摘要:

Chemical Society Reviews Editorial Board Professor H. W. Kroto FRS (Chairman) Professor M. J. Blandamer Dr. A. R. Butler Dr. E. C. Constable Professor B. T. Golding Professor M. Green Professor D. M. P. Mingos FRS Professor J. F. Stoddart Consulting Editors Dr. G. G. Balint-Kurti Professor S. A. Benner Dr. J. M. Brown Dr. J. Burgess Dr. N. Cape Professor A. Hamnett Dr. T. M. Herrington Professor R. Hillman Professor R. Keese Dr. T. H. Lilley Dr. H. Maskill Professor Dr. A. de Meijere Professor J. N. Miller Professor S. M. Roberts Professor B. H. Robinson Dr. A. J. Stace Staff Editors Mr. K. J. Wilkinson Dr. J. A. Rhodes Dr. M. Sugden University of Sussex University of Leicester University of St. Andrews University of Cambridge University of Newcastle upon Tyne University of Bath Imperial College London University of Birmingham University of Bristol Swiss Federal Institute of Technology, Zurich University of Oxford University of Leicester Institute of Terrestrial Ecology, Lothian University of Newcastle upon Tyne University of Reading University of Leicester University of Bern University of Sheff ield University of Newcastle upon Tyne University of Gottingen Loug h boroug h U n iversity of Technology University of Exeter University of East Anglia University of Sussex Royal Society of Chemistry, Cambridge Royal Society of Chemistry, Cambridge Royal Society of Chemistry, Cambridge It is intended that Chemical Society Reviews will have the broad appeal necessary for researchers to benefit from an awareness of advances in areas outside their own specialities.Deliberate efforts will be made to solicit authors and articles from Europe which present a truly international outlook on the major advances in a. wide range of chemical areas. It is hoped that it will be particularly stimulating and instructive for students planning a career in research. The articles will be succinct and authoritative overviews of timely topics in modern chemistry. In line with the above, review articles will not be overly comprehensive, detailed, or heavily referenced (ca.30 references), but should act as a springboard to further reading. In general, authors, who will be recognized experts in their fields, will be asked to place any of their own work in the wider context. Review articles must be short, around 6-8 journal pages in extent.In Consequence, manuscripts should not exceed 20-30 A4/American quarto sheets, this length to include text (in double line spacing), tables, references, and artwork. An Instruction to Authors leaflet is available from the Senior Editor (Reviews). Although the majority of articles are intended to be specially commissioned, the Society always considers offers of articles for publication. In such cases a short synopsis, rather than the completed article, should be submitted to the Senior Editor (Reviews), Books and Reviews Department, The Royal Society of Chemistry, Thomas Graham House, Science Park, Milton Road, Cambridge CB4 4WF. @ The Royal Society of Chemistry, 1993 All Rights Reserved No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form, or by any means, electronic or mechanical, photographic, recording, or otherwise, without the prior permission of the publishers. Typeset by Servis Filmsetting Ltd. Printed in Great Britain by Blackbear Press Ltd.

ISSN:0306-0012    

DOI:10.1039/CS99322FX005

出版商:RSC    

年代:1993

数据来源: RSC

ISSN:0306-0012    

DOI:10.1039/CS99322BX007

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

ISSN 0306-001 2 CSRVBR 22(2) 73-141 (1 993) Chemical Society Reviews Volume 22 Issue 2 Pages 73-1 41 April 1993 troperty scales can bc es Scales of Solute Hydrogen-bonding: Their Construction and Application to 8plication to a number c Physicochemical and Biochemical Processes By Michael H.Abraham (pp. 73-83)nical processes A-H + B=A-H...H The construction of scales of hydrogen-bond acidity and hydrogen-bond basicity, using 1:1 hydrogen-bondcomplexation constants in tetrachloromethane, is set out. These 1:l scales can be used to develop ‘effective’ ate there have been pu or ‘summation’ scales of hydrogen-bond acidity and hydrogen-bond basicity that are appropriate for f solute hydrogen-bond 11 are of any real prac situations in which a solute is surrounded by an excess of solvent molecules, Together with other solute descriptors that have also been developed, the summation scales of solute hydrogen-bonding can be incorporated into LSERs and QSARs that can be applied to numerous physicochemical and biochemical processes, and can be used to assess quantitatively the role of hydrogen-bonding in these processes.Polymer-Micelle Interactions: Physical Organic Aspects By Josephine C. Brackman and Jan B. F. N. Engberts (pp. 85-92) This review presents a summary of attempts to characterize the morphology of the complexes formed between ionic and non-ionic surfactants and water-soluble polymers. It is now generally accepted that complex formation involves the binding of micelles to the macromolecule.This binding process modifies the size and properties of the micelles and affects the Gibbs energy of the polymer. A complex combination of interaction forces provides the driving force for complexation. Criteria for polymer-micelle interaction are discussed. Particular emphasis is placed on the role of charge and structure of the surfactant headgroup and on the effect of polymers on the micellar structure and properties of cetyltrimethylammonium salts. MELDOLA LECTURE. Reactions of Group 13 Alkyls with Dioxygen and Elemental Chalcogens: From Carelessness to Chemistry By Andrew R. Barron (pp. 93-99) This article details the path by which an accidental reaction of In(But), with dioxygen led to the synthesis of stable Group 13 alkylperoxides, as well as the first examples of Group 13 alkyldisulfides and chalcogenide clusters.The synthesis of these new compounds is presented and the reaction chemistry of the alkylperoxide compounds with simple phosphorus substrates is also described. In the case of the chalcogenide compounds a unique topological rearrangement is presented. Finally, a schematic view of the reaction of a Group I3 alkyl with elemental chalcogens is discussed. How Do Diesel-fuel Ignition Improvers Work? By P. Q. E. Clothier, 8. D. Aguda, A. Moise, and H. 0.Pritchard (pp. 1 01-1 08) A wide-ranging study of the effect of free-radical initiators, especially i-octyl nitrate and di-t-butyl peroxide, on the spontaneous ignition of diesel fuel is reported. The important kinetic processes induced by these additives occur in the gas phase, and not in the liquid droplets.Di-t-butyl and i-octyl nitrate peroxide work in different ways: the nitrate is involved in the ignition process itself, whereas the peroxide competes for sulfur-containing inhibitors in the fuel. The Role of NMR in Boron Chemistry By David Reed (pp. 109-1 16) This article shows, with use of examples, how NMR spectroscopy can help in the study of boron hydride chemistry. The examples highlight some of the range of problems that can be analysed using combinations of one- and two-dimensional NMR techniques. Motion of Sorbed Gases in Polymers By Wen- Yang Wen (pp. 11 7-126) The mobility of sorbed gases in polymers is an interesting scientific problem with important technological implications. Motional mechanisms for penetrant diffusion are considered to depend strongly on whether the polymer matrix is in the rubbery or glassy state.The dual-mode model, which has been employed by a large number of polymer chemists and engineers, is now under scrutiny by some investigators. Early results of a few NMR experiments and computer simulations are beginning to shed new light on the motional dynamics 11 of both the penetrants and polymer chains. Thermodynamic and Related Studies of Amphiphile + Water Systems By Michael 1. Davis B: formed under (pp. I 27-1 34) xessure. is: Modest sized amphiphiles play an important role as aqueous co-solvents. Thermodynamic studies of their aqueous mixtures appear to indicate that they are capable of mimicking the important schemes of self- = xA(G,:rAf RTln.aggregation that are known to be characteristic of detergents and lipids. Attention is focused upon the alkyl poly(ethy1ene glycol) monoether + water systems. The composition dependence of their thermodynamic d G:,,B are the res properties is discussed in terms of both graphical and numerical analyses and possible interpretations of the effects of molecular aggregation. The Solubility of Gases in Water-Alcohol Mixtures By Robert W. Cargill (pp. 135-1 41 ) The solubility of gases in liquids can now be measured to a high level of precision and accuracy. Analysis of data from studies on several gases dissolved in aqueous alcohols has allowed AHo and ASo values to be calculated for the equilibrium.A two-structure model for water has been used to obtain a temperature coefficient for the mole fraction of ‘clusters’ present in water, and the gases have been classified according to their hydrophobic interactions. These hydrophobic interactions have been interpreted in terms of size and shape of the solute molecules. Articles that will appear in forthcoming issues include CENTENARY LECTURE. The Pursuit of Selectivity in Radical Reactions A. L. J. Beckwith Discovery and Development of Anthracycline Antitumour Antibiotics J. W. Lown Some Recent Synthetic Routes to Thioketones and Thioaldehydes W. M. McGregor and D. C. Sherrington The Nature of the Ammonium and Methylammonium Halides in the Vapour Phase: Hydrogen Bonding versus Proton Transfer A.C.Legon Electrolytes in Binary Solvents: Experimental Approach S. Taniewska-Osinska Computer Simulations on Aqueous Solutions of Some Non-Electrolytes K. Nakanishi Structural Systematics in Molecular Inorganic Chemistry A. G. Orpen Biosynthetic Incorporation of Non-Natural Amino Acids into Proteins J. Brunner The Physiological Role of Nitric Oxide A. R. Butler and D. L. H. Williams Low Oxidation States of Indium D. G. Tuck Chemical Society Reviews (ISSN 0306-0012) is published bi-monthly by The Royal Society of Chemistry, Thomas Graham House, Science Park, Milton Road, Cambridge, CB4 4WF, England. All orders accompanied with payment should be sent directly to The Royal Society of Chemistry, Turpin Distribution Services Ltd., Blackhorse Road, Letchworth, Herts., SG6 1HN, U.K.NB Turpin Distribution Services Ltd., distributors, is wholly owned by The Royal Society of Chemistry. 1993 annual subscription rate E.C. 290.00, U.S.A. $198.00, Canada &104.00+ GST, Rest of World g99.00. Customers should make payments by cheque in sterling payable on a U.K. clearing bank or in US.dollars payable on a U.S. clearing bank. Second class postage is paid at Jamaica, N.Y. 11431. Air freight and mailing in the U.S.A. by Publications Expediting Inc., 200 Meacham Avenue, Elmont, New York 11003. U.S.A. Postmaster: Send address changes to Chemical Society Reviews, Publications Expediting Inc., 200 Meacham Avenue, Elmont, New York 11003. All other despatches outside the U.K. by Bulk Airmail within Europe and Accelerated Surface Post outside Europe. PRINTED IN THE U.K. Members of the Royal Society of Chemistry may subscribe to Chemical Society Reviews at E30.00 per annum; they should place their orders on the Annual Subscription renewal forms in the usual way.

