are two vacant dn orbitals in >si 7 (a 1 (b ) Fig. 5. Schematic molecular conformations : (a) in-plane structure and (6) out-of-plane structure of phenyldisilanes. For details, see text. --- 3drr R- - Si - Si-Me I I M e M e Fig. 6. Schematic energy-state diagram for the intramolecular charge-transfer 2pn* + 3dn ( k c t ) of aromatic disilanes. R = phenyl or naphthyl. a disilane, perpendicular to each other. In a planar phenyldisilane, one d7t orbital will be perpendicular to the plane of the ring and thus conjugated with the aromatic n orbitals, the other not. The effect of twisting the phenyl group (or the disilanyl group) through 90" will be to break the conjugation with the one dn orbital, and it would be expected to allow conjugation with the other if the CA,-Si-Si bond were straight.However, the system cannot overlap with the other dz orbital by twisting since the bond angle of C,,-Si-Si is estimated to be ca. 1 loo, considering that the Si atom has tetrahedral sp3-hybrid orbitals. That is, the electronic overlap between 7t and dn systems is dominant for a planar structure [fig. 5(a)], but not for an out-of-plane (i.e. twisted) conformation [fig. 5(b)]. It is possible for 1,2 and 4 to make an in-plane-like conformation by rotation around the C-Si bond axis but not for 3 and 5. Therefore, the intramolecular c. t. fluorescence of phenyldisilanes (or polysilanes) is attributed toH.SHIZUKA et al. 35 1 the emission from the '(2pn, 3dn) state produced by the 2pn*+3dn intramolecular charge transfer. The c.t. emissions of naphthyldisilanes (7 and 8) may be ascribed to those from the l(2pn, 3dn) state by considering their features, which are similar to those of phenylsilanes. The 2pn*-3dn interaction resulting in formation of the l(2pn, 3dn) c.t. state is caused by (1) the increase in electron-donating power of the aromatic ring upon excitation, because of the promotion of an electron from the 2pn to 2pn*; ( 2 ) the disilanyl group (or polysilanyl group) provides a vacant 3dn orbital, which may be produced by the linear combination of 3d atomic orbitals, with a relatively low energy level compared with that of a single 3d atomic orbital [i.e.the electron-withdrawing power of the disilanyl (or polysilanyl) group increases]; and ( 3 ) the l(2pn, 3dn) energy level is lowered by polarisation energy because of the interaction between the c.t. state species and polar solvent molecules. Thus, the c.t. emission from the l(2pn, 3dn) state can be accounted for by the schematic energy diagram shown in fig. 6. DYNAMIC BEHAVIOUR OF THE '(2pn, 3dn) C.T. STATE The lifetimes of the excited phenyldisilanes at room temperature were very short (< 1 ns) and nanosecond time-resolved experiments with phenyldisilanes (1, 2 and 3) in EPA rigid matrices at 77 K were carried out. The fluorescence response function Ifm(t) of the normal emission at 290 nm for 1 and 2 [fig. 7(b) and (c), respectively] in an EPA rigid matrix at 77 K consisted of fast and slow decay components Iim(t) and Ifm(t), respectively.The slow decay function Ifm(t) at the delay time t > 40 ns shows a single-exponential decay (20. , ns for 1; 20. ns for 2), and the shape of the Zfm(t) function at t < 40 ns can be determined by the convolution method.34 The fast decay function Igm(t) is derived by subtraction of Irm(t) from Ifm(t) under ideal conditions. The I ; , (t) function is very close to the lamp function I, (t) of the D, pulser with a 6 function [fig. 7 (a)], showing the existence of the short-lived component ( < 1 ns for 1 and 2) in addition to the long-lived component (ca. 21 ns) in the l(n,n*) state. This fact can be understood by considering the molecular structure in the excited state of disilanes.That is, both in-plane-like and out-of-plane conformations exist in the locally excited state l(n, n*) 'B, of 1 (or 2) in EPA glass at 77 K. There is no conformational change in either the ground or the excited states under such conditions. The intramolecular 2pn* + 3dn c.t. occurs effectively for the former conformation resulting in fast decay, but not for the latter, which has a relatively long lifetime (ca. 21 ns). This finding shows the rapid formation of the c.t. state via the l(n, n*) lB, state of 1 or 2, which has a molecular conformation favourable for intramolecular c.t. even in EPA glass at 77 K. The fluorescence quantum yields mim and atm for the short- and long-lived components can be determined to be 0.06, and 0.02, for 1 and 0.10 and 0.07 for 2, respectively, using the following equations: and352 INTRAMOLECULAR CHARGE TRANSFER IN AROMATIC SILANES where Qfm denotes the fluorescence quantum yield for the shorter-wavelength emission in EPA glass at 77 K, as listed in table 3 .Note that ( 1 ) the short-lived component in the l(n,n*) state is greater than the long-lived one, showing efficient intramolecular 2pn* -+ 3dn c.t. even in an EPA rigid matrix at 77 K, and (2) the value of the @[m/Q;m ratio (2.1) for 1 is greater than that (1.4) for 2, as can be expected from their molecular structures. As for 3, which has an out-of-plane molecular 0 20 G O 60 80 0 20 LO 60 80 tlns tlns Fig. 7. Fluorescence response functions Z, ( t ) of 1 (b), 2(c) and 3(d) monitored at 290 nm in EPA glass at 77 K and the lamp function Zl ( t ) (a).Z:,(t) and Itrn ( t ) are the fast- and slow-decay components of the normal fluorescence, respectively. For details, see text. conformation, a single-exponential decay at 290 nm was observed in EPA glass at 77 K (Tfm = 31 . ns), as shown in fig. 7(4, indicating no intramolecular charge transfer. The difference in the decay features for 1, 2 and 3 is ascribed to their molecular conformations. The decay data are fairly consistent with the emission spectra in fig. The fluorescence response function Ict(t) at the longer wavelengths (@ NN 350 nm) corresponding to the c.t. emission was very close to the I i m ( t ) function, indicating the existence of a fast intersystem crossing k.t. C1(2pn, 3dn)]-t3(n, n*). This fast intersystem crossing process (kist in fig.8) may be due to the small energy gap ( 1 . 1 x lo3 cm-l) between them in addition to large 1.s. coupling with 3d character in the c.t. state where appreciable s.t. mixing is probably caused by vibronic S.O. coupling between the states. The 3(2pn, 3dn) state may exist energetically just below the l(2pn, 3dn) state. However, the intersystem crossing process between '(2pn, 3dn) and 3(2pn, 3dn) c.t. states is forbidden according to Lim's rule.43 The relaxation processes for the excited state of 1 (or 2) can be accounted for by the scheme as shown in fig. 8. In benzene and benzene derivatives the S, state ['(n, n*) l&] deactivates via fluorescence (kfm), internal conversion (kim) and indirect (kf,,) and direct (k&) intersystem crossing.33 When the disilanyl group is introduced into the benzene ring 5 (b).Table 3.Normal (v&) and c.t. (v:) fluorescence maxima, the 0-0 transition energies for S1(n, n*)+S, (@,) and T, (n, n*)-+S, emission quantum yields for normal (afm) and c.t. (@,J fluorescences and for normal phosphorescence (OP), normal fluorescence lifetimes for short (rim) and long (r:,) lived species and normal phosphorescence lifetimes (zp) of phenyldisilanes in EPAa glass at 77 K rf b , c rFmb,d F sample /lo3 cm-, /lo3 cm-l /lo3 cm-1 /lo3 cm-l Qfm @imb /@Fmb /ns /ns @ct @p rpd/s E 2 1 34.0 36.3 28.5 27.3, 0.09 0.06, 0.02, 2.1 < I 20., 0.09, 0.35 0.20 F - - - - - 0.62, 0.21 z 4 33.0 35.0 28.3, 28.0 0.03 - - - 6 1 18 0.036 0.41, 0.07 % m %n G, i% %, @imb fm 2 34.2, 35.0 27.6, 27.1 0.17 0.10 0.07 1.4 61 20., 0.03 0.42 0.39 31 .4 3 33.8, 34.9, - 26-95 0.296 - - 5 33.6, 34.7, - 27.0 0.17, - - 27., - 0.15 0.06 a EPA = ether: isopentane: alcohol = 5: 5: 2. For details, see text. Experimental errors within f 20%. Experimental errors within & 5%. w wl w3 54 INTRAMOLECULAR CHARGE TRANSFER IN AROMATIC SILANES such as 1 (or 2), the 2pn* + 3dn c.t. process (kct) is predominant in comparison with other processes: i.e. kct > k,, + kic + kf,, + kfsc. The fast intramolecular c.t. takes place effectively in the S, state having an in-plane-like structure in EPA glass at 77 K, followed by rapid intersystem crossing (kist) to produce the T, state [3(n, n*)], as described above. A similar tendency was obtained for 4 and 5. These data are summarised in table 3.In contrast, in non-polar MP (methylcyclohexane: isopentane = 3 : 1) glass at 77 K, relatively long lifetimes with single-exponential decays were obtained as 34 (l), 24 (2) and 30 (3) ns. The relatively long lifetimes are because the efficiency of transfer to the c.t. state is reduced in non-polar MP glass at 77 K.23 The phosphorescence spectra correspond to the 3B1, + lA,, radiative transition in benzene. The phosphorescence lifetimes of 1,2 and 3 in EPA glass at 77 K were short (0.20, 0.39 and 0.21 s, respectively), compared with that (8.8 s) of toluene, because of perturbation by the disilanyl group in the 3(n, n*) s, (14, 9' Fig. 8. Schematic energy-state diagram for the relaxation processes in the excited state of phenyldisilanes (1 and 2).For details, see text and ref. (33). MECHANISM OF THE l(2pn, 3dn) C.T. STATE COMPARED WITH THAT OF THE T.I.C.T. The intramolecular c.t. state or exciplexes in the electron donor (Dbacceptor (A) system of D-(CH,),-A (n 3 0) have been extensively s t ~ d i e d . ~ ~ - ~ ~ For the case of n = 0 (i.e. the D-A system), a strongly dipolar twisted intramolecular charge-transfer (t.i.c.t.) state leading to a longer-wavelength emission is known as an intramolecular c.t. conformation, where the n systems of D and A are twisted with respect to each ~ t h e r . ~ ~ + ~ ~ - ~ ~ Twisting or internal rotation during the lifetime in the excited state is needed for t.i.c.t. f ~ r m a t i o n . ~ ~ - ~ ~ For instance, c.t. emission is observed for the dimethylaminobenzonitrile (DABN) derivative held rigidly in the perpendicular position, but not for the corresponding derivative with a rigid planar s t r u c t ~ r e .