ISSN:0306-0012    

DOI:10.1039/CS99322FP009

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

Scales of Solute Hydrogen-bonding Their Construction and Application to P hysicoc hem ica I and Bioc hem ica I Processes Michael H. Abraham Chemistry Department University College London 20 Gordon Street London WCl H OAJ I Introduction It is well recognized that there are numerous processes in chemistry and biochemistry in which hydrogen-bonding plays an important r61e. For example gaseous butanone is more soluble in water at 298 K than is gaseous butane by a factor of around 1.7 x lo4;at least part of this increased solubility must be due to butanone-water hydrogen-bonding. Or consider the inhibition of firefly luciferase activity by aqueous butanone or aqueous butane. Now the increased solvation of butanone over butane in water leads to butane being the more potent by a factor of 74. In order to understand and to interpret these effects quantitatively it is necessary to separate out the various possible solute-solvent interactions and to establish numerical scales for solute properties such as hydrogen-bond acidity and hydrogen- bond basicity. It is the purpose of this review to show how these solute property scales can be established and then to illustrate their application to a number of selected physicochemical and biochemical processes. To date there have been put forward but few quantitative scales of solute hydrogen-bond acidity or hydrogen-bond basi- city that are of any real practical use. Sherry and Purcell,' suggested that the enthalpy of the hydrogen-bond complexation reaction (equation l) AH",could be expressed as the product of parameters characteristic of the acid and the base components. Some years later Raevsky et al. following the general plan of Sherry and Purcell and also Ioghansen,2 developed a very comprehensive set of solute parameters based on equation 2.3In this equation EA is the hydrogen-bond acidity of a given solute EB is the hydrogen-bond basicity of a given solute and dHAB is the standard enthalpy of reaction 1 in tetrachloromethane for the particular AB pair in kJ rnol-l AHiB= 22.5 EAEB (2) To standardize the scale EA is taken as -1.OO for phenol and EB is taken as + 1.OO for diethylether. Although equation 2 is very useful for the correlation and prediction of AH" values it is not so helpful in LSER or QSAR studies for the particular reason that most processes investigated in these studies are Gibbs- energy related and are not enthalpy related. It seems more Michael H. Abraham studied chemistry at the Northern Polytechnic London (B.Sc.) and then at University College London (Ph.D.). He was Reader at the University of Surrey and is now research fellow at University College London. D.Sc. (London) in 1974. He has published a book several reviews and over 200 research papers and was awarded the I991 Ebert Prize of the American Pharmaceuti- cal Association. relevant to construct solute scales that are derived from logK or AGO values in the examination of Gibbs-energy related processes. Together with scales based on equation 2 Raevsky et al. also derived corresponding Gibbs energy quantities using equation Now CA = -1.00 for phenol and CB = + 1.00 for diethy- lether; the constant 5.46 is taken as -AC" for the phenol- diethylether complexation in tetrachloromethane. Unfortuna- tely the methodology based on equation 3 suffers from a considerable disadvantage in the definition of zero acidity or zero basicity.Thus if the hydrogen-bond acid cyclohexane were to complex with the hydrogen-bond base diethylether we might expect K -+ 0 and hence dGo+ a,leading to a value of -co for CA for cyclohexane.* A similar difficulty has bedevilled the pKHB scale of solute hydrogen-bond basicity set up byTaft et and defined as logK for the complexation of bases with 4-fluorophenol in tetrachlor- omethane at 298 K (equation 1; AH = 4-fluorophenol). Again it is impossible to define an origin or zero point for the scale; in other words pKHB for nonbasic compounds such as alkanes cannot be identified. 2 Construction of the Hydrogen-bond Scales Faced with these difficulties we set out to construct scales of solute hydrogen-bond acidity and hydrogen-bond basicity using logK values for reaction 1 in tetrachloromethane so that the scales would be Gibbs-energy related. We started with solute hydrogen-bond acidity,s and set out IogKvalues for a series of acids against a given reference base. We were able to analyse data for no less than 45 reference bases so that we had 45 such series of 1ogKvalues. We then found that if 1ogKvalues for acids against a given reference base were plotted vs. logK values for acids against any other reference base we obtained a series of straight lines that intersected near a 'magic point' of -1.1 log units when K values were calculated on the molar scale (see Figure 1).This enabled us to construct 45 equations all of the form logK (series of acids against reference base B) (4)= L,logK,H + D where LB and DB characterize the base and where the logK2 values now characterize the series of acids. All 45 equations were constrained to pass through the magic point (-1.1 -1.1). Examples of equation 4 for a relatively weak base tetrahydro- furan (THF) and a relatively strong base dimethylsulfoxide (DMSO) are,5 logK (acids againstTHF) = 0.8248 logK,H -0.1970 (5)n = 23 p = 0.9960 sd = 0.089 logK (acids against DMSO) = 1.2399 logK,H + 0.2656 n = 51 p = 0.9947 sd = 0.096 (6) * Note in proof. Raevsky et al. (Quant. Structure-Activity Relat. 1992 11,49) have now introduced a constant term in equation 3.This avoids the zero point difficulty and puts Raevsky's method onto the same basis as Abraham's method. 73 CHEMICAL SOCIETY REVIEWS 1993 3 HMp7 2 log K (ref Base) / 1 Et2S Figure 1 Plots of logK (acids against reference base) vs loga showing the magic point In the above equations and elsewhere n is the number of data points p is the correlation coefficient and sd the standard deviationThe logKf values that characterize the various acids now form a quantitative scale of hydrogen-bond acidity defined through the 45 equations 4 Not only is the scale Gibbs-energy related but there is now a natural zero-point all compounds with zero hydrogen-bond acidity can be assigned log@ = -1 1 units It is convenient to shift the origin from -I 1 to zero itself and at the same time to compress the scale somewhatThis can be done by converting logKg to ay a = (logK'J + 1 1)/4 636 (7) so that equation 7 is the defining equation5 for the solute hydrogen-bond acidity ay Now all compounds with zero aci- dity i e with logK2 = -1 1 have ay= 0 In an exactly similar way logK values for series of bases against a reference acid A again formed a set of lines through the same magic point and led to 34 equations of the form,6 LogK (series of bases against reference acid A) (8)= LA lOgKg + DA Now logKg characterizes the hydrogen-bond basicity of the series of bases and can again be transformed into a more convenient scale = (logKi + 1 1)/4 636 (9) Some values of the hydrogen-bond solute parameters thus obtained are listed inTable I Within families there are connec- tions between hydrogen-bond acidity or basicity and full proton transfer acidity or basicity but these relationships collapse across families of solutes For example a? is larger for phenol than for simple carboxylic acids and Syis larger for DMSO than for Et,N A rather simple relationship exists' between a7/37 and the logK value for reaction 1 in tetrachloromethane at 298 K 10gK=7 354 a? /32 -1094 (10)n= 1312 p=09956 sd=O09 Table 1 Some valuesa of the solute hydrogen-bond acidity and basicity parameters a? and /37 and the effective parametersh CaF and 1/3y Solute aY PY ZaY LPY Alkanes 0 00 0 00 0 00 0 00 Alkenes 0 00 0 07 0 00 0 07 CH,Cl 0 13 0 05 0 10 0 05 CHCI,cc1 0 20 0 00 0 02 0 00 0 15 0 00 0 02 0 00 R2O Acetone 0 00 0 04 0 45 0 50 0 00 0 04 0 45 0 49 RCOR 0 00 0 48 0 00 0 51 DMSO 0 00 0 78 0 00 Acetonitrile 0 09 0 44 0 07 0 32 Nitromethane 0 12 0 25 0 06 0 31 Et,N 0 00 0 67 0 00 0 79 Benzene 0 00 0 14 0 00 0 14 PhCl 0 00 0 09 0 00 0 07 PhOMe 0 00 0 26 0 00 0 29 PhCOMe 0 00 0 51 0 00 0 48 PhCN 0 00 0 42 0 00 0 33 PhNH 0 26 0 38 0 26 0 41 Water 0 35 0 38 0 82 0 35 Methanol 0 37 0 41 0 43 0 47 Ethanol 0 33 0 44 0 37 0 48 Phenol 0 60 0 22 0 60 0 30 RC0,H 0 54 0 42 0 60 0 45 Taken from references 5 and 6 R = n dlkyl Ref 13 and unpublished work Although the form of equation 10is similar to that of Raevsky's equation 3 the constant ( -1 094) in equation 10 is now related to the magic point and hence to the zero origin Equation 10 is particularly useful in calculating values of ay or /3? from logK values when either a? or ,8yis known The usefulness of the entire analysis using the magic point was well illustrated by examination* of logKvalues obtained by Hine et a1These workers plotted logK(bases against a reference acid imide) YS logK(bases against 4-fluorophenol) and obtained a set of randomly intersecting lines Figure 2 * Such a plot implies that the relative hydrogen-bond acidity of imides varies accord- ing to the actual reference base especially for weak bases where logK values are below around 1 0 log units However if Hine's logKvalues are plotted according to equation 8 with the various lines forced through the magic point Figure 3 is obtained Now the relative hydrogen-bond acidity of the imines always remains the same no matter what is the reference baseThis of course is a condition for the establishment of any general scale ** The a? and /3? solute scales have been set up using 1ogKvalues for complexation in tetrachloromethane and it is of some interest to establish whether these scales can be used in other solvents Abboud et a1 O determined logK values for complexa- tion of acids with the reference base pyridine N-oxide in cyclo- hexane at 296 5K logKpyoThey found quite a good correlation between ayand logKpyo equation 1 1 which indicates that the a? (and /33scales might be useful with other solvents than tetrachloromethane 0.7 = 0 185 lOgKp,o + 0 069 n = 22 p = 0 993 sd = 003 However a comparison of scales set up with solvent 1,1,1-trichloroethaneTCE and a? and @ does suggest that some family dependencies may make conversion between these scales rather difficult,' within solute families conversions can again * We use the general basicity scale logpd in Figure 2 but this makes little difference ** We note that there are some (known) exceptions wlth regard to our solute scales In particular combinations of weak N-H acids with pyridine bases are excluded SCALES OF SOLUTE HYDROGEN-BONDING-M H ABRAHAM 5I 4 /b1 *lL’1 Figures 2 and 3 Plots of logK (bases against reference acids) vs logaThe reference acids are (a) 2-ethyl-2-methylsuccinimide,(b) 2-chloro-3-methylmaleimide (c) 2,3-dichloro-2-methylsuccinimide,and (d) tetrafluorosuccinimide be madeThus for phenols and alcohols (where NMP = N-methylpyrrolidinone),’ logK (against NMP inTCE) = 0 870 logK,H + 0 70 n=21 ~~0986sd=O 13 (12) We have therefore established rather comprehensive scales of solute hydrogen-bond acidity and hydrogen-bond basicity that can be extended using logK values in other solvents as shown above and also through incorporation of more recent data in tetrachloromethane itself l2 But all the ayand @values we have refer to 1 1 complexation equation 1 It is by no means obvious that such values are relevant to the solvation situation in which a solute is surrounded by solvent molecules and hence undergoes multiple hydrogen-bonding In order to test this we set up a number of multiple linear regression equations as shown below and ‘back-calculated’ the solute hydrogen-bond parameters to check whether or not the ‘effective’ or ‘summation’ solute hydrogen-bond acidity and basicity (Ca? and Cp$ could be taken as the 1 1 ay and /3? scales l3 In the event we were fortunate enough to find that in general the 1 1 scales could be used as solute parameters even when the solute was surrounded by a large excess of solventThere were exceptions and inTable 1 we compare the 1 1 values with the summation values13 for a number of representative solutes all of the monofunctional type In general multifunctional solutes cannot be dealt with in terms of 1 1 complexation constants and at present can only be handled through the back-calculation of summation values 3 Application of Hydrogen-bond Scales Before these summation scales can be applied to any given solvation process it is necessary to formulate some model of solvation We use a simple cavity model in which the process of dissolution of a gaseous solute in a solvent involves (1) the endoergic creation of a cavity in the solvent and (11) incorpora- 75 tion of the solute in the cavity with consequent setting up of various exoergic solute-solvent interactions Each of these interactions will require a relevant solute parameter or descrip- tor * After considerable preliminary work (see e g ref 14) the following solute descriptors were selected R,is an excess molar refraction that can be calculated from refractive index or can rather easily be estimated ~yisthe solute dipolarity-polariza- bility obtained to date from gas-liquid chromatography (GLC) of solutes on polar stationary phases,l and py are the solute hydrogen-bond acidity and solute hydrogen-bond basicity (where appropriate Cay and Z/3ymust be used but we retain the simpler nomenclature) L’ is the solute gas-hexadecane parti- tion coefficient at 298K and V is McGowan’s characteristic volumeTwo general linear solvation energy relationships (LSERs) can be constructed from these parameters,’ log SP = c + rR + ST? + aay + bp? + vV (14) In these equations the dependent variable log SP refers to some property of a series of solutes in a fixed phase (or phases)Thus SP could be L the gas-liquid partition coefficient for a series of solutes in a given liquid or it could be P,the partition coefficient for a series of solutes between water and say octanol In the case of biological properties where SP can be some biological response as an LC, equations 13 and 14 then represent two new families of quantitative structure-activity relationships (QSARs1 Equation 13 is the simpler and can be applied to processes involving gas +condensed-phase transferThe terms YR sn? aay and bpy represent specific solute<ondensed phase interac- tions respectively dispersion dipoledipole or dipole-induced- dipole plus some polarizability interaction solute acid-solvent base and solute base-solvent acid The logL16 term includes both general dispersion interactions and the endoergic cavity term at the moment it seems not possible to separate out the important cavity term on its own The constants in equation 13 are found by the method of multiple linear regression analysis and serve to characterize the particular condensed phase Y is the tendency of the phase to interact through 7~-and n-electron pairs s is the phase dipolarity-polarizability a is the phase hydrogen-bond basicity (because a basic phase will interact with an acidic solute) b is the phase hydrogen-bond acidity and I is a measure of the ability of the phase to distinguish between or to separate homologues in any homologous series For processes within condensed phases equation 14 is used but now the various constants will reflect differences between the phases If SPis Pact the water-octanol partition coefficient then for example the a-constant will be a measure of the difference in basicity between water and octanolThe vV term in equation 14 will include both differences in cavity effects and differences in general dispersion interactions the v-constant now being a measure of the difference between the hydrophobicity or lipo- philicity of the two phases 3.1 Application to Gas +Condensed Phase Processes Equation 13 has been applied to several large sets of gas-liquid chromatographic (GLC) data t For the retention of a series of solutes on a given stationary phase at a given temperature we can take SP to be VGthe specific retention volume or L(K) the gas+liquid partition coefficient or even t the adjusted retention time Details of the application of equation 13 to data * ,411 the processes we shall consider involve d series of solute molecules in d given solvent or solvents Hence the solvent(s) properties remain constant and it is not necessary to hdve to deal with solvent parameters Only the solute is vdried and hence only solute parameters are needed -f For multifunctional solutes we should really refer to ZU~and /?!jthe effective or summation hydrogen bond acidity or basicity In the examples that follow it should be recognized that a!j and 8’ do indeed refer to Cur and Z/?!j where appropriate obtained by Lafford et a1 and by McReynolds' are inTable 2 l3 Poole et a1 l9 have obtained retention data at 394K on a variety of phases including some novel molten salts Some Table 2 Application of equation 13 to GLC data at 393 K" Stdtionary phase c r s a / n p sd Carbowaxh -201 025 126 207 0429 199 0997 007 DEGSh -177 035 158 184 0383 199 0997 007 PPEh -251 014 089 067 0547 199 0997 006 TCEPh -169 0 26 193 1 88 0 365 199 0998 006 ZE7h -199 -041 146 0 77 0432 199 0995 007 Apiezon J -048 024 0 15 0 13 0596 165 0999 002 Pluronic L 72 -0 54 0 09 0 93 142 0 529 163 0 998 0 03 ,411 the phases are non acidic hence h = 0 "Data from ref 17 Note that IogSP = l0gL -IogL (decdne)The abbreviations are DEGS diethylenglycol succinate PPE polyphenyletherTCEP tricydnoethoxypropdne ZE7 zonyl E 7 Datd from ref 18 SP = Vc examples of analyses of Poole's data are in Table 3 Examination of Tables 2 and 3 shows how any GLC stationary phase can now be characterized by the constants in equation 13 Thus squalane and apiezon J are almost nonpolar and nonbasic (s and a are small or zero) but are very good at separating homologues (1 is large) Phases such as DEGS andTCEP are quite dipolar dnd quite basic (s and a are large) whereas ZE7 is dipolar but not very basic and the molten salts are dipolar and very basic (a = 3 4 for the 4-toluene sulfonate and the methane sulfonate) Not only can GLC stationary phases be characterized through equation 13 but so can any solvent phase For example we have applied equation 13 to gas-liquid partition coefficients in a number of amides at 298K 2o Using the most recent solvation equation 13 we find for N-formylmorpholine (NFM) dnd also for tri(2-ethylhexylphosphate) (2-EHP) at 298 K the equations logL (in NFM) = -0 53 + 2 5777; + 4 32ay + 0 730 logL' (15) n = 45 p = 09949 sd = 007 logL (in 2-EHP) = -0 07 -0 26R + 0 91ny + 3 74ay + 0 955 logL'6 (16) n= 22 p= 09978 sd = 004 Since neither of these solvents is acidic the 6-constant is zero However both are quite dipolar and NFM particularly is highly basic with a = 4 32 units * The toxicity of gases and vapours is a considerable environ- mental problem One procedure for the estimation of such toxicity is to measure the upper respiratory tract irritation * Note that the constants obtained dt 298K in equations 15 and 16 cdnnot directly be compared with those at 393KTables 2and 3 because dn increase in temperature invariably results in a marked decredse in s u and h CHEMICAL SOCIETY REVIEWS 1993 caused by vapours to mice We have shown21 that a QSAR for the toxicity of nonreactive compounds can be obtained using our new hydrogen- bond parameters and a recalculation based on equation 13 yields -IogFRD 10 96 + 0 8 In7 + 2 55ay + 0 722 10gLi6 n= 39 p=O987 sd=O 12 (17) This QSAR is useful for the prediction of toxicity of nonreactive compounds but also shows that the receptor site is reasonably polar (s = 0 81) and has a somewhat lower hydrogen-bond basicity than (2-EHP) Equation 17 can be contrasted with a preliminary equation for the solubility of gaseous solutes in water at 298 K. logL (in wdter) = -1 28 + 0 87R + 2 70777 + 4 Ola? + 4 80gY -0 210 10gLi6 (18) n = 350 p = 0 9952 sd = 0 19 It is clear that the receptor site cannot be any sort of aqueous environment since equation 18 is completely different to equa-tion 17 Many other examples of the application of equation 13 can be given for example to the solubility of gaseous solutes in or to the solubility of gaseous solutes in blood and other biological but the examples given here show the widespread use of this general solvation equation in the correla- tion of gas +condensed phase processes One other useful feature of equation 13 is that it enables the contribution of specific solute solvent interactions to the overdll logL values to be calculatedThus for a solute with a knownTY value the contribution of the solute solvent dipolarity-dipolar- ity term will simply be given by the snyproduct Calculations on these lines are in Tables 4 and 5 for gaseous solubility in NFM Table 4 An analysis of solute-solvent interaction contributions to IogL in NFM at 298 K Solute C rR2 my aay bpy /logLI6 Butane -053 000 000 000 000 1 18 Hexane -053 000 000 000 000 191 Benzene -053 000 134 000 000 203 Butanone -053 000 180 000 000 167 Ethanol -0 53 000 108 160 000 108 Table 5 An analysis of solute-solvent interaction contributions to logL in water at 298 K Solute c rR ST? uaY b# IlogL'6 Butane -128 000 000 000 000 -034 Hexane -128 000 000 000 000 -056 Benzene -128 0 53 140 000 067 -0 59 Butanone -128 0 14 189 000 245 -048 Ethanol -I28 021 113 148 230 -031 Table 3 Application of equation 13 to GLC data of Poole et a1 ,at 394 4 K" Stationary phase c r5-a I n P sd Squdlane -0 20 0 12 0 02 -0 10 0 581 39 0 999 0 03 ov 11 -0 30 0 10 0 54 0 17 0 516 39 0 999 0 03 OV 225 -0 51 0 02 121 0 96 0 462 39 0 998 0 03 Tetrabutylammonium 4-toluene sulfonate -0 62 0 01 1 66 3 36 0 440 34 0 989 0 11 Tetrabutylammonium methane sulfonate -0 63 0 09 1 60 3 41 0 437 34 0 990 0 10 Tetrabutylammonium picrate -0 54 0 10 156 1 42 0 445 36 0 994 0 06 Datd from ref 19 SP = L All the phases are non acidic hence h= 0 SCALES OF SOLUTE HYDROGEN-BONDING-M H ABRAHAM and waterThe important dispersion interaction term is not explicitly given in equation 13 because the ZlogL16 term includes both a positive dispersion interaction effect towards logL and a negative cavity term effect towards logLThe analysis of Abra- ham and Fuc~s,~~ and also cavity calculations based on Pierot- ti’s scaled particle theory,2 both show that dispersion interac- tions are very large and are nearly always the dominant positive interaction on logL Possibly the most extraordinary feature of equation 18 is the negative dependence of logL on logL16 whereas for all nona- queous solvents we have studied the I-constant is quite posit- ive *This is a manifestation of the hydrophobic effect Large gaseous solutes become very soluble in nonaqueous solvents but in water are no more soluble than small solutes Both SPT calculations and the analysis of Abraham and Fuchs for hexa- decane solvent suggest that the main reason for the peculiar behaviour of water lies in the cavity effect This becomes much more endoergic as the solute size increases than do cavity effects in nonaqueous solvents and leads to a large negative contribu- tion to logL Additionally the dispersion interaction term in nonaqueous solvents becomes more exoergic as solute size increases than does the dispersion term in waterThe dispersion interaction effect leads to more positive contributions to logL for larger solutes in nonaqueous media than with larger solutes in water We can now see from the breakdown in Table 5 exactly what are the factors that lead to increased aqueous solubility of gaseous butanone over gaseous butane at 298K The two main terms are and bpywhich contribute 1 89 and 2 45 log units to the more favourable solubility of butanone These correspond to extra exoergic Gibbs energies of 11 0 and 14 0 kJ mol- l due to dipole-dipole and hydrogen-bond (solute base-solvent acid) interactions in the case of butanone If we consider solvent NFMTable 4 the increased solubility of gaseous butanone over butane is mainly due to the STY term corresponding to an exoergic Gibbs energy of about 10 3 kJ mol-l favouring buta- none through dipole-dipole interactions The construction of scales of solute hydrogen-bonding together with other solute descriptors thus leads via equation 13 and also equation 14 to a quantitative assessment of the rde of hydrogen-bonding in various processes 3.2 Application to Processes in Condensed Phases There have been very many applications of equation 14 and so we shall merely give a few examples 01 recent ones In principle any kind of partition between two condensed phases can be examined and SP in equation 14 can be k‘ the HPLC capacity factor or Pa liquid-liquid partition coefficient for example We have already analysed the very important water-octanol parti-tion coefficient Pact using an earlier equation26 and we now recalculate the correlation using equation 14 logPocT = 0 08 + 0 58Rz -1 09,r + 0 03ar -3 40/3r+ 3 8lV (19) n= 584 p=O996 sd=O 13 Equation 19 shows that increase in solute size V favours wet octanol whereas increase in solute dipolarityTY or solute hydrogen-bond basicity SF favours the aqueous layer Solute hydrogen-bond acidity has little influence on logPOc Since the constants in equation 19 refer to differences in properties of the two phases concerned we can deduce that the hydrogen-bond basicity of water and wet octanol are almost the same Similar analyses can be carried out for numerous water-solvent partitions There are a large number of biochemical and toxicological processes that involve aqueous solutes interacting with a given * Exactly the sdme results dre obtained if solute volume V is used instead of the logLI6 parameter system In principle the general solvation equation 14 could be applied to any such process As an example we can quote the work of Franks and Lieb2 on the inhibition of firefly luciferase activity by aqueous nonelectrolytes Application of equation 14 to the data of Franks and Lieb leads to the regression -logEC = 0 58 + 0 72R2 -3 44/37 + 3 77Vx n=42 p=O989 sd=O33 (20) The crucial factors that determine the potency of aqueous solutes are thus the solute volume that increases inhibition and solute hydrogen-bond basicity that decreases inhibition * Since the solute hydrogen-bond acidity plays no part it can be deduced that the target site(s) must be of about the same hydrogen-bond basicity as water itself On the other hand the large negative b-constant in equation 20 indicates that the target site(s) must have a relatively poor hydrogen-bond acidity cf equation 19 Similar results are obtained for the potency of aqueous nonelectrolytes in inducing general anaesthesia in animals Again the general anaesthetic site must be of about the same hydrogen-bond basicity as water but significantly less acidic 28 We can now deal with the other example mentioned in the introduction namely that the potency of butane as an inhibitor is some 74 times that of butanone towards firefly luciferase activityThe two main terms governing potency as -logEC, are the bpyand the vVx terms Since V for butane and butanone is almost the same the diminished potency of butanone is due to increased hydrogen-bonding with water as compared to the target site(s) With /3? = 0 5 I for butanone the bfl? term contri- butes 3 44 x 0 51 = 1 75 log units (or a factor of 56) equivalent to a Gibbs energy of 10 0 kJ mol-I in favour of butane over butanone as an inhibitorThus most of the factor of 74 can be related to hydrogen-bonding of butanone with water 4 Summary Scales of hydrogen-bond acidity and basicity have been set up using formation constants in tetrachloromethaneThese a? and pyscales are Gibbs-energy related and are unique for such scales in that they incorporate ‘zero points’ It is found that a? and p? can be used as the basis of general scales that include ‘effective’ or ‘summation’ hydrogen-bond acidities and basicities for use as solute parameters or descriptors in LSER and QSAR equations Such equations that include also various other solute descriptors can be used to correlate and to interpret a wide variety of physicochemical and biochemical processesThe two general LSER and QSAR equations 13 and 14 can in principle be applied to any process involving gas +condensed phase transfer equation 13 or to any process within condensed phases equation 14The main requirement is sufficient values of the dependent variable logSP that span a variety of solute type Usually at least five data points are needed for each explanatory variable and the variety of solute type is necessary so that the explanatory variables or descriptors cover as wide a range as possible and most importantly are not subject to significant cross-correlations In order that these criteria be met it is necessary to have available the relevant solute descriptors for a rather large number of solutes In order to aid workers who wish to test equations 13and 14 we set out the solute descriptors for a reasonably large number and variety of solutes It is not feasible to list all the available parameters for example we have values of logL’ for 100 alkanes and have recently listed all the parameters for 120 alkylaromatic hydrocarbons 29 but hope- fully those given inTables 6 7 and 8 will be sufficient for many purposes Note that in these tables the parameters are effective or summation values where appropriate raylor et al 30 have recently discovered an unwelcome com- plication in that the hydrogen-bond basicity of certain solutes in water-solvent partitions seems to vary with the particular water-solvent system We find that for a large number of solutes including all those given in Tables 6 and 8 Cp? values are 78 CHEMICAL SOCIETY REVIEWS 1993 Table 6 Values of logL16 and V for inorganic gases and Table 7 Descriptors for solutes with varying xp2alkanesa Solute R2 777 1. 1/37 1s;Solute logL'6 vx Aniline 0 955 096 026 041 050 3934 Helium -1741 0 068 a-Toluidine 0 966 092 023 045 059 4442 Neon -1575 0 085 m-Toluidme 0 946 095 023 045 055 4463 Argon -0 688 0 190 p-Toluidine 0 923 095 023 045 052 4452 Krypton -0211 0 246 2-Chloroaniline I033 092 025 031 040 4674 Xenon 0 378 0 329 3-Chloroaniline I 053 1 10 030 030 036 4909 Radon 0 877 0 384 4-Chloroaniline 1 060 113 030 031 035 4889 Hydrogen -1 200 0 109 2-Methoxyaniline 0 988 100 023 050 067 4818 Oxygen -0 723 0 183 3-Methoxyaniline 1027 122 025 055 068 5023 Nitrogen -0 978 0 222 4-Methoxyaniline 1 050 1 10 023 065 072 4949 Methane -0 323 0 250 Methyl 4-aminobenzoate 1078 152 032 0 59 064 Ethane 0 492 0 390 Ethyl 4-aminobenzoate 1 040 152 032 0 59 0 64 Propane 1 050 0 531 n-Propyl4-aminobenzoate 1030 150 0 32 059 064 n-Butane 1615 0 672 n-Butyl4-aminobenzoate I 020 I47 032 0 59 0 64 2-Methylpropane 1 409 0 672 Pyridine 0 631 084 000 052 047 3022 n-Pentane 2 162 0 813 2-Methylpyridine 0 598 075 000 058 048 3422 2-Methylbutane 2 013 0 813 3-Methylpyridine 0 631 081 000 054 044 3631 2,2-Dimethylpropane 1820 0 813 4-Methylpyridine 0 630 082 000 0 54 043 3 640 n-Hexane 2 668 0 954 2,3-Dimethylpyridine 0 657 0 77 000 062 0 50 4045 2-Methylpentane 2 503 0 954 2,4-Dimethylpyridine 0 634 0 76 000 063 049 4006 3-Methylpentane 2 581 0 954 2,5-Dimethylpyridine 0 633 0 74 000 062 049 3 986 2,2-Dimethylbutane 2 352 0 954 2,6-Dimethylpyridine 0 607 0 70 000 063 049 3760 2,3-Dimethylbutane 2 495 0 954 3,4-Dimethylpyridine 0 676 085 000 062 048 4317 n-Heptane 3 173 I095 3,5-Dimethylpyridine 0 659 0 79 000 060 044 4214 2-Methylhexane 3 001 1095 2-Ethylpyridine 0 613 071 000 059 049 3844 3-Methylhexane 3 044 1 095 3-Ethylpyridine 0 640 079 000 057 047 4093 3-Ethylpentane 3 091 1 095 4-Ethylpyridine 0 634 080 000 057 047 4 124 2,2-Dimethylpentane 2 796 1 095 2,3-Dimethylpentane 3 016 1095 2,4-Dimethylpentane 2 809 1 095 3,3-D1methylpentane 2 946 1 095 2,2,3-Trimethylbutane 2 918 1095 constant and can be used in equations that describe any n-Octane 3 677 1236 gas -+condensed phase process and any water-solvent partition 2,2,4-Trimethylpentane 3 106 1236 n-Nonane 4 182 I377 processThere are however solutes for which the general py(or n-Decane 4 686 I518 Zp$ descriptor has to be modified for certain water-solvent n-Undecane 5 191 1659 partition processes specifically those involving solvents that n-Dodecane 5 696 1 799 contain a rather high proportion of water when saturatedThus n-Tridecane 6 200 1 940 for the solutes inTable 7 the alternative /3!jvalue must be used in n-Tetradecane 6 705 2 081 LSER equations that describe water-octanol water-isobuta-n-Pen tadecane 7 209 2 222 nol and possibly also water-butyl acetate and waterdiethy1 n-Hexadecane 7 714 2 363 ether partitioning Note that in the water-octanol LSER equa-n-Heptadecane 8 218 2 504 tion 19 only solutes with an invariant p2 value @For Z/33haven-Octadecane 8 722 2 645 n-Nonadecane 9 226 2 786 been included n-Eicosane 9 731 2 927 n-Heneicosane 10 236 3 068 n-Docosane 10 740 3 208 4.1 A Note on the Calculation of V n-Tricosane 11 252 3 349 In the various equations we have given Vx(in cm3 mol-I/100) has been calculated by the procedure of McGowanI6 in which (1 All these solutes have zero R values Units of V are (cm3 mol ')/I00 atom constants are simply summed and 6 56 cm3 mol-l subtracted for each bond noting that all bonds (single double Table 8 Values of solute descriptors used in equations 13 and 14 Solute R2 777 1.7 1/37 logL'6 Cyclopentane 0 263 0 10 0 00 0 00 2 477 Met h ylcyclopentane 0 225 0 10 0 00 0 00 2 816 Cyclohexane 0 305 0 10 0 00 0 00 2 964 Methylcyclohexane 0 244 0 10 0 00 0 00 3 323 Cycloheptane 0 350 0 10 0 00 0 00 3 704 Methylcycloheptane 0 300 0 10 0 00 0 00 4 034 C yclooctane 0 413 0 10 0 00 0 00 4 329 trans-Hydrindane 0 439 0 20 0 00 0 00 4 467 cu-H ydrindane 0 439 0 25 0 00 0 00 4 635 Adamantane 0 667 0 66 0 00 0 02 5 095 trans-Decalin 0 467 0 23 0 00 0 00 4 984 cu-Decalin 0 544 0 25 0 00 0 00 5 156 Ethene 0 107 0 10 0 00 0 07 0 289 Propene U 103 0 08 0 00 0 07 0 946 But-1-ene 0 100 0 08 0 00 0 07 1491 Pent- l-ene 0 093 0 08 0 00 0 07 2 047 SCALES OF SOLUTE HYDROGEN-BONDING-M Table 8 Continued Solute Hex-1-ene Hept- 1-ene Oct-1-ene Non- 1-ene Dec-I -ene Undec-1-ene Dodec-1-ene Buta-l,3-diene 2-Methylbuta-I ,3-diene 2,3-Dimethylbuta- 1,3-diene Cyclopentene 1-Methylcyclopentene Cyclohexene 1-Methylcyclohexene Cycloheptene 1-Methylcycloheptene Ethyne Propyne But-1-ynePent- I-yne Hex-1-yne Hept-1-yne Oct-1-yne Oct-2-yne Non-1-yne Dec-1-yne Dodec-1-yne Fluoroethane I -Fluoropropane 1-Fluorohexane I-Fluorooctane Dichloromethane Trichloromethane Tetrachloromethane Chloroethane 1 I-Dichloroethane 1,2-Dichloroethane I I 1-Trichloroethane 1,1,2-Trichloroethane 1-Chloropropane 2-Chloropropane 1-Chlorobutane 1-Chloropentane I-Chlorohexane 1-Chloroheptane 1-Chlorooctane Chlorocyclohexane 1,1 -Dichloroethene CIS-1,2-DichIoroethene trans-1,2,-Dichloroethene Bromomethane Di bromomet hane Tri bromomet hane Bromoethane 1-Bromopropane 1-Bromobutane 1-Bromopentane 1-Bromohexane 1-Bromoheptane 1-Bromooctane 1-Bromononane Bromocyclohexane Iodomethane Diiodomethane Iodoethane I -1odopropane 1-1odobutane 1-1odopentane 1-Iodohexane Halothane Teflurane Diethylether Di-n-propylether Di-n-butylether H ABRAHAM 79 logL'6 0 078 0 08 0 00 0 07 2 572 0 092 0 08 0 00 0 07 3 063 0 094 0 08 0 00 0 07 3 568 0 090 0 08 0 00 0 07 4 073 0 093 0 08 0 00 0 07 4 533 0 091 0 08 0 00 0 07 5 023 0 089 0 08 0 00 0 07 5 515 0 320 0 23 0 00 0 10 1 543 0 313 0 23 0 00 0 10 2 101 0 352 0 23 0 00 0 14 2 690 0 335 0 20 0 00 0 10 2 402 0 330 0 20 0 00 0 10 2 864 0 395 0 20 0 00 0 10 3 021 0 391 0 20 0 00 0 10 3 483 0 414 0 22 0 00 0 10 3 626 0 430 0 22 0 00 0 10 3 957 0 190 0 25 0 21 0 15 0 150 0 183 0 25 0 12 0 15 1 025 0 178 0 23 0 12 0 15 1520 0 172 0 23 0 12 0 12 2 010 0 166 0 23 0 12 0 10 2 510 0 160 0 23 0 12 0 10 3 000 0 155 0 23 0 12 0 10 3 521 0 225 0 30 0 00 0 10 3 850 0 150 0 23 0 12 0 10 4 019 0 143 0 23 0 12 0 10 4 537 0 133 0 23 0 12 0 10 5 657 0 052 0 35 0 00 0 10 0 559 0 034 0 35 0 00 0 10 I071 0 000 0 35 0 00 0 10 2 951 0 020 0 35 0 00 0 10 3 850 0 387 0 57 0 10 0 05 2 019 0 425 0 49 0 15 0 02 2 480 0 458 0 38 0 00 0 00 2 823 0 227 0 40 0 00 0 10 1678 0 322 0 49 0 10 0 10 2 316 0 416 0 64 0 10 0 11 2 573 0 369 0 41 0 00 0 09 2 733 0 499 0 68 0 13 0 08 3 290 0 216 0 40 0 00 0 10 2 202 0 177 0 35 0 00 0 12 1 970 0 210 0 40 0 00 0 10 2 722 0 208 0 40 0 00 0 10 3 223 0 201 0 40 0 00 0 10 3 777 0 194 0 40 0 00 0 10 4 282 0 191 0 40 0 00 0 10 4 772 0 448 0 48 0 00 0 10 4 016 0 362 0 34 0 00 0 05 2 110 0 436 0 61 0 11 0 05 2 439 0 425 0 41 0 09 0 05 2 278 0 399 0 43 0 00 0 10 I 630 0 714 0 67 0 10 0 10 2 886 0 974 0 68 0 15 0 06 3 784 0 366 0 40 0 00 0 12 2 120 0 366 0 40 0 00 0 12 2 620 0 360 0 40 0 00 0 12 3 105 0 356 0 40 0 00 0 12 3 611 0 349 0 40 0 00 0 12 4 130 0 343 0 40 0 00 0 12 4 663 0 339 0 40 0 00 0 12 5 090 0 336 0 40 0 00 0 12 5 560 0 615 0 54 0 00 0 16 4 395 0 676 0 43 0 00 0 13 2 106 1453 0 69 0 05 0 23 3 857 0 640 0 40 0 00 0 15 2 573 0 634 0 40 0 00 0 15 3 130 0 628 0 40 0 00 0 15 3 628 0 621 0 40 0 00 0 15 4 130 0 615 0 40 0 00 0 15 4 620 0 102 0 38 0 15 0 05 2 177 0 070 0 21 0 20 0 02 1370 0 041 0 25 0 00 0 45 2 015 0 008 0 25 0 00 0 45 2 954 0 000 0 25 0 00 0 45 3 924 80 CHEMICAL SOCIETY REVIEWS. 1993 Table 8 Continued Solute logL'6 Methoxyflurane 0 109 0 67 0 07 0 14 2 864 Isoflurane -0 240 0 50 0 I0 0 10 1576 Enflurane -0 230 0 40 0 12 0 13 1750 Fluroxene 0 183 0 30 0 00 0 27 1 600 Acetaldehyde 0 208 0 67 0 00 0 45 1230 Propionaldehyde 0 196 0 65 0 00 0 45 1815 Butyraldehyde 0 187 0 65 0 00 0 45 2 270 Iso-butyraldehyde 0 144 0 62 0 00 0 45 2 120 Pentanal 0 163 0 65 0 00 0 45 2 851 Hexanal 0 146 0 65 0 00 0 45 3 357 Heptanal 0 140 0 65 0 00 0 45 3 865 Octanal 0 I60 0 65 0 00 0 45 4 361 Nonanal 0 150 0 65 0 00 0 45 4 856 Propenal 0 324 0 72 0 00 0 45 I656 trans-But-2-ene-1-a1 0 387 0 80 0 00 0 50 2 570 2-Methylpropenal 0 400 0 70 0 00 0 50 2 180 Propanone 0 179 0 70 0 04 0 49 1696 Butanone 0 166 0 70 0 00 0 51 2 287 pent an-2-one 0 143 0 68 0 00 0 51 2 755 Hexan-2-one 0 136 0 68 0 00 0 51 3 262 Heptan-2-one 0 123 0 68 0 00 0 51 3 760 Octan-2-one 0 108 0 68 0 00 0 51 4 257 Nonan-2-one 0 119 0 68 0 00 0 51 4 735 Decan-2-one 0 108 0 68 0 00 0 51 5 245 Undecan-2-one 0 101 0 68 0 00 0 51 5 732 Dodecan-2-one 0 103 0 68 0 00 0 51 6 167 Cyclopentanone 0 373 0 86 0 00 0 52 3 221 Cyclohexanone 0 403 0 86 0 00 0 56 3 792 Cycloheptanone 0 436 0 86 0 00 0 56 4 376 Methyl formate 0 192 0 68 0 00 0 38 1285 Ethyl formate 0 146 0 66 0 00 0 38 1845 n-Propyl formate 0 132 0 63 0 00 0 38 2 433 n-Butyl formate 0 121 0 63 0 00 0 38 2 958 n-Pentyl formate 0 101 0 63 0 00 0 38 3 488 Methyl acetate 0 142 0 64 0 00 0 45 1911 Ethyl acetate 0 106 0 62 0 00 0 45 2 314 n-Propyl acetate 0 092 0 60 0 00 0 45 2 819 n-Butyl acetate 0 071 0 60 0 00 0 45 3 353 n-Pentyl acetate 0 067 0 60 0 00 0 45 3 844 n-Hexyl acetate 0 056 0 60 0 00 0 45 4 351 n-Heptyl acetate 0 050 0 60 0 00 0 45 4 865 n-Octyl acetate 0 029 0 60 0 00 0 45 5 364 Vinyl acetate 0 223 0 64 0 00 0 43 2 152 Ally1 acetate 0 199 0 72 0 00 0 49 2 723 Acetonit rile 0 237 0 90 0 07 0 32 1739 Proprioni trile 0 162 0 90 0 02 0 36 2 082 1-Cyanopropane 0 188 0 90 0 00 0 36 2 548 1-Cyanobutane 0 177 0 90 0 00 0 36 3 108 1-Cyanopentane 0 166 0 90 0 00 0 36 3 608 1-Cyanohexane 1-Cyanoheptane 1-Cyanooctane 1-Cyanononane 1-Cyanodecane 0 159 0 162 0 159 0 156 0 154 0 90 0 90 0 90 0 90 0 90 0 00 0 00 0 00 0 00 0 00 0 36 0 36 0 36 0 36 0 36 4 089 4 585 4970 5 460 5 940 Ammonia 0 139 0 35 0 14 0 62 0 680 Methylamine Ethy lamine n-Propylamine n-Butylamine 0 250 0 236 0 225 0 224 0 35 0 35 0 35 0 35 0 16 0 16 0 16 0 16 0 58 0 61 0 61 0 61 1300 1677 2 141 2 618 n-Pent ylamine 0211 0 35 0 16 0 61 3 139 n-Hexylamine n-Octylamine 0 197 0 187 0 35 0 35 0 16 0 16 0 61 0 61 3 655 4 520 Dimethylamine Diethylamine Di-n-propylamine Di-n-but ylamine Di-n-pent ylamineTrimethylamineTriethylamine Nitromethane 0 189 0 154 0 124 0 107 0 099 0 140 0 101 0 313 0 30 0 30 0 30 0 30 0 30 0 20 0 15 0 95 0 08 0 08 0 08 0 08 0 08 0 00 0 00 0 06 0 66 0 69 0 69 0 69 0 69 0 67 0 79 0 31 1 600 2 395 3 351 4 349 4 570 1620 3 040 1892 Nitroethane 0 270 0 95 0 02 0 33 2 414 1-Nitropropane 1-Nitrobutane 0 242 0 227 0 95 0 95 0 00 0 00 0 31 0 29 2 894 3 415 1-Nitropentane 0 212 0 95 0 00 0 29 3 938 SCALES OF SOLUTE HYDROGEN-BONDING-M H ABRAHAM 81 Table 8 Continued Solute logL'6 1 -Nitrohexme 0 203 0 95 0 00 0 29 4 416 Nitroc yclohexane 0 441 0 97 0 00 0 31 4 826 Formamide 0 468 1 30 0 62 0 60 Acetamide 0 460 130 0 54 0 68 Proprionamide 0 440 130 0 55 0 68 Butanamide 0 420 130 0 56 0 68 N-Methylformdmide 0 405 130 0 40 0 55 N-Methyldcetamide 0 400 130 0 40 0 72 N-Methylpropanamide N N-Dimethylformdmide 0 380 0 367 1 30 131 0 40 0 00 0 71 0 74 3 173 N N-Dimethyldcetdmide 0 363 133 0 00 0 78 3 717 Formic dcid 0 300 0 60 0 75 0 38 Acetic dcid 0 265 0 65 0 61 0 44 1 750 Propdnoic dcid 0 233 0 65 0 60 0 45 2 290 Butdnoic md 0 210 0 62 0 60 0 45 2 830 Pentdnoic dcid 0 205 0 60 0 60 0 45 3 380 Hcxdnoic acid 0 174 0 60 0 60 0 45 3 920 Heptdnoic dcid 0 149 0 60 0 60 0 45 4 460 Octdnoic dcid 0 150 0 60 0 60 0 45 5 000 Nondnoic md 0 132 0 60 0 60 0 45 5 550 Dccdnoic dcid 0 124 0 60 0 60 0 45 6 090 Undecdnoic dcid 0 100 0 60 0 60 0 45 6 640 Dodecdnoic dcid 0 083 0 60 0 60 0 45 7 180 Chlorodcetic dcid 0 373 108 0 74 0 36 Dichlorodcetic dcid 0 481 1 20 0 90 0 27 Trichloroncetic acid 0 589 133 0 95 0 28 Wdter 0 000 0 45 0 82 0 35 0 260 Met hd 11 01 0 278 0 44 0 43 0 47 0 970 Ethdnol 0 246 0 42 0 37 0 48 I485 Propdn-1-01 0 236 0 42 0 37 0 48 2 031 Propdn-2-01 0 212 0 36 0 33 0 56 1 764 Butdn-1-01 0 224 0 42 0 37 0 48 2 601 Pentdn-1-01 0 219 0 42 0 37 0 48 3 106 Hexdn-1-01 0 210 0 42 0 37 0 48 3 610 Hcptdn-1-01 0211 0 42 0 37 0 48 4 115 Octdn-1-01 0 199 0 42 0 37 0 48 4 619 Nondn-1-01 0 193 0 42 0 37 0 48 5 124 Decan-1-01 0 191 0 42 0 37 0 48 5 628 Undecan-1-01 0 181 0 42 0 37 0 48 6 130 Dodecdn-1-01 0 175 0 42 0 37 0 48 6 640 Cyclopentdnol 0 427 0 54 0 32 0 56 3 241 Cyclohexmol 0 460 0 54 0 32 0 57 3 758 Cycloheptdnol 0 513 0 54 0 32 0 58 4 407 Addmdntdn-1-01 0 850 1 20 0 32 0 56 Prop-2-en-1-01 0 342 0 46 0 38 0 48 I951 fr a/i~-But-9-en-l-ol 0 350 0 44 0 38 0 48 2 618 2.2,2-Trifluoroethanol 0 015 0 60 0 57 0 25 1224 HeXdflUoropropdn-1-01 -0 240 0 55 0 77 0 10 1392 Dodecnfluoroheptan-1-01 -0 640 0 50 0 65 0 22 3 089 Et hy 1t hio1 0 392 0 35 0 00 0 24 2 173 n-Pr op y 1t hi o1 n -Pent 4 1t h I o1 n-Hexylthiol n-Hept 4 It hiol ti -N ony1th i ol n-Decyl t hiol Ally1 thiol 11-BUt y l t hl 01 II-Octylt hi01 0 385 0 382 0 369 0 361 0 357 0 353 0 347 0 342 0 542 0 35 0 35 0 35 0 35 0 35 0 35 0 35 0 35 0 41 0 00 0 00 0 00 0 00 0 00 0 00 0 00 0 00 0 00 0 24 0 24 0 24 0 24 0 24 0 24 0 24 0 24 0 20 2 685 3 111 3 624 4 133 4 635 5 270 5 790 6 318 2 654 Dimethyl sulfide Diethyl sulfide Di-n-p ro p y 1 sulfide Di-n-but yl-sulfide Sulfur hexdfluoride C'i r b on d i su1fideTrimethyl phosphateTriethyl phosphateTri-n-propyl phosphate 0 404 0 373 0 358 0 345 -0 600 0 877 0 113 0 000 -0 050 0 38 0 38 0 38 0 38 0 20 0 21 1 10 1 00 1 00 0 00 0 00 0 00 0 00 0 00 0 00 0 00 0 00 0 00 0 29 0 32 0 32 0 32 0 00 0 07 1 00 1 06 I 15 2 238 3 104 4 120 4 950 -0 120 2 353 4 750 Tri-n-but)] phosphate BenzeneTo1uenc Ethyl ben7ene o-X ylei1e i7i-X ylene -0 100 0 610 0 601 0 613 0 663 0 623 0 90 0 52 0 52 0 51 0 56 0 52 0 00 0 00 0 00 0 00 0 00 0 00 121 0 14 0 14 0 15 0 16 0 16 2 786 3 325 3 778 3 939 3 839 82 CHEMICAL SOCIETY REVIEWS 1993 ~ Table 8 Continued Solute Crs? logL'6 p-Xylene 0 613 0 52 0 00 0 16 3 839 n-Propylbenzene 0 604 0 50 0 00 0 15 4 230 n-But ylbenzene 0 600 0 51 0 00 0 15 4 730 1,3,5-Trimethylbenzene 0 649 0 52 0 00 0 19 4 344 Pentamethylbenzene 0 850 0 66 0 00 0 20 5 798 Hexamethylbenzene 0 950 0 72 0 00 0 21 6 557 Styrene 0 849 0 65 0 00 0 16 3 856 Phenylethyne 0 679 0 58 0 12 0 24 3 692 Ally1 benzene 0 717 0 60 0 00 0 22 4 136 Di pheny lme t hane 1 220 1 04 0 00 0 28 6 313 Biphenyl 1360 0 99 0 00 0 22 6 014 Naphthalene 1 340 0 92 0 00 0 20 5 161 Anthracene 2 290 134 0 00 0 26 7 568 Phenanthrene 2 055 1 29 0 00 0 26 7 632 Fluoro benzene 0 477 0 57 0 00 0 10 2 788 Chlorobenzene 0 718 0 65 0 00 0 07 3 657 1,2-Dichlorobenzene 0 872 0 78 0 00 0 04 4 518 1,3-Dichlorobenzene 0 847 0 73 0 00 0 02 4 410 1,4-Dichlorobenzene 0 825 0 75 0 00 0 02 4 435 Benzyl chloride 0 821 0 82 0 00 0 33 4 384 Bromobenzene 0 882 0 73 0 00 0 09 4 041 1,2-Dibromobenzene 1190 0 96 0 00 0 04 5 456 1,3-Dibromobenzene 1 170 0 88 0 00 0 04 5 327 1,4-Dibromobenzene 1 150 0 86 0 00 0 04 5 324 Benzyl bromide 1014 0 98 0 00 0 20 4 672 Iodobenzene 1 188 0 82 0 00 0 12 4 502 Methylphenylether 0 708 0 75 0 00 0 29 3 890 Ethylphenylether 0 681 0 70 0 00 0 32 4 242 Benzaldeh yde 0 820 1 00 0 00 0 39 4 008 Acetophenone 0 818 101 0 00 0 48 4 501 Ethylphenylketone 0 804 0 95 0 00 0 51 4 971 Benzophenone 1 447 150 0 00 0 50 Methyl benzoate 0 733 0 85 0 00 0 46 4 704 Ethyl benzoate 0 689 0 85 0 00 0 46 5 075 n-Propyl benzoate 0 675 0 80 0 00 0 46 5 718 n-Butyl benzoate 0 668 0 80 0 00 0 46 6 210 Phenyl acetate 0 661 1 13 0 00 0 54 4 414 Dimethyl phthalate 0 780 I41 0 00 0 88 6 051 Benzonitrile 0 742 111 0 00 0 33 4 039 Benzylamine 0 829 0 88 0 10 0 72 4 319 Nitrobenzene 0 871 111 0 00 0 28 4 557 2-Nitrotoluene 0 866 111 0 00 0 27 4 878 3-Ni trotoluene 0 874 1 10 0 00 0 25 5 097 4-Nitrotoluene 0 870 111 0 00 0 28 5 154 Benzamide 0 990 1 50 0 49 0 67 5 767 Formanilide 0 970 1 40 0 50 0 50 Acetanilide 0 870 1 40 0 50 0 67 Benzoic acid 0 730 0 90 0 59 0 40 2-Methylbenzoic acid 3-Methylbenzoic acid 4-Methylbenzoic acid Phenol 0 730 0 730 0 730 0 805 0 90 0 90 0 90 0 89 0 60 0 59 0 60 0 60 0 34 0 38 0 38 0 30 3 766 o-Cresol 0 840 0 86 0 52 0 30 4 218 rn-Cresol 0 822 0 88 0 57 0 34 4 310 p-Cresol 2-Fluorophenol 0 820 0 660 0 87 0 69 0 57 0 61 0 31 0 26 4 312 3 453 3-Fluorophenol 0 667 0 98 0 68 0 17 3 842 4-Fluorophenol 2-Chlorophenol 3-Chlorophenol 4-Chlorophenol 2-Bromophenol 3-Bromophenol 4-Bromophenol 2-Iodophenol 3-Iodophenol 4-Iodophenol 2-Methoxyphenol 3-Methoxyphenol 4-Methoxyphenol 2-Cyanophenol 3-Cyanophenol 4-Cyanophenol 2-nitro phenol 0 670 0 853 0 909 0 915 1037 1 060 1080 1360 1370 1380 0 837 0 879 0 900 0 920 0 930 0 940 1015 0 97 0 88 106 1 08 0 90 115 117 1 00 1 20 122 0 91 117 1 17 133 155 163 1 05 0 63 0 32 0 69 0 67 0 35 0 70 0 67 0 40 0 70 0 68 0 22 0 59 0 57 0 74 0 77 0 79 0 05 0 23 0 31 0 15 0 20 0 31 0 16 0 20 0 35 0 18 0 20 0 52 0 39 0 48 0 33 0 28 0 29 0 37 3 844 4 178 4 773 4 775 4 526 5 144 5 135 4 964 5 528 5 492 4 449 4 803 4 773 4 531 5 181 5 420 4 760 SCALES OF SOLUTE HYDROGEN-BONDING-M Table 8 Continued Solute 3-Nitrophenol 4-Nitrophenol 1-Naphthol 2-Naphthol Benzyl alcohol Thiophenol Benzenesul fonamide N-Methylbenzenesulfonamide N N-Dimet h ylbenzenesulfonamide Furan Benzofuran Tetrahydro fur an 2-methyltetrahydro furan 1.4-Dioxdne Benzodioxan Paraldehyde Piperidine N-Methylpipendine N-Ethylpipendine Pyrrole N-Methyipyrrole Pyrazine 2-Methylpyrazine Pyrimidine Thiophene Benzo(b)thiophene Dibenzothiophen Thiazole H ABRAHAM CPY logL’6 157 0 79 0 23 5 692 172 0 82 0 26 5 876 1 05 0 61 0 37 6 130 1 08 0 61 0 40 6 200 0 87 0 33 0 56 4 221 0 80 0 09 0 16 4 110 155 0 55 0 80 1 50 0 30 0 82 1 50 0 00 0 86 0 53 0 00 0 13 0 83 0 00 0 15 4 355 0 52 0 00 0 48 2 636 0 48 0 00 0 53 2 820 0 75 0 00 0 64 2 892 1 07 0 00 0 35 4 971 0 68 0 00 0 68 3 169 0 46 0 10 0 69 3 304 0 39 0 00 0 70 3 330 0 32 0 00 0 63 3 729 0 73 0 41 0 29 2 865 0 79 0 00 0 31 2 923 0 95 0 00 0 62 2 920 0 90 0 00 0 64 3 254 1 00 0 00 0 65 2 837 0 56 0 00 0 15 2 819 0 88 0 00 0 20 5 174 131 0 00 0 18 7 575 0 80 0 00 0 45 Chem Soc 1988,110,8534 8 M H Abraham,P L Grellier D V Prior J J Morns,P JTaylor and R M Doherty J Org Chem 1990,55,2227 9 J Hine S Hahn and J Hwang J Org Chem 1988,53,884 10 J -L M Abboud K Sraidi M H Abraham,and R WTdft J Org Chem 1990,55,2230 11 M H Abraham P P Duce D V Prior D G Barrett J J Morris and P JTaylor J Chem Soc PerkinTrans 2 1989 1355 12 C Laurence M Berthelot E Raczynska J -Y Le Questel G Duguay and P Hudhomme J Chem Res (s) 1990,250 13 M H Abraham G S Whiting R M Doherty and W J Shuely J Chromatogr 1991,587,213 229 14 M H Abraham G S Whiting R M Doherty and W J Shuely J Chem Soc PerkinTrans 2 1990 145 1 15 M H Abraham P L Grellier and R A McGiIl. J Chem Soc PerkinTrans 2 1987 797 16 M H Abraham and J C McGowan Chromatographia 1987 23 243 17 F Patte M Etcheto and P Laffort Anal Chem 1982,54,2239 18 W 0 McReynolds ‘Gas Chromatographic Retention Data’ Pres- tonTechnical Abstracts Evanston I11 1966 19 B R Kersten S K Poole and C F Poole J Chromatogr 1990 500,329 20 M H Abraham G S Whiting R M Doherty and W J Shuely J Chem Soc ,PerkinTrans 2 1990 185 1 21 M H Abraham G S Whiting Y Alarie J J Morris P JTaylor R M Doherty R WTaft and G Nielsen Quunt Structure-Activitj Relat 1990,9 6 22 M H Abraham G S Whiting R M Doherty W J Shuely and P Sakellariou Polymer 1992 33 2 I62 23 M H Abraham and P K Weathersby unpublished results 24 M H Abraham and R Fuchs J Chem Soc PerkinTrans 2 1988 523 25 R A Pierotti Chem Rev 1976,76 717 26 M H Abraham G S Whiting R Fuchs dnd E J Chambers J Chem Soc PerkinTrans 2 1990,291 27 N P Franks and W R Lieb Nature 1984,310,599 1985,316,349 28 M H Abraham W R Lieb and N P Franks J Pharm Sci ,1991 80,719 29 M H Abraham and G S Whiting J Chromatogr 1992,594,229 30 D E Leahy J J Morris P JTaylor and A R Wait J Chem Soc PerkinTrans 2 1992 705 1 050 1 070 1520 I520 0 803 1 000 1130 1 100 1 100 0 369 0 888 0 289 0 241 0 329 0 874 0 136 0 422 0 318 0 300 0 613 0 559 0 629 0 629 0 606 0 687 1323 1959 0 800 or triple) are counted the same For complicated molecules it is time consuming to count the number of bonds but I find that this number is given by the algorithm B=N-I+R where B IS the number of bonds in the molecule N is the total number of atoms and R is the number of ringsThus for cyclohexane or benzene R = 1 and for cyclohexylbenzene diphenyl or naphthalene R = 2 So far I have found no exception to this rule Because Vxis so easily calculated values are given only inTable 6 partly as examples and partly to list those for the rare gases. References A D SherryandK F Purcell J Phys Chem ,1970,74,3535,A D Sherry and K F Purcell J Am Chem SOC 1972,94 1853 A B IoghansenTheor Exp Khim (USSR),1971,7,302 0 A Raevsky and V P Novikov Khimico-farm Z 1982. 16 583 0 A Raevsky V V Avidon and V P Novikov Khimico-farm Z 1982. 16.968.0 A Rdevsky V Yu Grigoriev and V P Soloviov Kliir?iico-farrnZ 1984 18 578 0 A Raevsky in ‘QSAR in Drug Design dndToxicology’ ed D Hadzi and B Jerman-Balzic. Else- vier Amsterddm 1987 R WTdft. D Gurka L Joris P von R Schleyer and J W Rdkshys J An? Chem SOL 1969,91,4801 J Mitsky L Joris and R WTdft J Am Chem Soc 1972,94 3438 M H Abraham P P Duce P L Grellier D V Prior J J Morris and P J Taylor Tetrahedron Lett ,1988,29 1587 M H Abraham P L Grellier D V Prior P P Duce J J Morris,andP J Taylor,J Chem Soc Perhin Trans 2 1989 699 M H Abrdhdm. P L Grellier D V Prior J J Morris P J Taylor C Lawrence and M Berthelot Tetrahedron Lett 1989 30 2571 M H Abraham P L Grellier D V Prior J J Morris and P J Tdylor J Chem Soc PerhinTrans 2 1990 521 M H Abrahdm. P L Grellier D V Prior R W Taft J J Morris P J TdylOr C Laurence. M Berthelot R M Doherty M J Kamlet J -L M Abboud K Sraidi and G Guiheneuf J Am

ISSN:0306-0012    

DOI:10.1039/CS9932200073

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

Polymer-Micelle Interactions Physical Organic Aspects Josephine C. BrackmanS and Jan B. F. N. Engberts" Department of Organic Chemistry University of Groningen Nijenborgh 4,9747 AG Groningen The Netherlands 1 Introduction Studies of the interaction between non-ionic water-soluble polymers and micelles have their roots in biochemistry for they originated from the study of protein-surfactant interaction. Polymer-micelle interaction3 in turn may now serve as a simpli- fied model for biological binding processes for instance to cell membranes. An important difference between proteins and non- ionic polymers is the complete absence of charged groups in the latter. Consequently polymer-micelle interaction results from an accumulation of relatively weak binding forces which makes the association process even more intriguing. At an early stage of the research in this field it was recognized that in the polymer-micelle complex the properties both of the micelles and of the polymers are mutually m~dified.~.~ To mention the most important aspects with respect to industrial applications the solubilization power as well as the viscosity of an aqueous solution of polymer-bound micelles is higher than that of the separate surfactant and polymer solutions.2This commercial interest is reflected in the fact that many of the early reports on polymer-micelle interaction originated from indus- trial research laboratorie~.~~~-~ The properties of the polymer- micelle complex are very well appreciated in formulations for paints and coatings in cosmetic products and in laundry detergents. Polymer-micelle interaction also plays a role in tertiary oil recovery. Although the applications of polymer-micelle complexes are numerous many problems are still unsolved. Particularly the question of how the precise chemical structure of the surfactant and the morphology of the unperturbed micelle are related to the tendency for association with polymers poses a challenge for chemists.The very limited choice of surfactants as well as dubious generalizations3 in the study of polymer-micelle inter- action certainly obscure this problem. For example the belief that in general anionic micelles interact with polymers cationic micelles hardly and non-ionic micelles not at all was deduced with sodium n-alkylsulfates predominantly sodium n-dodecyl- sulfate (SDS) as representatives of anionic surfactants with n- alkyltrimethyl-ammonium bromides as representatives of catio- nic surfactants and with n-alkoxypoly(ethy1ene oxide) ethers as representatives of non-ionic surfactants. Not only is the genera- lization unwarranted but also the rationalization behind it is hampered by the limited choice of surfactants. ~ ~~ Josephine C. Brackman was born in Haarlem in 1962. A member of Professor J. B. F. N. Engberts's group at the University of' Groningen she completed her Ph.D research in 1990 on the interactions be- tween water-soluble polymers and surfactants. 1.1The Development of the Polymer-Micelle Model The recognition of the interaction between non-ionic water- soluble polymers and surfactants occurred decades later than the notion that surfactants proper form aggregates. But the morphology of the micelle attracted minimal comment until the eighties despite the enormous number of articles devoted to the properties of micelles. In contrast the morphology of the polymer-surfactant complex puzzled chemists from around 1955 when the pioneering work of Saito4 was published till the end of the seventies when an NMR study of the poly(ethy1ene oxide) (PEO)/SDS system by Cabane9 firmly established the contemporary model. However many intriguing questions remain. In particular the relationship between the chemical structure of surfactant and polymer and their propensity for interaction and also the dominant driving force for interaction are still debated in the literature. In 1957 Saito4 published the first extensive study on polymer- surfactant complexation.Two major observations were (i) an increase in viscosity of an aqueous poly(vinylpyrro1idone) (PVP) solution upon addition of SDS and (ii) an increase in solubilizing power of an SDS solution upon addition of PVP.Though it was suggested that the aggregation of sufactant molecules in the presence of polymer resembled normal micellization Saito proposed that at a low surfactant-to-polymer ratio the surfac- tant molecules bind individually to the polymer (as in the case for protein-surfactant interactions at low surfactant concent- ration).This binding was thought to occur by dipolar interac- tion of the surfactant headgroups with polar sites on the polymer while the surfactant chain was thought to lie parallel to the polymer chain. At a higher degree of saturation the location of the alkyl chain would be altered. However Saito4 wisely stated that the structure of the polymer-micelle or polymer- surfactant complex had not yet been clearly established. The major concept to emerge in the following decade sum- marized in Breuer's and Robb's review arti~le,~ was the picture of individual molecules along the polymer with some kind of micellization occurring above the critical micelle concentration (cmc) of the surfactant in pure water.lO,ll Many details of polymer-micelle interaction were revealed in that period including the finding that complexation takes place even below the normal cmc.l0This has long been used as Jan B. F. N.Engberts was born in Leiden in 1939. He obtained his Ph.D degree in I967 under Professor J. Strating at the University of Groningen. After post-doc- toral research at the Universitjj of Amsterdam with Projessor Th.J.de Boer he was appointed as Professor of General Chem- istry at the University of Gron-ingen in 1978. He is fascinated by the unique solvent properties of water and his research is focused on organic chemistry in aqueous solutions. He has authored and co-authoredsome 250 publications on solvent eflects reaction mechanisms and surfac t an t aggrega t ion. 85 support for individual binding ' It was also found that above a minimum molecular weight (MW) of the polymer the interac- tion is independent of MW,7 t? lo and that a certain saturation takes place at increasing surfactant concentration OThe importance of hydrophobic interactions2 in polymer-surfactant complex formation was deduced from the stronger interactions obtained as polymer hydrophobicity4 O and surfactant alkyl chain length were increased lo Much later in 1987 measure- ments of heat capacities and apparent molar volumes also revealed a shift of these thermodynamic properties upon addi- tion of polymer in the direction of enhanced hydrophobic association Using cmc values for a series of homologous sodium alkyl sul- fates C,H2,+ ,OSO,Na (n = 10 11 12) either with or without added PVP Shinoda6 derived the Gibbs energy of trans-ferring a CH group from the aqueous solution to either the aggregate (polymer-micelle complex) or the micelle to be in both cases 1 1 kT He deduced from these data that C,H,,+ 1-OS0,Na molecules absorbed on PVP are in contact with each other and are not uniformly distributed on the PVP macromole- cule right from the initial stages of absorption In the same year Lange5 commented on the discrepancy between the increase in viscosity upon polymer-surfactant complexation which indi- cates coil expansion and the increased solubilizing power of the polymer-surfactant complex which involves a compact struc- ture of the complex He also stressed the cooperative nature of the complex formation which is apparent from the existence of a critical concentration for its formationThese arguments appeared to require the notion that micelles bind to the polymer -but initially another explanation was advanced to explain the new results The idea that surfactants bind to polymers in clusters took hold7 and the next issue gradually became apparentTokiwa and Tsujii (1973)' assumed without any discussion that the surfactant micelles encompass portions of the polymer chain Shirahama,' however in 1976 suggested that a binding of the polymer at the micellar surface (above the cmc) occurs leading to a stabilization through reduction of the core-water contact but he did not yet believe in the existence of micelles below the unperturbed cmc He also predicted lower aggregation numbers for the polymer-bound micelles ' In 1977 Cabane9 established the polymer-micelle model as it is now quite generally accepted (see for instance the excellent review article of Goddard3) Figure 1 gives a schematic represen- tation of the model Cabane studied the PEOjSDS system using I3C 'H and 23Na NMR Only the first three carbon atoms of SDS counted from the SO headgroup exhibit 3C chemical shifts consistent with the presence of PEO Cabane9 concluded that in the polymer-micelle complex the major part of the alkyl Figure 1 Schematic representation of a polymer-mlcelle complex according to NagarajanThe probably more realistic representation(a)' 5e from 1985 shows the development of the model Figures (a) and (b) reproduced by permission from references 15e and 1% respectively CHEMICAL SOCIETY REVIEWS 1993 chain resides in an environment indistinguishable from a normal micelle which is a micellar coreThe first three carbon atoms are influenced by the polymer because the polymer binds at the micellar surface which in an unperturbed micelle (according to Cabane) is one third occupied by -SO groups The other two thirds of the surface contain primarily the first chain segments The NMR signals of the polymer are barely influenced by complexation with micelles That is interpreted as an indication that only a fraction of the polymer is actually absorbed onto the micellar surface whereas the rest protrudes as loops in the aqueous surroundingsThis was anticipated because total absorption and thus a restricted mobility of the polymer chain would be very unfavourable for entropic reasons Cabane9 also mentions two sound common sense reasons why PEO should bind at the micellar outer sphere First PEO does not dissolve in hydrocarbons and will therefore not penetrate into the hydro- phobic micellar interior Second most probes that have been used to study micellar properties are quite hydrophobic but still bind at the micellar surface Therefore it is likely that even rather hydrophobic polymers like poly(propy1ene oxide) (PPO) poly- (vinylmethylether) (PVME) partially acetylated poly(vinyla1co- hol) (PVA-Ac) and hydroxypropyl cellulose (HPC) will also seek the outer layer of the micelle Very recently Kwak et a1 ' published an NMR study on the system w-phenyldecanoatejPE0 and concluded that PEO resides in the interior of the micelleThe conclusions were based on H aromatic ring current-induced shifts of the PEO protons However the argument hinges on the (debatable) assumption that the phenyl moieties do not fold back to the surface of the micelles Two additional indications that both hydrophilic polymers and relatively hydrophobic polymers bind to the micellar sur- face are the smaller aggregation numbers of polymer-bound micelles14l7 2o and the variation in interaction tendency with headgroup charge ' (vide supra) The decrease in aggrega- tion number was initially only documented for the system PEO/ SDS,14 l7 21 PVPjSDS l7 l8 2o 21 and PVA/SDS17 but has recently also been reported for PPO/SDS19 21 and in our studies for CTAB in the presence of PPO and PVME (Section 4)This is in accord with the presence of the polymer at the micellar surface whereas solubilization in the core is expected to lead to an increase in aggregation number The effect of short- chain and long-chain alcohols and alkanes on the aggregation number of micelles supports these considerations 22 23 Short-chain alcohols which reside at the micellar surface decrease the aggregation number whereas alkanes which reside in the core increase the aggregation number 22 The finding that cationic surfactants usually interact more weakly with polymers than anionic surfactants do (which will be discussed in detail below) also points to a location of the polymer in the same region as the headgroups whatever the origin of the difference is Gilanyi and Wolfram in 198 1 began an endeavour to find a quantitative model for the prediction of binding isotherms and critical concentrationsTheir model was based on the mass- action law for micellization l7 Like all other models for polymer-micelle interaction' 24 25 considered so far the pre- dictions were checked with experimental data on SDS micelles bound to the hydrophilic polymers (PVA PVP,and PEO) The models of RuckensteinZ4 and Nagarajan' were checked on the system PEO/SDS and the model of Evans2 on the system PEO/ Cu(DS) 26 Gilanyi and Wolfram" also made the important point that the formation of regular (free) micelles may take place at a surfactant concentration below the saturation concent- ration of the polymer since the activity of the surfactant rises as the polymer is loaded with micelles l4The activity of the surfactant ions may thus reach the critical value for formation of free micelles before binding of micelles to the polymer is completed Several authors have studied the influence of polymers on the properties of probe molecules bound to micelles These results indicate a more open and water-penetrated structure of the polymer-bound micelles POLYMER-MICELLE INTERACTIONS-J C BRACKMAN AND J B F N ENGBERTS Details concerning the polymer-micelle interaction3 have been provided by techniques like NMR self-diffusion electric birefringence ultrasound absorption and the use of surfactant- ion selective electrodes Nevertheless a consistent explanation for the influence of the precise chemical structure of surfactant and polymer on the interaction tendency and a quantitative model that is applicable to more systems than PEO/SDS alone are still lacking A clarification of just these problems is of the utmost importance for the further understanding of polymer- micelle interaction and development of the modelThis review will focus on recent work carried out in our laboratory and is aimed at obtaining a better understanding of the relation between the chemical structure of both surfactant and polymer and the propensity for polymer-micelle interaction 2 Criteria for Polymer-Micelle Interaction A reduction of the cmc due to the presence of polymer fails to be the ultimate criterion for polymer-micelle attraction It is certainly decisive in one sense that is if a reduction takes place it definitely points to polymer-micelle association How- ever one should consider the situation that binding of the polymer does not affect the stability of the micelle significantly but that only the polymer is stabilized in the binding process For example it was believed for a long time that polymers do not bind to non-ionic micelles However our studies2' on the clouding behaviour and Krafft temperatures of PPO solutions in the presence of micelles formed from n-octyl 8-D-thioglucopyra- noside (OTG) (I) provide evidence that the micelles do interact with the polymer qH20H HYH I OH n-Oct yl p-D-1hioglucopyranoside This remarkable association between PPO and OTG micelles was definitely confirmed by microcalorimetric measurements When an OTG solution was injected into the PPO solution the microcalorimetric response curve consisted of an endothermic peak followed by an exothermic peak (Figure 2)This phenome- non is attributed to rapid endothermic polymer-micelle associa- tion near the injection point followed by a slower disintegration of the complex and dilution of the surfactant molecules in the entire solution *'The total dilution enthalpies shown in Figure 3 are summations of the areas of the endothermic and exothermic peaks The curve for OTG dilution in H20 can be characterized by three regions In the pre-micellar region I the injected micelles disintegrate completely and the enthalpy change for de-micelli- zation and loss of intermicellar interactions is recorded Region I1 is the transition region around the cmc In the post-transition region 111 the injected micelles remain intact and only a very small enthalpy change for reduction of intermicellar interaction is measuredThe enthalpy of micellization calculated as the difference in dilution enthalpy between region I and 11 is + 4 5 kJ mol a normal value for a non-ionic surfactant Comparison of the curve for the PPO solution with the curve for H,O reveals that PPO exerts only a small endothermic effect on the pre-micellar enthalpy of dilution Furthermore the transi- tion region is located in the same concentration range indicative of an unchanged cmc However a clear endothermic effect 15 min-11 40pW Exotherm v 17 Figure 2Top Microcalorimetric response curve upon (ref 27) injection of a concentrated OTG solution into a PPO solution with the final OTG concentration remaining below the cmc The numbers refer to the titration steps I e 9 corresponds to the ninth titration step (see also Figure 3) Each response consists of an endothermic and an exothermic peak Bottom similar data but now the final OTG concentration is beyond the cmc Note the increase of the endothermic signal relative to that shown in the top partThe exothermic effect has disappeared completely beyond titration step 18 Signal noise is caused by the stirrer Temperature 25 "C + 4 3 kJ mol-I is observed in the post-transition region of the PPO solution We contend that this value represents the enthalpy of interaction between PPO and OTG micelles Since the Gibbs energy of micellization of OTG IS unchanged by the presence of PPO the endothermic interaction enthalpy is apparently compensated by a positive entropy change This AH/ AS compensatory behaviour probably originates largely from the release of water molecules from the hydrophobic hydration shells of the polymer upon interaction with the micelles The different behaviour of PEO and PPO most likely reflects the difference in Gibbs energy of transfer of the polymer from water to a more apolar environment PPO is more soluble in hydrocarbons than in water contrary to PEO In this context we also refer to subsequent studies on HPC 28 The question arises whether PPO interacting with OTG micelles resides at the micellar surface like PEO in the system PEO/SDS or deeper in the micellar coreThe latter possibility is not likely for the reason mentioned previously but additional evidence is called for Aggregation numbers may give a clue because if PPO resides in the core an increase in aggregation number is expected instead of the usual decrease found in most polymer-ionic micelle complexes We have made an attempt to measure aggregation numbers of OTG micelles in the absence and presence of PPO using static fluorescence quenching We obtain an aggregation number of 156 f10 for OTG micelles which is rather high compared to the values of 68-84,29 or 87,30 for 8-D-n-octylglucoside (with an ether instead of a thio linkage) determined by light scattering and sedimentation techniques In the presence of 0 5 g dL-' of PPO we find a value of 96 f3 Although the exact values may be slightly in error we submit that the trend is obvious and points to location of PPO in the outer region of the micelleThus PPO/OTG interaction proba- bly resembles the classical PEO/SDS association The most important conclusion is however that polymer-micelle interac- tion is not necessarily accompanied by a reduction in cmc It goes without saying that the driving force for polymer- micelle interaction is a reduction in Gibbs energy of the total system Still it is worthwhile to note that both stabilization of CHEMICAL SOCIETY REVIEWS 1993 -4 -6 0 3 6 9 12 Final concentration of OTG in mmol I -' 15 Figure 3 Enthalpy of dilution as a function of the final OTG concent-ration in water or in an aqueous solution of PPO at 25 "C(U)in water cmc = 8 05 x in PPO solutions (V) exothermic effect (V)endothermic effect (0)summation of exothermic and endothermic effectsThe numbers (9-12 17-21) correspond with the titration steps indicated in Figure 2 (Taken from reference 27 ) the micelle proper and a reduction in the Gibbs energy of the (hydrated) polymer may provide the major contribution to the total free energy for the formation of polymer-bound micelles 3The Role of the Charge and Structure of the Surfactant Headgroup Micellar charge whether positive or negative definitely pro- motes micelle stabilization upon binding of polymers It is not however a prerequisite for association as hds long been propa- gated The major effects of surfactant charge on the stabiliza- tion of polymer-bound micelles are a contribution from the reduction in electrostatic repulsion due to the smaller size of the bound micelles and the influence of charged groups on the hydration sheaths of polymers The former effect is operative for both negatively and positively charged surfactantsThe latter effect does in practice depend on the sign of the charge since only a limited choice of charged groups can be used as head- groups of a surfactant 3.1 The Effect of Headgroup Charge on Polymer-Micelle Interaction n-Dodecyldimethylamine Oxide3' Dipolar n-dodecyldimethylamine oxide (DDAO) belongs to an interesting class of non-ionic surfactants sometimes referred to as the semi-polar class The headgroup charge of DDAO in the micelle can be easily varied between 0 and 1 via variation of the pH(equation 1) No drastic change in structure and volume of the headgroup is involved The effect of charge variation on the cmc and other micellar properties has been investigated in some detail 3 I I Cmc Values Cmc values (obtained by the pH-method developed for phos- phate surf act ant^^^) are listed inTable 1 The degree of protona- Table 1 Cmc valuesa of DDAO at various degrees of protonation in the absence and presence of polymers 8' Polymerh 00 0 24 0 47 0 75 0 98 - 17 1 53 1 80 2 54 4 74 PPO - 1 46 133 163 - PVMEPEO 10 k - 156 161 1 70 184 2 08 2 55 - InmM Polymer concentration tn 0 9gdL Calculated from the pH dt the cmc using pK = 5 0 tion (p 5 0 01) is calculated from pH at the cmc using a pKA value of 5 0 It can be adjusted by varying the pH of the concentrated DDAO stock solution The pH at the cmc and thus p is not noticeably affected by the presence of polymer The obtained cmc values for DDAO in H,O are relatively low in comparison with those reported previously A similar obser- vation was made in case of the phosphate surfactants 32 Presu-mably the method responds to even the first stage of aggregation Within the limits of reproducibility (5 YOat /3 = 0 24 2 YOat p = 0 5 and 0 75) the cmc at /3 = 0 24 is not affected by the presence of polymers At higher degrees of protonation the cmc is decreased in the presence of PPO and PVME but not influenced by the presence of PEO It is tempting to conclude from the cmc data that the stabilization of the micelles by PPO and PVME increases with increasing micellar charge However d more quantitative conclusion should be based on a compari- son of Gibbs energies of micellization in the presence and absence of polymer In a first approximation the Gibbs energy of micellization is related to the cmc expressed in mole fraction units according to equation 2 'The change in standard Gibbs energy of the micelles due to the binding of a polymer is given by equation 3 in which cmcp represents the cmc in the polymer solution l2The quantity dG& pol -AG; denotes the change in standard Gibbs energy when 1 mole of surfactant molecules is transferred from unperturbed micelles to polymer-bound micelles plus the change in free energy of the polymer induced by this process Values for dG& pol -A Gklc (Table 2) confirm the intuitive conclusion from the cmc data namely that polymer-induced stabilization is more pronounced at higher ACi = RT In (cmc) (2) dGi pol -dG& = RT In (cmc,/cmc) (3) POLYMER-MICELLE INTERACTIONS-J C BRACKMAN AND J B F N ENGBERTS Table 2 dG; pol -dG& for DDAO micelles,a at various degrees of protonation in the presence of polymers 8' Polymerh 0 24 0 47 0 75 PPO -01 -08 -11 PVME 01 -02 -05 PEO 10 k 01 01 00 In kJ mol estimated error 0 1 kJ mol Polymer concentration ca 0 9 g dL I Calculated from the pH at the cmc using pK = 5 0 micellar chargeThis seems to agree with current views on polymer-micelle interaction We contend however that although indeed the interaction with the ionic surfactant is stronger than with the non-ionic surfactant any rationalization based on headgroup volume is misplaced Protonation will hardly influence the size of the headgroup but the hydration shell will be affectedThis is expected to lead to a larger (hydrated) size of the cationic headgroup Apparently the size of the cationic headgroup will not be much different from that of a trimethylammonium group We propose that the increase in stabilization of the micelles by interaction with polymers at increasing micellar charge stems from an enhanced reduction of electrostatic repulsion Particularly at higher micellar charge the formation of smaller polymer-bound micelles will be favoured since electrostatic repulsion is diminished whereas the increased hydrocarbon-water contact area is stabilized by the polymer Since hitherto the influence of charge has only been studied by comparing polymer-micelle interaction for SDS CTAB and Triton X-100,3 I e for surfactants with completely different headgroups too much emphasis has been placed on headgroup structure and size instead of on the role of charge proper 3 1 2 Aggregation Numbers3' It should be stressed that one should be careful not to link the oemrrenrc of polymer-micelle interaction too heavily to the stabilization of the micellesTherefore aggregation numbers have been measured to decide whether or not the absence of a reduction of the cmc points to the complete absence of polymer- micelle interaction The aggregation numbers of DDAO micelles at various degrees of protonation were determined by static fluorescence quenching 33 34 Our data (Table 3) on DDAO in the absence of polymer agree with those reported in the literatureThe aggrega- tion numbers of DDAO in water show a decreasing trend with increasing ,!? Enhanced electrostatic repulsion may account for the observation The slightly higher aggregation number at p = 0 47 compared to those at = 0 24 and 0 75 would be in accord with inter-headgroup hydrogen-bonding being maximal Table 3 Aggregation numbers of micelles of DDAO at various degrees of protonation in the absence and presence of polymers [surfactant] Polymer" mM 00 - 30 75 - 20 76 PPO 20 55' PVME 20 57' PEO 10 k 20 73' Polymer concentrdtion ta 0 5 g dL Ph 024 047 075 098 70 72 70 66 70 73 73 67 46 43 38 - 46 42 37 - 67 71 73 - * Cdhlated from the pH at the cmc using pK = 5 0 Calculated on the assumption that the cmc in the presence of polymer equals that in H,O The effect is too small however to exclude the possibility of an experimental artifact We emphasize that the possibility of systematic errors that may obscure a comparison is appreciably higher within a horizontal row ofTable 3 than within a vertical column The data in Table 3 nicely illustrate that an unperturbed cmc may have different origins In the case of DDAO/PEO at various degrees of protonation the unperturbed numbers (within confi- dence limits) obviously indicate the absence of interaction In the case of DDAOjPPO and DDAO/PVME at low degree of protonation in contrast the reduction in aggregation number definitely suggests polymer-micelle association Interestingly this interaction does not lead to stabilization of the micelle (because of the negligible effect on the cmc) most likely because of counteracting contributions to the total Gibbs energy from the changes in Gibbs energy of surfactant molecules and polymer upon transferring a mole of surfactant molecules from normal to polymer-bound micelles Steric hindrance between the hydrated non-ionic headgroups and polymer segments will be unfavourable whereas in the case of PPO or PVME the transfer of polymer segments to the micellar phase will be favourable Furthermore there will be no favourable loss of electrostatic repulsion like at higher ,8 The decrease in aggregation number in the presence of PPO and PVME becomes more pronounced at higher pThis is anticipated since a reduction in electrostatic repulsion by increasing the surface-to-volume ratio of the micelles will be more important at higher micellar chargeThe influence of PPO and PVME on the aggregation number is within the confidence limits equal even though dG& pol -dC& is clearly more negative for PPO than for PVME This may be rationdlized in terms of stronger hydrophobic interaction for PPO compared to PVME A slight difference in morphology of the polymer- micelle complex due to the lower molecule weight of PPO (MW 1000) compared to that of PVME (MW 27000) should also be taken into account 4 The Influence of Polymers on the Micellar Architecture of Cetyltrimet hylammoni um Salts3" Cetyltrimethylammonium salts (CTAX) particularly the bro- mide and chloride are by far the most widely studied cationic surfactantsThe formation of viscoelastic solutions at extremely low concentrations (ca 10 M) in the presence of salicylate anions is an especially fascinating phenomenon Notwithstand- ing these interesting properties CTAX salts as wel! ds the relatively few other cationic surfactants that have been investi- gated have a poor reputation in the field of polymer-micelle chemistry This stems from the fact that they give only signifi- cant interaction with rather hydrophobic polymers though recently a modest propensity for binding to more hydrophilic polymers has been detected 36 Three explanations have been advanced for the origin of the weakness of the interactions between polymers and cationic surfactants (I) the bulkiness of the cationic headgroup (11) a positive charge (videznfra) on the polymer and (111) a difference in interaction of cations and anions with the hydration sheath of the polymerThe first explanation was suggested by Sait~,~ and later adopted by Nagarajan The bulkiness of the headgroup of most cationic (and non-ionic) surfactants is assumed to hamper the presence of a polymer at the hydrocarbon-water interface Furthermore a bulky headgroup quite effectively shields the hydrophobic core Thus the stabilization of the core- water interface by the polymer is less relevant in that case than for micelles having a core less shielded by headgroups Small angle neutron scattering studies indeed reveal that the trimethyl- ammonium headgroup in micelles of n-tetradecyltrimethyl- ammonium bromide does not even leave enough space for penetration of water molecules between the headgroups 37 There are however several observations which cannot be reconciled with bulkiness playing a major roleThese include (I) DAC and protonated n-dodecyldimethylammineoxide (DDAOH+) micelles (vide supra) do not interact with PEO PVP and PVA or only very weakly and (11) the finding that the bulkiness of the hydrophobic polymers does not prevent inter- action with for instance CTAB which has a voluminous tri- methylammonium headgroup The second explanation 38 involves electrostatic repulsion with a proposed slightly positive charge on the polymersThis charge is thought to originate from protonation of the ether oxygens in the case of polyethers and the amide moiety in the case of PVP The pH dependence of the interaction between SDS and PEO was used by Schwuger8 to support this view Moroi and Sait~~~ used the same concept to explain the difference between DTAB and SDS in mixed micelle formation with non- ionic micelles of the poly(oxyethy1ene)alkylether type The very low pK value of an ether or amide [pK (CH,CONH,)H + = 0 31 however raises serious doubts about the importance of protonation at neutral pH The third explanation is based on the different influence of cations and anions on the hydration sheath of the polymer and thus of headgrouppolymer interactionTo support this view Wittelg refers to the work of Na~per,~~ who studied the role of electrolytes in the flocculation behaviour of polymers The decrease in clouding temperature of PEO by the addition of salts also indicates the more pronounced influence of anions com- pared to cations The ion-polymer interactions are usually thought to occur via hydration shell overlap effects Probably the size of the headgroup and the interaction of the headgroup with the hydration sheath of the polymer are the main reasons for weak interaction of cationic micelles with polymers The electrostatic repulsion between polymer and micelles may modify interactions at low pH but does not seem to be decisive under neutral conditions Overall only an appre- ciably hydrophobic polymer cdn overcome these fdctors by d favourable Gibbs energy of transfer of polymer segments from the aqueous to the micellar phase and interact also with cationic (and non-ionic) micelles 4.1The Sphere-to-Rod Transition of CTATS~~ So far only the interaction of spherical cationic micelles with polymers has been discussed but certain cetyltrimethylammo- nium salts are well known for the formation of rodlike (or wormlike) micelles Using the salicylate salt Hirata et af 40 have published electron micrographs of these rods Although their results have been criticized and are most likely artifacts asso- ciated with chemical staining the use of cryo-transmission electron microscopy avoids these artefacts and direct imaging of the rods has become feasible 41 From these direct images of the rods it appears that the diameter (45 to 60 A) is an agreement with expectations 42 Surprisingly Nagaragan' 5c is the only author previously to have considered rodlike micelles in the study of polymer-micelle interactions He predicted theoretically that rodlike micelles of SDS formed in the presence of NaCl would be transformed to polymer-bound ellipsoidal micelles in the presence of PEO One of the counterions that is able to induce the formation of rod-like micelles from cetyltrimethylammonium surfactants is tosylate (Ts) Sepulveda and co-~orkers~~ first introduced CTATs for the measurement of the degree of dissociation of CTAX in which X represents inorganic counterions Later they studied the rheology of solutions of CTATs and of other CTAX surfactantsThey also reported cmc values degees of dissocia- tion and the transfer Gibbs energy for the counterion from water to the micelle 43 The tosylate ion is less rod-inducing than the salicylate (Sal) ion As a result globular micelles of CTATs are initially formed above the cmc (2 6 x lop4M) 43bThese micelles start to grow at a critical rod concentration (crc) of around 15 mM Thus CTATs provides the possibility of studying the sphere-to-rod transition and the influence of polymers on the concentration at CHEMICAL SOCIETY REVIEWS. 1993 which this transition takes place In the case of CTASal rod-like micelles are formed directly above the cmc We have used viscosity measurements to obtain the concen- tration at which the sphere-to-rod transition of CTATs micelles occurs in the absence and presence of PVME For these rheological measurements we used a shear-viscometer that can be equipped with different measuring devices one having cone- and-plate geometry and the other cylindrical geometry Provided that the rheometer is also equipped with a special sensor the former device allows the measurement of first normal stress differences indicating viscoelasticity as well as shear stress from which the apparent viscosity can be calculated according to equation 4 apparent viscosity = shear stress/shear rate (4) The latter device only allows the measurement of shear stress but produces more accurate data Usually the apparent viscosity of a solution of rod-like micelles drops rapidly when the shear rate is increased Only at low shear rates (or at very high shear rates) is the viscosity Newtonian that is independent of shear rate InTable 4 these low shear (Newtonian) viscosities are listed for solutions con- taining various concentrations of CTATs in H,O and in the presence of 0 25 and 0 5 g dL-' of PVME (measured with cylindrical geometry) 35 For the highly viscous solutions shear rates as low as 6x 10 s have been used It is hard to associate the sphere-to-rod transition to a well-defined concen- tration since the viscosity increases non-linearly with the CTATs concentration (Figure 4)The viscosity of a 15 mM CTATs solution in H,O is already four times as high as that of water (1 cP) At 18 mM CTATs a first normal stress difference indicat- ing viscoelastic behaviour and thus the presence of rods can be observed above a shear rate of 476 s-(using cone-and-plate geometry) Such viscoelastic behaviour can also be observed visually as the recoil of trapped air bubbles when a swirling motion of the solution is abruptly stopped From 20 mM CTATs onwards thixotropic behaviour is definitely displayed using a cone-and-plate measuring device (between 119 s-and 476 s-l) and from 25 mM CTATs onwards using a cylindrical measuring device (between 60 s and 119 s ')Thixotropic behaviour is the occurrence of a decrease in apparent viscosity with increasing time under shear and is revealed in this case after a stepwise increase in shear rateThe thixotropy as well as the viscoelasticity and non-Newtonian behaviour are indicative of changes in the internal structure of the solution These changes originate from alignment and disruption of the rod-like micelles by shear forces 44 Although the transition concentration for CTATs cannot be clearly defined it seems obvious from Figure 4 that the presence of PVME shifts the sphere-to-rod transition to higher concent- rations However there may be a pitfall in this alluring conclu- sion In 1985 Hoffmann et af 45stated 'that all theories which try to explain the viscoelastic properties of micellar solutions on Table 4 Apparent viscosities of CTATs in aqueous solutions in the absence and presence of PVME at 25 "C 7app1 Pa s [CTATs] mM H2O O25gdl 'PVME 05gdl 'PVME 10 0 0017 0 0016 0 0017 15 0 0046 20 0 016 0 0047 0 0040 25 0 047 0 013 30 0 53 0 081 0 021 35 5 09 0 24 0 067 40 128 0 32 45 3 55 125 50 10 01 2 54 POLYMER MICELLE INTERACTIONS-J C BRACKMAN AND J B F N ENGBERTS 500 400 a0 S 300 cv1 0 2 200 100 0 0 10 20 30 40 50 [CTATos] mM Figure 4The viscosity at low shear rates (Newtonian behaviour) of CTATs in H,O (0),0 25 g dL-' PVME (m) and 0 5 g dL PVME (0)at 25 "C measured with cylindrical geometry (Extrapolation of the lines is based on the data from Table 4 ) models that are based on the existence of well-defined rods without taking into account the transient nature of the micelles sooner or later must fail' He illustrated this statement with the behaviour of n-tetradecylpyridinium salicylate and n-tetradecyl-ammonium salicylateThese compounds have similar cmc values critical rod concentrations and light scattering behav-iour suggesting that the micellar structuresand the interactions between them should also be similar In fact the viscosities of aqueous solutions of these two compounds differ by almost two orders of magnitude The differences between the structural relaxation times of the micelles was shown to lie at the origin of this difference For these surfactants the relaxation time stems from the kinetics of formation and dissociation of the micelle whether stepwise per monomer or via coalescence or fragmen-tation of the entire micelle and not from the rotation of the rods Since this relaxation time may be influenced by the presence of additives42 such as n-butanol or n-pentanol it is conceivable that the shift in concentrationwhere the viscosity increase of the CTATs solution takes place caused by PVME is also due to these kinds of effects and not to a shift in concentration of the sphere-to-rod transition However we submit that this is not the case (vide znfra) and that indeed a shift in transition concent-ration upon PVME addition takes place We propose that PVME preferentially binds to spherical micelles of CTATs for which the surface-to-volume ratio is more favourable for inter-action with the polymer Headgroupheadgroup repulsion and headgroup-absorbed polymer repulsion will be less compared to those for polymer-bound rod-like aggregates while the extra hydrocarbon core-water contact is stabilized by PVME When the CTATs concentration exceeds the saturation concentration of PVME free micelles will be formed which grow into rods upon increasing the concentration 46 4.2The Polymer-inducedTransition from a Non-Newtonian to a Newtonian F1~id~~q~~ Cetyltrimethylammonium salicylate (often prepared by addi-tion of sodium salicylate to CTAB solutions) is the archetype of a cationic s~rfactant~~that forms rod-like micelleseven in dilute (ca M) solutions At higher concentrations CTASal solutions become viscoelastic and behave in a strongly non-Newtonian mannerThe maximurn in viscosity lies at a [Sal-]/ [CTA+] ratio below A second maximum in viscosity is observed in the presence of an excess of salicylate ions for n-tetradecylpyridinium micelles 44 One does not need special apparatus to observe the high viscosity and viscoelasticity of such a curious mixture It is also easily seen that the presence of 0 5 g dL-l PVME or PPO completely eliminates the gel-like properties and reduces the viscosity to about that of water Addition of the more hydrophi-lic polymers PEO or PVP does not induce such a transition Although the change in the properties of the CTASal solutions induced by PVME or PPO strikes the eye rheological measure-ments were performed to quantify the effect The same shear viscometer as used in the study of CTATs was usedThe (apparent) viscosities of micellar CTAB solutions in the absence and presence of sodium salicylate polymers and low molecular weight additives are listed in Table 5 These values have been obtained using a measuring device with cylindrical geometry The CTAB/NaSal solutions whether or not in the presence of PVP ethanol or t-butanol and to a slightly lesser extent CTAB/NaSal/PEO (20k) exhibit genuine non-Newto-nian behaviour z e the apparent viscosities vary dramatically with changing shear rate (Table 5) Table 5 The effect of sodium salicylate and several monomeric and polymeric additives on the viscosity of a micellar CTAB [CTABI [Nasal] mM mM additiveu 25 25 PVME 25 PEO 25 15 25 15 PVME 25 15 PPO 25 15 PEO 25 15 PVP 25 15 EtOH 25 15 t-BuOH * [Additive] = 0 5 g dL Shedr rate = 0 2985 s viscosity CP 1 08 f0 02 1 510 f0 0006 1 26 f 0 02 277Lh8 8( 1 630 f0 006 I 080 f0 006 274,h 16 9' 28 1 7,h I5 4' 3055,h8 I' 2213.h 8 9 Shear rate = 477 6 s By contrast the apparent viscosities of CTAB/NaSal in the presence of PVME or PPO and of CTAB solutions without Nasal are orders of magnitude lower and are independent of shear rate indicative of Newtonian behaviour This polymer-induced transition from a non-Newtonian to d Newtonian fluid is like in the case of CTATs attributed to preferential binding of spherical rather than rod-like micelles onto the hydrophobic polymersThis is completely consistent with the reduction in aggregate size of CTAB micelles in the presence of PVME and PPO and the shift to higher surfactant concentrations for the sphere-to-rod transition of CTATs by PVMEThe hydrophilic polymers PEO and PVP do not bind CTAX micelles and therefore do not exert dramatic effects on the rheology of a solution of these aggregates 5 Conclusion Polymer-micelle interaction depends on several properties of the surfactant molecule such as the chemical nature geo-metry and charge Unfortunately an arbitrary combination of these factors is not easily obtained For instance it is hard to find an anionic surfactant without an oxygen-rich head-group or a cationic surfactant with many oxygen atoms and without a quaternary nitrogen atomTherefore it is not yet possible to formulate general rules concerning the relative importance of the properties mentioned above Cationic anio-nic and non-ionic surfactants have all been shown to undergo polymer-micelle interaction on the premise that the polymer is sufficiently hydrophobic The first non-ionic and cationic surfactant that interacts substantially with for instance PEO has still to be reported For cationic micelles a betaine like RN(Me,)CH,COOH may be found to interact with PEO in view of the favourable interaction between the ether linkage and the COOH moiety 92 6 References I L M Klotz in ‘The Proteins’ Vol I Part B ed H Neurath and K Bailey Academic Press New York 1953,p 727 2 M M Breuer and I D Robb Chem Ind (London) 1972,13,530 3 E D Goddard Colloids Surf 1986,19,255 4 S Saito KolloidZ 1957 154 lg 5 H Lange Colloid Polym ,Sci ,1971,243 101 6 H Arai M Murata and K Shinoda J Colloid Interface Sci ,1971 37,223 7 FTokiwa and K TSUJII Bull Chem Soc Jpn 1973,46,2684 8 M J Schwuger J Colloid Interface Sci ,1973,43,491 9 B Cabane J Phys Chem 1977,81 1639 10 M N Jones J Colloid Interface SCI 1967,23 36 11 (a)S Saito and M Yukawa J Colloid Interface Sci 1969,30 21 1 (b)S Saito and M Yukawa Kolloid Z -Z Polym ,1969,234 1015 12 K Shirahama and N Ide J Colloid Interface Sci 1976,54,450 13 G Perron J Franqoeur J E Desnoyers and J C T Kwak Can J Chem 1987,65,990 14 B Cabane and R Duplessix Colloids Surf 1985 13 19 15 (a) R Nagarajan and B Kalpakci in ‘Microdomains in Polymer Solutions’ ed P Dubin Plenum Press New York 1985 p 369 (b) R Nagarajan Colloids Surf 1985 13 1 (c) R Nagarajan Adv Colloid Interface Sci 1986 26 205 (d) R Nagarajan and B Kalpacki Polym Prepr (Am Chem Soc Div Polym Chem ) 1982,23(1) 41 (e)R Nagarajan J Chem Phys 1989,90 1980 16 Z Gao R E Wasylishen and J CT Kwak J Colloid Interface Sci 1990,137 137 17 (a)T Gilyani and E Wolfram Colloids Surf 1981 3 181 (b) T Gilyani and E Wolfram in ref 15(a),p 383 18 (a)R Zana J Lang and P Llanos in ref 15(a),p 357 (b)R Zana J Lang and P Llanos Polym Prep (Am Chem SOC ,DIV Polym Chem ) 1982,39(1) 39 19 (a)F M Witte and J B F N Engberts Colloids Surf 1989,36,417 (b)F M Witte Ph D Thesis University of Groningen 1988 20 E A Lissi and E Abuin J Colloidlnterface Sci ,1985 105 1 21 F M Witte and J B F N Engberts J Org Chem ,1987,52,4767 22 A Malliaris. J Phys Chem ,1987,91,6511 23 R Zana S Yiv C Strazielle and P Llanos J ColloidInterface Sci 1981,80,208 CHEMICAL SOCIETY REVIEWS. 1993 24 E Ruckenstein G Huber and H Hoffmann Langmuir 1987 3 382 25 D F Evans,D J Mitchel1,and B W Ninham J Phys Chem ,1984 88,6344 26 CTreiner and D Nguyen J Phys Chem ,1990,94,2021 27 J C Brackman,N M vanOs,andJ B F N Engberts Langmuzr 1988,4,1266 28 F M Winnik Langmuir 1990,6 522 29 H Hoffmann and G Huber Colloid Surf 1989,40 18 1 30 K Kameyama and T Takagi J Colloid Interface Sci 1990 137 1 31 J C Brackman and J B F N Engberts Langmuir 1992,8,424 32 J C Brackman and J B F N Engberts J Colloid Interface Sci 1989,132,250 33 N JTurroandA Yekta J Am Chem SOC 1978,100,5951 34 G G Warr and F Grieser Chem Phys Lett 1985,116,505 35 J C Brackman and J B F N Engberts Langmuir 1991,7,2097 36 K Shirahama A Himuro and N Takisawa Colloid Polym Sci 1987,265,96 37 S S Berr E Caponetti J S Johnson Jr ,R R M Jones and L J Magid J Phys Chem 1986,90 5766 38 Y Moroi H Akisada M Saito and R Matuura J Colloid Interface Sci 1977,61 233 39 D H Napper ‘Polymeric Stabilisation of Colloidal Dispersions’ Academic Press London 1983 p 141 40 H Hirata Y Kanda Y Sakaiguchi Bull Chem Soc Jpn ,1989,62 246 1 41 P K Vinson and Y Talmon J Colloidlnterface Sci ,1989,133,288 42 H Hoffmann and G Ebert Angeu Chem Int Ed Engl 1988,27 902 43 (a)D Bartet C Gamboa and L Sepulveda J Phys Chem 1980 84,272 (b)C Gamboa and L Sepulveda J Phys Chem 1989,93 5540 (c) C Gamboa and L Sepulveda J Colloid Interface Sci 1986,113,566 44 H Rehage and H Hoffmann J Phys Chem 1988,92,4712 45 H Hoffmann H Lobl H Rehage and I Wunderlich Tenside Deterg 1985 22 6 46 J C Brackman Ph D Thesis University of Groningen 1990 47 J C Brackman and J B F N Engberts J Am Chem SOC 1990 112,872