~ ~ ~ 51 In contrast, the exciplexes of the sytem with n 2 1 tend to produce c.t. complexes with spatial overlap.44 In the present system of D (phenyl group)-A (disilanyl group), twisting or internal rotation around the C-Si bond axis in an EPA rigid matrix at 77 K is not expected during the lifetime of the excited state, and the Franck-Condon state molecular conformation in the rigid glass at 77 K may be the same as that in the ground state. C.t. emission in EPA glass at 77 K was observed for 1 with a planar or planar-like STATEH. SHIZUKA et al. 355 structure, but not for 3 with a twisted (perpendicular) structure, as described above.This finding shows that twisting or internal rotation in the excited state is not necessary for l(2pn, 3dn) c.t. fo-mation of phenyldisilanes, although internal rotation around the C-Si bond axis in the compounds enhances formation of the l(2pn, 3dn) state in fluid media. For naphthyldisilanes (7 and 8), the c.t. emission was observed only in fluid polar media, but was not appreciable in a MP or EPA rigid matrix at 77 K and fluid non-polar solvents. These features for 7 and 8 suggest that the c.t. state might be produced by twisting in the excited state according to the t.i.c.t. mechanism.49 The discrepancy in the c.t. emission properties between phenyl and naphthyl disilanes can be understood by taking into account the differences in the energy levels of the l(z, n*) 'B, (lLb) states (i.e.the 0-0 transition energies of the donor groups). The 0-0 transition energy (36.2 x lo3 cm-l) for 1 is much greater than those for naphthyl- disilanes (31.1 x lo3 cm-l for 7 and 30.8 x lo3 cm-l for 8). For the energy level of the 3dn vacant orbital of the acceptor (disilanyl group), there is no difference between them. As a result, the 2pn*+3dn c.t. occurs more effectively in 1 than in 7 and 8. The 2pn* -+ 3dn c.t. for 7 and 8 needs some stabilisation energy due to interaction between the newly produced '(2pn, 3dn) c.t. state and polar solvent molecules, whereas the c.t. process for 1 occurs appreciably in a fluid non-polar solvent. The polar solvents play an important role in lowering the energy level of the l(2pn, 3dn) state by solvation, as stated above.No intramolecular c. t. emission was observed for polycyclic aromatic disilanes (11, 12 and 13: aromatic rings 3 3) whose l(n, n*) energy levelst are less than 30 x lo3 cm-l. This result indicates that the energy level of l(n, n*) should be higher than 30 x lo3 cm-l in order to produce the l(2pn, 3dn) state from l(n, n*). It is known that the energy levels between the donor and acceptor groups are very important for the ground intramolecular c.t. The present system is such a case in the excited state. The energy of the t.i.c.t. state can be approximately given 59 where Ip* is the ionisation potential of the donor, Ea is the electron affinity of the acceptor parts of the twisted molecule, C is the Coulomb interaction energy and AEs is the solvent stabilisation energy depending on the temperature and on the polarity of both solvent and solute in the t.i.c.t.state. We examined whether eqn (7) might hold for aromatic disilanes or not. By changing the donor properties the values of Etict might be expected to be dependent upon those of ($ = Ip -go), where ID and E,*, represent the ionisation potential in the ground state and the 0-0 transition energy of the donor group. The $ valuest were estimated to be 4.49 (benzene), 4.13 (naphthalene), 4.41 (phenanthrene), 4.07 (anthracene) and 4.22 eV (pyrene). There is no distinct relation between the c.t. emission properties and the $ values in the present system. The mechanism for l(2pn, 3dn) intramolecular c.t. formation is different from the t.i.c.t.mechanism. Thus, it is concluded that the interaction between 2pn* and a vacant 3dn orbital is possible when the 2pz* energy level is greater than 30 x lo3 cm-l (the number of aromatic rings is 1 or 2) and there is some electronic overlap between the 2pn* orbital of the donor (aromatic ring) and the vacant 3dn orbital of the acceptor (disilanyl). This l(2pn, 3dn) intramolecular c. t. state is related to the photochemical reactions of aromatic disilanes ; that is, the photochemical reactions originate from the c.t. state.l3~ 2 3 9 24 t The 0-0 transition energies for 11, 12 and 13 are 28.6, 24.4 and 26.8 x lo3 cm-', respectively. 1 The Zp and values were taken from the data in ref. (35).356 INTRAMOLECULAR CHARGE TRANSFER IN AROMATIC SILANES CONCLUSIONS (1) Broad and structureless fluorescence spectra of phenyldisilanes and naphthyl- disilanes with large Stokes shifts are observed and are attributed to the emissions from the intramolecular c.t.states with large dipole moments. Evidence that the c.t. emission originates from the '(2pn, 3dn) state produced by 2pn* -+ 3dn intramolecular charge transfer is demonstrated by the effect of steric twisting on the emission. (2) For aromatic monosilanes and polycyclic disilanes (aromatic rings 2 3), no c.t. emission is observed even in polar solvents. (3) The 2pn*-3dn charge transfer to produce the '(2pn, 3dn) c.t. state occurs for the following reasons: (a) the energy separation between the 2pn* and vacant 3dn orbitals becomes small upon excitation: the energy level of the l(n, n*) 'B, (or lLb) state should be > 30 x lo3 cm-l (aromatic rings = 1 or 2), (b) the disilanyl group (or polysilanyl group) provides a vacant 3dn orbital with a relatively low energy level compared with that of a single 3d atomic orbital and (c) the '(2pn, 3dn) energy level is lowered by solvation energy in polar media.(4) The fast formation (< 1 ns) of the l(2pn, 3dn) c.t. state of phenyldisilane (1) from the locally excited state of l(n, n*) 'B, (or lLb) occurs effectively, followed by the rapid decay ( < 1 ns) of intersystem crossing l(2pn,3dn) + 3(n, n*), even in an EPA rigid matrix at 77 K. ( 5 ) The 2pn* -+ 3dn intramolecular charge transfer takes place with electronic overlap between the 2pn* orbital of the aromatic ring (the electron donor) and the vacant 3dn orbital of the disilanyl group (the electron acceptor), and internal rotation or twisting during the lifetime of the excited state is not necessary for the intramolecular c.t.process, although such twisting enhances the c.t. process. The intramolecular c.t. mechanism for l(2pn, 3dn) formation is different from the t.i.c.t. mechanism. This work was supported by a Scientific Research Grant-in-Aid from the Ministry of Education of Japan (no. 56540246 and 57307007). We thank one of the reviewers for his helpful comments. E. G. Rochow, An Introduction to the Chemistry of the Silicones (Wiley, New York, 2nd edn, 1951). C. Eaborn, Organosilicon Compounds (Butterworths, Boston, 1960). H. Gilman, W. H. Atwell and G. L. Schwebke, Chem. Ind., 1964, 1063; J.Organomet. Chem., 1964, 2, 369. (a) R. West, Ann. N. Y. Acad. Sci., 1974, 239, 262; (6) R. West and E. Carberry, Science, 1975, 189, 179. H. Gilman, W. H. Atwell and F. K. Cartledge, Adv. Organornet. Chem., 1966, 4, 1 . B. G. Ramsey, Electronic Transitions in Organometalloids (Academic Press, New York, 1969), chap. 4. L. E. Gusel'nikov, N. S. Nametkin and V. M. Vdovin, Ace. Chem. Res., 1975, 8, 18. M. Ishikawa, Pure Appl. Chem., 1978,550, 1 1 . M. Gielen (Freund, Tel-Aviv, 1979), vol. 4, p. 7. M. Ishikawa and M. Kumada, Adv. Organomet. Chem., 1981, 19, 51. l2 (a) H. Shizuka, H. Obuchi, M. Ishikawa and M. Kumada, J. Chem. Soc., Chem. Commun., 1981,405; (b) H. Shizuka, Y. Sato, M. Ishikawa and M. Kumda, J. Chem. SOC., Chem. Commun., 1982,439. l 3 M. Ishikawa, H.Sugisawa, T. Fuchikami, M. Kumada, T. Yamabe, H. Kawakami, K. Fukui, Y. Ueki and H. Shizuka, J. Am. Chem. Soc., 1982, 104, 2872. l4 R. West, J . Organomet. Chem., 1965, 3, 314. l5 (a) H. Sakurai, H. Yamamori and M. Kumada, Chem. Comrnun., 1968, 198; (b) H. Sakurai and M. Kumda, Bull. Chem. SOC. Jpn, 1964,37,1894; (c) H. Sakurai, K. Tominaga and M. Kumada, Bull. Chem. SOC. Jpn, 1966,39, 1279. * M. Kumada and K. Tamao, Adv. Organomet. Chem., 1966,6, 19. lo M. Ishikawa and M. Kumada, Reviews on Silicon, Germanium, Tin and Lead Compounh, ed.H. SHIZUKA et al. 357 l6 (a) C. G. Pitt, J. Am. Chem. SOC., 1969,91, 6613; (b) C. G. Pitt, M. M. Bursey and P. F. Rogerson, J. Am. Chem. SOC., 1970,92, 519. l7 J. Nagy and J. Reffy, J. Organomet. Chem., 1970, 22, 565.H. Bock and H. Alt, J. Am. Chem. SOC., 1970,92, 1569. l9 E.g. H . Schmidbaur, J. Am, Chem. SOC., 1963, 85, 2336. 20 E.g. R. D. Cowell, G. Urry and S. J. Weissama, J. Am. Chem. SOC., 1963, 85, 822. 21 H. Soffer and T. De Vries, J. Am. Chem. SOC., 1951, 73, 5817. 22 V. Vaisarova and V. Chvalovsky, Collect. Czech. Chem. Commun., 1968, 33, 859. 23 H. Shizuka, H. Obuchi, M. Ishikawa and M. Kumada, to be published. 24 M. Ishikawa, M. Oda, N. Miyoshi, N. Fabry, M. Kumada, T. Yamabe, K. Aka$ and K. Fukui, J. 25 H. Gilman and C. D. Lichtenwalter, J. Am. Chem. SOC., 1958, 80, 608. 26 M. Ishikawa, S. Katayama and M. Kumada, to be published. 27 M. Ishikawa, K. Nakagawa and M. Kumada, J. Organomet. Chem., 1979, 178, 105. 28 F. C. Whitmore, L. H. Sommer, P. A. Digiorgio, W.A. Strong, R. E. Wan Strien, D. L. Bailey, 2B R. A. Benkeser, W. Schroeder and 0. H. Thomas, J. Am. Chem. SOC., 1958,80,2283. 30 A. Weissberger, E. S. Proskauer, J. A. Riddik and E. E. Troops Jr, Organic Solvents (Interscience, 31 W. H. Melhish, J. Phys. Chem., 1961, 65, 229. 32 J. N. Damas and G. A. Crosby, J. Phys. Chem., 1971, 75, 991. 33 H. Shizuka, Y. Ueki, T. Iizuka and N. Kanamaru, J. Phys. Chem., 1982,86, 3327. 34 E.g. K. Tsutsumi, S. Sekiguchi and H. Shizuka, J. Chem. SOC., Faraday Trans. I , 1982,78, 1087. 35 J. B. Birks, Photophysics of Aromatic Molecules (Wiley-Interscience, London, 1970). 36 H. Shizuka, H. Hiratsuka and M. Ishikawa, to be published. 37 (a) S. Suzuki and H. Baba, Bull. Chem. SOC. Jpn, 1967,40,2199; (6) H . Suzuki, T. Fujii and K. Sato, 38 N. Mataga, Bull. Chem. SOC. Jpn, 1963, 36, 654. 3g K. Tsutsumi and H. Shizuka, Z. Phys. Chem. (Wiesbaden), 1978,111, 129. 40 (a) S. Tobita and H. Shizuka, Chem. Phys. Lett., 1980,75, 140; (b) H. Shizuka and S. Tobita, J. Am. 41 E. Lippert, 2. Naturforsh., Teil A, 1955, 10, 541; 2. Elektrochem., 1957, 61, 962. 42 N. Mataga, Y. Kaifu and M. Koisumi, Bull. Chem. SOC. Jpn, 1955, 28, 690; 1956, 29, 465. 43 B. T. Lim, S. Okajima, A. K. Chandra and E. C. Lim, Chem. Phys. Lett., 1981,79,22. 44 N. Mataga and M. Ottolenghi, Molecular Association, ed. R. Foster (Academic Press, London, 1979), 45 R. S. Davidson, Molecular Association, ed. R. Foster (Academic Press, London, 1975), vol. 1, p. 215. 46 F. C. de Schryver, N. Boens and J. Put, A&. Photochem., 1977,10, 359. 47 W. Klopffer, Organic Molecular Photophysics, ed. J . B. Birks (Wiley-Interscience, London, 1973), 48 P. Froehlich and E. L. Wehry, Modern Fluorescence Spectroscopy, ed. E. L. Wehry (Plenum Press, 49 Z. R. Grabowski, K. Rotkiewicz, A. Siemiarczuk, D. J. Cowley and W. Baumann, Nouu. J. Chim., 50 K. Rotkiewicz, K. H. Grellmann and Z. R. Grabowski, Chem. Phys. Lett., 1973, 19, 315. 51 K. Rotkiewicz, Z. R. Grabowski, A. Krowczynski and U. Kuhle, J. Lumin., 1976, 12/13, 877. 52 Z. R. Grabowski, K. Rotkiewitz and A. Sierniarczuk, J. Lumin., 1979, 18/19, 420. 53 E. M. Kosower, Ace. Chem. Res., 1982, 15, 259. 54 D. Huppert, H. Kanety and E. M. Kosower, Chem. Phys. Lett., 1981, 84,48. 55 Y. Wang, M. McAuliffe, F. Novak and K. B. Eisenthal, J. Phys. Chem., 1981,85, 3736. 56 D. Huppert, S. D. Rand, P. M. Rentzepis, P. F. Barbara, W. S. Struve and Z. R. Grabowski, J. 57 W. Rettig, J. Phys. Chem., 1982, 86, 1970. 58 S. Nagakura and J. Tanaka, J. Chem. Phys., 1954,22, 236. 59 R. Rettig, J. Lumin., 1980, 26, 21. Am. Chem. SOC., 1979, 101, 4612. H. K. Hall, E. W. Pietrasza and G. T. Kerr, J. Am. Chem. SOC., 1946,68, 475. New York, 1955). Bull. Chem. SOC. Jpn, 1972, 45, 1937. Chem. SOC., 1982, 104, 6919. vol. 2, p. 1. vol. 1, p. 357. New York, 1976), vol. 2, p. 319. 1979, 3, 443. Chem. Phys., 1981,75, 5714. (PAPER 3/510)

ISSN:0300-9599    

DOI:10.1039/F19848000341

出版商:RSC    

年代:1984