ISSN:0306-0012    

DOI:10.1039/CS9932200085

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

MELDOLA LECTURE* Reactions of Group 13 Alkyls with Dioxygen and Elemental Chalcogens: From Carelessness to Chemistry Andrew R. Barron Chemical Lab0 ra tories, Harvard University, Cambridge, Massachuse tts 02I 38, U.S.A. 1 Introduction Since the atmospheric oxidation of group 13 organometallics is usually uncontrollably fast, often resulting in spontaneous combustion, chemists handling alkyl derivatives of the group 13 metals aluminium, gallium, and indium have traditionally made every effort to limit oxidative decomposition by carrying out reactions under an inert atmosphere. If the supply of oxygen is restricted, however, alkoxide compounds are formed as a result of auto-oxidation via unstable alkylperoxide intermediates (equation l).l., M-R [M-OOR] -M-OR (1) The literature on auto-oxidation of organo-aluminium com-pounds is extensive, and has been thoroughly reviewed., In contrast, studies on the auto-oxidation of organo-gallium and indium compounds has been a recent development, the majority of work having been conducted by Alexandrov and co-worker~.~While these researchers have studied the reactivity of a variety of gallium and indium trialkyls with dioxygen and organic peroxides, have successfully isolated several examples of gallium and indium alkylperoxide compounds (e.g., Me,MOOR M = Ga, In; R = Me, But), and have studied the reactions of gallium and indium alkylperoxides with a number of organic substrate^,^ no structural data have been available for a group 13 alkylperoxide compound.Our interest in the reactivity of organo-gallium and indium compounds with dioxygen and consequently the elemental chalcogens arose as a result of experimental carelessness. Hav- ing investigated the photochemical decomposition of the hybrid organo-gallium compounds GaMeJBu'), -we were attempt- ing to prepare the indium analogues for study.5 However, our efforts to do so were severely hindered by difficulties in the preparation and handling of the alkyl precursor In(But),, a ~~~ ~~ Professor Andrew R. Barron received his B.Sc. (Hons) and A.R.C.S. degrees from Imperial College, University of London in 1983. He remained at Imperial to conduct his Ph.D and D.I.C. research under the supervision of Professor Sir Geoflrey Wilkin- son, graduating in 1986.After a brief sojourn as a postdoctoral ussociute with Professor Alan H. Cowley at the University of Texas, Austin, in 1987 he joined the faculty at Harvard University, where he is cur-rently an Associate Professor. In 1991 he received the Meldo- la medal and prize for his work with Group 13 organometal- lics, and is a recipient of a DuPont Young Faculty Fellowship and an Alcoa Prize Fellowship. His research inter- ests span both the inorganic chemistry and materials science of aluminium, gallium, and indium. 93 yellow crystalline solid which is extremely light sensitive, decom- posing in seconds under ambient light.6 The proven stability of the Lewis base complexes of group 13 trialkyl as compared to the parent compounds prompted us to synthesize the quinuclidene (quin) adduct of In(But),.The Lewis acid-base complex In(But),(quin) is a colourless crystalline solid which shows no propensity to photochemical decomposition. The accidental oxidation of a sample of In(Bu'),(quin) stored in pentane solution for six months in an inadequately greased Schlenck flask, resulted in the deposition of large colourless crystals, that were determined by X-ray crystallography to be the dimeric indium alkylperoxo compound [(BU~),I~(~-OOBU~)],!~ The stability of [(But),In(~-OOBut)]2, and its gallium analogue, towards both thermolysis and further oxidation prompted our investigation into the reactions of M(But), with dioxygen and also (the related reactions) with the elemental chalcogens.A review of these studies as well as related results from other laboratories are presented below. 2 Reactions of M(But), with Dioxygen The interaction of M(But), (M = Ga, In) with an excess (1 atm) of dry oxygen leads to the formation of the alkylperoxo com- pounds [(But),M(p-OOBut)], in essentially quantitative yield (equation 2).7,8 Alexandrov et al. have rep~rted~,~ that the thermally sensitive alkyperoxo compounds [Me2M(p-00R)], (M = Ga, In; R = Me, But) may be prepared by the reaction of MMe, and ROOH (equation 3). Attempts to synthesize [(But),Ga(p- OOBut)], and [(But),In(p-OOBut)]2 by an analogous route have, however, failed. MMe, + ROOH -)[Me2M(p-0OR)l2 + MeH (3) The t-butyl substituted compounds [(But),M(p-OOBut)], (M = Ga, In) are moisture sensitive, but are stable indefinitely under an atmosphere of dry oxygen at ambient conditions.The enhanced stability of [(But),M(p-OOBut)], as compared to the methyl derivatives is undoubtedly due to the steric hindrance provided by the group 13 t-butyl groups (see below). Alkylperoxo moieties are known to act as monodentate (TI) (l), bidentate (y2) (2), and bridging (p,) (3) ligands to transition metals. O * Delivered at the Dalton Division Symposium dt the Scientific Societies' Lecture Theatre, London, on 2 April 1992 Since the existence of dimers involving bridging alkoxides is a common feature in group 13chemistry, it would, therefore, be logical to expect the alkylperoxo group in [(But),M(p-OOBut)], (M = Ga, In) to ligate in a similar fashion.The p,-bridging mode of co-ordination of the t-butyl peroxide group and the dimeric nature in the solid state has been confirmed by X-ray crystallography, for both the gallium8 and indium7 compounds. The molecular structure of [(But),Ga(p-OOBut)], is shown in Figure 1; the indium analogue is iso-structural. The molecules consist of two (But),M fragments bridged by two p,-t-butylper- 0x0 groups, oriented in an eclipsed-staggered conformation, presumably so as to minimize lone-pair repulsion on the oxygen atoms. A similar t-butyl peroxide geometry was found in the X-ray structural study of the related [(C~,CCO,)P~(~-OOBU~)]~. Figure 1 The molecular structure of [(Bu(),Ga(p-OOBu')], complex Group 13 peroxides have previously been observed to undergo an auto-oxidation reaction in which one of the remain- ing alkyl groups is oxidized to an alkoxide by the peroxide ligand via either intra- (equation 4) or inter-molecular (equation 5) oxygen tran~fer.~ R,MOOR -+ RM(OR), R,MOOR + MR3-+2R,MOR A rapid inter-molecular oxygen transfer (cf.equation 5) appears to occur during the oxidation of Al(But), since the mono- alkoxide [(But),Al(p-OBut)], is the only product isolated. , However, no intra-molecular oxygen transfer reaction is observed for [(Bu~)~G~(~-OOBU~)],, even at elevated tempera- tures. Instead, heating in benzene solution results in the forma- tion of the mono-alkoxide dimer [(But),Ga(p-OBut)], with the concomitant oxidation of benzene to phenol (equation 6).8 Since the dimeric structure of [(But),Ga(p-OOBut)], is undoub- tedly similar to that of the less hindered methyl derivatives, the CHEMICAL SOCIETY REVIEWS, 1993 alcohols, and esters has been in~estigated,~ the isolation and crystallographic characterization of [(Bu~),G~(C~-OOBU~)]~, and the observation that [(But),Ga(p-OBut)], is the only gallium- containing side-product during the oxidation of benzene to phenol, provide an excellent system for gaining a better under- standing of the reactivity of group 13 alkylperoxides with organic substrates.In order to demonstrate the oxidative ability of [(But),Ga(p-OOBut)], we have investigated the model oxi-dation of phosphines to phosphine oxides.l4 3.1 Phosphines and Triphenylarsine The interaction of [(But),Ga(p-OOBut)], with two molar equi- valents of PPh, (in the absence of water) results in the formation of the Lewis acid-base adduct [(But),Ga(p-OBut)(O=PPh3)]in near quantitative yield (equation 7), hydrolysis of which gives uncomplexed O=PPh, as the only phosphorus-containing spe- cies, as determined by 31P NMR spectr~scopy.~~ + 2ER3 ~entane[(B~~)~Ga(p-00Bu')], 25 "C 2[(But),Ga(OBut)(O=ER3)J E = P.As (7) Similar phosphine and arsine oxide complexes [(But),Ga-(OBu')(O=ER,R')] (E = As, R = R' = Ph and E = P, R = Ph, R' = Me; R = R' = Et; R = R' = Bun; R = R' = Pr'), are obtained from the reaction of [(But),Ga(p-OOBut)], with AsPh, or the appropriate phosphine (cf.equation 7).14 Spectro-scopic data for these compounds are consistent with monomeric Lewis acid-base complexes of the (But),Ga(OBut) moietv (4). This structure has been confirmed for [(But),Ga(OBut) (O=AsPh,)] by X-ray crystallography (see Figure 2). l4 Figure 2 The molecular structure of [(But),Ga(OBut)(O=AsPh3)]. But '0I absence of significant auto-oxidation for [(Bu~),M(~-OOBU~)]~ E = (M = Ga, In) suggests that the oxygen transfer for less sterically hindered alkyl peroxides is inter- and non intra-molecular. We presume, therefore, that this enhanced stability towards auto- oxidation is due to the steric bulk of t-butyl substituents on the metal, which preclude the close approach of the two molecules of [(But),Ga(p-OOBut)], required for inter-molecular oxygen transfer. 3 Reactivity of [(Buf),Ga(p-OOBuf)], While the reactivity of group 13 alkyl peroxides with aldehydes, P, AS (4) The triphenylphosphine oxide complex may be prepared by an independent synthesis involving the phosphine oxide clea- vage of the Ga,O, unit in [(But),Ga(p-0But)], (equation 8). This_ reaction is indicative of the strongly Lewis-basic nature of O=PPh,, since the dimeric alkoxide is not cleaved by THF or pyridine.Triphenylphosphine oxide also cleaves the Ga-0-Ga bridges in [(But),Ga(p-OOBut)], to give the Lewis acid-base complex [(But),Ga(p-OOBut)(O=PPh,)l (equation 9). '0 MELDOLA LECTURE-A R BARRON ~[(Bu'),G~(~-OBU~)]~+ O=PPh, -+ [(Bu'),Ga(OBu')(O=PPh,)] (8) $[(B~')~Ga(p.-00Bu')],+ O=PPh, +[(But),Ga(OOBut)(O=PPh3)] (9) Unfortunately difficulties in isolating [(But),Ga(OOBut)-(O=PPh,)] precluded its structural characterization by X-ray crystallography However, based on spectroscopic characteriza- tion we propose that [(Bu~),Ga(OOBut)(O=PPh3)]exists as a four-coordinate monomer (5), and is the first example of a terminal alkyl-peroxide of gallium 'B uI 0 3.2 Phosphites In contrast to the interaction of phosphines with [(Bu'),Ga(p- OOBu')],, the oxidation of phosphites, P(OR), (R = Ph, Me), does not result in adduct formation, but a quantitative yield of [(But),Ga(p-OBut)], and the appropriate phosphate ~[(Bu~>,Ga(p-OOBut)],+ P(OR), --+ +[(But),Ga(p-OBu')], + (O=P(OR), (lo) This lack of adduct formation is in accord with the weaker donor ability of phosphates as compared to phosphine oxides 3.3 Diphosphines Whereas [(BU')~G~(~-OOBU~)],reacts under ambient con-ditions quantitatively with two molar equivalents of either PR, or P(OR),, under corresponding conditions only partial reac- tion occurs with 1,2-bzs-(diphenylphosphino)ethane [dppe Ph,P(CH,),PPh,] Interaction of [(But),Ga(p-OOBut)], with one equivalent of dppe, in pentane or benzene, results in the oxidation of half an equivalent of the dppe and the formation of the mixed alkoxide/alkylperoxide compound [(But),Ga(p-OBut)(pu-OOBut)Ga(Bu'),1(equation 1 1) When the reaction is carried out with two equivalents of [(Bul),Ga(p-OOBut)],, [(B~~),Ga(p-OBu~)(p-00Bu~)Ga(Bu~)~] the phosphineand oxide (dppeO,) are the only products isolated The structure of [(But),Ga(p-OBut)(p-OOBut)Ga(Bu'),l con-sists of two (But),Ga fragments bridged by one p,-t-butylperoxo and one p2-t-butoxo group The geometries of which are analo- gous to those found in [(But),Ga(p-OOBut)], and [(But),Ga(p- OBut)12 respectively l4 In contrast to the results observed for dppe, the reaction of two equivalents of [(B~~)~Ga(p-00Bu~)],with 1 ,I-his-(dipheny1phosphino)methane(dppm, Ph,PCH,PPh,) does not yield the free phosphine oxide, but a mixture of four products (equation 12), of which only one contains phosphorus This may readily be separated from the other products by fractional crystallization, and its structure has been deter- mined by X-ray crystallographic analysis (Figure 3) to be [(Bu1),Ga{(0)P(Ph),CHP(Ph),~}] This compound is also prepared by the direct reaction of Ga(But), with dppmO, n Figure 3 The molecularstructure of [(Bu'),Ga((O)P(Ph),CHP(Ph),O)] [(Bu'),Ga((O)P(Ph),CHP(Ph),OJ]2[(But),Ga(p-OOBut)], + + __.[(Bu~),Ga(p-OBut)(p-OOBut)Ga(But)~]dPPm + ~[(BU~),G~(~-OBU~>]~+ Bu'OOH (12) 4 Reaction of Ga(Buf), with Elemental Chalcogens Whereas the often violent reaction of group 13 alkyls with dioxygen has, for obvious reasons, been a subject of both practical concern and academic interest for many years, few studies have dealt with their reactions with the elemental chalco- gens The direct reaction of sulfur with trialkyl-aluminium compounds was first reported, in the patent literature,' to yield, after hydrolysis, small quantities of alkenethiols Further studies showed that, in contrast to the reactions with oxygen, elemental sulfur inserts into only one of the possible three aluminium- carbon bonds to afford good yields of dialkyl-aluminium thiolates l6 AIR, + E -R, Al(ER)E=S Se Reactions under more forcing conditions led to the formation of RSR and RSSR,' but the aluminium-containing products were not isolated Similar results have been reported for the interac- tion of aluminium alkyls with powdered selenium metal (cf equation 13) l8 Since the reaction of group 13 alkyls with dioxygen to give alkoxide products has been shown to occur via alkylperoxide intermediates, e g equation I,' l9 and the propensity for catenation is increased with the heavier group 16 elements, an interesting question may be posed is it possible that alkyldichal- cogenides are formed as transient intermediates in the reaction between group 13 alkyls and elemental chalcogens? Given our isolation of the thermally robust alkylperoxides of galliums and indi~m,~from the oxidation of the tri-t-butyl compounds (see above), this suggests that the use of sterically hindered alkyl substituents, such as t-butyl, should allow for the isolation of the chalcogenide analogues, i e ,equation 14 In addition, the presence of multiple allotropic forms of the chalcogenides raises a further question does the identity of the products from the reaction of group 13 alkvls with the elemental chalcogens depend on the allotropic ,form of the chalcogen employed? In order to address these questions, and to further understand the reaction chemistry of group 13 alkyls with the group 16 elements, we investigated the reaction of Ga(But), with elemen- tal sulfur, selenium, and tellurium.20 4.1 Sulfur The interaction of Ga(But), with cycloocta-sulfur (S,) at ambient temperatures results in the formation of a mixture of two colourless crystalline compounds, separable by fractional crystallization.They have been characterized by NMR spectros- copy and mass spectrometry as the bis-alkyldisulfido bridged compound [(But),Ga(p-SSBut)], (6) and the mixed bridge com- pound [(BU~),G~(~-SBU~)(~-SSB~~)G~(B~~)~](7).,O In the absence of a large excess of sulfur, [(But),Ga(p-SSBut)], slowly decomposes at room temperature to give the mixed bridged species and elemental sulfur.SBu' ButI I Thermolysis of [(But),Ga(p-SSBut)], in the presence of an excess of elemental sulfur yields the cubane compound [(But)- Ga(p3-S)I4 (equation 15),,O which may also be prepared by the reaction of Ga(But), with H2S via the hydrosulfido complex [(But),Ga(p-SH)], (equation 16)., The aluminium cubane compound [(But)Al(p3-S)I4 has been prepared by analogous routes, however, the alkyldisulfide or the hydrosulfide com- pounds were not observed.22 Ga(Bu'), + S,(xs) -,[(Bu')Ga(p,-S)I, + Bu'S(S),SBu' (1 5) n = 1, 2, 3, 4 Ga(But), + H,S -,[(But),Ga(p-SH)], (16)-,[(Bu1)Ga(p3-S)l4+ Bu'H The molecular structure of [(But)Ga(p3-S)], has been deter- mined by X-ray crystallography (Figure 4),, and consists of a distorted cubane core of four gallium atoms and four p,-sulfido groups.Given the analogies observed for gallium and iron aqueous chemistry it is perhaps not surprising that the gallium homologue of the Fe,S, cubes prepared as synthetic represen- tations for ferrodoxins is isolable. What is notable, however, is that [(But)Ga(p3-S)], [Ga-S = 2.359(3)w] is structurally similar not to the isoelectronic [(~-C,H,)Fe(p.,-S)], [Fe-S = 2.20(8)-CHEMICAL SOCIETY REVIEWS. 1993 2.26(4)81],,, but to the mixed valence (formally FelllFell,) trianions, [(X)Fe(p3-S)],3 -[Fe-S = 2.297(6)-2.351(9)A].24 Thermogravimetric analysis data indicate that [(But)Ga(p3- S], sublimes completely above 225 "C,at atmospheric pressure, making it suitable as a single source precursor for gallium(I1) sulfide.However, repeated sublimation at atmospheric pres- sure results in its conversion from a tetramer into an octamer [(BU~)G~S],.~~Although unable to obtain crystallographic structural data for the octamer, we proposed that, based on the chemistry of tin oxides,26 it adopts a drum structure (S), consisting of two fused 8-membered Ga,S4 cycles. . ,JBu Prolonged thermolysis of [(But)Ga(p3-S], in refluxing hexane results in its near quantitative conversion to a single new species, [(But)Ga(p3-S)I7 (Figure 5).27 The heptamer may also be formed in excellent yield (>SO%) by the solid state thermolysis of [(But)Ga(p3-S)I4 in a sealed tube at 175 0C.27Dissolution of [(But)Ga(p3-S)],(n = 4,7,S) in pyridine results in the formation of [(But)Ga(p-S)py], (see Figure 6), which is converted to the tetramer by sublimation, but forms the hexamer (Figure 7) during solid-state thermolysis. These topological rearrange- ments observed for [(But)Ga(p3-S)lx (as summarized in Scheme 1) are, if not unique in cluster chemistry (other rearrangements being accompanied by changes in speciation), certainly the most extensive and reversible for a single well-characterized species.It is likely that such cage transformations may be possible for a wider range of main group clusters, e.g.the iso-electronic aminoalanes, (RAINR'),, and the alkyl alumoxanes, (RAIO),. 4.2 Selenium In contrast to the results observed for s8, the reaction of Ga(But), with the analogous selenium allotrope, Sea, does not yield a stable alkyl diselenide, but results in the direct formation of the Ga,Se, cubane.20 Thus, the reaction of Ga(But), with red v Figure 4 The molecular structure of [(But)Ga(p3-S)I4. Figure 5 The molecular structure of [(Bu')Ga(p,-S)],. MELDOLA LECTURE-A R BARRON Figure 6 The molecular structure of [(Bu')Ga(p-S)py], V Figure 7 The molecular structure of [(Bu')Ga(p,-S)], Scheme 1 (1) A, hexane, 3-4 days, (11) sublimation, 1 atm, (111) A, pentane, 12h, (IV) A, pentane, (v) pyndine, mins, (vi) pyridine, 1-2 days, (vii) A, pyridine, (viii) vacuum sublimation, (ix) A, solid state selenium (Se,, a-cubic form), in pentane at room temperature, affords in essentially quantitative yield (based on 77Se NMR) Bu'SeSeBu', and [(Bu')Ga(p,-Se)], (equation 16) The cubane compound may also be prepared in near quantitative yield from the reaction of Ga(But), with H,Se in pentane at ambient temperatures, presumably via a hydroselenolate intermediate (equation 18) 28 Ga(Bu'), + Se, +[(Bu')Ga(p,-Se)], + Bu'SeSeBu' (17) ooc4Ga(But), + 4H,Se ~entane[(Bu')Ga(p,-Se)], + 8Bu'H (18) It should be noted that the alkylgallium selenide [(Et)GaSe], formed on the thermal decomposition of the triethylsilyseleno- late (equation 19) was proposed to have a highly polymeric structure based on its high thermal stability (m p 360-410 "C, decomp ) and low solubility in non-coordinating solvents 29 [(Et),Ga(p-SeSiEt,)], [(Et)GaSe], + GaEt, + Se(SiEt,), (19) Treatment of Ga(But), with metallic grey selenium (Se,) in pentane results in the formation of the selenolate compound as the only gallium-containing product (equation 20) 2o Subse-quent conversion of the selenolate into the cubane may be accomplished under more forcing conditions M(Bu')~+ E, +[(Bu'),M(~-EBu')]~ (20) M = Al Ga E = Se Te The cubane compounds [(But)M(p3-Se)14 (M = Al, Ga) may also be prepared by the reaction of selenium metal directly with the liquid trialkyl 22 4.3 Tellurium Unlike either sulfur or selenium, tellurium exists as d single allotropic form, that of a silvery-white semi-metallic trigonal structure isomorphous with grey selenium Thus, as may be expected, the reaction of M(But), (M = Al, Ga) with tellurium metal at room temperature proceeds in an analogous manner to that of grey selenium, resulting in the formation of the appropri- ate telluride dimer (cf equation 20) 2o The first example of an organogallium compound containing a Ga-Te bond, recently reported by Beachley and co-worker~,,~was prepared by the reaction of Ga(CH,Bu'),CI with the appropriate lithium telluride 2Ga(CH,But),CI + 2LiTePh Et20 -[(ButCH,),Ga(p-TePh)],T 2LiC1 (21) The reaction of either M(Bu'), or [(Bu'),M(p-TeBut)], (M = Al, Ga) with an excess of elemental tellurium in refluxing toluene proceeds in an analogous manner to that of its selenium counterparts yielding, in addition to a stoichiometric quantity of the organoditelluride, the telluride cubane (equation 22), which is spectroscopically similar to its sulfur and selenium analogues O 4M(But), + Te(xs) -[(But)M(p3-Te)], + 4Bu'TeTeBu' M=Al Ga (22) 5 A Possible Reaction Pathway of Group 13 Trialkyls with the Elemental Chalcogens The homogeneous nature of the auto-oxidation reaction of group 13alkyls has made it possible to obtain detailed mechanis- tic data concerning the reaction pathway Unlike organo- aluminium compounds the oxidation of the second and the third M-C bonds is not typical for gallium and indium trialkyls Thus, as noted in the Introduction, the controlled auto-oxi- dation of MR, (M = Ga, In) yields the dialkyl mono-alkylper- oxides, making any subsequent mechanistic study simpler Alexandrov has proposed that the auto-oxidation of gallium and indium trialkyls occurs by the following concurrent reaction R,MOO' + R'MR3 + 02 -+ [R3M 021< R,MOOR Unfortunately the heterogeneous nature of the reaction of MR, with the elemental chalcogens as well as the insolubility and uncharacterized nature of the products have precluded any equivalent study to those on the oxidation reaction However, as 98 discussed above, use of the sterically hindered t-butyl group allows for the isolation of a number of the intermediates, which in turn allows for the postulation of the reaction pathways Em (E = Se, Te) pentane, 25'C I E, (E= Se,Te) 1(Bu"l+GaBu' + (Bu')~G~(EEBu') I Em (E =Se,Te)I Scheme 2 Isolated species are highlighted 5.1 Metallic Selenium and Tellurium (Scheme 2) The initial reaction of M(But), with metallic selenium (tellur- ium) results in the insertion of a single Se(Te) atom into one of the Ga-C bonds to yield the isolable selenolate (telluro- late) 2o 22 At elevated temperature this product reacts further with the chalcogen to give an unstable transient alkyl diselenide (telluride), which decomposes to the di-t-butyl diselenide (tellur- ide) and a reactive organogallium fragment Dialkyl ditellur- ides have been shown to eliminate tellurium metal on thermoly- sis (equation 24)31 but under the conditions employed for the synthesis of [(But)Ga(p3-Te)], the reverse reaction does not occur RTeTeR --+ RTeR + Te(meta1) (24) Thus, the di-t-butylditelluride (and selenide) must be formed as a direct reaction product and not vza subsequent reactions with excess elemental chalcogen The organometallic fragment formed then reacts further to give the resulting selenide (tellur- ide) It is highly unlikely that this gallium fragment will be as written, z e Ga(But), but the formation of such a species on the chalcogen metal surface cannot be discounted, and we note that a number of examples of (RM), have been isolated for aluminium 32 E8 (E= S, Se) /I E = Se S8, toluene reflux v]+GaBu' I F Scheme 3 Isolated species are highlighted 5.2 Cyclo-Octasulfur and cyclo-Octaselenium (Scheme 3) The observation that [(But)Ga(p-SSBut)], is the first detectable product in the reaction of Ga(Bu'), with s8 poses the following question concerning the mechanism of sulfur atom insertion does the reaction proceed via the concerted insertion of two CHEMICAL SOCIETY REVIEWS, 1993 sulfur atoms to give the disulfide directly (equation 25), or a stepwise insertion of a single sulfur atom giving the thiolate which reacts further to give the disulfide (equation 26)7 Ga-R -?L.Ga-S-S-R (25) Ga-R -Ga-S-R Ga-S-S-R (26) If the latter, z e equation 26, is the preferred pathway then it is reasonable to expect that the thiolate [(But),Ga(p-SBut>l2 should react with excess sulfur to give both the disulfide compounds However, under analogous reaction conditions to those employed for the formation of the disulfides directly from Ga(Bu'),, z e pentane and room temperature, no reaction is observed, even over an extended period of time Thus, unlike the case for the metallic elements, the reaction of Sa(Se8) results in the insertion of two sulfur (selenium) atoms In the case of selenium the resulting alkyl diselenide is unstable and undergoes further reaction as shown in Scheme 3 The alkyl disulfide compound is stable, undoubtedly due to the high stability of polysulfur species, and only reacts with excess sulfur under forcing conditions, to give unstable alkyl polysulfides whose decomposition yields the dialkyl polysulfide Acknowledgments I thank all the graduate students, post- doctoral fellows, and undergraduates who have been in my research group at Harvard over the last five years, for their dedicated and skilful efforts, for their intellectual contributions, and for suffering my more bizarre ideas, in carrying out our research In addition, collaborations with Professors S G Bott (University of North Texas) and D L Lichtenberger (Univer- sity of Arizona) and Drs M Stuke (Max-Planck-Institut fur Biophysikalische Chemie) and J W Ziller (University of Cali- fornia, Irvine) have proved both invaluable and pleasurable 6 References 1 P B Brindley, 'The Chemistry of Peroxides', ed S Patai, Wiley, London, 1983 2 A G Davies, 'Organic Peroxides', ed D Swern, Vol 2, Chapter 4, Wiley, London, 1971 3 T G Brilkina and V A Shushunov, 'Reactions of Organometallic Compounds with Oxygen and Peroxides', Nauka, Moscow, 1966, 226 4 YuA AlexandrovandN V Chikinova, J Organomet Chem ,1991, 418, 1 5 (a)W M Cleaver, A R Barron, Y Zhang, and M Stuke, Appl Surf Sci , 1992, 54, 8, (b)Y Zhang, W M Cleaver, M Stuke, and A R Barron, Appl Phys A, 1992,55,261 6 D C Bradley, D M Frigo, M B Hursthouse, and B Hussain, Organometallics, 1988, 7, 1112 7 W M Cleaver and A R Barron, J Am Chem Soc ,1989,111,8966 8 M B Power,W M Cleaver,A W Apblett,A R Barron,andJ W Ziller, Poljhedron, 1992, 11, 477 9 Y A Alexandrov, N V Chikinova, G I Makin, N V Kovnilova, and V I Bregadze, Zh Obsch Khim , 1978,48,467 10 H Mimoun, 'Comprehensive Coordination Chemistry', ed G Wilkinson, R D Gillard, and J A McCleverty, Pergammon, Oxford, 1988, Vol 6, Chapter 1 11 D C Bradley, Adv Chem Ser , 1959, 23, 10 12 H Mimoun, R Charpentier, A Mitscher, J Fischer, and R Weiss, J Am Chem Soc 1980,102, 1047 13 M R Mason and A R Barron, unpublished results 14 M B Power, J W Ziller, and A R Barron.Organometallics, in press I5 (a)H Jenkner and A -G Kali-Chemie, Ger Pat 103 1306 (1958), Chem Abstr , 1960,54,17269 (b)R E Leech and J E Knap, Union Carbide Corp , US Pat 2998455 (1960), Chem Abstr , 1962, 56, 2333 16 L I Zakharkin and V V Gavrilenko, Izv Akad Nauk SSSR Otdel Khim Nauk, 1960, 1391 17 L I Zakharkin and V V Gavrilenko, Bull Acad Sci USSR Div Chem Sci , 1960, 1294 18 A P Kozikowski and A Ames, J Org Chem , 1978,43,2735 19 (a)J D Odom, in 'Comprehensive Organometallic Chemistry', ed MELDOLA LECTURE-A.R. BARRON G. Wilkinson, F. G. A. Stone, and E. W. Abel, Pergamon, Oxford, 1982, Vol. 1, Chapter 4. (b)J. J. Eisch, in ref. 19(a), Chapter 6. 20 M. B. Power, J. W. Ziller, A. N. Tyler, and A. R. Barron, Organome-talks, 1992, 11, 1055. 21 M. B. Power and A. R. Barron, J. Chem. Soc., Chem. Commun., 1992, 1315. 22 A. H. Cowley, R. A. Jones, P. R. Harris, D. A. Atwood, L. Contreras, and C. J. Burek, Angew. Chem., Int. Ed. Engl., 1991,30, 1143. 23 T. Toan, B. K. Teo, J. A. Ferguson, J. J. Meyer, and L. F. Dahl, J. Am. Chem. Soc., 1977,99,408. 24 P. K. Mascharak, K. S. Hagen, J. T. Spencer, and R. H. Holm, Innorg. Chim. Acta, 1983,80, 157. 25 A. N. MacInnes, M. B. Power, and A. R. Barron, Chem. Muter., 1992, 4, 11; A. N. MacInnes, M. B. Power, A. R. Barron, P. P. Jenkins, and A. F. Hepp, Appl. Phys. Lett., 1993, 62, 71 1. 26 R. R. Holmes, Acc. Chem. Res., 1989, 22, 190. 27 M. B. Power, J. W. Ziller, and A. R. Barron, Organometallics, 1992, 11, 2783. 28 (a) T. B. Rauchfuss, in ‘The Chemistry of Organic Selenium and Tellurium Compounds’, ed. S. Patai, John Wiley, New York, 1987, Vol. 1. (b)F. Bottomley and R. W. Day, Organometallics, 1991, 10, 2560. 29 N. S.Vyazankin, M. N. Bochkarev, and A. I. Charov, J. Organomet. Chem., 1971,27, 175. 30 M. A. Banks, 0. T. Beachley, Jr., H. J., Gysling, and H. R. Luss, Organometallics, 1990, 9, 1979. 31 A. E. D. McQueen, M. B. Parker, J. B. Mullin, and D. J. Cole- Hamilton, Chemtronics, 1989,4, 264. 32 C. Dohmeier, C. Robl, M. Tacke, and H. Schnockel, Angew. Chem., Int. Ed. Engl., 1991, 30, 564.

ISSN:0306-0012    

DOI:10.1039/CS9932200093

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

How Do Diesel-fuel Ignition Improvers Work? P. Q. E. Clothier B. D. Aguda A. Moise and H. 0. Pritchard Department of Chemistry York University Downsview Ontario Canada M3J 1 P3 1 Introduction An important descriptor of diesel fuel is its Cetane Number this is an indicator of the time delay between injection and sponta- neous ignition of fuel in a standard diesel engine running under specified conditions; the shorter the ignition delay the higher the cetane number. Thus those groupings of atoms within a hydro- carbon molecule that are beneficial in conferring a resistance to spontaneous ignition in a gasoline i.e. a high octane number are undesirable when they occur in a diesel fuel and vice versa. The cetane number scale uses two standard compounds cetane (n-hexadecane) defined as 100 and heptamethylnonane defined as 15 so that assuming linear mixing a 1:l mixture would have a cetane number of 57.5 a 1:2 mixture would have one of 43.3 etc. Long-chain paraffins tend to have high cetane numbers e.g. n-dodecane = 80 n-tridecane = 83 in addition to cetane itself = 100.On the other hand hydrocarbons containing benzene rings tend to have very low cetane numbers,' e.g. diphenyl = 2 1 diphenylmethane = 1 I and 172-diphenyleth-ane = 1. Extremely low cetane numbers are also found for hydrocarbons with a benzene ring carrying short-chain substi- tuents e.g. xylene = -10 and m-di-iso-propylbenzene = -12 but as the side chain becomes longer the cetane number rises to 26 for n-hexylbenzene and to 50 for n-nonylbenzene. Substances containing fused rings also exhibit very low cetane numbers e.g. a-methylnaphthalene = 0. A corollary is that the minimum spontaneous ignition temperatures for aromatic hydrocarbons are higher than for non-aromatiw2 Legislated National Standards usually require that the cetane Baltazar Agudu was born in Pasuquin Ilocos Norte Philippines in 1956; he obtained his B.Sc. in Agricultural Chemistry from the University of the Philippines at Los Banos in 1978 andhis Ph.D. in Chemistr-v,from the University of Alberta in 1986 on the charac- terization of steady states in reaction networks. Peter Clothier um born in Belleville South Africa in 1956 and came to Canada in 1973. He obtained his B.Sc. in Chemistry from Lakehead University in 1981 and his Ph.D. from York University in 1990 principally for an experimental study of the thermal explosions of gaseous methyl isocyanide. Avygdor Moise was born in Bucarest Romania in 1954 and emigrated to Canada in 1975. He obtained his B.Sc. in Chemistry and Physics from York University in 1980 and his Ph.D. in 1989 modelling the thermal ignition of methyl isocyanide in the gas phase. number of commercial diesel fuel shall exceed a certain value say 40. Most diesel engines do not perform well with fuels of cetane number below this for example in cold weather the difficulty of starting a cold engine increases as both the cetane number of the fuel and the temperature decrease. Diesel fuel is a distillate boiling between 150"C and 350 "C but this is not a unique description other fractions boiling within this range but meeting different secondary specifications comprise naphtha jet fuel kerosene and so on; an approximate classification is shown inTable 1. Within certain narrow limits the relative amounts of gasoline diesel and/or jet fuel and of heavier (heating) oils that can be Table 1 Classification of distillate fuels Classification Approximate boiling range Gases < 30°C Straight run gasoline 30-200 "C Light naphtha 30-1 I0 "C Heavy naphtha 80-200 "C Middle distillate fuels 135-360 "C Kerosene and jet fuel 145-280 "C Diesel fuel 160-330 "C Light fuel oil 2 15-360 "C Heavy fuel oil 290-400 "C Huw Pritchard was born in Bangor Wales in 1928 and studied Chemistry in the University of Manchester obtaining his B.Sc. in 1948 and his Ph.D. in 1951 for research in thermochemistry and quantum chemistry. He remained on the Faculty in Manchester until 1965 when he moved to the newly established York Univer- sity inToronto. In addition to thermochemistry and quantum chemistry his interests have included the application ojelectronic digital computers to problems in chemistry such as vibration- rotation relaxation and thermal explosions and the study offree-radical and of unimoleculur reactions in the gas phase. He was awarded a D.Sc. from the University of Manchester in 1964 and elected Fellow of the Royal Society of Canada in 1979. The present work was supported through a Co-operative Research and Development Grant to H.O.P. fundedjointly by the Natural Sciences and Engineering Research Council of Canada and by Esso Petroleum Canada. 101 obtained by simple fractional distillation are fixed and if there is a mismatch between this and what the market demands then either shortages will ensue or the legislated standards will not be met.3To compensate for this refinery processes have been developed whereby heavier 'gas oil' fractions are subjected to catalytic cracking and hydrogenation to give more gasoline and distillate but these cracked materials tend to be aromatic in nature; consequently they make good gasolines but poor diesel fuels. Mass-spectrometer analysis4 shows that they are much richer in alkylbenzenes as well as in 2- and 3-ring aromatic compounds; it requires much more severe hydrogenation con- ditions to saturate the aromatic rings. At the same time some countries have undertaken massive projects to extract oils from shale and tar-sands deposits typically bitumen is separated from the rock or sand and cracked at high temperature to form lighter materials; hydrogenation is then used to reduce both the nitrogen and sulfur content and to help to stabilize the cracked products. Again these materials possess much more aromatic character than does diesel oil from conventional sources; typical assays are shown inTable 2. Table 2 Comparison of composition (in wt%) of conventional diesel fuel with those of synthetic materials Conventional Diesel fuel Cracked Gas oil Synthetic Diesel Fuel Paraffins 39 19 17 Naphthenes Alkylbenzenes 2-Ring Aromatics 3-Ring Aromatics 34 18 8 1 16 34 28 3 37 36 8 2 Furthermore 'H and 13C NMR studies on the alkylbenzenes show that for the same molecular weight those found in conventional fuel have relatively few (1-3) longer aliphatic side-chains whereas those from synthetic sources have many (4-6) shorter side-chains several of them perhaps being methyl groups4Thus it is this preponderance of short side-chains rather than the increase in the aromatic content itself that is the principal cause of the lower cetane numbers for synthetic diesel fuels; also the high content of naphthalenic compounds in the cracked oil fractions makes them unsuitable for augmenting the diesel fuel and jet fuel supply.* A temporary palliative to the diesel fuel quality problem is the addition of ignition promoters either organic nitrates or organic peroxides but if the cetane numbers of the available fuels were to continue to decline in the future the cost of the additive would become significant for example to raise the cetane number of diesel fuel from 35 to 40 by addition of iso-octyl nitrate would add about 10% to the untaxed cost at present-day price^,^ and release many unwanted tonnes per annum of NO into the atmosphere in major urban areas.Thus a better understanding of how these additives work is becoming essential. 2 Present Knowledge of Additive Behaviour A wartime study on the effect of 72 different additives (including 26 nitrates and 9 peroxides) on a selection of 10 different fuels found quite generally that nitrates and peroxides caused substantial increases in cetane number whereas for other types of additive no clear patterns emerged for example a few amines accelerated ignitions weakly others were slightly inhibiting.The general consensus from engine measurements is that a given additive is not equally effective in all fuels that the lower the cetane number of the fuel the more additive is needed to bring * We note in passing that high aromatic content is undesirable in jet fuels principally because the higher luminosity of the flames and their greater propensity for soot formation adversely affect engine reliability and maintenance costs. CHEMICAL SOCIETY REVIEWS. 1993 about the same increase in cetane number and that the presence of an ignition improver may reduce the cycle-to-cycle variability in engine behavio~r.~-~-~ One of these studies,' using four additives and three different fuels concluded that the improve- ment in cetane number correlated with the number of free radicals produced by the thermal decomposition of the additive during the pre-ignition period. There are severe disadvantages inherent in trying to study chemical kinetic processes in a reciprocating engine not only is it a dirty environment with oil droplets abraded metal soot and with combustion residues left from the preceding ignition stroke but the temperature and pressure are changing all the time. A more controlled way is to use an instrumented steel bomb held at a fixed temperature and charged with a known pressure of air into which the fuel is injected as a spray; in this way it is convenient to observe the variation of ignition delay with temperature pressure and fuel formulation and also to determine the minimum temperature at which each fuel will ignite. O There are many examples where ignition of fuel droplets has been studied under controlled conditions in a shock la including one in which the effects of additives were studied.' lh The shock tube method is limited however because if one wants to simulate the low-temperature conditions that occur in trying to start a cold engine the costs for driver gases become prohibitive. Another powerful method of determining ignition delay is the use of a rapid compression-ignition machine but its use is limited to gases' and volatile fuels such as octanes. Griffiths and co-workers have made an extensive study of the effect of isopropyl nitrate and of di-t-butyl peroxide on the spontaneous ignition of n-butane in the 700-900 K range and of di-t-butyl peroxide on the ignition of methanol. '2a,14aThey argue strongly that the most important factor is the heat released by the combustion of the additive in the pre-ignition phase and to some extent the number and nature of the radicals formed; they estimate that a temperature rise of the order of 60 "C would be necessary to cause the same acceleration in ignition rate as the presence of 1% of di-t-butyl peroxide.14a.b In our engine (see below) with a cylinder capacity of 460 ml and an injection of 0.015 ml of liquid fuel per stroke the complete combustion of di- t-butyl peroxide present in 1 % concentration if released homo- geneously would raise the gas temperature by about 15 "C;9b but di-t-butyl peroxide is effective even at 0.25% or corresponding to less than a 4 "C rise surely within the cycle-to- cycle variability and therefore insignificant. We have made engine measurements with several additives but at lower temperatures than usually found in a normally operating diesel engine.g Ignition delay fell monotonically with additive concentration (up to 10% additive) with no correlation between efficiency and the rate of thermal decomposition of the additive nor with its normal heat of combustion but this may not be a contradiction since Griffiths and co-workers suggest that it is the heat of partial oxidation that is important.12" Thus the present state of our understanding is very spotty there is general agreement that organic nitrates and organic peroxides stimulate ignition in a diesel engine and that the more additive the shorter the ignition delay. Other things being equal the effect of a given additive is different for different fuel Oa and different for the same fuel under different engine conditions.6 Several plausible correlations between effi- cacy and fundamental kinetic or thermal properties have been suggested but none of them is without exception. 3 Experimental Details Since our primary interest originally was in the cold-starting problem there were only two previously tried choices -to use a bomb as discussed above,1° or to use an electrically driven refrigerated diesel engine. s' We chose instead a simpler option of running one cylinder of a car engine in diesel mode with the diesel injector replacing the spark plug thus with a compression HOW DO DIESEL-FUEL IGNITION IMPROVERS WORK7-H 0 PRITCHARD ET AL ratio of 7 5 and a block temperature of 100°C compression temperatures in the 350-400 "C range typical of those encoun- tered when trying to start a cold high-compression diesel engine could routinely be achieved 9a Our engine does not mimic the true conditions too closely -for although the peak temperatures are about right the pressures are a factor of 2 or 3 too low the temperature gradients are different because the cylinder walls are at 100"C instead of say -20 "C,and of course hot rather than cold fuel is sprayed into the heated air Nevertheless probably because ignition delays follow rather similar patterns as a function of temperature at different pressures Oa we found a good correlation between ignition delay and cetane number CN = 91 -6 ~in the region 3 dT d~8 ms ignition delays of between 1 and 2 ms would be observed for these same fuels in a proper diesel engine In our engine the ignition delay is quite sensitive to the diameter of the air intake to the instrumented cylinder and this was chosen to make the ignition delay for ordinary diesel fuel about 7 5 ms Figure 1 shows pressure traces for diesel fuel and for fuel with 1 % of iso-octyl nitrate (Zethylhexyl nitrate) , I" 1. I Tune Figure 1 Ignition traces for (a) neat diesel fuel (b) diesel fuel containing 1% of iso-octyl nitrate Upper traces cylinder pressures at top-dead- centre (first peak of each doublet) the excess pressure above baseline is -12 5-13 atm (180-190 psi) Lower traces composite of timing markers generated from spark signals and pressure at diesel injector pressure peaks are -240 atm (3500 psi) Ignition delay is taken to be the time interval between the initial downslope of the injector pressure trace and the beginning of the steep pressure rise of the cylinder pressure trace Three different types of fuels were studied regular diesel fuel cetane number 42-45 low-sulfur low-nitrogen kerosene cetane number 35-40 and HT-CLHO (Hydro-Treated-Cata- lytic-Light-Heating-Oil) cetane number 12- 15The former was purchased locally but the latter two were supplied by Esso Petroleum Canada each material was acquired in substantial quantity but as to be expected from the experience of others every batch exhibited a different ignition delay and sensitivity of that delay to the presence of various additivesThis variability in stock fuels although undesirable could not be avoided since each engine run required a minimum of 300 ml of fuel to wash out the fuel pump and connecting lines and to record data from 15-20 ignition strokes In the case of the kerosene and heating oil samples they were blended with cetane in order to bring the ignition delays into our accessible range 4 Which Manifestation of Aromatic Character is Important? There are two properties possessed by aromatic hydrocarbons which could render them more difficult to ignite than paraffins The first is the ability of the benzene ring to add oxygen atoms to form a phenol e g O(,P) + benzene -,phenol O(,P) + naphthalene - a-or p-naphthol thereby interfering with the growing chains in the pre-ignition phase,16 the rates of many such reactions have now been studied and they have rather small activation energies l7The other is that the benzylic hydrogen atom adjacent to the benzene ring is readily abstractable and since the resulting benzyl radical is very unreactive we have an alternative mecha- nism for breaking the chains necessary to initiate the ignition It is relatively easy to distinguish between these two In our earlier study,9a we found that diethyl ether admitted to the air intake would stimulate the ignition of diesel fuel but that this stimulation was less strong if the fuel contained toluene ethyl- benzene behaves similarly but benzene and naphthalene do not exhibit any interference with these ether-assisted ignitions Thus suspicion falls more heavily on the benzyl groups than on the mere presence of the aromatic rings 5 Does the Important Additive Chemistry Occur in the Liquid or the Gas Phase? It is widely accepted that the important kinetic processes which lead to spontaneous ignition in a diesel engine take place in the gas phase,' reduction of the fuel spray velocity below a critical value and evaporation occur on a shorter time scale l8 How-ever the additives are unstable substances and it is possible that their oxidation or other reaction in the liquid droplets could generate sufficient heatgb to cause a micro-explosion a common occurrence in higher-temperature sprays alternatively reac- tive species initially formed in the higher-density liquid phase could survive through the evaporation process First we present some additional circumstantial evidence in support of the conventional view that without additives the important spontaneous ignition processes occur in the gas phase We observed extreme sensitivity of the ignition delay to the presence of very small amounts of other gases introduced into the air-intake stream entrainment of 20 ml/min of either C1 or of NO in the air stream to the instrumented cylinder reduced the ignition delay for diesel fuel from 7 4 to 5 2 ms Since the air flow per cylinder at 800 rpm in a 6-cylinder 2 75 litre engine is approximately 180 l/min these additives amount to just over 100 ppm Likewise ozone present in a concentration of about 10 ppm reduces the ignition delay from 7 4 to 6 5 ms a similar effect with ozone has also been reported independently by others recently *O It seems very unlikely that these minute traces of reactive gas could affect any chemistry that was occurring in the liquid phase Convincing evidence that the additives also function in the gaseous phase was obtained from the following experiments In trying to synthesize peroxy radicals in situ we made use of the sequence of reactions di-t-butyl peroxide -,2Bu'O' Bu'O' -,CH,' + acetone CH,' + HCl + CH + C1' C1' + C4H90-0 C4H9+ HC1+ 'C4H8 0-0 C4H9 'C,H 0-0 C4H -,C,H,(isobutene)+ '0-0C,H,(Bu'OO') HCl was produced by thermal elimination from t-butyl chloride The results were as follows ignition delay for the fuel was 7 4 ms falling to 5 9 ms with 1% of di-t-butyl peroxide added and to 5 1 ms when both di-t-butyl peroxide and t-butyl chloride were present however if the peroxide was present in the fuel and 20 ml/min of either gaseous t-butyl chloride or of HCl was included in the air stream the ignition delay fell to 4 5 msThe same amount of t-butyl chloride on its own either in the fuel or in the air stream gives a smaller improvement in ignition delay to the 6 3-6 5 ms range A similar conclusion can be reached by considering the effect of iso-octyl nitrate on the ignition delayThe thermal decompo- sition of iso-octyl nitrate leads to the formation of NO some formaldehyde and some rather inert radicals 9bThe equivalent amount of formaldehyde produced by the thermal decompo- sition ofs-trioxane present in the fuel only improves the ignition delay to about 6 4ms whereas 1YOof iso-octyl nitrate in the fuel causes the ignition delay to fall to 49 ms With untreated fuel and the same volumetric flow in the air stream of either NO or of NO as would have been produced by the complete decompo- sition of 1Yo of iso-octyl nitrate the ignition delays were within 0 1 ms of that found for iso-octyl nitrateThis suggests not only that iso-octyl nitrate works primarily in the gas phase but that the principal function+ of the iso-octyl nitrate is to generate NO2 6 One Mechanism or Many? As our investigation proceeded we began to unearth hints that the additives worked in different ways whereas the previous focus has always been to findTHE mechanism by which additives stimulate ignition Figure 2 shows the results of ignition measur- ements on untreated diesel fuel on fuel containing 1% of di-t- butyl peroxide 1YOof iso-octyl nitrate and 0 5% of eachThe trace for the neat fuel exhibits considerable cycle-to-cycle varia- 1%ION (4.9 ms) 0.5% ION + 0.5% DTBP (5.5 ms) X 1%DTBP (5.9 ms) X neat diesel fuel (7.4 ms) Ignition Delay qgn/ms Figure 2 Distribution of ignition delays for neat diesel fuel diesel fuel containing 1% di-t-butyl peroxide 0 5% each of iso-octyl nitrate and of di-t-butyl peroxide and 1% of iso-octyl nitrate each cross repre-sents one ignition Notice the smaller spread in the times for the fuel containing both additives t If one conjectures that the peroxide labelled P5inTable 3 were to decompose thermally into CO and a BulO. radical then the third fragment would be the same octoxy radical as is initially produced in the thermal decomposition of iso-octyl nitrate since this peroxide is very much less effective than is iso-octyl nitrate this supports the notion that it is the NO that is important CHEMICAL SOCIETY REVIEWS 1993 bility as to a lesser extent do those with 1% of additive however with the mixed additives the cyclic variation is consi- derably reduced Hence there must be at least two factors leading to variability in ignition delay one affected by iso-octyl nitrate the other by di-t-butyl peroxide In our survey of the effects of additives we found one commer- cial sample of low-sulfur kerosene (lamp fuel) for which di-t- butyl peroxide gave absolutely no improvement in speed of ignition whereas iso-octyl nitrate did also if di-t-butyl peroxide was added to the fuel containing iso-octyl nitrate there was no change in ignition delay confirming that these two additives act independently of each other Eventually we were able to repro- duce this behaviour by treating our sample of low-sulfur low- nitrogen kerosene with silica gel to reduce further its aromatic sulfur and nitrogen content 21Thus it appears that di-t-butyl peroxide is not involved at all in the hydrocarbon ignition chemistry at our low temperature but that it counteracts some sulfur or nitrogen compounds present in the fuel A third useful piece of evidence comes from chemical treat- ment of the fuels which will be discussed in more detail below When HT-CLHO is treated with ozone its cetane number is raised from 12 to about 15 likewise if it is treated with lead tetra-acetate the cetane number also rises to near 15 However if the fuel is treated successively with both reagents in either order the cetane number goes to about 18 It is clear that there must be two kinds of inhibiting material present one affected by ozone and the other by lead tetra-acetate 7 Other Evidence from the Use of Mixed Additives Figure 2 appears to show that the effect on the ignition delay of using a half-and-half mixture of iso-octyl nitrate and di-t-butyl peroxide is midway between that of using either additive on its own On the other hand it has been reported recently that a mixture of iso-propyl nitrate and di-t-butyl peroxide is more effective than either additive by itself implying some synergy between the actions of the additives 14( We found several striking counter-examples as shown inTable 3 apart from the last entry there appears to be strong interference between the reactions of the various pairs of peroxides the interference between di-t-butyl peroxide and the peroxide P4 was also reported without comment by Griffiths and co-workers 14( Table 3 Ignition delays for peroxide additives alone or mixed with di-t-butyl peroxideThe ignition delay for the untreated fuel is 7 4ms Peroxide 1% DTBP 1% Pn 0 5% of each PI 5 9 ms 5 2ms 66ms P2 5 9 ms 54ms 64ms P3 5 9 ms 5 8 ms 64ms P4 5 9 ms 6 1 ms 6 9 ms P5 5 9 ms 6 1 ms 5 8 ms Key to peroxides DTBP = di-t-butyl peroxide P1 = t-butyl perbenzoate* P2 = 1,l -di-(t-butylperoxy)-3,3,5-trimethylcyclohex~ne*P3 = n-butyl-4,4-bis(t-butylperoxy)-vdlerateP4 = 2,5-dimethyl-2,5-di-(t-butylperoxy)-hexaneP5 = 0 O-t-butyl-0-(2-ethy~hexyl)-monoperoxy-cdrbondle Notice that the two peroxides marked with dn asterisk were included in our previous st~dy,~~ but with that fuel sample they were somewhat less effective than wds di-t-butyl peroxide whereas now they are quite superior This interference between pairs of peroxides essentially des- troys the thesis that the heat of combustion of the additive partial or otherwise is an important contributor to the ignition process the mixture will not have a lower heat of combustion than if the two substances were burned separately HOW DO DIESEL-FUEL IGNITION IMPROVERS WORK"-H 0 PRITCHARD ET AL 8 Modelling Calculations The modelling of the combustion of simple hydrocarbons is now an active area of research with much emphasis on the difference between branched and unbranched hydrocarbons in relation to engine knock 22 Less well advanced are the topics of aromatic and of ignition of hydrocarbons at hydrocarbon ~xidation,~ the low temperature of interest to us here In order to investigate what might be the mechanisms by which alkyl nitrates and peroxides stimulate ignition in the very-low temperature regime we undertook to study their possible effects on the ignition of methane at 400 "C Bone and Gardner found that at 1 atm and 400"C the induction time for ignition of a 2 1 mixture of CH 0 was about 20 minutes but in the presence of NO it was reduced to a few seconds We used the 35-step mechanism of Seshadri and which appears to work well over a wide temperature range down to 1300 K and computed an ignition delay of 9300 seconds at 1 atm and 900K but no ignition at 800 K Since Bone and Gardner admit that their gases contained impurities and since the calculated ignition delays will be extremely sensitive to small uncertainties in the assumed rates under these conditions we regarded this as a satisfactory result upon which to base an exploratory study of the effect of additives on low-temperature ignition Extending the mechanism26 to allow for the addition of 0 1 mole O/O of NO the ignition delay at 900 K fell to 36 seconds in fact ignition would now occur down as far as 700KThe principal mechanism by which NO does this is the abstraction of an H atom from CH to give HNO which then dissociates into OH + NO the NO is recycled to NO and the OH takes part in the ignition process whether the initial additive is NO HNO or NO does not make a great deal of difference and they can probably be regarded as equivalent HNO is also readily formed in a rapid sequence whereby NO successively strips H atoms from any intermediate CH,O first to give CHO and then CO since formaldehyde is formed in the thermal decomposition of iso-octyl nitrate,96 this probably contributes to its superiority as an ignition improver We next modelled the effect of di-t-butyl peroxide on these ignitions using the set of reaction rates given by Griffiths and Phillips 27The ignition delays with 0 1 mole Yo of additive were 11000 140 27 and 8 seconds at 600 700 800 and 900K respectively these are to be compared with a delay of nearly 10000 seconds at 900 K in the absence of additive In all cases there is an instantaneous rise of about 20T caused by the oxidation of the di-t-butyl peroxide followed by a two-stage Ignition most clearly separated near 800K here there is an additional rise of 100°C during the first two seconds with ignition occurring 25 seconds later We conclude that although the oxidation of di-t-butyl peroxide is highly exothermic and causes an initial temperature rise most of the effectiveness of di- t-butyl peroxide stems from chemical origins varying the rate of decomposition of di-t-butyl peroxide had no effect which is sensible since all of the peroxide is decomposed in less than a millisecond at any of these temperatures The ignition delay is by far the most sensitive to the rate assumed for the reaction CH + CH,COCH -* CH + CH,COCH the latter radical being a source of ketene CH and HO but only above 800 K 120 In fact a methane-oxygen mixture con- taining 0 2 mole YO of acetone starting at 830K follows virtually the same track as the 800K 0 1 mole YOdi-t-butyl peroxide case thus the 100"C rise which is a major cause of the acceleration comes from the reactions of acetone These superficial studies of the effect of additives on the very- low temperature ignition of methane and oxygen which were carried out by using the CHEMKIN routines,2s provide some This suggests that acetone itself should accelerate theignitions it does both in our engine and also under normal diesel operating conditions but it is about an order of magnitude less efficient than di-t-butyl peroxide pointers as to how NO and NO (and indirectly alkyl nitrites and nitrates) accelerate ignition they also tend to support our observation that di-t-butyl peroxide does not by itself acceler- ate hydrocarbon ignition in our low temperature regime More- over they demonstrate conclusively that the acceleration that results from the simultaneous presence of di-t-butyl peroxide and HC1 has nothing to do with the possible formation of Bu'OO. radicals but is caused by the presence of C1 atoms 9 ChemicalTreatments of the Fuel There are several reagents that are specific for benzylic hydrogen atoms 29 We chose lead tetra-acetate and found that stirring 300 ml of diesel fuel with 3 g of lead tetra-acetate at 55 "C for 3 hours reduced its ignition delay from 7 4 to 6 3 ms correspond- ing to an increase In cetane number from about 44 to about 50 The unused lead tetra-acetate was treated with formic acid and from the volume of CO recovered it was estimated that about 0 2-0 3 mole YOof the fuel had been reacted on the assumption that its average molecular weight is about 250 Analysis of the fuel showed that there was no change in sulfur content but that the lead content increased from undetectable to about 6 ppm even up to 40 ppm when the lead tetra-acetate was synthesized in situ from red lead and glacial acetic acid We examined the reactions of several aromatic hydrocarbons under the same conditions among them phenyl-cyclohexane m-di-isopropylbenzene 9-methylanthracene 9-ethylfluorene In general very little occurred for example with 1-phenylhex-ane about 1YOwas oxidized to 1-phenyl-1-hexanone and about 0 1% to 1-phenyl-1-hexanol neither of which would be suffi- cient to cause the improvement in ignition quality that was observed 9 We will return to the question of what lead tetra- acetate does to the fuel later but it is clear that it neutralizes In some way some very powerful ignition inhibitor that is present in 0 2-0 3 mole YOconcentration It is known that passing ozone through diesel fuel will improve its cetane number by some 5-10 units with at the same time a moderate reduction in sulfur content 30 With our samples of diesel fuel and heating oil ozone is completely absorbed initially until about 5 g of ozone per litre of fuel has been consumedThey turn dark in colour and deposit black resinous and odorous materials which are soluble in acetone or glacial acetic acid but not in hydrocarbons ozonized heating oil even yields a copious precipitate when mixed with cetane for the engine tests! On analysis the deposits were found to contain 3- 4 times as much sulfur as did the original fuel we also confirmed the reduction in sulfur content of the fuel itself as shown in Table 4 Conversely low-sulfur low-nitrogen kerosene absorbs very little ozone it remains clear and gives pale brown transpar- ent deposits Kittrel et a/ used a complicated recovery process,3o but we simply ozonized the fuel filtered off the precipitated material and made the ignition delay measurementsThis procedure has its disadvantages as the ozonized fuel continues to put down sticky residues upon standing over long periods (months) thus without careful filtration immediately before use and diligent cleaning of the injector pump immediately after use blockages in the pump or the pintle injector can occur We presume that Table 4 Effect of ozone on ignition delay and sulfur content of diesel fuel Function Base fuel 0 25 g OJlitre 4 g O,/litre Total sulfur 0 21% 0 2 1%o 0 12% Ignition delayThiol 74ms 12 PPm 74ms 6 PPm 57ms 4 PPm $ Note also that any significdnt degree of oxidation of the fuel itself would hdve the undesirable result of reducing its hedt of combustion this is mainly due to the formation of peroxides as both diesel fuel and heating oil still contained 15 ppm of peroxide 4months after ozonization The ignition delay is unaffected at low levels of ozonization but at about 4 g OJIitre the improvement in ignition delay corresponds to an improvement in cetane number from about 44 to about 54 further ozonization causes a lengthening of the ignition delay to some 6 2-6 4ms representing an apparent reduction in cetane number Also unlike the heating oil case the effect of ozone and lead tetra-acetate are not cumulative treatment with both reagents in either order gives intermediate ignition delays of about 6 0 ms it is clear that several conflicting effects must be at work Another useful observation is that the treatment of diesel fuel with DDQ (2,3-dichloro-5,6-dicyano-p-benzoquinone)at room temperature reduces the ignition delay by about 1 ms about the same as does lead tetra-acetate At room temperature this reaction can only be a dehydrogenation and suggests that perhaps the principal result of treatment with lead tetra-acetate which we have found has little effect on a selection of alkylben- zenes could be a dehydrogenation tooTreatment with both lead tetra-acetate and DDQ in either order is no better than either treatment on its own Whilst not exactly a 'chemical' treatment it has been known for a very long time that adsorbents like silica gel and alumina will remove polar materials from diesel fuel Filtration of our diesel fuel through silica gel yields an almost colourless liquid having the same ignition delay as that which had been treated with lead tetra-acetate or with DDQTreatment of this filtered material with either lead tetra-acetate or DDQ did not produce any further reduction in ignition delay showing that the silica gel had removed all of the inhibiting materials with which they would reactThe adsorbed materials could be completely reco- vered by Soxhlet extraction with boiling acetone to give a dark odorous liquid addition back to the filtered fuel restored it to its original appearance odour and ignition delay Similar results were obtained with alumina as the adsorbent 10 A Search for Possible Inhibitors The results ofTable 4clearly indicate that sulfur compounds should be screened for potential ignition inhibitors also we should not overlook the possibility of finding a hydrocarbon inhi bitor 10.1 Hydrocarbons The least readily ignitable of the aromatic hydrocarbons we mentioned earlier had slightly negative cetane numbers Such numbers are usually deduced by blending the substance with cetane measuring the cetane number of the mixture and calculating the unknown cetane number by a linear mixture rule formula Since our inhibitor is present in only 0 2-0 3% concentration this would mean that we are looking for a hydrocarbon with a blending cetane number of (say) -400 which is quite unheard of We considered dibenzosuberane to be a possible candidate this molecule has three benzyl groups and could easily be dehydrogenated to form a conjugated bridge between the two benzene rings Addition of 1% of dibenzosu- berane to diesel fuel increased the ignition delay by less than 0 1 ms well within the repeatability of our measurementsThis means that dibenzosuberane cannot have a blending cetane number much worse than that of m-di-isopropylbenzene some- where in the -10 to -20 range and we conclude tentatively that no hydrocarbon exists that could be sufficiently inhibiting to match our requirements 10.2 Sulfur Compounds We screened a series of readily available sulfur-containing compounds (all present in 0 4% concentration) and found that generally various thiols and sulfides were beneficial e g benzyl CHEMICAL SOCIETY REVIEWS 1993 mercaptan p-thiocresol thiophenol diphenyl sulfide and thianthrene with accelerations of 0 2 0 5 0 4 0 3 and 0 5 ms respectively On the other hand available benzothiophenes were found to be quite strong inhibitors e g benzothiophene (thia- naphthene) and dibenzothiophene 0 8 and 0 6 ms longer delay times respectively More interestingly dibenzothiophene sul- fone had no effect on the ignition delay The question then arises as to whether dibenzothiophene-like molecules are affected either by the additives in the fuel or by the treatments which we have found to be beneficial? Some experiments designed to provide answers to these questions are described in the next three sections 11 Additives versus Inhibitors The experiments described here were performed with one of the fuels as described above which was unaffected by the presence of I YOof di-t-butyl peroxideThe addition of 1 % of iso-octyl nitrate to this fuel shortened the ignition delay by 0 9 ms and addition of 0 4% of dibenzothiophene to the fuel lengthened it by 0 4msThen addition of this amount of dibenzothiophene to the fuel containing iso-octyl nitrate also increased its ignition delay by 0 4ms whereas it did not change at all that of the fuel containing di-t-butyl peroxideThus it is clear that di-t-butyl peroxide is able to cancel completely the inhibiting effect of dibenzothiophene whereas iso-octyl nitrate cannot this is further clear evidence that these two additives work in different ways and at the same time provides a clue as to what di-t-butyl peroxide does 12 ChemicalTreatments versus Inhibitors It is known that ozone reacts quantitatively with dibenzothio- phene to form first the sulfoxide and then the ~ulfone,~~ dibenzothiophene+ 20 +dibenzothiophene sulfone + 20 which we have found not to be an inhibitor We found that fuel containing dibenzothiophene when treated with just sufficient ozone to complete this reaction exhibited the same ignition delay as did the untreated fuel thus ozone can remove dibenzothio- phenes from the fuel However treatment with either lead tetra- acetate or DDQ does not counteract the inhibition by di benzo t hiop hene As a result of the interactions between dibenzothiophene and di-t-butyl peroxide described above another form of chemical treatment suggests itself that of pre-heating the fuel with di-t- butyl peroxide Diesel fuel containing 1YOof di-t-butyl peroxide was heated under reflux at 150 "Cfor a sufficient length of time to have decomposed all of the peroxide A small amount of brown residue was produced which contained about four times as much sulfur as did the original fuel correspondingly the sulfur content of the fuel fell to about 95% of its original value and its cetane number was increased to about 601The amount of peroxide used in the pre-treatment had to be reduced to 0 1% before the ignition delay of the treated fuel equalled that of the fuel containing 1 YOof di-t-butyl peroxide as an additive in fact a noticeable improvement (0 3 ms) in ignition delay was mea- sured when only 0 01% of di-t-butyl peroxide was used in the pre-treatmentThese results suggest that the fuel contains a small fraction of sulfur-containing material which is very strongly inhibiting but that the preponderance of the sulfur compounds present are benign at least as far as ignition delay is concerned 13 How Does Di-t-butyl Peroxide Interact with Di benzothiophene? Little is known about the reactions of dibenzothiophene with free radicals in the gas phase However phenyl radicals pro- duced by thermal decomposition of benzoyl peroxide in molten dibenzothiophene at 1 10"C yield monophenyl-substituted HOW DO DIESEL-FUEL IGNITION IMPROVERS WORK"-H 0 PRITCHARD ET AL dibenzothiophenes and a small amount of the sulfone 32 With the sulfur atom numbered as 5 and the substitution positions in the benzene ring numbered 4 3 2 1 with increasing distance from the S atom phenyl substitutions were 28% 21% 12% 3 1YO,respectively We heated a 2 5 1 molar ratio mixture of di-t-butyl peroxide and dibenzothiophene dissolved in benzene in a stainless steel bomb at 150"C until all of the peroxide had decomposed After reaction the product was found to contain a large amount of unreacted dibenzothiophene a small amount of sulfone and a mixture of monomethyl-substituted dibenzothiophenes We were not able to resolve the 2-and 3-is0mers,~~ but they amounted only to about 18% of the product with the 1-and 4- isomers dominating at about 38 and 44% respectively the proportions are different from those of phenyl case In view of the small yields it does not seem that these reactions can account for the effects of di-t-butyl peroxide as described in the two preceding sections 14 Conclusions It should be remembered that the results proved here obtain in a rather low-temperature regime roughly equivalent to that found when starting a cold diesel engine at temperatures below 0 "C and in the absence of direct experimental confirmation they should only be assumed to be transferable to the normal diesel regime with some caution We have found quite conclusively that the important chemistry contributed to the spontaneous ignition process by the ignition-improver additives takes place exclusively in the gas phase We have also found equally conclusively that there is no single generic mechanism by which these additives reduce the ignition delay for whereas both iso-octyl nitrate and di-t-butyl peroxide may accelerate the ignition of ordinary diesel fuel about eq~ally,~ we have found certain low-nitrogen low-sulfur fuels for which iso-octyl nitrate causes the anticipated improve- ments but for which di-t-butyl peroxide has no effect We infer in fact that iso-octyl nitrate acts vza the formation of NOz mainly on the hydrocarbon itself whereas di-butyl peroxide acts on some inhibitors present in the fuel most likely sulfur- containing inhibitors however only a small fraction of the total sulfur content resides in inhibiting molecules Here our primi- tive modelling studies were of help in showing how NO which is formed in the thermal decomposition of iso-octyl nitrate can accelerate ignition by abstracting hydrogen from the hydrocar- bon to form HNO Presumably this is true for all aliphatic nitrates and nitrites with differences in efficiency among them stemming from variety in behaviour of the alkyl radicals formed alongside the NO Likewise with di-t-butyl peroxide as already noted by Griffiths and co-workers,l our calculations reveal an almost instantaneous temperature rise caused by combustion of the peroxide but below about 800K until reactions involving acetone become important it is insufficient to stimulate rapid ignitionThus it is not surprising that one can find certain fuels for which the ignition delay in our low-temperature regime is unaffected by the presence of di-t-butyl peroxide However for lack of knowledge of the products and/ or mechanisms of decomposition of other peroxides we cannot speculate on what causes the observed differences in behaviour between them We have located one general type of sulfur-containing inhibi- tor the benzothiophenes Much of the sulfur content of our diesel fuels resides in dibenzothiophenes which have survived the catalytic hydrogenation used to reduce the sulfur content this is because they have bulky alkyl groups in the 4,4'-positions which interfere with the action of the catalyst It seems probable therefore that dibenzothiophenes in general are ignition inhi- bitors -just as they pick up oxygen atoms from ozone they readily pick up 0atoms in the pre-ignition phase thereby killing the chains required to cause the ignition However hydrocarbon fuels contain an enormous range of types of sulfur compounds some of them also containing nitrogen or oxygen and inhibition is by no means restricted to the one type that we have identified In fact we have not yet made a positive identification of any substance that is present in the fuel that is a potent inhibitor and intriguing amongst the inhibitors present are those that are so readily deactivated by lead tetra-acetate or by DDQ possibly by dehydrogenation 15 Future Requirements On an industrial scale ozonization is not a practical proposition because of the huge quantities that would be required the low yield of ozone per kilowatt-hour of electrical input and because of the unstable state in which it leaves the fuel the stability requirements for long-term storage under a wide range of conditions of heat and cold are very demanding Nor is treatment with lead tetra-acetate (or equivalently red lead and acetic acid) because it introduces lead into the fuel for example recent Canadian legislation requires that the lead content of a fuel shall not knowingly be increased Nor would the simple device of generating C1 atoms in the pre-ignition phase be acceptable it is now known that the presence of chlorine compounds in fuels leads to the formation of and their formation would be especially likely with highly aromatic fuelsThus at the present time there seems to be no alternative but to seek a yet better understanding of the problem An important task is to identify some of the inhibitors that actually exist in the fuel we need to be able to identify say by GC-MS some substances that are destroyed on treatment with lead tetra-acetate or with DDQ and possibly some new sub- stances formed as a resultThis is not an easy task the gas-phase chromatogram of diesel fuel comprises some 300-350 peaks,14b and we are searching for inhibitors that make up some 0 2-0 3% of the total however we are almost certainly not looking for just one or two substances but for series of similar substances perhaps even hundreds each present in concent- rations of tens of ppmThe silica-gel extracts offer a better chance but here again the chromatogram also gives more than 300 peaks the number of peaks observed is determined more by instrumental factors such as the threshold sensitivity of the integrator and the column resolution and by the boiling range of the sample than by the number of distinct substances present It would be useful to study the kinetics of the reactions of atoms and free radicals with dibenzothiophene carbazole and their homologues and to perform parallel molecular orbital calculations on structure and reactivity in these systems Also to elucidate the mechanisms for the thermal decomposition in the gas phase of some of the more esoteric peroxides shown inTable 3 with a view to explaining the interferences demonstrated there between pairs of additives We have not described here any results of nitrogen-containing compounds because the situation is not clear cut as was also the case in the initial study of ignition improvers,s cited earlier Simple amines (n-decylamine tri-n-butylamine pyridine indole) appear to be mildly inhibiting -perhaps not unexpected in view of the known anti-knock properties of amines also our simple modelling calculations show that NH inhibits the ignition of methane and oxygen at low temperatures Carba-zoles which are known to be present in these fuels seem to be ignition promoters some of them quite strong and amides and imides have no effect We are continuing to try to unravel these contradictory results 16 References 1 A D Puckett and B H Caudle 'Ignition Qualities of Hydrocarbons in the Diesel-fuel Boiling Range' United States Department of the Interior Bureau of Mines 1948 I C 7474 2 E M Goodger and A F M Eissa J Inst Energy 1987 84 3 G H Unzelman 'Diesel Fuel Demand -A Challenge to Quality' Report of Energy Economics Group Institute of Petroleum Lon- don October 1983 4 D E Steere andT J Nunn 'Diesel FuelTrends in Canada' SAE I08 Technical Paper Series 1979,790922 5 W E Robbins R R Audette and N E Reynolds SAE Quarterly Transactions 1951 5,404 6 C L Wong and D E Steere ‘The Effects of Diesel Fuel Properties and Engine Operating Conditions on Ignition Delay’ SAETechni- cal Paper Series 1982,821231 7T -M Li and R F Simmons ‘Twenty First Symposium (Inter- national) on Combustion’The Combustion Institute Pittsburgh 1988,~ 455 8T J Russell in ‘Gasoline and Diesel Fuel Additives’ ed K Owen John Wiley Chichester 1989 p 65 9 (a)P Q E Clothier,A Moise,andH 0 Pritchard,Combust Flame 1990,81 242 (b)and references therein 10 (a)R W Hurn and K J Hughes Znd Eng Chem ,1956,48,1904 (b) D H Hoskin C F Edwards and D L Siebers ‘Ignition Delay Performance versus Composition of Model Fuels’ SAETechnical Paper Series 1992,920 109 11 (a)V M Boiko V V Lotov and A N Papyrin Prog Astronaut Aeronaut 1990 132,205 (b)E K Dabora ibzd 1990 132 31 1 12 (a)T Inomata J F Griffiths and A J Pappin ‘TwentyThird Symposium (International) on Cornbustion’The Combustion Insti- tute Pittsburgh 1990 p 1759 (b)J D GabanoT Kageyama and F Fisson Prog Astronaut Aeronaut 1990,131,407 13 P Park and J C Keck ‘Rapid Compression Machine Measure- ments of Ignition Delays for Primary Reference Fuels’ SAETechni- cal Paper Series 1990,900027 14 (a) J F Griffiths A J Pappin and A N Reid in ‘Fuels for Automotive and Industrial Diesel Engines’ Institution of Mechani- cal Engineers 1990 p 101 (b)J F Griffiths A J Pappin M A R Al-Rubaie and C G W Sheppard ‘Fundamental Studies Related to Combustion in Diesel Engines’ Paper C433/018 Institution of Mechanical Engineers 1991 p 151 (c) M A R Al-Rubaie J F Griffiths and C G W Sheppard ‘Some Observations on the Effectiveness of Additives for Reducing the Ignition Delay Period of Diesel Fuels’ SAETechnical Paper Series 1991 912333 15 (a)N A Henein and C -S Lee ‘Autoignition and Combustion of Fuels in Diesel Engines under Low AmbientTemperatures’ SAE Technical Paper Series 1986 861230 (b) N A Henein ‘Cetane Scale Function Problems and Possible Solutions’ SAETechnical Paper Series 1987 870584 16 A B Lovell K Brezinsky and I Glassman ‘Twenty Second Symposium (International) on Combustion’The Combustion Insti- CHEMICAL SOCIETY REVIEWS 1993 tute Pittsburgh 1989 p 1063 17 H Frerichs MTappe and H Gg Wagner Ber Bunsenges Phys Chem 1990,94 1404 18 J Sato K Konishi H Okada andT Niioka ‘Twenty First Symposium (International) on Combustion’The Combustion Insti- tute Pittsburgh 1988 p 695 19 C K Law in ‘Droplets and Bubbles’ American Institute of Physics Conference Proceedings No 197 1989 p 321 20TTachibana K Hirata H Nishida and H Osada Combust Flame 199I 85 5 I5 21 (a)C JThompson H J Coleman HT Rall and H M Smith Anal Chem ,1955,27,175 (b)G U Dinneen J R Smith R A Van Meter C S Allbright and W R Anthoney ibzd 1955,27 185 22 (a) C K Westbrook J Warnatz and W J Pitz ‘Twenty Second Symposium (International) on Combustion’The Combustion Insti- tute Pittsburgh 1989 p 893 (b)C K Westbrook W J Pitz and W R Leppard ‘The Autoignition Chemistry of Paraffinic Fuels and Pro-knock and Anti-knock Additives A Detailed Chemical Kinetic Study’ SAETechnical Paper Series 199 1,9 123 14 23 D A Bittker Combust Sci Techno1 1991,79,49 24 W A Bone and J B Gardner Proc R Soc London Ser A 1936 154,297 25 K Seshadri and N Peters Combust Flame 1990,81 96 26 (a)M W Slack and A R Grillo Combust Flame 1981,40 155 (b) C -H Lin H -T Wang M C Lin and C F Melius Int J Chem Kinet 1990,22,455 27 J F Griffiths and C H Phillips Combust Flame 1990,81 304 28 J A Miller R J Kee and C K Westbrook Annu Rev Phys Chem 1990,41,345 29 M Hudlicky ‘Oxidations in Organic Chemistry’ ACS Monograph 186 1990 30 J E Kittrel S T Darian and P S Tam U S Patent applied for Chem Abs 1987,107 (20) 1799682 31 A Maggiolo and E A Blair ‘Ozone Chemistry and Technology’ ACS Advances in Chemistry Series 1959,21,200 32 E B McCall A J Neale andT J Rawlings J Chem Soc 1962 5288 33 J D Payzant T W Mojelsky and 0 P Strausz J Energy and Fuels 1989 3 449 34 See e g ,many papers in ‘Geochemistry of Sulfur Fossil Fuels’ ed W L Orr and C M White ACS Symposium Series No 429 1990 35 G H Eduljee Chem Brit 1988,24 1223