数据来源: RSC

摘要:

J. Chem. Soc., Faraday Trans. 1, 1984,80, 359-382 Canonical Chemical Theories Exemplified by the Methylolation of Urea and Melamine BY TADDESSE GEBREGIORGIS AND MANFRED GORDON* Department of Chemistry, University of Essex, Wivenhoe Park, Colchester C04 3SQ Received 14th April, 1983 Old and recent data on equilibria and kinetics of methylolation of urea and melamine are analysed. The theoretical framework affords a sequence, claimed to be canonical, of successive phenomenological approximations. The theory, expounded earlier, begins with two successive schemes whose essentials go back to early work by Pauling and by Flory. The two stages here suffice to show that all the data from eight laboratories, covering the six-membered family of methylol ureas and the ten-membered family of methylol melamines, are in quantitative agreement, a circumstance unsuspected by other workers in the field.The statistical- thermodynamic and kinetic principles involved are therefore expounded in more detail than before. Using them, very small substituent effects (e.g. 0.35 kJ mol-l) are deduced with high significance and accuracy. Owing largely to the recent measurements by Tomita, the methylol melamines are claimed currently to be the thermodynamically best characterised family in the chemical literature. A challenge arises for theoretical chemists to test quantum-theoretical techniques against the measured energetics of substituent effects in this family, and to exploit for other families the graph-theoretical analysis of molecular additivity which underlies the approximation schemes, which are already well tested.1. INTRODUCTION A large part of chemical theory, both fundamental and technical, rests on phenomenological data extracted from comparative experimental studies of classes of related compounds, e.g. homologous series, related oligomers or wider families like the alkanes. This work is addressed to experts in two fields which remain far apart despite appeals' to bring them together. On the one hand, polymer scientists interested in polycondensation mechanisms may welcome a demonstration that the application of old293 theories shows a wide spectrum of old and recent data on the families of polymethylol urea (fig. 1) and melamine (fig. 2) to be in apparently unsuspected quantitative agreement. On the other hand, the neglect by theoretical chemists of the unifying graph-theoretical canonical framework (cf.Kennedy et aZ.*) which allows chemical measurements on families of compounds to be analysed so as to yield the requisite fundamental parameters has been depriving them of the most systematic phenomenological framework for their labours. But for such neglect, theoreticians rather than technologists might well have discovered and explored the methylol melamines, which we claim below to constitute the thermodynamically best charac- terised family in the chemical literature. The structure of the paper is now summarised. In section 2 we explain once again the m e t h ~ d ~ - ~ of analysing the thermodynamic and kinetic measurements obtained during the last 35 years, mainly for technological interest by technologists, on the methylolation of urea and melamine, and in sections 3 and 4 we apply the method to the data.The discussion (section 5) shows that the recent experiments by Tomita@ allow for the first time the correction of non-ideal behaviour of the dilute aqueous 359360 CANONICAL CHEMICAL THEORIES ,NH CH,OH < ' 3 , 4 > ,NH CH,OH 2 /NHz%> o=c 7 o=c, 7 o=c 'NH, Km NHZ K , 0 \NH CH,OH r 001 1101 t 201 'symmetrical 1 o 2 1 'u n symmetrical' [I23 N (CH,OH) / o= c \N(CH,OH l 2 (041 Fig. 1. Reaction scheme for all possible methylolations and demethylolations of urea. The symbols [iJl are explained in the text below eqn (13). The equilibrium constants are labelled in the notations of Tomita and HiroseI6 ((K,, ,)) and Slonim et u1.12 [K,,].e t c . Fig. 2. Scheme of all possible methylolations and demethylolations starting with melamine (top left) and ending with the formation of hexamethylol melamine (bottom right).6 The symbols [iJl are explained in the text below eqn (1 3). The equilibrium constants are labelled in Tomita's n ~ t a t i o n . ~T. GEBREGIORGIS AND M. GORDON 36 1 systems to be applied to the calculated thermodynamic parameters. With this small but accurate correction, the historical and recent data are harmonised with remarkable accuracy by the analysis in terms of the linear first-shell substitution effect (FSSE), essentially the effect of neighbouring (' first-shell ') substituents on bonding equilibria. Turning to the wider theoretical framework, we recall in section 6 that this FSSE (invented by Pauling2) constitutes the first correction to be applied, within the systematic framework, to the classical bond-additivity approximation. Several successful applications are also listed, and the mathematically convergent canonical sequence of refinements beyond the FSSE is reviewed, with special reference to the gaseous alkane family.2. METHYLOLATION REACTIONS OF MELAMINE AND UREA: THEORETICAL ANALYSIS The comparison, on a unified and fundamental basis, of the results of the various sets of authors on urea and melamine formaldehyde addition reactions is now introduced by an elementary rederivation of the equations needed. The general statistical theromodynamics (Gordon and Scant1ebury;'O Gordon and Temple'l) is reduced to simple (but still evidently not obvious) terms in our case of methylolation reactions in which formaldehyde is merely a monofunctional reagent.The main items considered in the derivation of the two-parameter family of equilibrium distributions of methylol ureas or melamines, which successfully fit experimental results, are these: (1) The relation between statistical factors (in a given pair of forward and backward reactions) and the symmetry numbers of the reactants and products. (2) The stoichiometric relation between the two sets of equilibrium constants used in the previous literature. (3) The relation (cf. section 5.2.2) between the substitution effects upon equilibria with those upon kinetics of the forward and backward reactions concerned.Aldersley et aZ.' used the simple assumption of splitting the standard free-energy increments caused by the equilibrium substitution effect equally between the activation energies of the forward and backward reactions, i.e. increasing the forward activation free energy and decreasing the backward activation free energy by the same amount. This should be a good approximation to the truth, especially for the moderate effects encountered here. (4) The generalisation of the random Flory- Stockmayer (or bond additivity) reaction scheme by introduction of substitution effects, i.e. effects on the free energy of the interactions between several methylols in a given compound. This would require introduction of a single adjustable substitution- effect parameter; however, the data on methylolation of urea show that a single local substitution effect, restricted to interactions between two methylols borne by a single amino group, is insufficient.A second substitution effect involving methylols borne on the two distinct amino groups of a urea molecule is clearly discernible by the proper analysis. The linear approximation to the substitution effect is all the data can support. The accountancy is simplified by postulating a constant local substitution effect (within a given dimethylolamino group) and a linear general substitution effect involving any pair of methylols, whether borne on the same or on different nitrogens of a urea or melamine molecule. AG, denotes the free energy change due to the interaction of any pair of methylols under the general substitution effect and AG, that of any pair subject to the local effect.The corresponding factors in the partition functions, which furnish the statistical weights due to substitution effects in the equilibrium distribution, are (1) denoted by x2 = exp ( - AGJRT) y = exp (-AG,/RT). (2)362 CANONICAL CHEMICAL THEORIES The squaring of the parameter x was introduced historically to simplify some of the formulae and is otherwise trivial because x2 is treated as adjustable by optimisation (see below). The squaring was not extended to y by the early authors5p and we leave their definitions undisturbed. Consider the reactions RFk-,+F=RFk (k = 1, 2, ..., f) (3) with the equilibrium constants (for ideal solutions - activities are introduced later) : Here R is a urea or a melamine molecule (see fig.1 and 2), F is a formaldehyde molecule and f is the functionality and is equal to four for a urea molecule and six for a melamine molecule. Square brackets denote mole fractions. We treat first the simplest case and later the effects of various complications. If the unit R is totally symmetric (the symmetry group of the monomer R being S,, the symmetric group of orderf!), as is implied in Flory’sf-functional random condensation model, and if we assume also, as does Flory’s scheme, total randomness of the reaction, i.e. the bond additivity scheme applied with all bonds formed having equal bond free energies, then the normalised equilibrium distribution will be the binomial one :, where a is the average equilibrium degree of substitution k.If the symmetry is reduced, as in melamine (DSh) or urea (C,,), then the mole fraction [RF,] (k = 0, 1 , 2, . . . , f) may split for some k into a sum of mole fractions of structural is~mers.~ The proper statistical-mechanical generalisation of eqn ( 5 ) is [RF,,] = .A’-t~i; ak( 1 - ay-” (6) where [RFkp] is the concentration of thepth isomer with exactly k substituents F, and a,,isitssymmetrynumber; asalways,Nisthenormaliser. Forexample,N = 8forurea and 48 for melamine; and generally, by comparing eqn ( 5 ) and (6), for all k N2 t~;;=(g ( k = O , l , ..., f). P-1 (7) The fact that a fixed N satisfies ally+ 1 equations is a very special case of a geometrical theory fundamental for the ‘graph-like state’ developed by Gordon and Temp1e.l’ This theory applies as long as the bonds formed (here between R and F) are freely rotating.The appearance of pairs of isomers for RF, in urea formaldehyde, and for RF, to RF, in melamine formaldehyde, motivates the completely general reformulation of eqn (3) and (4): ERFk-11, + F [RFkI, (8) Kk, mn = [RFkln/[RFk-llm lF1 (9) where m and n are labels (m, n = 1, 2) for two specific isomers contemplated. Using eqn (6) we find (10) The ratios of the symmetry numbers of the nth isomer of RF, to the mth isomer of RFk-l, which forms the first factor on the right, may be divided into the phenomeno- Kk, mn = (Ok, n/Ok-1, m) (a/[F1(lT. GEBREGIORGIS AND M. GORDON 363 logical equilibrum constant &. mn to produce the fundamental equilibrium constant k , mn = Kk, mn(Ok-1, m/Ok, n ) .This was the nomenclature proposed by Gordon and Temple,ll because the removal of the symmetry factors on the right of eqn (10) reduces the phenomenological constant to one dealing with the actual free energy of bond formation. Slonim et aP2 have called the fundamental equilibrium constant Kequiv, meaning equivalent equilibrium constant. Besides, rather than by the ratio of symmetry numbers, they divided Kk, mn by the ratio ( q / r ) of the statistical factors in the backward and forward reactions of eqn (8), i.e. q is the number of ways of removing an F from [RF,], so as to produce [RFk_,],+F and r the number of ways in which F can be added to [RFk-Jm to produce [RF,], in the notation of Slonim et aZ. The fact that in all possible reactions, involving arbitrary numbers of products and reactants, the ratio of the symmetry numbers agrees with the ratio of statistical factors is a condition for the compatibility of kinetic, thermodynamic and statistical-mechanical theories of chemical reactions. This has been proved by elementary graph and group-theoretical arguments.ll The reader may convince himself that it is true, although not too obvious, by taking examples from fig.2. E.g. the trimethylol melamine [ 121 may add F to form the tetramethylol melamine [04] with q = 4, r = 1 ; the symmetry numbers are 16 and 4, also standing in the ratio 4: 1 . For the present study this equality of ratios may seem to have little significance except as a checking procedure. For all equilibria the use of symmetry numbers is preferred, because it leads simply to the calculation of the mole-fraction distribution [see eqn (13)].Strong arguments, apart from those based on economy, favour the use of a canonical set of equilibria, just sufficient to determine all relevant equilibria, such as those exemplified by the Kk,mn above. The canonical set is one analogous to the formation of compounds from the elements, long favoured in compilations of standard thermodynamic data. Thus we for the forhation of the pth isomer (p = 1, 2) of k-methylol melamine or k-methylol urea from melamine or urea (both denoted by R) and k formaldehyde molecules. Building the compound up in one step from its primary components, rather than by adding one formaldehyde at a time, as in the set of reactions of eqn (8) means that k - 1-1 where the product goes over a suitable sequence of intermediates.For instance, hexamethylol melamine can be synthesised by sequences such as (see fig. 2) [OO] -+ [ 101 -+ [02] + [ 121 --* [04] [ 141 -+ [06], [OO] + [lo] 4 [20] + [30] + [22] + [14] + [06] etc. Although the literature frequently presents results expressed in terms of non- canonical constants of type Kk,mn [eqn (9)] as in the work of Tomita9 and Slonim et aZ.,12 the Kcanon [eqn (1 l)] are to be preferred. This is exemplified by the extraction of much more information by Gordon and ScantleburylO using Kcanon from data on the system POCl, + P20, by Groenweghe et aZ.,13 who were content with non-canonical equilibria of the type given by eqn (9). The following argument for using sets of reactions and constants seems decisive even for such simple situations as presented by methylol ureas, where the advantage in terms of extracting the information from364 CANONICAL CHEMICAL THEORIES experimental data will be illustrated below.The argument equally applies to theories of equilibrium constants for the complexing of metals, as pointed out by a referee [cf. later ref. (20)]. The aim of the thermodynamic theory is to aid in building statistical-thermodynamic models which must contain suitable parameters. The parameters generally relate to distributions of compounds, e.g. alkane^,^ P,O,-type olig~mers,~ methylol ureas or melamines etc. These distributions are themselves the key to physical properties of polycondensates, e.g. their n.m.r.spectra,lO enthalpies, viscosities, gel points where appropriate, optical14 and elastic properties15 etc. The canonical set has the advantage that every such equilibrium refers to a single member of the distribution (and to the monomer or basic compound like urea), while the non-canonical set involves ratios of two oligomers etc. Evidently, the way to the distribution and its parameters from the canonical set is therefore easier, and distributions and their parameters (as distinct from mere tables of equilibrium constants) have been derived by this route. In particular, the fitting of spectral intensities to theoretical model distributions (see below, sections 3 and 4) or of theoretical molecular additivity ~ c h e m e s ~ v ~ ~ is based on linear combinations of concentrations of individual numbers of the distribution. The non-canonical equilibria (4) involve non-linear combinations, viz.ratios of concentrations; thus it is fortunate that they can be transformed to achieve linearity by the mapping into the canonical set in eqn (12), which must be used implicitly if not (as is preferred) explicitly. We derive the distribution given by Gordon et aL5 for methylol melamine equilibria (f= 6) and equally applicable to methylol ureas (f= 4): (13) (For metal-ligand equilibria, j should normally be set equal to zero.) Here nij is the equilibrium mole fraction of a compound, denoted (fig. 1 and 2) by [b], bearing i methylols on secondary amino groups and j on tertiary amino groups. Thus j is restricted to even values, 0, 2, 4 for urea and 0, 2, 4, 6 for melamine.Eqn (13) generalises eqn (5) from the random reaction model to one including the general linear substitution-effect parameter, x, and a local substitution-effect parameter, y , which governs the difference in reactivity between a primary and a secondary amino group towards formaldehyde [eqn (1) and (2)]. Taking first the simple local effect embodied in PI2, this scales the concentrations by a factor y for each of the j / 2 tertiary amino groups (which bear two methylols each) found in the molecule. The necessary renormalisation to unity is in any case contained in the factor A’”. The introduction of the factor yj’2 entails that the parameter called a in eqn (5) loses its meaning as the equilibrium fractional conversion of hydrogen sites to methylols, and this change is signalled by changing the notation in eqn (1 3) from a to y.Turning to the general linear substitution effect introduced through the factor d i + j ) * , it must be conceded that it does not bear the outward appearance of linearity. Its linear nature is apparent in relation to the non-canonical set of equilibria; the squaring of the exponent is the result of the transformation to the canonical set by summation in the logarithmic form of eqn (12). nij = Jaij-l x(i+j)2yj12yi+j( 1 - y)f-(i+i). For the purpose of the derivation of the factor di+j)‘ in eqn (1 3) we may set We may write Introducing these two factors on the right in place of xkZ into eqn (13), we see that the term xk is really trivial, because it can be ‘absorbed’ into a change in the parameterT.GEBREGIORGIS AND M. GORDON 365 y, which in any case has lost its physical significance as the degree of conversion. Following Gordon and Scantlebury3 we use the algebraic transformation xkyk( 1 - 7)e-k *y*k( 1 - y*)6-” (16) N* = [l +(x- 1)yy (17) and y* = yx/[l +(x- l>y]. (18) where N * is a constant, in turn to be absorbed in N : The physical significance of thus disposing of a factor xk is a shift in the arbitrary zero level of the energy of a compound containing k equal substituents. The physical meaning of the residual term (x2)k(k-1)/2 remaining on the right of eqn (1 5 ) [after our transforming the factor xk away by absorption in the other two parameters of the distribution (13)] is simple: it represents the introduction of a factor x2 for each of the distinct pairs, k(k- 1)/2 in number, of methylol groups in k-methylol melamine into the partition function.Thus a thermodynamic substitution effect of each such group on each of the others is implemented in what is, after all, a linear manner (linear in the number of pairs of substitutents which interact with each other to produce the general substitution effect). The meaning of the manoeuvre expressed by eqn (15)-(18) may also be made physically plausible. It shows that in view of the definitions (17) and (18) the distribution (1 3) is identical to the following distribution : (19) where the exponents of x2 and of y are the number of pairs of methylols interacting, respectively, in the general and local effects. The method adopted for determining optimal (and therefore approximate) values of x2 and y to fit a set of data yields the same results whether we use the simpler form, eqn (13), or instead eqn (19), notwithstanding the dependence of P and y* on x through eqn (17) and (18).Thus eqn (13) is a normalised family of distributions depending on x,y and y, and eqn (19) is the same family [using eqn (17) and (18)] parametrised by x, y and y*. The optimisation is conducted under two constraints: the trivial normalisation constraint and the constraint of the mean degree of substitution P to its measured equilibrium value P, E n i j ( i + j ) = P,. This equation allows y or y* to be eliminated, so that the family of distributions (13) or (19) is reduced to a 2-parameter sub-set, compatible with P,, depending only on x and y , and common to eqn (1 3) and (19).Since generally there will be only one member of this sub-set which optimises the fit to a given set of data, the optimum values of x and y are now independent of whether we optimise the fit of eqn (13) or (19), and eqn (1 7) and (1 8) have become irrelevant. [Note that in eqn (19) only the product NN* is determined by the optimisation, so N* has no separate significance.] n o . = NN*(x2)$(i+j)(i+j-l) j / 2 *i+j * 6-i-j 29 Y 7 (1-Y) En, = 1 (20) (21) 3. SUBSTITUTION EFFECTS ON METHYLOLATION OF UREA: COMPARISON OF THREE INVESTIGATIONS In this section we summarise the nature of the data obtained by de Jong, de Jonge and Eden and some others around 1950, by Tomita and Hirose16 (1 976) and by Slonim et (1978), and the reduction of the various data by the calculation of two 13 FAR 1366 CANONICAL CHEMICAL THEORIES Table 1.Summary of substitution-effect parameters calculated from published measurements on methylolation of urea and melamine ~~~ source and nature of data X Y 1 (a) methylolation of urea ref. (8) kinetics of dissappearance of formaldehyde6-8 assuming equal-splitting rule, pH 9.5, T = 45 OC 0.17 0.173 ref. (16) lH n.m.r. : equilibrium constants ref. (12) 13C n.m.r. : equilibrium constants, pH 9.8, T = 60 "C pH 7-7.7, T = 30 "C 1 (b) methylolation of melamine ref. ( 5 ) equilibrium constants, 45 OC ref. (8) equilibrium and kinetics, pH 9.5, T = 45 "C paper chromatography; 14C tracer as originally calculated, assuming solution ideality (fig.5 ) as corrected here, for non-ideality ref. (9) equilibrium constants, pH 9; 28, 38 and 48 OC high-speed liquid chromatography (+ lH n.m.r. for assignments) isomer ratios overall optimisation over 18 sets of six [F]/[M] ratios 0.805 f 0.04 0.250 f 0.06 0.837 f 0.04 0.199 & 0.05 1.125 1.01 1 1.07 0.5 0.6 1 f 0.03 0.28 - 0.269 & 0.032 1 .O (fixed) 0.327 & 0.046 1.07 0.280 & 0.01 8 substitution parameters x and y. The results of these calculations are compared in table l ( a ) and section 5. 