ISSN:0306-0012    

DOI:10.1039/CS9932200101

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

The Role of NMR in Boron Chemistry David Reed Department of Chemistry, University of Edinburgh, West Mains Road, Edinburgh EH9 3JJ 1 Introduction The study of the boron hydrides and their chemistry over the past 25 years has been greatly facilitated by the rapid advances made both in X-ray crystallography and in NMR spectroscopy. The former permits the study of molecular structure in the solid state, but yields no information about the structure of, or behaviour of, materials in solution, and also needs the isolation of appropriate single crystals to permit such analysis. NMR analysis can enable chemists to elicit structural infor- mation about molecules in solution and, in some cases, can give information about molecular dynamics (e.g. fluxionality). The purpose of this review is to show, using examples as appropriate, how the range of NMR techniques now available have been used in the analysis of boron hydrides and their derivatives.2 Applications of NMR to Boron Hydride Chemistry Boron possesses two isotopes, namely loB and -'-'B, both of which can be observed using NMR spectroscopy. Both possess the property known as nuclear spin, which makes such experi- ments possible, and in each case the spin, I, is >3(see Table 1). In common with most nuclei of I> 4,the dominant relaxation mechanism is quadrupolar, which means that relaxation times can be quite short -for llB typical T, values are < s, though where the boron is located in a symmetrical environment this could be up to 1 s.The result of this is that -'lB NMR Table 1 Some properties of Boron Nuclei vs. Proton Resonance Relative Natural YO Frequency Sensitivity Isotope Abundance Spin Z 2.35 T field vs. 'H 1 OB 18.83 3 9.3 0.02 IlB 81.17 312 32.1 0.165 'H 99.98 112 100.0 1.oo David Reed was born in Leeds, England in 1954. He obtained his BSc. (1975) and Ph.D. (1978) degrees from the University of Leeds, the latter degree being obtained under the supervision of Professor N. N. Greenwood and involving the study of the synthesis and characterization of metallaboranes. He then moved to the University of Strathclyde, where he worked on electrochemical studies of boron compounds. Since 1981 he has been associated with the High Field NMR Centre at the University of Edinburgh, where he has particular inter- ests in using highjield NMR to study solution structures and properties of both main-group and transition-metal cluster compounds.spectra tend to be made up of signals with linewidths at half the peak height, wt, typically in the range 30-100 Hz. This problem is overcome to some extent by the nucleus possessing a reason- ably large chemical shift range (ca, 50 ppm typically, though examples do exist outside this range). Furthermore, such line- widths do mean that only lJ(-"B--'H) couplings are usually observed, these being in the range 80- 160 Hz. Boron and proton nuclei have been used in the study of boranes, with -'lB being by far the more used of the two boron nuclei, both because of its sensitivity (see Table l), and in terms of its chemical shift dispersion (it has ca.3 xmore Hz/ppm than OB). As mentioned earlier, the -'B nucleus being quadrupolar leads to broad lines. As a result of this, some coupling infor- mation is lost, primarily -'J(-' B-l B). An additional problem often found with -''B NMR spectros- copy is that signal overlap is not uncommon, despite the relatively large chemical shift range possible. Furthermore, the analysis of -'H spectra of boranes is complicated by the coupling of the protons with the boron nuclei, particularly -'B which, having I =3/2, couples with the lH nucleus such that a four line multiplet of relative intensity 1:1:1:1 is observed for each proton bound to a boron.For 'bridging' hydrogens located between two boron atoms, the result is a seven line multiplet of relative intensities 1:2:3:4:3:2: 1. Some of the problems described above can be illustrated by looking at the llB, -'-'B('H} (Figure l), and 'H (Figure 2a) spectra of the selenaborane [NEt,H][B,,H, ,Se].-' The IIB spec- trum has all the coupling to the 'H nuclei retained, whilst the 'B('H) spectrum has it removed. Even in the latter case, no evidence of lJ( -'-'B--' 'B) is apparent. Also, the H spectrum is clearly very complex. I I'"'Ii"'l"~'l""1"''1""I" -5 -10 -15 -20 -25 -30 PPm Figure 1 192.48 MHz lIB{IH) (bottom) and 'lB (top) spectra of the [BloH,,Se-] anion, (14.1 T), and its structure. I09 I PPm Figure2 (a) 'H spectrum of [NEt,][B,,H, ,Se]; (b) IH(IIB) spectrum of "Et,"l,H, 1SeI; (c) (b) -(a).Broad band boron decoupling results in a much simplified H spectrum (Figure 2b), with all the cage protons giving rise to singlet signals, though these are partly hidden by signals arising from the organic cation. Such difficulties can be minimized by obtaining a difference spectrum (Figure 2c). The spectra described above do not yield much information regarding B-B or B-H connectivities, and a complete analysis of such a system would require such information. There have been some technical advances in recent years which have greatly boosted such studies: (i) The development of superconducting magnets has resulted in magnetic field strengths as high as 14.1 Tesla (to date).Increasing magnetic field strength has two major benefits, namely increased sensitivity and increased dispersion of signals. Table 2 shows the comparison of magnetic field strength (Tesla) with lH and IlB resonance frequencies (MHz). (ii) During the past 15 years or so there has been a rapid increase in the use of pulse sequences. Some of these have resulted in the development of two-dimensional NMR, an idea which was first proposed by Jeneer,2 and first demonstrated by Ern~t.~Since that time innumerable variations of such experi- ments have been described, primarily involving correlation of homonuclear or heteronuclear spin-spin couplings and homo- nuclear or heteronuclear NOE effect^.^ In the NMR study of boron hydrides, the main objective would be to establish B-H and B-B connectivities with relevant signal assignment, thereby assisting in structure elucidation.The determination of such information can be effected by a combi- nation of llB/lH, lB/llB, and lH/IH correlation experiments, as appropriate. To illustrate how such experiments can be used, the selenabor- CHEMICAL SOCIETY REVIEWS, 1993 -1 -2 -3 -4 Table 2 Comparison of lH and lB resonance frequencies at different magnetic field strengths (Tesla) Field Strength (TI Resonance Frequency IIB (MHz) Resonance Frequency 'H (MHz) 2.35 32.1 100 4.70 64.2 200 8.46 115.5 3 60 14.10 192.5 600 anes [BloHl lSe-]l and [B,H,Se,15 will be examined in detail.In both cases comprehensive assignment was achieved with B/IH and B/I B correlation experiments. The problem with the anionic species was one of assignment of the signals, the structure having been determined crystallographically. The dis- elenaborane was studied because there was a question about the molecular structure, with two possibilities as shown in Figure 3. Correlation of 'H signals with IlB signals can be effected in either of two ways, namely via the two-dimensional 'hetcor' experiment or by one-dimensional selective lH{ 'B) experi- ments. The former has not been extensively utilized by boron chemists, though the first illustrations of the application of two-dimensional techniques to boron hydrides were shown using this type of e~periment.~,~ The first example given was of its appli- cation to the carbaborane C2B5H7,6 which demonstrated the potential of the experiment, and this was followed up by two papers showing the application to decaborane( 14).'.* The result of such an experiment is a plot which correlates chemical shifts of the 'H nuclei, 6('H), with those of the boron nuclei, 6( B).Such information can be presented in either of two ways, namely as a stacked plot (as indicated in Figure 4a) or as a contour plot (as shown in Figure 4b). In practice, the former is of little value, as it is not easy to extract the chemical shift THE ROLE OF NMR IN BORON CHEMISTRY-D. REED 10 0 BH 0Se 10 Figure 3 The two possible structures of [B,H,Se,].information from such a presentation, while the contour plot provides an easy-to-read presentation of such data. The experi- ment uses delays which are derived from the value of J(X-H) (X = llB in this case) and as a result, the main problem associated with this technique is the difference in 'J(B-H) values between terminal (exo) B-H nuclei and bridging B-H-B nuclei (ca. 140 Hz and ca. 40 Hz respectively). Experimental para- meters can be modified to take this into account, but not always successfully: sometimes responses deriving from bridging nuclei are weaker than ideal. Such an experiment was used for the correlation of B and lH signals in [B,H,Se2], this having no bridging protons. The resulting plot, shown in Figure 5, allows clear correlation between the lH and 'B chemical shifts in this case. The second method of correlating 'H and ''B signals, and by far the most popular in practice, is via 'H observation with Figure 4 Examples of (a) a stacked plot two-dimensional spectrum and (b) the same data presented as a contour plot.1 -2.2 -2.4 -2.6 I -2.8 I € -3.0Q-1--.Q I -3.2 -I --3.4 T -. -3.6I I -3.8 I I4.0 ,"~',''~','''','~'~,'~'~I'"'I~"~1 ~'~~ 0 -5 -10 -15 -20 -25 -30 -35 PPm Figure 5 An 'B/'H two-dimensional correlation experiment on [B,H,Se,] carried out at 8.46T (360.1 MHz for 'H; I1 5.5 MHz for IlB). selective l B decoupling. Many modern FT instruments possess the capability for this type of experiment. The experiment is carried out by observing 'H while simultaneously irradiating at a frequency corresponding to an individual llB signal.A spectrum is also acquired where decoupling power has been applied at an off-resonance position, thus providing an effecti- vely undecoupled spectrum. Subtraction of the latter from the former provides a decoupling difference spectrum. Such an experiment is repeated for all the B frequencies determined from the 'B spectra. The results of this type of experiment for the [BloHllSe-] anion are shown in Figure 6. Particularly noteworthy is the clear bridging proton response on decoupling at the boron frequency corresponding to the 'B signal at 6 -17.6 ppm. Boron-boron connectivity data can be extracted by using perhaps the most commonly applied of the two-dimensional NMR experiments that have emerged over the last fifteen years, namely the COSY experiment (from Correlation Spectros- copy).The use of 'B-l 'B two-dimensional correlation spec- troscopy was initially demonstrated by Grimes et a/.,showing its application to the cobaltaborane [6-CpCo(B,H, 3)] and the carbaborane [2,3-(C2H,)C,B4H6].9 This was followed by a more comprehensive overview of its applications, as well as of some limitations. O Similar conclusions were drawn from other work described at that time.' The examples quoted in these papers led to a number of observations about the application of 'B COSY which, for brevity, will be outlined here prior to more detailed discussion of some examples.(i) There is no evidence for couplings betweeen non-adjacent boron nuclei. (ii) Boron nuclei connected by bridging hydrogens are less likely to show COSY responses, although examples do exist where such responses are found. (iii) Where boron nuclei exhibit shorter relaxation times it is more difficult to get good COSY spectra. (iv) Boron signals often overlap coincidentally, leading to interpretation problems. Some early examples of' 'B COSY experiments suggested that boron nuclei linked by bridging hydrogens did not exhibit coupling9~''though contradictions to this idea were quickly found, namely [~-(C~H,)COB~H,] and [nid0-1,2,3-(C6Me6)Fe- (c2 H s 2c2B3H s 1.' The result of a COSY experiment carried out on [BloHl ,Se-] is shown in Figure 7.Analysis of such data is relatively straight- forward. The one-dimensional spectrum is mirrored along the diagonal, while those borons which are coupled may give rise to off-diagonal responses (cross peaks) at (Bx,By) and (By,Bx). The analysis of the data in Figure 7 was greatly assisted by the CHEMICAL SOCIETY REVIEWS. 1993 1 D Figure 6 A series of selective lH{IIB} difference spectra for [B,,H,,Se-1, carried out at 14.1T. An l1B{IH} spectrum shows the B decoupling sites. fact, derived from the selective IH{IIB) data, that the llB signal due to B(9,lO) is found at 6-17.6. Also, of the two signals possessing relative area one, only that at 6 -14.6 shows coupling to B(9, lo), and hence must derive from B(5).Thus the signal at 6-35.6 must arise from B(1). The highest frequency signal, at 6-5.6, shows coupling to all others, and is thus due to B(4,6). The signal at 6 -9.2 couples to that of B(9,10), and must arise from B(8,l I), leaving the last signal at 6-16.3being due to B(2,3). One noteworthy feature of the COSY is that an expected correlation between B(2,3) and B(8,ll) has not been observed; for this there is no apparent explanation. Table 3 summarizes the B and H assignments observed for [B,,H Se-] derived from the experiments outlined. The one-dimensional IlB and llB{lH)spectra of [B,H,Se,J are shown in Figure 8, with six boron environments of relative intensities 1:2:2:2:1 :1. These data would be consistent with either of the structures in Figure 3, both of which possess a plane of Table 3 Assignments of the llB and 'H NMR spectra of the [BloH,,Se-] anion 6("B) WH) Assignment -5.6 2.37 436 -9.2 2.05 8,ll -14.6 1.83 5 16.3 1.67 223 -17.6 1.26 9,lO -35.6 1.05 1 -4.11 bridge (9,lO) 1 VF 1 1 L -30--25--204 0 9,lO -15 ae -lOj 0 e 43 1""1""1""1""I""I""I"-5 -10 -15 -20 -25 -30 -35 Figure 7 An IlB COSY Experiment on [B,,H,,Se-1, carried out at 14.1T.THE ROLE OF NMR IN BORON CHEMISTRY-D. REED 5 6 -5 -io -i5 -20 -25 -30 -35 PPm Figure 8 192.48 MHz ' B{'H) (bottom) and ''B (top)NMR spectra of [B,H,Se,] carried out at 14.1 T. symmetry through B(1), B(3), and B(10). The llB COSY spectrum of this molecule (Figure 9) provided both signal assignment and structural determination, the latter on the basis of the proposal that non-adjacent 'B nuclei do not give rise to cross peaks.1 cQ f10 -301 -25--20-o -5 -io -is -20 -25 -30 -35 Figure 9 An * 'B COSY experiment on [B,H,Se,] carried out at 14.1 T. Thus if (I) were the correct structure, then B(1), B(3), and B(10) should each display 3,2, and 2 correlations respectively. Were the structureconsistent with (11),then B( I), B(3) and B( 10) should each show 3,3, and 1 connectivities respectively. Figure 9 shows that two of the area 1 signals, namely A and F, have cross peaks associated with them, whilst E only shows one cross peak. Thus (11) appears to be the structure. From this it is clear that E is the signal from B( lo), meaning D must derive from B(5,6).Signal D in turn couples with F, so F must arise from B( 1). Signal F also couples to B, which must be ascribable to B(2,4), leaving C as B(7,8). Thus [B9H9Se2] adopts structure 11, and its ' ' B and H assignments are summarized in Table 4. A general point, alluded to earlier, is that often in 'B COSY experiments cross peaks do not show up. In some cases there is Table 4 Assignments of the 'B and 'H NMR spectra of [B9 H9 Se2 1 S(1'B) 6('W Assignment 2.1 3.63 3 1.6 3.48 224 -0.6 3.02 738 -1.0 3.33 576 -8.6 2.68 10 -34.6 2.33 1 no apparent reason (as in the example earlier with [B, ,HI, Se-I). Sometimes, however, such behaviour can be rationalized in terms of the rapid relaxation properties exhibited by some B nuclei.Such quadrupolar induced relaxation can be made less efficient by sample heating, thereby improving chances of success. However, the thermal instability of many compounds often precludes this measure. In many cases the techniques discussed so far are adequate for the complete assignment of 'B and 'H resonances. Occasio- nally, however, they are not adequate. Problems of rapid B relaxation, 'missing' cross peaks or overlapping ' B signals occur all too often. Such problems can be overcome in some cases by the use of H-'H COSY with simultaneous broad band llB decoupling, the first application of which was to the rhodaborane [5-(C,HS)-5-Rh-nido-B,H,31, Figure 10.' This example highlighted a number of useful features which supple- ment data gathered from other experiments (Figure 11).Thus, in addition to showing the presence of ,J(H-B-B-H) between exo cage protons, couplings between bridging protons and exo protons were also shown, including those involving the Rh-H- B. A number of problems relating to this type of experiment were also defined. For example, the lB chemical shift range is often several thousand hertz, and it may not be possible to effect full decoupling over the whole chemical shift range as a result. This problem is, ofcourse, exacerbated at higher field strengths. Also, other ligands which contain organic components may be present in the molecule: these tend to give much sharper signals in the H spectrum than do decoupled cage protons.This can mean 'rl noise' interfering with the resulting 'H-lH COSY, this problem showing up as streaks along one of the two dimensions. Figure 10 Structure of [5-(C,H5)-5-Rh-nido-B,H,,I. Finally, it would be inappropriate to ignore the uses of NMR in looking at fluxional processes in boron hydride derivatives. Bridgelendo hydrogen fluxionality was studied for the systems [B9Hl,L] (L = SMe,, SEt,, MeCN, (Me,N),CN, BH,CN-, SCN-).l4>l6 Of these, the structure of the CH,CN derivative has been determined crystallographically. Interestingly, the NMR properties of these were shown to differ between the neutral [B9H1 ,L] complexes and the anionic [B9H1 ,L]-exam-ples with the notable exception of the neutral (Me,N),CN derivative, whose properties were like those of the anionic species quoted.l5 The room temperature llB and llB('H} spectra of B,H,,SEt, (Figure 12) are typical of the neutral species, whilst those of [B9H13 NCS-] (Figure 13) are typical of the anionic derivatives.In the former case, the llB spectrum comprises six doublets, labelled A-F. In the latter there are five CHEMICAL SOCIETY REVIEWS, 1993 -3 -0 0Oo 0 -2OQO -1 0 E* I v 0 00 0 5 lz 2 3 4 ,““I’”’r.’--I’ 4.0 3.0 2.0 1.0 0.0 -1.0 -2.0 -3.0 6(’HYPPm Figure 11 400 MHz ‘H COSY (llB decoupled) experiment on [5-(C,H,)-5-Rh-nido-B,H 131. (Reproduced from reference 13.) doublets and one singlet (signal E). In both cases signal E was ascribed to B(4), which was subsequently confirmed by llB COSY experiments.” The differences in the llB spectra of the neutral and anionic species were taken to imply that the proton on B(4) was fluxional with the bridging protons in the latter (anionic) cases, but not so in the neutral.This was supported by ‘H(”B) NMR data (Figures 14 and 15), which show distinct signals corresponding to bridging hydrogens at ca. 6 -3.50 in the neutral cases, whereas in the anionic examples a broad signal corresponding to 5 protons was found at ca. 6 -1.4. Thus, the anionic examples quoted showed fluxionality between the two bridging hydrogens and the ‘endo’hydrogens on B(4), B(6), and B(8), whereas the static form was observed for the neutral derivatives. The static form of the IH(l ‘B) spectra of the NCS-derivatives was reached if the data were acquired at 203 K.Thus the energy barrier to such intramolecular hydrogen exchange processes was shown to be dependent on the ligand used. A rather different type of temperature dependent phenome- non was described for the closo metalladicarbaborane derivative [1-Ph-3-(r-C9H,)-3,1,2-closo-CoC,B9H ,,I, whose crystallogra- phically determined structure is given in Figure 16.18 In this example the interest lay in the behaviour of the phenyl and the indenyl ligands. The variable temperature ‘H NMR spectra, showing the aromatic region, are given in Figure 17. The spectrum recorded at 298 K showed signals which were consis- tent with seven indenyl protons but only three phenyl protons. Subsequent low temperature experiments showed the presence of two further signals, which were readily ascribable to ortho protons from the phenyl ring.One of these, labelled A, moved significantly to lower frequency with decreasing temperature. In A B CDE F I 7-20 0 -20 -40 Figure 12 115.5MHz llB (botom) and IIB{IH}(top) NMR spectra of [B,H,,SEt,] (8.46T), and its structure. 20 10 0 -10 -20 -30 -40 Figure 13 115.5 MHz llB (bottom) and ‘B{’HJ(top) NMR spectra of [B,H,,NCS-] (8.46T). addition the signal labelled B also moved to lower frequency as the temperature decreased. The explanation of such behaviour is that the phenyl group is undergoing a slowing of rotation around the C( 1)-C( 11) axis with decreasing temperature.Thus, at room temperature the rate of exchange of the ortho protons will be about the same as their ‘H NMR separation in Hz.This results in coalescence of the signals which, when they are widely separated, means they will ‘vanish’. On cooling, their rate of 115THE ROLE OF NMR IN BORON CHEMISTRY-D. REED 3 r I4.0 2.0 0 -2.0 PPm Figure 14 360.13 MHz 'H{"B} NMR spectrum of [B,H,,SEt,] (8.46T). Figure 16 Structure of [1-Ph-3-(n-C9H7)-3,1,2-closo-CoC2B,H,,]. f r ---4:O 2:o b -2.0 -4.0 PPm Figure 15 360.13 MHz lH[l'B) NMR spectrum of [B,H,,NCS-] (8.46T). Figure 17 Variable temperature 'H NMR spectra of [I-P~-~(T-C~H~)-~,~,~-c~oso-COC,B~H~,~. CHEMICAL SOCIETY REVIEWS, 1993 exchange has become less, so the individual signals can be observed The low frequency shifts of A and B can be explained if the preferred low temperature conformation is consistent with the crystal structure Thus H(12) and H(35) (A and B respecti-vely) are each sitting above aromatic T systems, and are conse- quently deshielded 3 Summary The previous section has shown how various NMR techniques have been applied to a selection of boron hydride derivatives It has shown that different types of data can be extracted from such systems Thus, signal assignments can be carried out by use of appropriate correlation experiments It was also demonstrated that where structural data were not available, such information could be deduced from suitable correlation experiments, using the specific example of [B,H,Se,] Additionally, examples were given that showed how data could be extracted about systems which exhibit dynamic processes in solution NMR can therefore be shown to both supplement and, in suitable cases, substitute for X-ray crystallographic studies Acknowledgements I would like to thank Dr A R Butler for advice and discussion about the manuscript, Drs T R Spalding and J H Morris for provision of many of the samples discussed, and the S E R C for the provision of both the 8 5 T and 14 1T instruments 4 References 1 D Reed, G Ferguson, B L Ruhl, 0 Ni Dhubhghaill, and T R Spalding, Polyhedron, 1988,7, 117 2 J Jeneer, Ampere International Summer School, Basko Polje, Yugoslavia (1 97 l), unpublished 3 W P Aue, E Bartholdi, and R R Ernst, J Chem Phjis, 1976,64, 2229 4 Several good reviews have been published, e g G A Morris, Magn Reson Chem , 1986, 24, 371, R Benn and H Gunther, Angew Chem Int Ed Engl, 1983,22, 350 5 D Reed, T R Spalding, paper in progress 6 D C Finster, W C Hutton, and R N Grimes, J Am Chem Soc , 1980,102,400 7 I Colquhoun and W McFarlane, J Chem Soc Dalton Trans, 1981,2014 8 D P Burum, J Magn Reson, 1984,59,430 9 T L Venable, W C Hutton, and R N Grimes, J Am Chem Soc , 1982,104,4716 10 T L Venable, W C Hutton, and R N Grimes, J Am Chem Soc , 1984,106,29 11 D Reed, J Chem Research (S), 1984, 198 12 K Base, Collect Czech Chem Commun , 1983,48,2593 13 X L R Fontaine and J D Kennedy, J Chem SOC Chem Commun , 1986,779 14 G B Jacobsen, J H Morris, and D Reed, J Chem Soc Dalton Trans , 1984,415 15 D G Meina, J H Morris, and D Reed, Polyhedron, 1986,5, 1639 16 G B Jacobsen, D G Meina, J H Morris, C Thomson, S J Andrews, D Reed, A J Welch, and D F Gaines, J Chem Soc Dalton Trans, 1985, 1645 17 F E Wang, P G Simpson, and W N Lipscomb, J Chem Phys , 1961,35, 1335 18 Z G Lewis, D Reed, and A J Welch, J Chem Soc Dalton Trans , 1992,731