3.1. DATA OF DE JONG et d.,'7a CROWE AND LYNCH17b AND LANDQUISTL7' These data were collated by Aldersley and Gordon,* who calculated the substitution parameters in table 1. The data included rate constants of forward and backward methylolation reactions, and hence equilibrium constants for four reactions, condensed into three equations (because there are two isomers of UF,, see fig.1): k, kl k2 U + F s U F U F + F e UF, k3 ksl UF, + F f UF,.T. GEBREGIORGIS AND M. GORDON 367 These had been measured in buffer solutions at pH 7 and 35 O C . The substitution effects on kinetic constants were taken to be related to those acting on equilibria by the assumption of ‘equal splitting’ [see point (3) in section (2)] between forward and backward rate processes, leading for instance to the appearance of a term in y112 in the forward rate constant k, [eqn (25)] and a term in y-l12 in the backward rate constant k3$ [eqn (28)], the complete set of four equations being k, = k,[ 1 /2 + b1/,/4)] x (25) k,. = 2k1f x-, (26) k, = k, y1j2x2/2 (27) k,.= kl!( 1 + 2~-’/,) x-,. (28) The numerical factors (1/2, 1/4, 2) in various terms represent the reciprocal statistical factors (see above). For instance, eqn (25) signifies that the overall measured forward reaction of eqn (23) consists of one term k,x/2 and one term k,y1/,x/4; the first term postulates that the rate constant k, for the first step of the U/F reaction is modified, when proceeding to the formation of [20]-isomer of UF, (symmetrical dimethylol urea, fig. 1) in a second step by a statistical factor 1/2 and by the factor x, which arises from the substitution effect in respect of formation of one pair of methylols in UF,. Note that by the equal-splitting rule, the factor x2 for each such pair in the corresponding equilibrium constant ( K ) is taken to split ‘equally’ into a factor x in the forward constant k, [eqn (25)] and a factor x-l in the reverse constant k,! [eqn (26)].The statistical factor 1/2 in the term k,x/2 of eqn (25) arises because two hydrogen sites are available to form the [2O]-isomer from UF, which is half the number available in the formation of UF by adding F to U [eqn (22)]. Clearly this statistical factor must be taken into account in comparing the phenomenological measured rate constants k, and k, in eqn (25). Similarly, the term klXy1/2/4 in eqn (25) predicts the rate constant for forming the [02]-isomer (two methylols attached to tertiary amino nitrogen) of UF, (fig. l), with y112 the kinetic substitution effect in respect of forming a tertiary amino group (for which the equilibrium substitution effect is, by the equal-splitting rule, y).The factor x is again the general substitution effect affecting any pair of methylols and 1/4 the statistical factor. Eqn (26) contains a single term, because the measured constant k,! refers to the loss of either of the two methylols (hence the statistical factor 2) from the pure [20]-isomer in solution, not from an equilibrium mixture of the two isomers. The two unknowns x and y were extracted by optimisation to fit the four equations [eqn (25)-(28)] by Aldersley and Gordon to the old data on the rate constants. They used a general and very simple graphical optimisation method reproduced in fig. 3. Plotting the two unknowns logarithmically as variables, the four equations give four plots that should intersect in a single point if the theory were exactly obeyed.Considering the expanded scales in fig. 3, this ideal is nearly fulfilled for the five rate constants whose data were most reliable [eqn (25)-(27)]. Even the plot for the less reliable k3< [eqn (2811 does not pass too far from the point marked by a circle, chosen to represent the optimised fit. This point furnished the x and y values in the top row of table l(a). 3.2. DATA OF TOMITA AND HI ROSE^^ These authors showed that the compounds [lo], [20], [02] and [12] (see fig. 1) could be quantitatively determined by ‘H (100 MHz) n.m.r. spectroscopy in D,O. In this way they were able to obtain rate curves and equilibrium concentrations. They also 13-2368 1.6 1 . 2 0.4 0 CANONICAL CHEMICAL THEORIES 0 0.1 0.2 0.3 -log,, x Fig.3. Intersection procedure for deriving the substitution parameters x and y : (1) eqn (25); (2) eqn (26); (3) eqn (27); (4) eqn (28). Graphic data reproduced from ref. (8). Table 2. Effects of methylol groups on the equilibrium constants in methylolation of urea16 reaction scheme m, nu Kphenom Kfund -AG/kcal mo1-I U+F"UF 00 26.50 6.625 1.251 UF + F e FUF 10 5.357 5.35 1.110 UFF + F -c FUFF 20 5 2.5 0.606 UF+F* UFF 01 0.5 1 0 FUF + F -c- FUFF 1 1 0.467 0.467 - 0.504 a m = 0, 1, 2; n = 0, 1 . found that at high F/U mole ratios and high overall concentration the cyclic urones are important among the products. These are not thought to be important side products under the conditions in which the data considered in this paper were obtained, as seems borne out by the fit to the theory which neglects their formation.Tomita and Hirose determined the five equilibrium constants listed in column 3 of table 2 at 60 "C and pH 9.80 from the n.m.r. signal intensities for urea and its methylol derivatives other than the elusive tetramethylol urea [04]. The experimental error was ca. $- 20 %. From their phenomenological (overall) constants, Kphenom, we calculated (column 4) the fundamental equilibrium constants Kfund of bond formation by dividing out the appropriate statistical factors. As already explained (section 2), the statistical equilibrium factors are obtained from the ratio of the symmetry numbers of the urea derivatives featuring as reactant and product, respectively, or from the ratio of kinetic statistical factors as done by Slonim et a1.l2 The optimisation of x and y to fit the equilibrium constants measured by Tomita and Hirose was carried out exactly by the procedure we describe for the measurement by Slonim et aZ.[see eqn (30)-(34) below] and found to be x = 0.805 & 0.04 and y = 0.250 f 0.06 [see table l(a)].T. GEBREGIORGIS AND M. GORDON 369 3.3. THE DATA OF SLONIM et aP2 These authors presented a thorough 13C n.m.r. investigation which allowed them to calculate the equilibrium constants of all the methylolation reactions from the band intensities obtained in solutions buffered at pH 7-7.7, kept for 2 months to reach equilibrium at ca. 30 OC. They list, in their table 3 (reproduced as table 3), no doubt for the sake of convenience, the six non-canonical equilibrium constants for adding one formaldehyde molecule in each possible way to urea and its mono-, di- and tri-methylol compounds. Of course only five of these constants can be independent, since there are only five canonical equilibrium constants, corresponding to the five methylol ureas (table 2, column 1).The redundancy arises from the relation (given in their notation, which differs from ours, cf tables 2 and 3) where the first subscript denotes the number of methylols carried by the reactant on the distant nitrogen and the second that on the same nitrogen to which the formaldehyde is added to form a new methylol group; the superscript equiv means that the statistical factors have been removed. Eqn (29) is exactly verified by their measured values of the constants because the measured band intensities, from which the constants are calculated cancel out.The corresponding relation between their listed free-energy changes for these four reactions is necessarily also verified, although a slight rounding error here occurs in their table (table 3, last column). Table 3. Effect of methylol groups on the equilibrium constants in methylolation of urea', Kmn [R,(R,)NCON(R,)H + HOCH,OH T= R,(R,)NCON(R,)CH,OH + H,O] - AG /kcal Rl R, R3 m, na Kmn p q u i v mn rnoP H H H 00 1400 350 3.5 CH,OH H H 10 300 300 3.4 CH,OH CH,OH H 20 400 200 3.2 H H CH,OH 01 30 60 2.5 2.2 40 16 1.7 CH,OH H CH,OH 1 1 40 CH,OH CH,OH CH,OH 21 4 am=O, 1,2;n=0, 1. The calculation of optimum values of the substitution-effect parameters x and y , and the test of the model defined through their appearance in the distribution of eqn (1 3), proceeds as follows.Five equations link the measured free-energy values in table 3 to x and y , which are calculated from the equilibrium constants by the thermo- dynamic relation AG = -RTln K . Omitting the superscript from AGO for simplicity370 CANONICAL CHEMICAL THEORIES (standard states are discussed in section 5.2) and using subscripts in the notation of Slonim et al. just explained we have (30) (31) AGO,-AGOO= 2RTlnx+RTlny=-l.O (32) AGll-AGoo = 4RTIn x+RTlny = - 1.3 (33) AGZ1-AGOO = 6RTInxfRTlny = -1.8. (34) AG,, - AGoo = 2RT In x = - 0.1 kcal mol-l AG,,-AGO, = 4RTln x = -0.3 These equations are immediately written down, because the presence in the reactant of each methylol group adds 2RTIn x, and each tertiary nitrogen formed in the product adds RT In y, to the free energy of the methylolation reaction that adds one formaldehyde to that reactant.This simple linear behaviour was alluded to in eqn (1 3). The optimum values of x and y are found by least-squares fittings of eqn (30)-(34) and are entered in table 1 (a) and examined in section 5.1. 4. SUBSTITUTION EFFECTS ON METHYLOLATION OF MELAMINE: COMPARISON OF THREE INVESTIGATIONS 4.1. GORDON et al. ON THE THERMODYNAMICS OF THE SUBSTITUTION EFFECT The first quantitative evaluation of substitution effects in the methylolation of melamine, a system that scores over methylolation of urea by higher functionality and consequently a larger body of chemical species (fig.2), was published by Gordon et aL5 in a symposium held in 1965. By means of paper chromatography, they were able to resolve the nine methylol melamines into seven separate spots. The assignments amounted to individual spots for the five compounds containing one to three methylols, one spot comprising the two tetramethylol melamines [22] and [04] and one spot shared by penta- and hexa-methylol melamine. By radiometry based on the incorporation of [14C]formaldehyde, the relative concentrations of the compounds could be evaluated from the spots on the paper and compared with prediction based on eqn (13). As a result of optimising the fit over the radiation intensities thus determined for 28 spots, seven for four equilibrium mixtures of formaldehyde/ mela- mine mole ratios, [F]/[M] = 1, 3,6 and 10, the values of x = 1.125 and y = 0.5 _+ 0.1 were deduced.While limits of accuracy for x were difficult to determine (the general substitution effect being so small), the difference from unity (absence of the effect) did appear to be significant. 4.2 KINETIC SUBSTITUTION EFFECT (GORDON et aL6 AND ALDERSLEY et aL7) Determination of the substitution effect on the kinetics of the methylolation of melamine was perfected soon afterwards in two reports. The firsts of these served to show that the kinetics could be fitted, but to a first approximation only, by assuming the classical random polycondensation devoid of substitution effects, and that high-precision measurements would be needed, as well as the necessary computer programs for integrating the differential rate equations, if reliable substitution-effect parameters were to be determined. A careful and detailed investigation of this type was carried out by Aldersley et a1.’ Despite their excellent consistency in various tests applied by these authors, their values of y lie far from the concordant values of more recent investigations; however, this discrepancy is fully explained in sections 5.2.1 and 5.2.2 below.T.GEBREGIORGIS AND M. GORDON 37 1 4.3. TOMITA’S INVESTIGATION’ In 1977 Tomita published the most detailed and comprehensive investigation yet of the methylolation of melamine. He was able to harness advances in high-speed liquid chromatography (h.1.c.) which allowed all nine methylol melamines to be detected for the first time, preparatively separated on a small scale, identified by lH n.m.r.spectroscopy and quantitatively determined from h.1.c. peak areas. Unlike earlier w o r k e r ~ , ~ ~ ~ * he was able to take account of the amount of demethylolation occurring among higher methylol melamines during the process of analysis, although the effect was ‘almost negligible at short analysis times’. The whole distribution of products was measured intermittently up to equilibrium at seven feed ratios in the range 1 < [F]/[M] < 30, three temperatures (28, 38 and 48 “C), all at pH 9 and 1 /30 mol dm-3 melamine concentration. Tomita tabulated rate and equilibrium constants calculated directly from the concentrations in the reaction mixture, determined by h.1.c.’ of the species involved.The results were expressed as the twelve non-canonical equilibrium constants corresponding to the twelve equilibria seen in fig. 2. Of course, only nine of these can be independent. In fact, the three cycles of reactions as seen in fig. 2 allow us to write down one set of three linearly independent constraints which eliminate the redundancy : in Tomita’s notation. Eqn (35)-(37) are exactly verified by their measured values of the constants, because the constants are calculated from the same data, except for possible rounding errors. These errors here can exceed 5%; e.g. taking the constants on the left of eqn (35) from Tomita’s table 111, at [F]/[M] = 5 we find 16 x 1.0-1 x 39-’ x 2.3 = 0.944 instead of unity. Two ways of analysing Tomita’s excellent collection of data along the lines recommended in previous publications from this laboratory are now presented and compared, together with measures of the precision with which the substitution-effect parameters are determined.4.3.1 . THE ISOMER-RATIO TEST The isomer-ratio test is based on the fact (already exploited5 in 1965) that the equilibrium concentration ratio of two isomers is independent of the mean degree of substitution. Thus the local substitution-effect parameter can be determined with practically no mathematics and remarkable precision. We label 11, I, and I3 the ratios n02/n20, n12/n30 and n04/n,, of the concentrations of the unsymmetrical to the corresponding symmetrical isomers, in the accepted if confused nomenclature. (As will be relevant immediately, the ‘ unsymmetrical ’ isomer [02] has much higher symmetry than the ‘symmetrical’ one [20], and the same goes for [04] compared with [22], although the remaining case, [12] versus [30], is regular.) We see from eqn (13), and the fact that i+j is constant for a pair of isomers, that each of the three ratios Z, is simply given by I p = Y osym, ploasyrn, p where osym, and oasym, are precisely the symmetry numbers of the ‘symmetrical’ and ‘unsymmetrical’ members of the pth pairs of isomers just referred to.Since these372 CANONICAL CHEMICAL THEORIES numbers are easily determined (table 4), measurement of I p allows y to be calculated. Tomita's six runs at the highest [F]/[M] ratios between them give, at each of the three temperatures, ca.12 data pairs suitable for estimating y from eqn (38). The mean results for each isomer pair over the different available [F]/[M] ratios are listed, at each temperature, in table 4. The mean resulting y values for the three isomer pairs agree splendidly at each temperature, and the grand average of = 0.269 has a standard deviation of only 0.032. Table 4. Calculated local substitution effect ( y ) from ratio of isomersg for the mean degree of substitution per melamine (P,) in the range 0.45 < P, -= 5 recipro- cal sym- code (see fig. 2 number local substitution effect ( y ) assignment metry calculated from eqn (38) [iJ1 for notation) (oG1) 28 OC 38 "C 48 OC 02 20 12 30 04 22 average = grand average = 7 = 0.269 & 0.032 = 0.234k0.012; 0.294k0.017; 0.281 k0.026T.GEBREGIORGIS AND M. GORDON 373 4.3.2. OPTIMISATION BASED ON EQN (1 3) The result of the simple isomer ratios suggested that a complete data-fitting exercise would be useful. Accordingly, tables 5(a), (6) and ( c ) record the optimisation based on eqn (1 3) of x as well as y over all six equilibrium distributions at each of the three temperatures (28,38 and 48 O C , respectively). The theoretical concentrations of all 10 species, not merely those that have a pair of isomers, are reported, together with the experimental values of Tomita. For x = 1, i.e. assumed absence of the general substitution effect, the minimum standard deviations (ok, k = 1, 2, . . . , 18) of the experimental mole fractions from the theory were found by optimisation using eqn (13) over y (by varying y to give Pm) for each of the eighteen sets of equilibrium distributions.The standard deviations were calculated using ok = (21 - Z [nij(calc) - nij (expt)12 x (39) where n comprises the nine possible methylol melamines plus free melamine (n = 10). In the vicinity of the absolute minimum standard deviations, the eighteen optimum y values gave an average and standard deviation of 7 = 0.327 k 0.046. The eighteen standard deviations ok gave an average 5 = 0.017. When this p is compared with the isomer-ratio method reported (J = 0.269 k 0.032) we see that the agreement is within the experimental error inherent in the two standard deviations found. The previous work suggested that x is unity or just a little larger [see table 1 (b)], and to reduce from 0.327 to 0.269, i.e.to bring it closer to the direct determination from isomer ratios, clearly requires x to be increased above unity. In fact the minimum o was found for each of the six equilibrium distributions by trial- and-error optimisation of x, and y fixed at 0.327, varying x at intervals of 0.1. In four of the six cases the minimum occurred, at x = 1 . 1 and in the other two the minimum occurred at x = 1 . Accordingly, the optimum x value is estimated to be 1.07, in good agreement with the previous work. Again, with x fixed at 1.07 a parallel calculation gave = 0.280 k0.018 and 0 = 0.014. The large reduction in the standard deviations of p (from 0.046 to 0.018) and of t~ (from 0.017 to 0.014), as a result of changing x from 1.00 to 1.07, indicates that this small general substitution effect (AG, = -2RT In 1.07 = -0.349 kJ mol-l) is well supported by Tomita’s measure- ments.In any given equilibrium distribution the measured mole fractions cover a range which may exceed one decade. The calculated mole fraction distributions vary over a thousand-fold, but very small mole fractions are of course returned as zero by the chromatograph. 5. CONCLUSIONS ON METHYLOLATION OF UREA AND MELAMINE The main conclusion to be drawn is that the work of all the authors, starting 35 years ago with Crowe and Lynch,17b where it can reasonably be compared, is in quantitative agreement. 5.1. METHYLOLATION OF UREA The x and y values here computed [table l(a)] from recent IH and I3C n.m.r. equilibrium measurements on methylol ureas by Tomita and Hirose and by Slonim et a2.agree within one or two standard deviations (as here estimated) with the old chemical data listed below them, according to the analysis by Aldersley and Gordon.* Thus essentially the results of the recent work were predictable, since x and y completely determine the distributions involved [eqn (1 3)]. A glance at table 4 showsTable 5. Optimisation fitting of eqn (1 3) to Tomita’ss measured equilibrium distribution of methylol melamines (a) T = 28 O C i j 00 10 20 02 12 30 22 04 14 06 i j [F]/[M] = 3, P, = 1.49 [F]/[M] = 5, P, = 2.13 [F]/[M] = 7, P, = 2.71 theory theory theory x = 1.0 x = 1.07 y = 0.327 y = 0.280 experiment x = 1.07 x = 1.0 y = 0.327 y = 0.280 experiment x = 1.07 x = 1.0 y = 0.327 y = 0.287 experiment 0.151 0.373 0.306 0.025 0.041 0.084 0.017 1.38 x 10-3 1.13 x 10-3 2.53 x 10-5 0.175 0.354 0.288 0.020 0.040 0.094 0.024 1.71 x 10-3 2.46 x 10-3 1.02 x 10-4 [F]/[M] = 10, P, = 3.29 0.159 0.353 0.316 0.019 0.043 0.111 0 0 0 0 0.048 0.233 0.348 0.028 0.089 0.181 0.069 5.65 x 10-3 8.80 x 10-3 3.74 x 10-4 0.064 0.226 0.321 0.023 0.079 0.184 0.082 0.0 15 5.82 x 10-3 1.06 x 10-3 [F]/[M] = 15, P, = 3.89 0.05 1 0.223 0.343 0.020 0.076 0.196 0.090 0 0 0 theory theory x = 1.0 x = 1.07 y = 0.327 y = 0.280 experiment x = 1.0 y = 0.327 y = 0.280 experiment x = 1.07 w 4 P __ 0.013 0.021 0 0.1 18 0.140 0.103 12 0.277 0.265 0.278 0.023 0.019 0.0 0.122 0.103 0.098 3 0.248 0.240 0.264 c) 0.163 0.169 0.169 F B 0.013 0.0 12 0 0.036 0.048 0.05 1 0 3 1 c) 0 2.62 x 10-3 5.49 x 10-3 9 r [F]/[M] = 30, P, = 4.49 theory 0 x = 1.0 x = 1.07 y = 0.327 y = 0.280 experiment 00 10 20 02 12 30 22 04 14 06 2.62 x 10-3 0.036 0.163 0.0 13 0.122 0.248 0.277 0.026 0.103 0.013 5.46 x 10-3 0.048 0.169 0.012 0.103 0.240 0.265 0.019 0.1 18 0.02 1 0 0.055 0.174 0 0.098 0.262 0.258 0.013 0.120 0.021 3.47 x 10-4 8.33 x 10-3 5.44 x 10-3 0.067 0.087 0.178 0.348 0.028 0.228 0.049 9.86 x 10-4 5.62 x 10-3 0.014 0.079 0.077 0.181 0.322 0.023 0.23 1 0.067 0 0 0.088 0 0.072 0.2 18 0.316 0.0 17 0.228 0.062 2.84 x 10-5 1.24 x 10-3 1.47 x 10-3 0.018 0.043 0.087 0.310 0.025 0.368 0.145 1.11 x 10-4 2 .6 4 ~ 10-3 1.80 x 10-3 0.025 0.042 0.097 0.29 1 0.021 0.350 0.170 0 0 0.024 0 0.038 0.111 0.273 0.017 0.383 0.154(b) T = 38 "C [F]/[M] = 3, Pm = 1.32 [F]/[M] = 5, P, = 2.04 [F]/[M] = 7, P, = 2.50 theory theory theory x = 1.07 x = 1.0 x = 1.07 x = 1.0 X = 1.07 x = 1.0 i j y = 0.327 y = 0.280 experiment y = 0.327 y = 0.280 experiment y = 0.327 y = 0.327 experiment 00 10 20 02 12 30 22 04 14 06 0.197 0.402 0.274 0.022 0.030 0.062 0.010 8.49 x 10-4 5.78 x 10-4 1.07 x 10-5 0.222 0.379 0.260 0.0 19 0.03 1 0.072 0.015 i .i o x 10-3 1.34 x 10-3 4.68 x 10-5 0.176 0.396 0.275 0.022 0.036 0.074 0 0 0 0 0.057 0.244 0.350 0.029 0.082 0.168 0.059 4.83 x 10-3 6.93 x 10-3 2.71 x 10-4 0.075 0.245 0.323 0.023 0.073 0.172 0.071 0.0 12 5.04 x 10-3 7.89 x 10-4 0.046 0.269 0.331 0.024 0.078 0.178 0.074 0 0 0 [F]/[M] = 15, P, = 3.65 theory theory ij x = 1.00 x = 1.07 y = 0.327 y = 0.280 experiment x = 1.00 y = 0.327 y = 0.280 experiment x = 1.07 0.02 1 0.033 0 0.141 0.155 0.192 0.312 0.293 0.304 9 0.025 0.020 0.113 0.096 0.097 0 8 l! 5 theory 5 0.230 0.224 0.242 iz 0.125 0.133 0.126 2 0.022 0.032 0.029 5; 0.010 9.49 x 10-3 0.010 1.35 x 10-5 3.08 x 10-3 0 [F]/[M] = 30, P, = 4.43 z 8 x = 1.00 x = 1.07 y = 0.327 y = 0.280 experiment 00 10 20 02 12 30 22 04 14 06 5.63 x 10-3 0.058 0.212 0.017 0.126 0.257 0.229 0.019 0.068 6.75 x 10-3 0.0 10 0.073 0.21 1 0.01 5 0.106 0.248 0.226 0.016 0.083 0.012 0 0.088 0.217 0 0.108 0.260 0.2 16 0.0 15 0.082 0.014 8.2x 10-4 8.16 x 10-3 0.016 0.099 0.104 0.2 12 0.331 0.027 0.172 0.030 2.03 x 10-3 7.9 x 10-3 0.024 0.111 0.089 0.210 0.308 0.022 0.182 0.044 0 0 0.124 0 0.098 0.255 0.288 0.02 1 0.174 0.040 3.78 x 10-5 1.54 x 10-3 1.72 x 10-3 0.02 1 0.047 0.096 0.3 19 0.026 0.356 0.132 1-41 x 10-4 3.16 x 10-3 2.04 x 10-3 0.029 0.044 0.105 0.298 0.021 0.340 0.157 0 0 0.027 0 0.044 0.130 0.279 0.023 w 0.336 VI 4 0.161Table 5 (continued) (c) T = 48 "C w 4 o\ [F]/[M] = 3, P, = 1.28 theory theory theory x = 1.0 x = 1.07 x = 1.0 x = 1.07 x = 1.0 x = 1.07 [F]/[M] = 5, P, = 1.88 [F]/[M] = 7, P, = 2.33 ij y = 0.327 y = 0.280 experiment y = 0.327 y = 0.280 experiment y = 0.327 y = 0.280 experiment 00 10 20 02 12 30 22 04 14 06 0.210 0.408 0.265 0.022 0.028 0.057 9.12 x 10-3 7.46 x 10-4 4.84 x 10-4 8.56 x 0.233 0.383 0.253 0.018 0.029 0.068 0.014 9.96 x 10-4 1.17 x 10-3 3.93 x 10-5 0.236 0.360 0.282 0.0 17 0.034 0.072 0 0 0 0 0.077 0.284 0.348 0.028 0.070 0.142 0.043 3.45 x 10-3 4.28 x 10-3 1.43 x 10-4 0.098 0.279 0.321 0.023 0.064 0.149 0.054 3.84 x 10-3 7.82 x 10-3 4.59 x 10-4 0.089 0.275 0.330 0.03 1 0.065 0.156 0.055 0 0 0 [F]/[M] = 10, P, = 2.81 [F]/[M] = 15, P, = 3.39 -~ theory theory x = 1.0 x = 1.07 i j y = 0.327 y = 0.280 experiment x = 1.00 x = 1.07 y = 0.327 y = 0.280 experiment 0.03 1 0.045 0 0.177 0.186 0.253 5 0.333 0.309 0.3 16 0.027 0.022 0 0 3 0.103 0.089 0.098 F 0.218 0.209 0.097 0.108 0.096 5 0.015 0.023 0.0 18 7.63 x 10-4 1.94 x 10-3 0 cl F 3 E cl cl 0.208 0 E theory Ei 7.89 x 10-3 7.73 x 10-3 4 [F]/[M] = 30, P, = 4.23 x = 1.0 y = 0.327 y = 0.280 experiment x = 1.07 00 9.98 x 10-3 10 0.088 20 0.256 02 0.021 12 0.124 30 0.253 22 0.183 04 0.015 14 0.044 0.0 17 0.103 0.250 0.018 0.105 0.245 0.186 0.013 0.057 0 0.123 0.257 0 0.106 0.259 0.184 0.013 0.058 1.93 x 10-3 0.029 0.144 0.012 0.1 18 0.240 0.294 0.024 0.120 4.21 x 10-3 0.039 0.152 0.01 1 0.099 0.234 0.279 0.020 0.134 0 0.052 0.1 56 0 0.104 0.246 0.261 0.015 0.131 8.97 x 10-5 3.00 x 10-3 2.74 x 10-3 0.034 0.061 0.125 0.341 0.028 0.31 1 3.06 x 10-4 5.75 x 10-3 3.11 x 10-3 0.044 0.057 0.133 0.316 0.023 0.302 0 0 0.042 0 0.054 0.163 0.300 0.019 0.302T. GEBREGIORGIS AND M.GORDON 377 that any successful comparison between the two n.m.r. investigations would have to inject the symmetry numbers, or equivalently the statistical factors, as done in one of the original data sets12 but not in the other.9 5.2. METHYLOLATION OF MELAMINE 5.2.1 . EQUILIBRIUM MEASUREMENTS AND ACTIVITY EFFECTS Turning to melamine methylolation, there is again agreement within experimental error on the parameter x we extract from the data of three teams [table 1 (b)].However, the much cruder equilibrium measurements of 1965 show rather poor agreement of the parameter y with the highly accurate result of Tomita, and the agreement is worsened further if the subsequent thermodynamically and kinetically determined y value (0.61) is considered, even though Aldersley et al. estimated narrow error limits (kO.03) comparable with those now found in Tomita’s work [bottom of table 1 (b)]. These discrepancies are removed in section 5.3, where the older equilibrium data are harmonised with Tomita’s by corrections, calculated objectively from the latter work for the equilibrium between melamine and monomethylol melamine in terms of non-ideal solution behaviour : Indeed, Tomita’s data contain additional information on the activity term yFe yoo/ylo, which can be extracted especially accurately in terms of the model underlying eqn (1 3) based on linear substitution-effect parameters x and y .Thus while no, and n,, were measured (along with [F],, the equilibrium concentration of formaldehyde), at low values of P,, they became too small for measurement [tables 5(a)-(c)] at high Pm; however, by fitting the equilibrium data of the measurable mole fractions to eqn (1 3), with the success shown in these tables, smoothed values of no, and n,, were obtainable by calculation at all concentrations. They agreed well with measured values, where there were available, but are believed to be more accurate. Eqn (40) implies a standard state at which the activity coefficients all equal unity, which is normally chosen at infinite dilution in water.Extrapolation to infinite dilution is afforded by the data of Aldersley et al. and of Tomita only for the species formaldehyde, while at that limit the solvent water remains ‘contaminated’ with a small amount of melamine, uiz. 0.05 or 0.032 mol dm-3, respectively. This is because these authors report runs at these two initial melamine concentrations, while they varied the initial formaldehyde (or the ratios R 3 [F]/[M], initial formaldehyde to initial melamine). We therefore formally take as our solvent solutions of pure melamine in water at either of the two low concentrations (by extrapolation to R = 0). That a further extrapolation to infinite dilution of melamine would not measurably affect the calculated activity-coefficient term is suggested by fig.4, since the extrapo- lation of Kid to R = 0 (and hence [F], = 0) gives intercepts for Aldersley et al. and for Tomita which are clearly consistent despite their difference in the concentrations of melamine. The values of Kid plotted in fig. 4 were calculated from eqn (40), using the values of no, and n,, calculated [tables 5(a)-(c)] for x = 1, y = 0.327, and [r;le from the known initial [F] and measured value Pm, which Tomita showed to be in good agreement with his direct measurements of [F],. Since Ktherm is given by the intercepts of the ideal-solution equilibrium constants Kid at R = 0 in fig 4, the activity term Y[FIe yoo/ylo follows directly from eqn (40) and is plotted against R in fig.5. Here the squares represent the measurements by Aldersley et aL8 of [F], tested against theory in two different ways. The solid squares are based on the values x = 1, y = 0.61 which these authors found on the assumption, which they were forced to make at the time, that ideal-solution behaviour prevails. We see378 CANONICAL CHEMICAL THEORIES that this leads to the finding YIFle Y ~ ~ / Y ~ ~ = 1 with remarkable accuracy, thus apparently confirming the ideal-solution assumption. However, the open squares represent the same early data when Kid is calculated using the values which adequately fit Tomita's data, x = 1, y = 0.327. The data of Tomita, calculated with the same x and y, are shown in the other three curves with open symbols in fig. 5 for his three temperatures.The data of both laboratories are seen to be in good agreement by this searching test, while the fitting of two different hypotheses to the older data (closed and open squares) merely signifies that, within the margin of their accuracy, the change in y can be 'absorbed' by changes in Kid through non-ideal solution effects. Since Tomita's improved spectrum of measurements, extending to the concentrations of all the species present at equilibrium, have allowed us to determine x and y without the assumption of ideality, and afford independent determinations of Kid(R), the open-symbol plots determine the activity term unambiguously. 0.10 I 1 I I I I I I I 0 2 L 6 a R Fig. 4. Plot of ideal demethylolation equilibrium 'constant' against the initial ratio of formaldehyde to melamine concentration [eqn (40) and text]. The intercepts give the thermo- dynamic constants (Kid/mol dm-3) for Tomita's data: (a) Kid = 0.156 (T = 28 "C), (6) 0.1864 (38 "C) and (c) 0.233 ( T = 48 "C); and for the data of Aldersley and Gordon:* ( d ) 0.222 (T = 45 "C).Once non-ideality has thus been proved to affect the canonical equilibrium [eqn (40)] producing monomethylol melamine, the only equilibrium to contribute asymptotically as R 0, the question arises as to how the remaining canonical equilibria [eqn (1 l)] forming the eight higher methylol melamines (fig. 2) are affected by deviations from ideality. Here the consistency of theory and experiment, i.e. the prediction of concentrations of these eight concentrations at several Pm values from only two parameters x and y (table 5), indicates that, within the accuracy of measurement, the free-energy changes due to non-ideal-solution behaviour of a methylol melamine with r methylols varies linearly with r, a most reasonable finding.Since x is very close to unity (table l), such residual non-ideal-solution effects are likely to be exceedingly small. In other words, it seems very unlikely that there are substantial solution effects which almost exactly cancel the true general substitution effects on bond energies of methylol melamines in the gas phase. We see from table 1 that the values of the weaker general substitution parameterT. GEBREGIORGIS AND M. GORDON 379 x are significantly different for urea (0.8) and melamine (1.07), while the much stronger local parameter y does not differ significantly (cf values derived from Tomita’s work, 0.250 0.06 and 0.280 f 0.018, respectively, which are now in agreement with those of the other workers).This finding confirms exactly the theoretical expectations, because the general effect ranges over different, if distant, substituents, while the local effect in urea (--NH,). and melamine affects small but identical immediate neighbourhoods Q.81 I I 1 I 0 3 6 9 12 R Fig. 5. Plot of ratio yLFIe yoo/ylo of activity coefficients against initial ratio R of formaldehyde to melamine concentration. For data8* a see text. The temperature and pH ranges explored by different authors vary. Slight trends with temperatures from 28 to 48 OC in the values of y calculated here from Tomita’s data on [M]/[F] are perhaps just discernible in table 4, but not certainly significant.The changes over the pH range 7-9 in table 1 (a) for the urea methylols produce no measurable difference in the values of x and y we deduce from the data of Tomita and Hirosels and Slonim et aZ.12 This welcome robustness of the parameters to changes in conditions is no doubt due to the fact that x and y measure differences in free energies, or ratios of rate constants, due to quite small disturbances of structures or processes, whose fundamental characteristics are the same throughout the range of family members. 5.2.2. KINETIC MEASUREMENTS As regards the apparently quite adequate fitting by Aldersley et aZ. of their rate measurements to the parameters x and y, first determined by them from the much more reliable fittings of equilibrium data, we have performed computer checks on which we comment below in relation also to Tomita’s interpretation of his new (more extensive) measurements .The ten differential rate equations concerned in the fittings of three rate curves by computer integrations reduced the set of twenty-four rate constants to two, plus the two further parameters x and y (treated by the equal-splitting rule). Tomita wrote the same equations with their whole set of twenty-four rate constants (of which three are, however, redundant) and he did not factor out the appropriate statistical factors. He commented: ‘it is almost impossible to resolve these equations without finding a380 CANONICAL CHEMICAL THEORIES convenient method of involving a suitable plan of experiments which may permit the omission of some troublesome terms'.His attempt to extract individual rate constants by graphical devices, based on neglecting terms in the differential equations, is considered far too hazardous. The hazards are emphasised by the systematic procedure of injecting two basic rate constants (monomethylation equilibrium) plus 0, 1 , 2, . . . parameters in the canonical sequence: already with a total of four constants their rate measurements are now found to be at least as well fitted with y = 0.28 as with 0.61. In addition, we have also now noticed that the equal-splitting rule can be relaxed over a wide range of splitting proportions as a fifth parameter between the backward and forward free energies of activation, without substantially affecting the apparent fit.Kinetic substitution-effect parameters can evidently not be determined experimentally at present with an accuracy in any way comparable to that inherent in measurements of parameters govering equilibrium substitution effects. 6. THE SIGNIFICANCE FOR THEORETICAL CHEMISTRY Theoretical chemistry needs a sound framework of ' phenomenological ' theory. We briefly indicate here some parallel historical examples in which the same framework which underlies the present work has helped to reduce more or less extensive bodies of data to meaningful consistency. The framework itself is based on abstract analysis of additivity as applied at the level of molecular graphs. The classical notion of molecular additivity of properties within families of compounds needs no detailed exposition here; suffice it to say that pioneering studies are currently extending this notion as far as tensor properties in liquid crystals formed by suitable families.19 The first two steps in the early development of the graph-theoretical formulation (and the present work has required no more than these two levels) of additivity were taken by Pauling,20 oiz.his bond-additivity scheme and then the FSSE, which he invented for the specific family2 consisting of haemoglobin and its four successive adducts of oxygen. Bjerrum21 developed the FSSE refinement of bond additivity independently for various complex ammine families. Pentaerythritol and its four lauryl esters have (third-order) esterification rate constants whose ratios are related by the one parameter of the FSSE scheme within experimental error.22 The examples so far discussed in this paper had one component of structure which was monofunctional within the families contemplated (CH,O, 0,, NH, and lauric acid).Much greater richness and subtlety accrues if we turn to polyfunctional families with infinite sets of members. Here the classical Fl~ry-Stockmayer~~ model of random f-functional polycondensation corresponds to Pauling's additivity of enthalpies of bond for- mation (termed equireactivity by Flory). This model was called the basic paradigm of chemistry by Gordon and Temple." The reader may picture the alkane family as four-functional polycondensates of methane (with elimination of hydrogen).Thermo- dynamicists need not worry whether equilibria can actually be established experi- mentally : free-energy tables are sufficient. While the methylol melamine family has only ten members, the infinite alkane family has about sixty members for which good thermodynamic data are available. The parameterisation can significantly be continued for the alkanes up to the 610th canonical scheme in the hierar~hy.~ The need for more constants is explained because of the closer crowding of substituents, compared with those involved in the general substitution effect in urea and melamine, leading to non-linear and higher-shell effects. Thermodynamic data on the alkanes have been analysed by Cox and P i l ~ h e r ~ ~ and reanalysed by Kennedy et aZ.4a9b While we have seen that free-energy changes are directly measurable in the systems U/F and M/F by determining the concentrationsT. GEBREGIORGIS AND M.GORDON 38 1 of all species of the family together in one system at the equilibrium points under various conditions, the corresponding free-energy effects in the alkane series have to be calculated from separate measurements of enthalpies and entropies of formation of individual members of the alkane family. Table 2 in ref. (4b) shows that the experimental errors in the enthalpies of the ten simplest alkanes (excluding methane itself) average to ca. 0.48 kJ mol-l. The fitting of parameters.for x calculated from Tomita’s equilibrium measurements on methylol melamines is found above to be equivalent to a total change in free energy of 0.35 kJ mol-l; that this small value can be accepted here as significant follows not from independent knowledge of the experimental errors (which are not available), but from the relatively large reduction in the standard deviation (see above) of the fitting achieved by introducing x.This reduction can hardly be attributable to chance effects of experimental errors on the 180 data points involved. The experimental errors in Tomita’s measurements should, accordingly, be substantially less than the 0.48 kJ mol-l for the enthalpies of 10 simple gaseous alkanes, which would justify the claim that the M/F family is better characterised. The availability of accurate measurements, for melamine and urea families, of very small free-energy increments (2RT In x and RT In y ) is a challenge to theoreticians. Quantum calculations to ‘predict’ these results would be helpful for future research on melamine and urea resins.Sources of bias in the evaluation of thermodynamic data, which has greatly hampered the development of quantum-theoretical treatments of the alkanes, have been analysed before.4b A milder example of the elimination of bias by graph theory here arises in the melamine family. Graph theory tells us that three independent cycles in the reaction scheme of fig. 2 make three of the equilibrium constants redundant. If the redundancy is not removed, but all twelve equilibrium constants are thrown into the optimisation, then bias is introduced into the para- meters (apart from the waste of computer time).This is seen because the measured concentration [ 121 affects four of the constants while the measured concentration [OO] determines only KM, 1, and is thus given less weight in optimisation. Accordingly, the statistician also would advise the chemist to use canonical equilibrium constants [eqn (1 I)]. In the alkane family, bias was attributable to the total neglect of specific substitution effects, which were thought to be negligible but which were found to be appreciable when calculated (not optimised) from existing data by a graph-theoretical application of Moebius inversion.4b Returning to melamine and urea, most of the experimental work performed in this field has been motivated by the search for understanding of the technology of urea and melamine formaldehyde resins.l The parameters like x and y , even when they deviate little from unity, can have substantial effects on technical properties.The effects are especially marked on the gel point of resins, which is sensitive to the underlying molecular distributions, and indeed has frequently served to measure the deviation of substitution parameters from The second phase of resinification, i.e. the cross-linking which converts methylols to bridges between two melamine nuclei and the elimination of water, has been studied26 by suitably extending up to the gel point the theoretical framework here applied to the methylolation stage. Support from the DSM Company in the Netherlands is gratefully acknowledged and we thank Dr R. Cowell for help with programing computations of the kinetics of methylolation reactions.382 CANONICAL CHEMICAL THEORIES B. Mayer, Urea-formaldehyde Resins (Addison-Wesley, London, 1979). L. Pauling, Proc. Natl Acad. Sci. USA, 1935, 21, 186. M. Gordon and G. R. Scantlebury, Trans. Faraday SOC., 1964, 60, 604. (a) M. Gordon and J. W. Kennedy, J. Chem. Soc., Faraday Trans. 2,1973,69,484; (6) J. W. Kennedy, M. Gordon, J. Essam and P. Whittle, J. Chem. SOC., Faraday Trans. 2, 1977, 73, 1281. M. Gordon, A. Halliwell and T. Wilson, The Chemistry of Polymerisation Processes, S. C. I . Monograph no. 20 (Society of Chemical Industry, London, 1965), p. 187. M. Gordon, A. Halliwell and T. Wilson, J. Appl Polym. Sci., 1966, 10, 11 53. ' J. W. Aldersley, M. Gordon, A. Halliwell and T. Wilson, Polymer, 1968, 9, 345. J. W. Aldersley and M. Gordon, J. Polym. Sci., Part C, 1969, 16, 4567. B. Tomita, J. Polym. Sci., Polym. Chem. Ed., 1977, 15, 2347. lo M. Gordon and G. R. Scantlebury, Proc. R. SOC. London, Ser. A, 1966, 292, 380. l1 M. Gordon and W. B. Temple, The Graph-like State of Matter and Polymer Science, in Chemical l2 I. Ya. Slonim, S. G. Alekseyevan, Ya. G. Urman, B. M. Arshava and B. Ya Aksei'rod, Vysokomol. l3 L. C. D. Groenweghe, J. H. Payne and J. R. Van Wazer, J. Am. Chem. Soc., 1960,82, 5305. l4 W. Burchard, Adu. Polym. Sci., 1982, 48, 1. l5 G. Dobson and M. Gordon, J. Chem. Phys., 1965, 43, 705. l6 B. Tomita and Y. Hirose, J. Polym. Sci., Polym. Chem. Ed., 1976, 14, 387. l7 (a) J. I . de Jong and J. de Jonge, Rec. Trau. Chim., 1952,71,890; J. I . de Jong. J. de Jonge and H. A. K. Eden, Rec. Trau. Chim., 1953,72,88; (b) G. A. Crowe Jr and C. C. Lynch, J. Am. Chem. SOC., 1948, 70, 3795; 1949, 71, 373 (c) N. Langquist, Acta Chem. Scand., 1955, 9, 1127; 1459; N. Landquist, Acta Chem. Scand., 1956, 10, 244; 1957, 11, 776. K. Koada, J. Chem. SOC. Jpn, Pure Chem. Sect., 1954, 75, 571. Applications of Graph Theory, ed. A. T. Balaban, (Academic Press, New York, 1976). Soyedin, Ser. A, 1978, 20, 1477 (translated in Polym. Sci. USSR, 1978, 20, 1661). lB P. J. Flory, E. Salz, B. Erman, P. A. Irvine and J. P. Hammel, J. Phys. Chem., 1981, 85, 3215. 2o L. Pauling, The Nature of the Chemical Bond (Cornell University Press, Ithaca, 2nd edn, 1940). 21 J. Bjerrum, Metal Ammine Formation in Aqueous Solution (P. Haase and Son, Copenhagen, 1957). 22 M. Gordon and C. G. Leonis, J. Chem. SOC., Faraday Trans. I , 1975,71, 161; 178. 23 P. J. Flory, J. Am. Chem. Soc., 1941,63,3083; 3091 ; 3096; W. H. Stockmayer, J. Chem. Phys., 1943, 24 J. D. Cox and G. Pilcher, Thermochemistry of Organic and Organometallic Compounds (Academic 25 For a recent example see K. DuSek and M. Ilavsky, Colloid Polym. Sci., 1980, 258, 605. 26 T. Gebregiorgis, PhD Thesis (University of Essex, 1982) and to be published. 11, 45. Press, London, 1970), p. 552. (PAPER 3/597)

ISSN:0300-9599    

DOI:10.1039/F19848000359

出版商:RSC    

年代:1984

数据来源: RSC