ISSN:0306-0012    

DOI:10.1039/CS9932200109

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

Motion of Sorbed Gases in Polymers Wen-Yang Wen Department of Chemistry Clark University Worcester Massachusetts 01610 U.S.A. 1 Introduction The mobility of sorbed gases in polymers is an interesting scientific problem with important technological implications. Considerable effort has been invested in understanding the sorption diffusion permeability and selectivity of gases in polymers. The impetus for this work comes from applications such as polymer membranes for gas separation and polymer barriers for packaging. Polymers can serve in various forms as barriers to protect inside matter from the outside environment. For example they are used for food packaging beverage bottling and coating and casing for cables and electronic components. Polymers may be designed for applications requiring controlled penetrant sorp- tion and transport such as dye penetration and binding in fibres and sheets containing absorptive fillers as well as drug delivery systems for medical purposes.The other important uses of polymers are in separations gas separations ultrafiltrations and reverse osmotic applications. Membranes and hollow tubes are used to separate various components of gas mixtures helium separated from natural gases hydrogen recovered from industrial gases and coal gases methane from carbon dioxide oxygen from nitrogen. etc. Before considering the interaction of gases and polymers let us remember that there are various physical states for a polymer crystalline state glassy state viscoelastic state rubbery state viscous-flow state melted state etc.Since most gases usually do not dissolve in crystals the crystalline state of polymers will be excluded from our further consideration. When the temperature of a glassy polymer is increased to its glass transition temperature Tg,the polymer changes to the viscoelastic state with a significant decreases in its elastic modu- lus (see Figure 1). This change at Tg,is described as a free- volume limited relaxation attributed to the onset of segmental motions inside the polymer. Following a wide-spread usage among polymer chemists we shall employ the word ‘rubbers’ to mean polymers at temperatures above Tg,which would include both ‘viscoelastic’ as well as ‘rubbery’ polymers.Since polymers are usually saturated with air or other gases they are evacuated to form gas-free samples. Let us consider what may take place when a gas-free film is exposed to a gas of pressure p at one side and to a vacuum at the other side. First the gas sorption (adsorption and absorption) will take Wen- Yang Wen obtained his B.S. degree from National Taiwan University in 1953 and his Ph.D. degree from the Univer- sity of Pittsburgh in 1957. He did his postdoctoral research at Pittsburgh with Professor H. S. Frank and then at North- western University with Pro- fessor I. M. Klotz. In 1962 he joined the faculty of Clark Un- iversity and has remained there since. His research interests include the structure of water and aqueous solutions cata-lytic pyrolysis of coal and tar and the dynamics of sorbed gas in polymers. 117 Elastic modulus Polysty rene (m.p 523 K) Glassy Viscoelastic cu-IEf + I \\viEs.tI I ‘5 I I I 333 373 41 3 453 493 T /K Figure 1 Plot of the elastic modulus of polystyrene vs.temperature. place at the polymer surface. Next the sorbed molecules will penetrate into the polymer matrix and begin the process of diffusion. After a certain period of time the penetrant molecules will reach the other side of the polymer film desorb at the surface and evaporate into the vacuum as gas molecules. In this way the gas permeation process can take place through the polymer. In a steady-state permeation experiment a gradient in the gas pressure is established and maintained across a polymer film.In this steady state the constant flux J is measured on the low- pressure side of the permeation cell. The permeability P is calculated from the equation. where dp is the pressure difference at the two sides of the film and 1is the film thickness. In a transient permeation experiment a sudden increase in the gas pressure is applied to one side of the film. We then measure as a function of time the amount Q of penetrant per unit film area that enters the low pressure side. At a certain time after the application of pressure the steady state is reached and Q,becomes proportional to time t. This time lag 8 is given by lim Q = J(t -0)I-rX as shown in Figure 2. When gases are sorbed in polymers under constant pressure the system is expected to reach a thermodynamic equilibrium without undue delay provided that the temperature is above the Tgof the polymer. On the other hand the system is not expected to be in a true equilibrium state if the temperature is below the Tg of the polymer.The sorption and transport properties of gases in polymers can therefore be divided into two categories depending on whether the system temperatures are above or below Tg. Penetrants may be sorbed gases organic vapours liquids dyes and other additives (sometimes called ‘diluents’). In this review we shall confine our attention to penetrants which are gases under ordinary conditions. 8 =I'(Time lag)6D / / >t oe Figure 2 Amount Qtof penetrant passing across unit area of a polymer film as a function of time t.2 Macroscopic Observation 2.1 Gas Solubility in Polymers As an example Figure 3 shows sorption measurements for CO in silicone rubber at 35 "C for pressure in the range of 1 to 54 atm.' The volume of the silicone rubber expands slightly with CO sorption and this dilation has been determined and taken into account in theircalculations. For pressure up to 20 atm Henry's law seems to be obeyed yielding the Henry's law constant KD,of 1.385cm3(STP)/(cm3.polym.atm).At pressures greater than 20 atm the c vs.p curve gradually deviates upward from Henry's law and the high solubility region can be described by a Flory-Huggin's treatment with an interaction parameter x of 0.75./O140 r Silicone rubber-C02 120 35 "C t 100 1 B 0 Sorption 20 a -Desorption Flory-hug gins ' ' Henry's Law ' O F0 1 200 400 " 600 " 800 I 1000 P (Psi4 (68 atm) Figure 3 Sorption isotherms for sorption and desorption of carbon dioxide in silicone rubber at 35 "C. (Reproduced by permission from Macromolecules 1986 19 2285.) It is important to note that the solubility data determined by sorption and desorption experiments are the same. This is a distinct characteristic of the gaslrubbery-polymer systems indi-cating that the systems are in thermodynamic equilibrium. The Tgfor the silicone rubber poly(dimethy1 siloxane) is -123"C and the temperature T of measurements for CO is 35"C cledarly T> Tg.CHEMICAL SOCIETY REVIEWS 1993 Markedly different sorption-desorption response is observed for CO in glassy polycarbonate (PC) at 35°C as shown in Figure 4.'Since the Tgof PC is 150"C,we have T < Tgfor this system. The desorption curve is distinctly different from the sorption curve showing an effect of hysteresis. The solubility (c) of CO is determined as a function of increasing pressure from about 2 to 60 atm then the value of c is determined as a function of decreasing pressure from about 60 to 2 atm. The observed difference between the sorption and desorption curves is recog-nized as a characteristic of the gas/glassy-polymer system; it indicates that these systems are not in thermodynamic equilibrium. 70 1 Pol ycarbonate-C02 60 50 Q,EX 40 02 a t-30 0 u c 20 10 0 0 200 400 600 800 1000 P (PW (68 atm) Figure 4 Sorption/desorption data illustrating the pronounced hystere-tic response observed in the carbon dioxide/polycarbonate system at 35 "C.(Reproduced by permission from Macromolecules 1986 19,2285.) Figure 5 gives the results of Lowell and McCrum2 for the solubility coefficient k plotted as a function of temperature for cyclopropane in linear polyethylene (T> Tg). The sharp increase in the values of k as the temperature climbs above the melting point T is attributed to the disappearance of the crystallites that existed in the linear polyethylene. 2.2 Gas Permeability The gas permeability of a planar membrane is defined as P = J; IlAp where J is the steady-state flux of the gas.The mean permeabi-lity P is the product of a mean diffusion coefficient r) (a kinetic factor) and a solubility-related function 3 (a thermodynamic factor) where (4) MOTION OF SORBED GASES IN POLYMERS-WEN-YANG WEN TI "C 180 160 140 120 100 80 60 1o3 1 8 I 1 I I t I I I I I I I 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 3.0 1O~IT Figure 5 Temperature dependence of the solubility constant k for C,H in linear polyethylene above and below the melting point. Measure- ments taken at successively increasing temperature. (Reproduced by permission from J. Polym. Sci. Part A-2 1971 9 1935.) D(c)is the effective mutual diffusion coefficient and c is the local sorbed gas concentration in the membrane.Ch and C1are the concentrations at the membrane interfaces when pressures ph and pIare maintained at these interfaces respectively (Ph > pl). The function S is defined by For example the values of I) for CO are nearly constant and independent of pressure (for Ap in the range of 1 to 10 atm) across a membrane of rubbery poly(dimethy1 siloxane) PDMS at 35 "C(T> Tg);3in contrast the permeability decreases appre- ciably with the initial increase of Ap across a membrane of glassy polycarbonate at 35 "C (Tg= 150"C).4 2.3 Diffusion and Diffusion Coefficients 2.3.1 Fundamental Relations Diffusion is the process by which the sorbed gas molecules are transported from one part of a polymer to another as a result of random motions of the molecules.The phenomenological theory of diffusion is based on the postulate that the flux of the penetrant is proportional to the concentration gradient mea- sured perpendicular to the polymer film plane where D is the diffusion coefficient. By considering the mass balance of an element of polymer volume it can be shown that the fundamental equation of diffusion takes the form of Fick's law given by In many penetrant/polymer systems D depends however on the concentration. In this case and also when the medium is not homogeneous so that D varies with the position (x,y,z) equation 7 becomes &/dt = d/dx(Ddc/dx)+ d/dy(Ddc/dy)+ d/dz(Ddc/dz) (8) When diffusion takes place effectively in one direction along the x-axis only equations 7 and 8 reduce respectively to l?c/l?t= DdZC/dX2 (9) dclzt = i?/dx(Ddc/dx) (10) The diffusion coefficient is a measure of the random molecular motions in that system.For example at 25 "C,the value of D for CO in N gas at 1 atm is 0.165 cm2 s-l but that in water is only 1.92 x lop5 cm2 s-l. In contrast the diffusion coefficients of CO at infinite dilution are 2.64 x lo-' cm2 s-' in PDMS but only 1.2 x lop8cm2 s-in polycarbonate glass at 35 "C. The so-called 'apparent diffusion coefficient' may be calcu- lated from the equation where 8 is the observed time lag and 1is the membrane thickness (see Figure 2). However D does not have a simple meaning since it is not directly related to a true molecular mobility. More often the effective diffusion coefficient D(c) is determined for comparison with other techniques that probe molecular motion.D(c)in equation 4 can be determined from permeability and solubility measurements using the following relations dI)/dp in the above expression is obtained from the pressure dependence of the mean permeability coefficient P,and dp/dc is evaluated from the solubility-pressure relation. D(c) is a func- tion of the penetrant concentration and depends on the nature of the sorbed gas/polymer system and the temperature. According to Stern's version5 of Fujita's free volume theory,6 the thermodynamic diffusion coefficient DT IS given by the expression where vf is the fractional free volume c$~ is the amorphous volume fraction of the polymer and Ad and Bdare characteristic parameters.The meanings of these parameters are not clear but are thought to depend on the size and shape of the penetrant molecule. It should be noted that equation I3 may be applied to sorbed gases in rubbery polymers but not to penetrants in glassy polymers. 2.3.2 Some Experimental Observations Diffusion coefficients at infinite dilution for five gases in polycar- bonate are shown in Figure 68 as a function of kinetic diameter at 35°C. The coefficients are derived on the basis of the dual- mode model with DD and DH referring to the diffusion coeffi- cients for Henry's law species and Langmuir species respectively (see Section 3.3.1). u)0 0.1 He CO2 Ar 0.001 2.6 2.8 3.0 3.2 3.4 3.6 3.8 Kinetic diameter from zeolite sorption (A) Figure 6 Diffusion coefficients for various gases in polycarbonate at 35 "Cas a function of the kinetic diameter (molecular sieving dimen- sion in zeolites).The solid points refer to the diffusion coefficients of the Henry's law species. The open points refer to the diffusion coefficients of the Langmuir species. (Reproduced by permisison from J. Membrane Sci.,1977,2,165.) 120 The plot of infinite dilution coefficients Do vs the van der Waals volume is shown for a variety of penetrants in natural rubber and glassy poly(vinylch1oride) in Figure 7 In this figure the effect on molecular mobility of penetrants is clearly contrasted for a rubber and for a glass In a rubber as the kinetic diameter of the penetrant size approaches that of the normal hydrocarbons C,H + ,with n = 3 4 or 5 the diffusivity reaches a plateau In the relatively rigid environment of the glassy polymer the size and shape of penetrants are well discriminated resulting in a wide spread of their mobility lo4 [ He 9 H2 A’ \.,CHI Q,0 Kr’ PVC 0 20 40 60 80 100 120 140 160 180 van der Waals volume (cc mole-’) Figure 7 Diffusion coefficients for a variety of penetrants in natural rubber at 25 “Cand rigid poly(viny1 chloride) at 30 “Cplotted against the van der Waals volumes of the penetrants (Reproduced by permission from ‘Material Science of Synthetic Mem- branes’ ACS Symposium Series 1985 269,28 ) 3 Theoretical Models 3.1 General Considerations We may consider the solubility and motion dynamics of pene-trants in polymers on various levels At the highest level we may wish to formulate a statistical mechanical description for the equilibrium and time-dependent properties of the whole system On this level the various effective potentials representing interac- tions among penetrant molecules and the polymer molecules are taken into account Thermodynamic equilibria may be con- sidered to be established during the ordinary laboratory time scale for the penetrant/rubber systems but not for the penetrant/ glass systems At a second level for the penetrant/glass systems we may employ a lattice model for polymers and formulate stochastic rate equations for the penetrant motion The main charcteristic of this model is that the nearest-neighbour interactions between the molecules are treated properly and double occupancy of the lattice sites is excluded At a third level for both glassy and rubbery polymers the sorbed gas transport over large distances and long times may be modelled by macroscopic diffusion equations Transport properties such as diffusion coefficients and permeabilities are considered to reflect the average properties of the sorbed gas and polymer The widely used free-volume models may also be grouped to the third level of the description 3.2 The Pace-Datyner Theory of Diffusion Pace and Datyner proposed a statistical mechanical model for diffusion of simple molecules in polymer lo They assumed that (1) a polymer consists of bundles of parallel chains with coordi- CHEMICAL SOCIETY REVIEWS.1993 nation numbers four and that a small molecule can pass along the axis of the interchain space (or tube) with no activation energy In addition they assumed that (11) the penetrant mole- cule can move perpendicular to this axis by the bending sepa- ration of two polymer chains which will require activation energy The second type of motion is slow and determines the translational diffusion coefficients of the penetrant The chains are supposed to interact according to the DiBenedetto poten- tial,I1 which asssumes the chain to be a sequence of Lennard- Jones centres (e g CH units) They derived lengthy expressions for the activation energy AE diffusion coefficient D,and the frequency of chain openings v that allow passage of penetrant molecules The expression for the diffusion coefficient contains a parameter that represents the mean-square jump length of the penetrant molecule D = < L2 > v/6 = D,exp(-AEIRT) Jump lengths L however cannot be independently estimated by the theory and the values of L implied from experimental diffusion coefficients are too long (400-1000 A) and physically unrealis tic Recently Kloczkowski and Mark’ examined the Pace- Datyner theory in detail and suggested several improvements They obtained a lower activation energy and a lower jump length but the values of L are still too high 3.3 Models for Gas/Glassy-polymer Systems 3 3 I Models Assuming No Plasticization of Polymers Mears’ proposed a ‘dual-mode’ picture in which sorbed mole- cules can remain in one of two surroundings in the glassy polymer The first surrounding or ‘mode’ is that of the dissolved or mixed state while the second mode is that of the trapped state in which the sorbed molecules reside in defects (often called ‘microvoids’ or ‘holes’) that were frozen into the glass at the transition temperature Tg Various investigators demonstrated that gases show distinctly non-linear isotherms in glassy polymers and represented them as a sum of a Henry’s law contribution (1 st mode) and a Langmuir contribution (2nd mode) where C is the total sorbed gas concentration CD is the concentration of dissolved molecules CHis the concentration of molecules trapped in holes kDis the Henry’s law constant CHis the concentration in the holes at saturation b is the hole affinity constant and p is the pressure of the penetrant in the gas phase A dual-mode transport theory is introduced to augment the sorption theory Paul and Koros revised the transport theory to allow partial mobility of the trapped penetrant molecules ’The total flux of penetrant in polymer glasses is given by where DD and DH are diffusion coefficients for the sorbed molecules in the dissolved and trapped modes respectively In addition the local equilibrium between the two modes are assumed to be established very rapidly These two modes are related by where K = Ch b/kDand Q = b/kD Combining equations 15 and 16 Paul and Koros derived the following expression for the flux where F = DH/DD The time dependence of the penetrant con- centration may be obtained by the combination of equation 17 and the continuity equation MOTION OF SORBED GASES IN POLYMERS-WEN-YANG WEN This transport model of Paul and Koros contains five para- meters and can be used to fit the permeability and diffusion data The values of Fobtained usually fall in the range of 0 04 to 0 6 One of the difficulties with the Paul-Koros model is that the flux equation (equation 15) does not contain terms that couple the two modes CD and CH The equation for the flux would contain terms that take into account the transport into and out of the holes Another difficulty with equation 15 is the impli- cation that the hole domains are large enough to sustain diffusion Frederickson and Helfand included the coupling terms which provide mobility to molecules that are trapped into holes l4 In addition to CD(x,t)and cH(x,t),the concentrations CD(x)and CH(x) are considered for the dissolved and trapped modes of sorption when they are fully saturated with the respective molecules Their derived expression for the tota penetrant flux 1s where u is a length that parameterizes the range of W,,(y) and Dkis the diffusion coefficient for penetrant within the kth phase WK&) are phenomenological coefficients that reflect the rate at which penetrant molecules in mode L are transferred into a region of mode K a distance y away The authors noted the presence of the terms contained in equation 19 that couple the two modes CD and CH Furthermore they noted that the 'bare' diffusion coefficients Db and D& have been renormalized by KA and A respectively The consequence of this is that even if penetrant molecules in the trapped mode have no intrinsic mobility (D&= 0) the exchange of molecules between modes provide mobility to trdpped molecules This is definitely an improvement of the Paul-Koros expressions With the assumption of the rapid local equilibrium equation 19 can be rewritten as where the effective concentration-dependent diffusion coeffi- cient for the dissolved species is The permeability can now be calculated explicitly where S = A/Db F = Dh/Db and yl = bpl,yh = bph Equation 25 reduces to the Paul-Koros expression for the permeability in the special case of S = 0 and PI = 0 In addition Fredrickson and Helfand14 treated the time lag and derived lengthy mathematical expressions which are said to reduce to the result of Paul and Koros as a special limiting case Within the limitations of the case for non-swelling polymer glasses their treatment seems to be quite satisfactory as a consistent dual-mode model for the penetrant transport 3 3 2 Models Which Consider Polymer Plasticization The dual-mode model of Paul and Koros assumes as does that of Fredrickson and Helfand that the penetrant solubility in the polymer is sufficiently low so that the plasticizing (swelling) effects can be ignored However such effects are measured and detected in a number of important physical applications There- fore the dual-mode model of Paul and Koros was extended by Stern et a1 to take into account the plasticization of the polymer by penetrants These investigators showed that the extended dual-mode model could be used to describe the solu- tion and transport of methanol benzene and acetone in ethyl cellulose and of water in poly(acrylonitri1e) as well as the solution of vinyl chloride in poly(viny1 chloride) Recently Zhou and Stern16 considered the effect of plasticiza- tion on diffusivity in the Henry's law mode and the Langmuir mode separately For convenience the authors assumed DDand DH to be exponential functions of CDand CH,respectively where DD(0)and DH(0)are the mutual diffusion coefficients in the limits of infinite dilution /3 and pHare empirical parameters that depend on the penetrant/polymer system and the tempera- ture and characterize the effects of plasticization on penetrant transport in the dissolved and trapped modes respectively The effective diffusion coefficient D,K is expressed by where The ratio DeK/D(0)is plotted in Figure 816 in accordance with equation 28 as a function of the dimensionless penetrant pres- sure bph(= uC,) for the case withpi = 0 As shown in the Figure 8 D,E will increase with increasing plasticization (BD,pH > 0) and decrease as a result of antiplasticization (/3D,/3H < 0) or clustering of penetrant molecules Both effects can occur simul- taneously each effect predominating in a different pressure range This is illustrated for fl= -0 05 for which the plot of D,K/D(O)passes through a maximum Such a maximum has been observed for water in poly(acrylonitri1e) 0 c225-c c g2'0-u a= g1 c $1 3 c g0 0 10 -2 lo-' loo 10' 1o2 Dimensionless penetrant pressure bph Figure 8 Ratio of diffusion coefficient Defi/DD(0)as a function of dimensionless penetrant pressure bph for different values of pD (Reproduced by permission from J Polym Sci Part B Polym Phys 1989,27,205 ) This model of Stern and his co-workers16 requires seven parameters kD 6,CH,DD(O),DH(O),pD,and BH This substantial number of parameters though it may be required by the complexity of the problem can inhibit wide applications of this model CHEMICAL SOCIETY REVIEWS 1993 4 NMR Investigations 4.1 The Spin-Echo Method with a Pulsed Field Gradient for Diffusion As a consequence of diffusion the sorbed molecules change their locations in the polymer sample.In an inhomogeneous magnetic field the component of the displacement in the direction of the field gradient changes the Larmor frequency of nuclear spins attached to the diffusing molecules. The amplitude of the spin echo is reduced as the molecules diffuse to greater distances during the interval T between the pulses. The spin-echo method is based on the difference between the amplitude of the echo with and without a field gradient G( = dH/dx) applied for a time 6.Although the main magnetic field (H x4T) is changed only slightly by the gradient which is typically about 0.02 T cm-l the high sensitivity of the NMR method can detect the changes of the Larmor frequency of nuclei moving more than about 10 nm. With the pulse sequence shown in Figure 9,17 the ratio R of spin-echo amplitude in the presence of a gradient A(G)to that with no gradient A(0)is R = A(G)/A(O)= exp[ -y2DG2s2(A-8/31] (30) Here y is the gyromagnetic ratio of the nucleus and d >> 6 is the diffusion time i.e. the interval between the times at which the consecutive gradient pulses are applied. The diffusion coefficient can be obtained from the slope of a plot of In R vs.Sz at constant G and A. ECHO ATTENUATION -'(GI = exp[-y2,,*G 2*DGj2*(A-S)]A (0) n/2 STIMU LATE D n ~-8-ECHOA(G)I' L Figure 9 Pulsed-gradient spin-echo method (a) and pulsed-gradient stimulated-echo method (b) for the determination of the self-diffusion coefficient. (Reproduced by permission from J. Polym. Sci Polym. Phys. Ed. 1985 23,387.) Zupancic et al. studied the diffusion of sorbed butane in linear poly(ethy1ene) (PE) as a function of the vapour pressure of butane. The PE has a weight-average molecular weight of 52000 and a volume of crystallinity of 0.714. The observed pressure dependence of the diffusion coefficient is an exponential func- tion of the vapour pressure p at 23 "C D = 3.77 x lo-* exp(1.125~) (31) where Dis in the units of cmz s-andp is between 0.3 and 3 atm.Zupancic et al. did not report the temperature dependence of the diffusion coefficient. 4.2 Proton Spin-Lattice Relaxation Studies of Poly(dimethy1siloxane) under High Pressure of Gases by Assink This molecular behaviour of PDMS exposed to gases pressur- ized up to 2040 atm was investigated by Assink.I8 Figure 10 shows the behaviour of proton TI as a function of pressure for the gases helium nitrogen and argon. At room temperature it is well above the temperature at which (T,), occurs so that decreasing the molecular motion (bringing it closer to the resonance frequency) decreases T,. Helium pressurization exhi- bits this typical pressure dependence of decreasing T with increasing pressure.In contrast a very small pressure depen- dence is exhibited by PDMS when nitrogen is used as the pressurizing medium and the opposite pressure dependence is shown by PDMS when argon is used as the pressurizing medium. 1.61.4 L Argon 1.2 1/@ 1.o h S. 0.8 L-0.6 Helium\,w0.4 I I I I 0 50 100 150 200 250 Pressure (MPa) Figure 10 Spin-lattice relaxation time of PDMS vs. pressure for the gases helium nitrogen and argon. (Reproduced by permission from J. Polym. Sci. Polym. Phys. Ed. 1974 12 228 1.) Before considering the mechanism involved we wish to exam- ine the time behaviour of TI following the gas solution process. If a sample of PDMS is suddenly exposed to a soluble gas its relaxation time should increase steadily from a small value to the equilibrium as the gas dissolves. It is known that the uptake of gas by the sample is proportional to tt for short times.If one assumes that the change of the rubber T is directly related to the concentration of the sorbed gas then one expects that it too should be proportional to ti as shown in Figure 11. Due to its small atomic size and large solubility helium gas applied to PDMS acts only as a pressurizing medium. In contrast two mechanisms are considered when nitrogen and argon gas pressures are applied to the polymer they are pressur- izing and plasticization mechanisms. The dissolved gas increases the internuclear distances of adjacent chains and decreases the effectiveness of their dipolar interaction thus increasing the relaxation time TI.This plasticizing effect of PDMS by argon is greater than that by nitrogen as shown in Figure 10.Figure 12lSshows the proton TI of PDMS as a function of temperature and three gas pressures at about 1 atm under helium pressure of 1362 atm and under argon pressure of 1362 atm. At atmospheric pressure the motion which results in the T minimum at -73°C corresponds to the motion responsible for the glass transition temperature. But the Tg of PDMS is -123°C as measured by differential scanning calorimetry or dynamic mechanical spectroscopy. These techniques are sensi- tive to low freqencies (x1 Hz) while the frequencies that are characteristic of TIare near the Lamor frequency of the protons at 90 MHz. The temperature associated with the glass transition depends upon the frequency of the technique used to measure it.The glass transition temperature of -73 "Cby NMR TImethod MOTION OF SORBED GASES IN POLYMERS-WEN-YANG WEN I 0 loo I 0 90 h v) 0 70 0 60 0 20 40 60 80 11 Time' (s2) Figure 11 Spin-lattice relaxation time of PDMS exposed to 957 atm nitrogen at time zero (Reproduced by permission from J Polym SCI Polym Phys Ed 1974 12,2281 ) 20 10 h u) K 05 02 -1 50 -1 00 -50 0 50 Temperature("C) Figure 12 Spin-lattice relaxation time of PDMS vs temperature for various conditions 0,helium at p = 1 atm A,helium at p = 1362 atm 0,argon at p = 1362 atm (Reproduced by permission from J Polym Sct Polym Phys Ed ,1974 12,2281 ) is 40 to 60 degrees above the Tgof -123"C by DSC or DMS methods The WLF equation can be used to correlate these temperature shifts with the frequency of observation As discussed above we know that helium gas behaves as a pressurizing medium and decreases the molecular correlation frequency of the PDMS We expect therefore that the pressuri- zation by argon is less distinct and is shifted to the left of the minimum for helium The minimum enables us to determine the temperature -61 "C at which the molecular correlation fre- quency is near the resonance frequency The smaller shift of -73 "Cto -61 "C must be interpreted as plasticization by the argon gas dissolving into the rubber and allowing motion at -61 "C which would otherwise be allowed only at the higher temperature of -35 "C 4.3 General Considerations of Relaxation Mechanisms and Motion of Gases Sorbed in Polymers Let us consider carbon- 13-enriched carbon dioxide As a pure gas CO may relax primarily by the spin-rotation (SR) mecha-nism through collisions When the gas is sorbed in a polymer the contributions of various relaxation mechanisms are expected to change For example simple liquids like CS have approxima- tely comparable relaxation contributlons from the chemical shift antisotropy (CSA) mechanism through rotations and the SR mechanism through collisions Carbon dioxide dissolved in a polymer matrix is in a con- densed phase and therefore should be closer to liquid CS than gaseous CO Although enriched CO is used in the polymer experiments homonuclear dipole-dipole (DD) relaxation is not likely since the sorbed gas is typically only several percent of the sample by weight when the gas pressure is about 10 atm (A good NMR signal in PC is obtained when the CO pressure is at or above several atmospheres ) However heteronuclear DD relax- ation between protons on the polymer and the carbon- 13 of the CO is highly probable since the protons have a significant concentration in the samples 2o Each relaxation mechanism provides access to a different type of motional information Spin-rotation relaxation depends on collisional changes in the rotational angular momentum quan- tum number and would provide a measure of the collision frequency of the sorbed gas in the polymer The CSA mechanism depends on the rotational correlation time and thus provides information on molecular rotation The intermolecular dipole- dipole relaxation mechanism depends on the translational motion of the sorbed gas molecule through the polymer matrix and on the distance of closest approach to the polymer protons This mechanism provides a short-range measure of the transla- tional diffusion coefficient of the penetrant molecule The nature of the polymer matrix is also important in influencing the mobility of the penetrant Diffusion of pene- trants through rubbers is expected to be characterized by a different formalism than diffusion through glasses Diffusion in rubbers may be treated as though the rubber is a simple liquid and the penetrant is dissolved in the polymer Molecular level information from spin relaxation can be used to test this approximation Penetrants in glassy polymers are usually des- cribed by the dual-mode model As discussed in Section 3 3 the penetrant in this model may be sorbed in either Langmuir sites or dissolved sites the latter being comparable to the mode of sorption in a rubber Different translational diffusion times are assigned to the two types of penetrants in a glassy polymer From this viewpoint one would expect changes in spin relaxa- tion as one changes from a glassy polymer to a rubbery polymer or as one traverses the glass transition of a given polymer A typical spectrum of CO gas sorbed in thin strips of glassy polycarbonate (PC) is shown in Figure 13,O along with a spectrum of the gas sorbed in powdered PC According to these spectra the resonance frequencies for the 13C02 sorbed in powdered PC and in PC strips are rather different Since relaxation of gas sorbed in the bulk polymer is our primary interest sufficiently thick samples (blocks or stacked disks) should be used so that a resolved resonance for the sorbed gas can be identified and measured 4.4 Relaxation Mechanisms for 13C0 in Silicone Rubber In order to characterize various group motions in polymers an increasing number of investigators are using a fractional correla- tion function of the type Cain et a1 also used this Williams-Watts type correlation 13c02in Polycarbonate powder I I I I I 1'1'1 I 130.0 128.0 126.0 124.0 122.0 120.0 118.0 116.0 114.0 PPm Polycarbonate film strips 1 '1 I I'I'I'I' I I 130.0 128.0 126.0 124.0 122.0 120.0 118.0 116.0 PPm Figure 13 Carbon-13 NMR spectra of 3C0 dissolved in powdered polycarbonate and 5-mm polycarbonate strips.The free gas reso- nance is at 124.2 ppm. Both spectra recorded at 62.9 MHz. Note the relative amounts of dissolved and free carbon dioxide are not equiva- lent for these two samples. (Reproduced by permission from J. Phys. Chem. 1990,94,2128.) function to depict motions of CO in silicone rubber.21 In equation 32 T~ is the central correlation time and sets the position of the distribution on the time axis.The parameter a which falls in the range of 0 < a < 1 determines the width of the distribution. The relaxation data modelling and fitting procedure begins with initial guesses for the values of six parameters ~~,(300K) the correlation time for spin rotation at 300 K; ~~~~(300K) the correlation time for CSA at 300 K; b the distance of the closest approach between a proton and a carbon-13 nucleus; AH the activation energy; and a the width parameter of the distribu- tion. These parameter values are adjusted in an effort to fit the experimental data and minimize the sum of squares (the least squares). The parameters of the best fit are = 1.6 xrs,(300K) = 8.8 x s; ~~~~(300K) s = 3.8 X m7S; dH = 1.6 kJ mOl-' (33TDD(~OOK) b= 3.9 x lo-' cm; a = 0.58 The average values for the dipole-dipole correlation time a id translational diffusion coefficient are given by (TDD) = XPkTk = 7.8 x s k (34) (D,)= xpkDk= 1.8 x 10-cm2 s-l k where pkis the weighting factor (or fractional population) for the kth relaxation time.Since Dk = b2/Tk the average diffusion constant corresponds to the harmonic average of the correlation time and vice versa. CHEMICAL SOCIETY REVIEWS. 1993 It is interesting to note that the obtained average of the self diffusion coefficient 1.8 x cm2s-l is close to the mutual ~diffusion coefficient of 2.64 x 10 cm2s-reported by Stern et al. from their permeability measurements of CO sorbed in silicon rubber. Given the complexity of the interpretation of the NMR data and the short-range viewpoint of the NMR experiment the agreement is better than would be expected.With regard to the spin rotation the correlation time ~SR(z8.8x s) found for *3C0 (10 atm) sorbed in silicone rubber turns out to be much closer to the value of the liquid CS (9.3 x s) than that for gaseous CO (4.6 x 10-lo s). As pointed out above Stern et al. reported the diffusion coefficient of CO in PDMS as 2.64 x cm2s-I at 35 OC. According to Figure 14 however the population fraction of CO molecules having these high values of diffusion coefficients is very small. Then what is a qualitative explanation for this apparent discrepancy? The process of CO permeation through the rubber under a pressure gradient may be imagined to take place in the following way.-___0.20 a = 0.58 0.15 6 0.10 -0.05 / '=-0 ----,---. L-c-::>-A ---.-The fractional population of CO with diffusion coefficients of lop6 to cm2 s-' is indeed very small but this small population of CO dominates transport. It is analogous to the important role played by the path of least resistance in a parallel circuit carrying a flowing current. The simple weighted averag- ing of diffusion coefficients in a system characterized by two or more diffusion coefficients leads to the parallel-circuit result. In relation to the sorbed gas diffusion problem at hand the internal states would correspond to sites in the polymer matrix with different sorbed gas mobilities. 4.5 Relaxation Mechanisms of 13C0 in Polycarbonate Glass For consideration of relaxation mechanisms for 3C0 in polycarbonate which is a glass at temperatures below 150 "C,a simple data set comparable to that for silicone rubber (Table 1) is shown in Table 2.Again a field dependence is apparent in the Table 1 Relaxation times and nuclear Overhauser enhancement for 13C0 sorbed in silicone rubber at 273 K and 11 atmZ0 Frequency/MHz 22.6 50.3 62.9 100 TI Is 9.7 11.7 13.1 T2lms 150 150 NOE 1.14 1 .o 1.o MOTION OF SORBED GASES IN POLYMERS-WEN-YANG WEN Table 2 Relaxation times and nuclear Overhauser enhancement for 3C0 sorbed in polycarbonate at 333 K and 5 atrnz0 Frequency/MHz 22 6 50 3 62 9 126 TIIs 46 36 33 22 T2lms 141 40 NOE 128 108 10 TIdata of Table 2 but one that is opposite to that of the silicone rubber system This type of field dependence is characteristic of CSA relaxation and specifically for the fast motion (short correlation time) regime A CSA contribution is an indication that rotational motion in this system must be slowed down relative to that in the silicone rubber This is not unreasonable considering that we have gone from the relatively high mobility of a rubbery matrix to that of a glass An NOE is also present in this system and therefore dipoledipole relaxation must also be a contributor In addition it is necessary to invoke a consider- able contribution from spin-rotation relaxation Cain et a1 23 presented a quantitative interpretation based on a two-site rapid exchange model In particular they follow the lattice model of Bendler Shlesinger and Weiss 24 This model contains at least 13 potentially adjustable parameters TsRand TCSA for the hole site TSR and TCSA for the dissolved site diffusion coefficients for hole-to-hole sites (DHH) for hole-to- dissolved (DHD),for dissolved-to-hole (DDH),and for dissolved- to-dissolved (ODD),distance of closest approach (b),activation energy for the hole site AHH activation energy for the dissolved site AHD,the mole fraction of dissolved sites (XD),and lastly the population of the dissolved site (PD)as a function of tempera- ture To obtain a credible interpretation a number of these potential parameters are set to the literature values and a number of others are interrelated and thereby eliminated as adjustable (Eventually the number of adjustable parameters are reduced to five ) This process does involve simplifying assump- tions which may or may not be too severe The parameters and the best-fit values obtained by Cain et a1 are listed below Ts,(dissolved) < I x 10 l4 s TCSA(dissolved)= 4 4 x 10 s TcsA(hole) = 4 3 x 10 s AHD = 10 5 kJ mol DHH= 1 x 10 locm2 A summary of constants entered into the fitting process may also be listed DDD=150x 10 'cm2s ' dHH=dHD+62=167kJmol b=32x10 scm XH =2 x 10 3(Tg-T) (35) Overall the interpretation of the spin relaxation data for 13C0 in glassy polycarbonate leads to a measure of local dynamics which is consistent with the conceptual basis of the dual-mode model Motion in the dissolved site is liquid-like and is similar to that observed in liquid CS and not too different from CO sorbed in silicone rubber Motion in the hole or Langmuir site is considerably slower in agreement with the proposed nature of this site The lattice model provides a connection between the microscopic view of NMR data and the macroscopic measurement of translational mobility derived from permeability The dissolved-to-dissolved type of diffusion is the dominant contributor to permeability from the lattice model viewpoint The hole-to-hole diffusion was negligible with a very small diffusion coefficient consistent with the assumption The two-site model used here is consistent with the dual-mode picture and does provide a substantially better quantitative interpretation than a distribution of correlation times associated with a skewed bell-shaped curve Unfortunately their data are not absolutely decisive in support of the dual-mode model It is the strong dependence of T on temperature that could be matched by a two-site interpretation and could not be matched with the distribution model 5 Computer Simulation Diffusion of small molecules in polymers may be simulated by computers using equilibrium or non-equilibrium molecular dynamics (MD) For example an equilibrium MD simulation of an assembly of n-alkane-like chains together with a low concent-ration of one-centre Lennard-Jones particles has been per- formed by Trohalaki et a1 This work is a logical extension of their work on n-alkanes without penetrants The potential energy as a function of bond length is given by where I is the equilibrium bond length The deformation of the bond angle 8 from its tetrahedral value 8 is governed by a potential function which is quadratic in cos8 The dihedral angle 4 is constrained to lie mainly in the trans and gauche states by a torsional potential 5 Ed = k 1ancosn+ nO Non-bonded interactions are evaluated by a truncated Len- nard-Jones potential ELJ= 4r*[(r*/rJ12 -(Y*/Y,~)~] + c for rIJ< 1 5 Y* (39)=o for Y > 1 5 Y* The simulation consists of integration of Newton's equations of motion of the individual segments and penetrant particles In their work the parameters kb I koand so on were chosen such that the chain model will mimic a polyethylene molecule and the segment represents a CH group The simulation system con- sisted of a cubic box containing 500 CH segments (25 chains) and four penetrant particles The simulations were performed at constant N V,and E with At of 10 fs A Cray Y-MP computer was employed and each simulation was run for 1O5 time steps for a duration of 1 ns The self-diffusion coefficient of a component in a mixture D (a= 1,2) can be calculated from the Einstein relation D,= ( 1/6Na) lim 2 < [B,(O) -B,(t)l2 > '*XI 0 where v is the velocity vector of a particle and N is the number of particles of type a Some of the results for self-diffusion coefficients are given in Table 3 for the case of a penetrant particle with r* = 0 418 nm The values of the diffusion coefficients found by simulation are somewhat higher for example for CO at 361K the simulation gives 17 8 x 10 cm2 s ',while the experimental Table 3 Self diffusion coefficients for polymer segment and penetrant with Y* = 0 418 nm25 T/K D(polymer)/lO cm2 s D(penetrant)/lO cm2 s 42 1 7 082 f0 462 28 23 f2 57 361 6 013 f0 208 17 81 f0 54 300 3 848 f0 431 13 38 f0 93 240 1 556 f 0 I12 7 16 f0 31 value is 4 44 x lops cm2 s-l The major factor most likely to have given rise to these discrepancies is the difference in liquid density At 361 K the simulated system has a density of 0 576 g ~m-~,whereas the experimental C system has a density of about 0 741 g cm-3 at 13 6 atm The much larger free volume available in the simulated system is considered by Trohalaki et a1 to be responsible for the increased diffusion coefficients Gusev and Suter26 have presented a theory for the solubility of small particles in static structures of host matrices This theory is based on the statistics describing a gas dissolved in a polymer with sites that can be occupied by at most one solute particle (a spatial Fermi gas) The theory has been applied to the methane solubility in computed matrices of amorphous polycar- bonate that cannot be plasticized The theory seems to yield the correct order of magnitude and the proper functional form of the sorption isotherm The computed distribution functions for the methane solubility in the glassy polycarbonate indicate that the sites with the values of l/b z 10 bar are of special importance in agreement with the dual-mode sorption model with bH =0 095 bar- However according to the authors the distribution of sites is of a form that does not render it amenable to a quantitative reduction to the shape predicted by the dual- mode model One may speculate tentatively either that there is only one very broad distribution of sites or that the distribution of the Langmuir mode is narrow while the distribution of the Henry’s law mode is broad 6 Conclusions Various workers have investigated gas sorption and transport for many years mostly from macroscopic viewpoints The time has come for researchers to characterize the motions of the sorbed gas in polymers on the molecular level The dual-mode model which has been employed by a large number of polymer chemists and engineers is now under scrutiny by some investi- gators Early results of a few NMR experiments and computer simulations are beginning to shed new light on the motional dynamics of both the penetrants and polymer chains The jury is still out and the outcome for the verdict on the dual-mode model is yet uncertain but the time for its unravelling cannot be too far away Acknowledgements The author wishes to acknowledge grant supports from the NSF (# DMR-9001678) and the Depart- CHEMICAL SOCIETY REVIEWS 1993 ment of the Army (# DAAL03-9 1-G-0207) His principal co- investigators are Alan A Jones and Paul T Inglefield 7 References 1 G K Fleming and W J Koros Macromolecules 1986 19,2285 2 P N Lowell and N G McCrum J Polym Scz Part A-2 1971,9 1935 3 S A Stern V M Shah and B J Hardy J Polym Sci Part B Polym Phys ,1987,25 1263 4 T A Barbari W J Koros and D R Paul J Polym Sci Part B Polym Phys ,1988,26,709 5 S S KulkarniandS A Stern J Polym Scr Polym Phjs Ed 1983 21,441 and 467 6 H Fujita Fortschr Hochpolym Forsch ,1961 3 1 7 W J Koros D R Paul and A A Rocha J Polym Sci Polym Phys Ed 1976,14,687 8 W J Koros A H Chan and D R Paul J Membrane Sci ,1977,3 I65 9 R T Chern W J Koros H B Hopfenberg and V T Stannett ACS Symp Ser ,1985,269,28 10 R J Pace and A Datyner J Polym Sci Polym Phys Ed 1979,17 437,453 and 465 11 A T DiBenedetto J Polym Scr ,1963 A-1 3459 12 A Kloczkowski and J E Mark J Polym Scr Part B Polym Phys 1989,27 1663 13 P Meares J Am Chem Soc ,1954,76,3415 14 G H Fredrickson and E Helfand Macromolecules 1985 18,2201 15 S A Stern and V Saxena J Membrane Sci ,1980,7,47,1982,12,6 16 S Zhou and S A Stern J Polym Sci Part B Polym Phys ,1989 27,205 17 I Zupancic G Lahajnar R Blinc D H Reneker and A Peterlin J Polym Sci Polym Phys Ed 1978,16 1399 18 R A Assink J Polym SCI Polym Phys Ed 1974 12 2281 19 H W Spiess D Schweitzer U Haberlen and K H Hausser J Magn Res 1971,5 101 20 E J Cain W -Y Wen R D Jost X Liu Z P Dong A A Jones and P T Inglefield J Phys Chem ,1990,94,2128 21 E J Cain A A Jones P T Inglefield R D Jost X Liu and W -Y Wen J Polym SCI Part B Polym Phys ,1990,28 1737 22 S A Stern V M Shah and B J Hardy J Polym Sci Part B Polvm Phjs ,1987,25 1263 23 E J Cam W -Y Wen A A Jones P T Inglefield B J Cauley and J T Bendler J Polym Sci Part B Polym Phys ,1991,29 1009 24 J T Bendler M F Shlesinger and G H Weiss Macromolecules,to be submitted 25 S Trohalaki,A Kloczkowski J E Mark,D Rigby,and R J Roe in ‘Computer Simulation of Polymers’ ed R J Roe Prentice Hall 1991 pp 220-232 26 A A Gusev and U W Suter Phys Rev A ,1991,43 6488

ISSN:0306-0012    

DOI:10.1039/CS9932200117

出版商:RSC    

年代:1993

数据来源: RSC

摘要:

Thermodynamic and Related Studies of Amphiphile + Water Systems Michael I. Davis Department of Chemistry University of Texas at El Paso El Paso Texas 79968-0513U.S.A. 1 Introduction The family of alkyl poly(ethy1ene glycol) monoethers has the generic formula C,H + .(OC,H4),.0H; conventionally and conveniently abbreviated to C,E,. The alkanols form the sub- family C,,E,. The members of this family of non-ionic amphiphiles with m >/ 6 and n 3 3 have been classified as detergents and shown to be capable of forming micelles. An extensive account of the investigations that have been made of the aqueous mixtures of these species and the conclusions that have been drawn from those investigations can be found in reference 1. The binary aqueous systems are characterized by three important features (i) critical micelle concentrations (cmc) (ii) miscibility gaps with lower critical solution temperatures (LCST) and (iii) formation of one or more liquid crystalline mesophases. For species with a common alkyl chain the cmc values are found to be insensitive to variation in the number of oxyethylene groups in the polar head.They are however highly sensitive to the length of the alkyl chain. The cmc values (moles- l) at 25 “C are well approximated by equation 1. log (cmc) = (2 -nz/2) (1) Studies of the C6E3+ water system have shown that its cmc decreases significantly as the temperature increases. The same trend is presumed to exist for the other systems. The lower critical solution temperatures increase significantly with the number of oxyethylene groups for species with a common alkyl chain and decrease with the length of the alkyl chain for species with the same polar head group.The miscibility gaps are asymmetric with the critical solution concentrations being only slightly higher than the cmc values. At temperatures just above the LCST both phases are micellar. The more amphiphile-rich phases contain a substantial mole fraction of water. Five types of liquid crystalline mesophase have been reported for C,,E + water systems but rarely are more than two or three found in any given system. The lattice symmetries of these mesophases reveal a progression of micellar shapes as the amphiphile mole fraction increases. The first mesophase is of cubic symmetry. It is a close-packed lattice of spherical micelles Born in London in 1936 Michael Davis MUY educated at Latymer Upper School and Universitjj College London. His post-graduate studies in the area ofgas phase electron diflraction were car- ried out at the University of Oslo under the direction of Odd Hassel. He joined the fa- cultj’ of the Universitjq ofTex-as at Austin in 1961 and has been a projessor of chemistry at the University of Texas at El Paso since 1971. His cur-rent research activities are focused upon the measure-ment analysis and interpre- tation of the thermodynamic properties of amphiphile + bcatev mixtures. with the interstices filled by water. This mesophase is only found for species in which the polar head group is relatively large compared to the alkyl chain. At somewhat higher amphiphile concentrations a hexagonal mesophase frequently exists. It consists of cylindrical micelles with water again filling the interstices.The third mesophase is also of cubic symmetry its proposed bicontinuous structure is described as a rather compli- cated three-dimensional network of interwoven ribbon shaped micelles. The fourth mesophase is that of the most general interest. Described as lamellar it consists of membrane-like amphiphile bilayers that are separated by layers of water. The fifth most amphiphile-rich mesophase is hexagonal and con- sists of cylindrical inverted micelles with water filling the central channels. The liquid crystalline mesophases ‘melt’ to form micellar mixtures. It is assumed that the geometries of the micelles thus formed are related to those of the mesophase.Thus cylindrical micelles are expected to break up into rods and the lamellar sheets into disks. Thermodynamic investigations appear to have been restricted to the aqueous mixtures of species with m 6 6 and n 6 3. This is largely because it is difficult to differentiate between the thermo- dynamic properties of the very dilute premicellar solutions of the larger amphiphiles and those of pure water. There is a further problem with the substantial cost of the larger detergent species. C,E + water systems with m > 6 and n < 3 or with m < 6 and n > 3 may be of interest but have not been widely studied with the exception of the higher alcohol + water systems. Thermodynamic studies of the aqueous mixtures of ionic detergents reveal sharp changes in the slope de,/d.xA at the critical micelle concentration (where Q is a molar or excess molar thermodynamic property of the system and xA is the amphiphile mole fraction). Graphical evidence for such changes of slope is particularly striking in plots of the excess apparent molar properties @$ (= QE/xA)or @Q (= @$ + Qz A where Qlr,A is the molar property of the pure amphiphile). A major change in the approach to the analysis and interpre- tation of the thermodynamic properties of smaller amphi- phile + water mixtures resulted from the publication of the apparent molar isobaric heat capacities of the 2-butoxyethanol (C4E,) + water ~ystem.~ Although there is no evidence that C4E1 is capable of forming stable micelles the profile of this data is essentially identical to that found for the ‘true’ detergent + water systems. A plot of that data is shown in Figure 1.The change in slope occurs at xA = 0.019 which is very close to the composition predicted for the critical micelle concentration of an alkyl poly(ethy1ene glycol) monether with m = 4. Since the appearance of reference 3 the results have been published of a variety of studies of the C4E1 + water and other related systems.This review will attempt to summarize the information that has been accumulated from those studies and to present some possible interpretations. Most of the evidence that will be examined will come from thermodynamic sources. 2Thermodynamic Experiments Descriptions of the experimental techniques that are most widely used for measuring excess Gibbs free energies excess enthalpies and excess volumes with appropriate references are to be found in reference 4.That review also discusses some results of the thermodynamic investigations of non-aqueous binary liquid systems. Most of the recent published data for the isobaric heat 127 32’3 A 3.90 n 3 A 3 286 A 260 240 220 n 200 180 160 l4I Figure 1 Apparent excess molar isobaric heat capacities of butoxyetha no1 in its aqueous mixtures Units J K * mol capacities of binary aqueous mixtures have been measured using Picker-type flow microcalorimeters Isobaric expansivities can be obtained from density measure- ments at several different temperatures Molar volumes of aqueous mixtures appear to be fairly well approximated by quadratic temperature dependence Consequently densities or excess molar volumes are needed for at least three but prefera- bly more different temperatures In recent years there has been an increased interest in the determination of either isothermal or isentropic compressibili- tiesThe role that isentropic compressabilities play in the thermodynamic characterization of binary aqueous mixtures is the subject of a recent review 3 Graphical Presentation of Data The thermodynamic properties of binary liquid mixtures are measured as a function of cornposition in order to identify regions over which significant changes occur in the patterns of molecular aggregation It is clearly desirable to generate visually sensitive graphical representations of the data Plots of a directly measured variable such as density or ultrasonic speed may be interesting but it is usually preferable to work with molar quantities A given data set Qm,can contain important information about its composition dependence that is not evident from a plot against mole fraction because of a substantial difference between the values for the two pure componentsThat problem can be alleviated by converting the molar quantities into their excess molar counterparts An excess molar property is defined by the expression QE = Qm -Q2 where Q$ is the molar property of an ‘ideal’ mixture of the same composition The thermodynamic definition of ideality for a binary liquid mixture A + B formed under the conditions of constant tem- perature and pressure is where GZ A and GZ are the respective molar Gibbs free energies (chemical potentials) of the two pure liquid components Equation 3 is equivalent to Raoult’s law provided that the vapours of the two liquids can be treated as ideal gases Expressions for other ideal molar properties can be derived by assuming that such relationships as exist between the properties CHEMICAL SOCIETY REVIEWS 1993 of pure substances (or of real mixtures) also hold true for ideal mixturesThus The ideal molar values of the volume enthalpy isobaric heat capacity (C,) and the quantities A = (6 V/ST),and K = -(6V/ 8qp are given by the mixing rule It should be observed that this is not the approach to defining ideal and thus excess molar properties found in most text- booksThat approach suffers from an inability to define ideal values for isochoric and isentropic properties like Cvand Ks That inability resulted in the appearance in the literature of a variety of different definitions of KsE bhich are inconsistent with equation 3 The problem was resolved by Benson and Kiyohara’ who adopted the logical approach that if a thermo- dynamic quantity such as Cv,can be evaluated from other properties for which the ideal molar values are unambiguously defined then the ideal molar values of that property are similarly accessibleThus since it follows that CCm= Cfm-T(Afm)2/K’$m (7) Benson’s approach provides access to the ideal and excess molar values of a substantial number of different molar proper- ties of binary liquid mixtures some of which have very interest- ing composition dependence * Plots of excess molar properties frequently fail to reveal all of the important features of their composition dependence A number of different strategies are available to project out some of the more subtle features It is sometimes useful to generate plots of the apparent (or excess apparent) molar properties of the amphiphilic compo- nents @PA and @bA tend to exhibit interesting composition dependence at low amphiphile concentrationsThat makes them particularly useful when one is seeking information about critical micelle concentrations or their analogues QQA is also useful when dealing with the thermodynamic properties of aqueous solutions of solids and sparingly soluble liquids An alternative to the excess apparent molar properties that are more informative about the composition dependence at higher amphiphile concentrations are the reduced excess molar properties One notes that QEd = a constant is a descriptor of the simplest type of non-ideal mixing An important characteristic of both the apparent and reduced properties is that they are derived from Q,, or QE,on a point-to- point basis and require no type of curve fitting or local smoothing of the data Somewhat more composition-sensitive are the partial excess molar properties Qi and Qb of the amphiphile and water respectively which are obtained from the equations (9) It is necessary to have a reliable method of evaluating the derivatives dQE/dxA at each datum pointThe method should involve local rather than global scrutiny of the form of the data In this laboratory we have chosen to use a modified cubic splines approach The conventional cubic splines procedure requires THERMODYNAMIC AND RELATED STUDIES OF AMPHIPHILE + WATER SYSTEMS-M I DAVIS that the fitted curve pass exactly through all of the experimental QEdatum points This means that all of the errors in the data set are retained and as a consequence the first derivative curve is likely to be quite erraticThe modification that has been adopted involves two cubic splines arranged asymmetrically about each datum point Each spline spans at least five and up to ten datum entries if the point density is sufficient Both splines are required to have the same values of QE and of dQE/dxA at the mole fraction of interest but can have differing second and third derivatives This approach provides a modest level of smoothing of the original data The first derivative curves that it provides generally appear to be satisfactory The major shortcoming that we have encountered has been a tendency to smooth out some of the more dramatic ‘real’ features of the data A comparison of some different ways of providing graphic displays of the composition dependence of thermodynamic data sets is provided by Figure 2These curves were derived from the isobaric excess molar heat capacity data of Benson et al for the 1-PrOH + water system It is seen that dcpE/dsA and Cpg exhibit more dramatic composition-dependence than do either @CE or C/7redE ‘40T T 240 0 -200 -160 0 -80 Figure 2 Excess molar isobaric heat capacities of I-PrOH + water Open circles excess apparent molar C of 1-PrOH Closed circles dCPE/dXA Open squares excess partial molar Cpof 1-PrOH Closed squares reduced excess molar C Units J K-’ molt’ 4 Curve Fitting Strategies The conventional strategy for fitting smooth curves to the excess molar property data of binary liquid mixtures involves the use of the Redlich-Kister equation O I In The coefficients C,are evaluated by least squares analysis and the optimum number of terms n is determined by standard regression criteria Equation 11 serves well for non-aqueous systems but has been found to be ill-suited for fitting the data of most C,,,E + water systems A substantial fraction of the investigations of binary aqueous systems has been devoted to water-rich mixturesThe most common approach has been to represent the apparent molar properties of the solute species by a polynomial function of its molal concentration This is the McMillan-Mayer equation @bA is the apparent and also the partial molar property of the solute at infinite dilution Several authors have devised group additivity rules which do a reasonable job of correlating the values for a substantial number of structurally related solutes l2 l3The explanation for the success of the additivity rules for Q = I/ and Cpis that each atomic group produces a characteristic modification to the pattern of self aggression of the solvent water molecules in their immediate vicinity The effects of the interactions between pairs of solute mole- cules is represented by b Wood and his co-workers have made extensive studies of b values for a large number of binary aqueous systems leading to the establishment of the Savage- Wood grouppair additivity rule l4 That rule is consistent with random mutual orientation of the solute pairs and indicates that b is more likely to be associated with overlap of the hydration co-spheres than with direct solute-solute contact The djQ A curves for C,E + water systems appear to possess quadratic dependence upon molality (or mole fraction) over a modest composition rangeThe quadratic coefficient c IS des-cribed as being related to cooperative association of the amphi- philic solutes What is being considered here is a ‘hydrophobic effect’ where the solute species tend to cluster in such a way as to minimize direct contact between their alkyl tails and water That results in enhanced contact between pairs of non-polar groups There have been two major approaches to extending the curve fitting process beyond the extreme water-rich concentration range Each is based upon a distinct concept of the significance of the cmc The ‘mass action’ approach treats a micellar aggregate as a polymeric species and thus deals with the equilibrium nAc-+A (13) It is argued that the equilibrium stays well over towards the monomer side at concentrations up to just below the cmc at which point it swings rapidly over to the polymer sideThis concept has been incorporated into an analytical model by Desnoyers and Roux l6 The ‘pseudophase’ approach treats the micellar aggregates as constituting a distinct dispersed microphase In that context the critical micelle concentration is the solubility of the micellar microphase This carries with it the implication of some type of discontinuity in the partial molar properties other than the chemical potentials at the cmcThe pseudophase approach has been incorporated into the semi-empirical four-segment compo- sition model Referring to the excess molar enthalpies of ethanol + water Franks and Ivesl* wrote of ‘the necessity for separate consider- ation of at least two composition ranges for they may involve quite separate problems ’Many authors in their interpretations of the physical properties of binary hydro-organic systems have drawn attention to what they regard as evidence for the existence of several discrete composition ranges As its name suggests the four segment model treats the total composition range of a binary non-ionic amphiphile + water system as being made up of four-separate parts (segments) Across each of the segments the excess molar properties are assumed to possess a distinct and relatively simple mole fraction dependence QEmust be continuous at each of the three segment junctionsThe version of the model described in reference 17 also requires continuity of dQE/dxA at two of the three junctions That version provides an acceptable quality of fit for the dvailable data sets closer scrutiny indicates it to be maybe more flexible than is necessary for many of the data sets and in such cases it has been found that further constraints can be intro- duced into the model without serious detriment to its curve fitting ability 5 Composition Dependence of the Patterns of Molecular Aggregation Before considering the algebraic details of the four-segment 130 CHEMICAL SOCIETY REVIEWS 1993 model it is proper to examine some of the conceptual aspects of hydrogen bonds to the solvent water moleculesThe thermody- treating these binary systems as having partitioned composition namic data indicate that this favours the more compact types of ranges Some justification for this approach may be found in the aqueous self aggregation but at the same time may bring about appearance of dHE/dxA for ethanol i-water (Figure 3) as enhancement of hydrogen bonding efficiency among their neigh- obtained from the data of reference 19 bouring water molecules There appear to be two distinct types of solute-solute interac- tion Hydrogen co-sphere overlap is thought to dominate at very low amphiphile concentrations but cluster formation becomes 4000~T 4000 progressively more important as the amphiphile concentration increases That situation seems to account for the appearance of the pwcurves Figure 4 shows the water-rich pwcurve for 1-propanol + water at 25 "C generated from the data of reference 22 -2000 rJ -2000tI. 7-4000i ,q -.-4000 TO2 -boo0 .-5000oO1loo'000 -3000 .. -8000 000 -1 GOCG -.-'0000 -l?OOO1 1-12000 -002110 01 02 03 04 05 Ob 07 08 09 10 x (EtOH) Figure 3 Excess molar dH/dXAfor EtOH + water Units J mol-' A water-rich segment extends up to the inflexion point in the vicinity of xA = 0 1 There is a central segment that extends from roughly xA = 35 to 65 where dHE/dxA is very nearly linear A similar linear feature is present in the d VE/dxA data of x (PrOH)I-PrOH + water 2oThree segment junctions may thus be assigned in the vicinities of x = 0 1 x2 = 0 35 and xg = 0 65 Figure 4 Excess partial molar volumes of water in I-PrOH + water splitting the total composition range into four parts Units cm3 mol It has been found that a satisfactory fit of water-rich excess 5.1 The Water-rich Segment 0 < xAdx molar property data can be obtained with a cubic function of xA Over the past twenty years our ideas about the structure of The segmentjunction YA = yl is the mole fraction above which liquid water have been strongly influenced by the results of that cubic equation is no longer satisfactoryThe water-rich computer simulations While there are some differences between segment may be limited for any one of several reasons (I) In the the conclusions drawn from molecular dynamics2 lU and Monte case that the amphiphile is truly capable of forming micelles the Carlo21h simulations the basic picture appears to be much the water-rich segment is limited by the cmc (11) Where as appears same to be the case for 2-butoxyethanol one is obliged to talk of One may view a water sample as a continuous network of micelle-like or pseudomicellar aggregates then Y might be relatively short-lived hydrogen bonds Both types of simulation termed a cmc analogue (111) There are systems like I-BuOH indicate that the fraction of water molecules with no hydrogen + water in which the amphiphile has limited solubility and the bonds at all is at any given instant very small and that the water-rich segment terminates with the formation of a bulk majority are involved in two or moreThe water molecules are phase separation The amphiphile-rich phase may contain a capable of participating in a wide variety of local schemes of self substantial amount of water At temperatures up to 10K above aggregation of which the ice-like pattern is the lowest in energy its lower critical solution temperature 2-butoxyethanol + water and the least compact A dynamic equilibrium may be envisaged shows evidence of a cmc analogue at a lower amphiphile between molecules in ice-like sites on the one hand and in a concentration than the onset of the miscibility gap 4u (iv) It is multitude of higher energy more compact alternative sites on improper to talk in terms of a cmc analogue for methyl or ethyl the otherThe equilibrium shifts to occupancy of the alternative amphiphiles Nevertheless the composition dependence of some sites as either the temperature or the pressure is increased This of the excess molar properties of their binary aqueous systems model provides a plausible rationale for the unusual tempera- possess features which are similar to but much less pronounced ture-dependence of both the isobaric expansivity and the isoth- than those of their more hydrophobic homologues l6 l7 A ermal compressibility possible explanation for the termination of the water-rich The presence of an amphiphilic solute has the effect of segment for the aqueous mixtures of these species is that the disrupting the structural equilibrium of the water molecules in volume fraction of the amphiphile has risen to a level where all of its immediate vicinityThe alkyl groups in common with the water is directly involved in solvation introduced a mass-action approach completely non-polar solutes are described as being structure Some years ago H~idt~~ making implying that their presence induces enhanced ice-like somewhat different from that of Roux and Desnoyers Her aggregation There is evidence for this in the positive values analysis addressed the question of the relative amounts of water found for Vk,in combination with negative values for Hh at involved in hydrophobic and hydrophilic hydrationThe con- very low amphiphile concentrations clusion reached was that the fraction of water molecules engaged The polar head groups are generally capable of forming in hydrophobic hydration falls off fairly sharply with increasing THERMODYNAMIC AND RELATED STUDIES OF AMPHIPHILE + WATER SYSTEMS-M I DAVIS amphiphile mole fraction whereas hydrophilic hydration persists up to high amphiphile concentrations 5.2 The Central Segment x2<xA<xg This is the segment across which dQE/dxA appears to maintain linear dependence upon mole fraction This observation led to the proposal of the original segmented composition model 24 Such simple composition dependence suggests the existence of a fairly uniform scheme of molecular aggregation spanning the entire segment It is pointed out in reference 1 that in detergent + water systems roughly equivolumar mixtures frequently give rise to a lamellar liquid crystalline mesophaseThe question may then be posed as to whether it is possible that what one observes in the central segments of the somewhat smaller amphiphile + water systems is a manifestation of the existence of some type of labile pseudolamellar analogue consisting of metastable disks 5.3 The Amphiphile-rich Segment x3<xA< 1.0 Figure 5 shows the partial excess molar volumes of water in the three systems EtOH + water,22 EtOEtOH + water,25 and EtOEtOEtOH + water 26 The profiles of the three sets of data are very similar up to an amphiphile mole fraction of 0 5 Above that point the form of the EtOH +water data differs very markedly from the other twoThis probably reflects significant differences between the modes of aggregation of the pure alkanols on the one hand and the ethoxylated ethanols on the other Tg5 .. -05 -1 0 -1 5 -3 Ot 1-30Q O9) 1-3 5 Figure 5 Excess partial molar volumes of water in C,E + water Open circles EtOH + water Closed squares -EtOEtOH + water Closed circles EtOEtOEtOH + water Units cm3 mol-I Alkanols tend to self-associate through the agency of hydro- gen bonding to produce relatively small labile clusters In solutions of alkanols in non-polar solvents there appears to be a tendency towards the formation of cyclic oligamers made up of four hydrogen-bonded monomers 27 It is probable that such oligomers play a significant role in the structure of the pure alkanols Solute water molecules would tend to be incorporated into the polar regions of such structuresThe significant changes in Vk with increasing water content may be an indication of significant variations in the mode of alkanol self-aggregation as the composition changes from pure alkanol to the limit of the central segment That no such significant changes occur in the Fwvalues for the other two systems might be taken as an indication that the self aggregative schemes of the pure ethers already resemble that of the central segment The question might be raised as to whether there is any evidence for the formation of inverse micelles in these systems If there were one might expect to observe some type of critical concentration analogous to the cmc of the water-rich region Such does not appear to be the case However it is possible that in some of the systems the central (pseudolamellar) segment contains either micellar aggregates or some type of metastable analogues If so one might expect to encounter some type of discontinuity in the thermodynamic data at the composition XA = x3 Some of the recent attempts to improve the four-segment model have revealed that while the constraint that d2QE/dxA2 should be continuous at xA = x3 has relatively little effect upon the curve fitting attributes of the model for most of the alkanol + water data sets its imposition is detrimental for the data of the alkoxylated ethanols + water 5.4The Intermediate (Transitional) Segment x1 <xA<x2 While it is possible to make some plausible suggestions as to the nature of the schemes of molecular aggregation which exist in the water-rich (premicellar) central (pseudolamellar) and amphiphile-rich segments it has proved to be more difficult to describe this intermediate or transitional segment For amphiphilic species which are capable of forming stable micellar aggregates xA = x1 is the concentration at which they first appear As the mole fraction of the amphiphile increases so the population of micellar aggregates rises and some type of equilibrium must exist between them and their saturated solu- tionThere are several points that need to be kept in mindThe phase diagrams and other studies of the detergent + water systems indicate that there are progressive changes in the micellar geometries Further the amphiphile concentration is not likely to rise very far above the cmc before all of the water present is directly involved in solvation making it inappropriate to talk in terms of a solution in the normal sense of the word Some help in gaining an appreciation of the nature of the aqueous mixtures of the more hydrophobic amphiphiles within this segment may be gained from a consideration of Figure 6 This shows the partial excess molar volumes of water in the C6E + water at 5 oc28and C6E3 + water at 15"c29 Above what is recorded as the cmc for C6E3 and what is presumably also that for C6E2 the P$ values remain virtually constant before beginning a fairly significant decreaseThat the cmc is lower in the case of C6E is thought to be due to the temperature difference rather than the size of its polar head group 05T Too' I .. I -0:c-3 ?G-1-025.-025 Figure 6 Excess partial molar volumes of water In C,E + water Open circles C,E + water Closed squares -C,E3 + water Units cm3 mol It may well be that in those systems there exists above the cmc a range of compositions where the water can maintain something approximating to bulk characteristics The addition of water would then result in some of the amphiphile going back into 'solution' Constant partial molar properties are attributes of phase or microphase separations 5.5The Four-segment Model Equations In the foregoing sections some comments have been made about the trends in the composition dependence of the thermo- dynamic properties of C,,,E,,+ water systems as either the alkyl chain length or the polar head group size is increased The four- segment composition model provides a reasonably effective tool for exploring the nature of such trends The excess molar properties of the water-rich mixtures dppear to possess cubic mole fraction dependence This is equivalent to a quadrdtic version of the McMillan-Mayer equation (equation 13) There are several different ways in which a cubic function of the amphiphile mole fraction might be formulated For the purpose of this review the following form has been adopted aw IS the excess appdrent (dnd also partial) moldr property of the amphiphile at infinite dilution and is edsily converted into A by the addition of Q,T,A hw plays much the same role ds does 6 of equation 13 and may be converted into b by multiplying by M,,,wjlOOO cw is regarded ds representing the effects of cluster formation dQE/duA appears to possess linear dependence upon mole fraction dcross the central (pseudolamellar) segment so that QE can be assigned quadratic dependenceThe following quadrdtic equation has been adopted In this context qw represents an excess molar property of some type of hypothetical standard state of water and as such may be regarded as a measure of the effects of the changes in the nature of aqueous self-aggregation qA plays a similar role for the amphiphile hLmust then represent the effects of the interac- tions between the two hypothetical states If there were a true phase-separation or a uniform microphase separation bL would be zero The six optimizable parameters of equdtions 14 dnd 15 form d minimum set For a few alkanol + water data sets this is dn adequate curve fitting model but for most of the alkdnol + water sets at least seven parameters are required The excess molar properties of the intermediate (transitional) segment appear to require an equation with quartic dependence upon mole fraction However for many data sets it is possible to require that both QE and dQE/duA are single valued (conti- nuous) at both xA = Y and Y and also that d2QE/dxX is single valued at xA = x2This eliminates all five of the coefficients of the quartic equation as independent model parameters For the amphiphile-rich segment the excess molar property data can be fitted to a cubic function of mole fraction For most of the alkanol + water systems it is possible to constrain QE dQE/duA and d2QE/dxA to be continuous at y3 which elimi- nates all three of the coefficients of the cubic equation as independent parameters It is interesting to note that for the C,,,E + water systems with n > 0 the constraint on dzQE/dxf is no longer conducive to an acceptable quality of fit In such cases one must add to the parameter set the excess apparent molar property of water at infinite dilution in the amphiphilic solvent The parameters of the four-segment model equations cdn be optimized using a linear least squares procedure provided that the compositions corresponding to the three segment junctions dre fixedThe determination of the optimum segment junction compositions forms a critical part of the analysis A rough idea CHEMICAL SOCIETY REVIEWS 1993 of the locations of the segment junctions may be gained from visual appraisals of plots of the dQE/dxA curves Using those estimates as a starting set one may then proceed to determine optimum values for each data sct by means of a 'simplex' procedureThis procedure tends to be vulnerable to blatant errors in the data and requires screening of the more obvious violations of internal consistency For the most part the data sets for a given system lead to a reasonably consistent estimate of the locations of the segment junctions Some results of four-segment analyses are set out inTables 1-7 Table la Amphiphile mole fractions for the segment junction x1 at 25 "C m= 1 2 3 4 6 I1 = 0 0 135 0 100 0070 1 009 0075 0050 0025 2 0 075 0 025 0 0035" 3 0 025 0 0025 The values for C,E2 + water correspond to a temperature of 5 "C These values are all for n-alkyl dmphiphiles The values found for 2-PrOH + water and iso-PrOEtOH + water are both higher than those given in the table That for t-BuOH + water is 0 050 Values are found to vary slightly from one type of data to another and with temperature for the C,E + water and C,E + water systems Table lb Amphiphile mole fractions for the segment junction y2 at 25 "C m= 1 2 3 4 6 I1 = 0 0400 0400 0250 1 0333 0333 0250 0250 2 0 333 0 333 0 250 3 0 333 0 300 Table lc Amphiphile mole fractions for the segment junction x3 at 25 "C m = 1 -7 3 4 6 11 = 0 0600 0667 0750 I 0 500 0 500 0 500 0 500 2 0 500 0 500 0 500 0 500 3 0 400 0 400 The largest number of data sets is for the excess molar volumesThe following tables give the values obtained for the six mdjor four-segment model parameters of the property VE Table 2a Values of the parameter uW for the excess molar volumes of n-alkyl C,,,E,,+ water systems at 25 "C (units cm3 mol I) n?= 1 2 3 4 6 I?=O -255 -362 -445 1 -424 -636 -802 -888 2 -8 44 -1073 -13 10 (5°C) 3 -1403 -1526 THERMODYNAMIC AND RELATED STUDIES OF AMPHIPHILE + WATER SYSTEMS-M I DAVIS The uw values may be converted into the apparent molar volumes of the amphiphiles at infinite dilution (Table 2b) by adding the molar volumes of the pure amphiphiles Table 2b Apparent molar volumes of C,E species at infinite aqueous dilution at 25 "C (units cm3 mol-I) m= 1 2 3 4 6 n=O 38 20 5506 7073 1 7502 91 07 10656 12294 2 127 98 160 39 3 196 34 228 75 These values correspond to an average increment of 36 5 cm3 mol-per OC,H group and 16 0 cm3 mol- 1 per CH group at a P-carbonThey conform fairly well to a simple group additivity scheme The aw values derived from the excess molar enthalpy data are dll negativeThere are substantial negative shifts as the size of the polar head group increases due presumably to hydrogen bonding between water and the ether groups There is a modest decrease ( -3 kJ mol- ') on increasing the chain length from I to 2 but no significant change thereafter The uw values for the excess molar isobaric heat capacities are all positive and gener- dlly of the same order of magnitude as the absolute molar heat capacities of the amphiphiles They increase with both the length of the alkyl chain and the size of the polar head group The aw values obtained from the KE data show similar trends to those for VE Table 3 Values of bw for the excess molar volumes data for C,E + water systems at 25 "C (units cm3 mol- ') m= 1 2 3 4 6 n=O -89 -304 -606 I -199 -421 -401 -1434 2 -44 1 -1800 -1225 (5°C) 3 -1666 -380 There dre no clear trends in the bw values (Table 3) associated with variation of the polar head group size but a very definite increase as the length of the alkyl chain increasesThe bwvalues for the excess molar enthalpy data are all positive They increase significantly with both the length of the alkyl chain and the size of the polar head group Both hydrophobic and hydrophilic hydration appear to lead to enthalpy decreases As a conse- quence overlap and disruption of the hydration co-spheres might be expected to be endothermic It is difficult to pick out any reliable trends among the bw values derived from the limited number of excess molar Cpdata sets Table 4 Values of cw for the excess molar volumes data for C,E + water systems at 25 "C (units cm3 mol ') m= 1 2 3 4 6 n=O 20 115 589 1 98 317 702 6760 2 405 6636 372500 (5°C) 3 61 17 615900 If bw is an indicator of the effects of hydration co-sphere overlap the rapid decrease in its VE values as the dlkyl chain length increases indicates that the major effect is to disrupt the pattern of ice-like aggregation Similarly the trends in the cw values which are regarded as being indicators of the effects of amphiphile clustering also reflect the diminution of the struc- ture-making effects of hydrophobic hydration There is really too little data for HEto state more than the cW values become more endothermic with increasing alkyl chain length Table 5 Values of qw for the excess molar volumes of C,,E + water systems (units cm3 mol-') m= 1 2 3 4 n=O -0019 -0559 -0386 1 -0458 -0731 -0453 -0214 2 -0 958 -0 536 3 -0 772 -0 264 -0 389 There is a clear tendency for qw (Table 5) to become more negative as the size of the polar head group increasesThe greatest decreases in the molar volume among species with a common polar head group occur when nz = 2 It is suggested that this parameter reflects the effect of the changes in the patterns of self aggregation of water from that of the pure liquidThe negative values would then correspond to a decrease in the level of ice-like aggregation The qw values that have been derived from the excess moldr enthalpy data are all negativeThey show the same kind of dependence upon alkyl chain length as do the values for VE and the same tendency to become more negative with incredsing polar head group size The values for the excess molar isobaric heat capacity data are all positive and tend to show the sdme trends as but are of opposite sign to the HEvalues Table 6 Values of qA for the excess molar volumes of C,,E + water systems (units cm3 mol-I) m= 1 2 3 4 6 n=O 0037 -0088 -0088 1 0068 0 584 0 507 0432 2 3 0 708 0 613 1252 1 086 (5°C) 1 149 There is a remarkable difference between the qA vdlues (Table 6) for the alkanols and the alkoxyethanols It is not clear whether this reflects differences in the modes of self aggregation of the two families in the mixtures or as pure liquidsThe values obtained for the HEdata are generally quite small Those for the alkanols are all negative while those for the alkoxyethanols are positive There are no well-defined trends among the CF values Table 7 Values of bL for the excess molar volumes of C,,E + water systems (units cm3 mol- ') m= 1 2 3 4 n=O -405 -298 -166 1 -416 -385 -345 -331 2 -3 76 -3 76 3 -4 70 6 -5 14(5"C) -4 55 There is a sharp increase in the bL values with increasing alkyl chain length for the alkanol+ water systems The values obtained from the HEdata for the alkanols range from -1 16kJ mo1-l for MeOH + water to 2 12 kJ mol-l for 1-PrOH+ water It is this variation which is manifested in the vastly different HEprofilesThe value of bL for the alkoxyethanols is consistently negative 6 Conclusions In the foregoing sections an examination has been made of some of the graphical and analytic perspectives of the thermody- namic data that have been accumulated for the aqueous mix- tures of the C,E family In conclusion it is fair to ask whether we are any closer to an understanding of the nature of the patterns of molecular aggregation which exist within such mixtures than we were 25 years ago The first claim that can be made is that our concepts have been polarized in the right direction by recognizing that the hydro- phobic effects which are responsible for the formation of micelles vesicles and membrane bilayers almost certainly play some role in determining the less dramatic patterns of molecular aggregation that exist in the aqueous mixtures of even quite humble amphiphiles In that context it is important to recognize the influence ofTanford 30 One may state with a fair degree of certainty that the structures of the aqueous mixtures are dictated by tendencies towards minimization of direct contact between water and the apolar hydrocarbon tails and by a tendency for water molecules to hydrogen bond to the lone pairs of electrons of the ether and hydroxyl oxygens In the formation of a micelle the separation of water from the alkyl groups is essentially complete All of the evidence points to 2-butoxyethanol (C,E,) forming aggregates which if not truly micellar bear considerable resemblance to micelles It is possible that such aggregates have no single well- defined geometry It is noted that the C,E + water systems show some of the same types of composition dependence features for their ther- modynamic properties as are associated with the existence of critical micelle concentrations Little mention has been made here of the t-butanol + water system but it also shows pseudo- micellar characteristics There is evidently a fair degree of hydrophobic clustering of the amphiphiles at premicellar concentrationsThis leads to substantial increases in the viscosities of the amphiphile + water mixtures and corresponding decreases in their diffusion coeffi- cients Equally evident is the fact that such clustering does not necessarily lead to the formation of stable micelles since many of the features of the composition dependence of the thermody- namic properties that are attributable to clustering are also present to a more modest extent for the methyl and ethyl systems Relatively little attention had been paid to the thermodyna- mic data at higher amphiphile mole fractionsThere appears to be evidence for the existence of a range of compositions which corresponds to the lamellar regions of detergent + water systems Questions concerning the patterns of aggregation in binary aqueous mixtures will not be properly answered until such time as it becomes feasible to carry out large-scale computer simula- tionsThat is not something that is anticipated in the immediate future It is possible however to examine the stabilities and preferred structures of small clusters using molecular modelling techniques such as those based upon the Allinger algorithms Some preliminary studies carried out in this laboratory reveal the fact that the C,E species are capable of existing in a wide variety of different conformations many of which are close CHEMICAL SOCIETY REVIEWS. 1993 enough in energy to the most stable of their number that they have significant populations The sequence of conformational energies in the gas-phase is likely to differ from those in the pure liquid states and those existing in aqueous mixtures Significant changes in conformation on going from the pure liquid to an aqueous mixture may well have a significant effect upon the values of the excess molar properties This review has dealt exclusively with a single family of non-ionic amphiphilesThis family has become the object of a great many investigations because of the possibility of varying both the alkyl chain length and the size of the polar head group In that sense it is believed that the lessons learned from the study of the aqueous mixtures of the members of this family serve to enhance our understanding of amphiphile + water mixtures in general The continuously increasing data base places progressively greater constraints upon conceptual models that we might invoke We may be able to discuss the structures of amphiphi- le + water mixtures somewhat more intelligently than before but we obviously still have a long way to go before our understanding of them will have reached a completely satisfac- tory level 7 References I ‘Physics of Amphiphiles Micelles Vesicles and Microemulsions’ ed V Digiorgio and M Corti North Holland Amsterdam 1985 2 J B Rosenholm R B Grigg and L G Hepler in ‘Solution Behavior of Surfactants’ ed K L Mittal and E J Fendler Plenum Press New York 1982 Vol 1 p 359 3 (a) G ROUX,International Data Series Ser B 1978 p 44 (b) G Roux G Perron and J E Desnoyers J Soln Chem ,1978,7,639 4 K N Marsh Annu Rep Prog Chem Sect C 1980,77 101 5 P Picker P A Leduc P R Phillip and J E Desnoyers J Chem Thermodyn 1971,3,631 P Picker Can Res Dev 1974,7 11 6 G Douheret and M I Davis Chem Soc Rev 1993,22,43 7 G C Benson and 0 Kiyohara J ChemThermodyn 1970 11 1061 8 M I Davis and G DouheretThermochim Acta 1991 190,267 9 G C Benson P J D’Arcy and 0 Kiyohara J Solution Chem 1980,9,93 1 10 0 Redlich and A T Kister Znd Eng Chem 1948,40,345 11 W G McMillan and J E Mayer J Chem Phys 1945,13,276 2 H Holland and E Vikingstad Acta Chem Scand Ser A 1976,30 182 3 C Jolicoeur and G Lacroix Can J Chem 1976,55,624 4 J J Savage and R H Wood J Solution Chem 1976,57,33 5 J E Desnoyers Pure Appl Chem 1982,54 1469 6 A H Roux and J E Desnoyers Proc Indian Acad Sci (Chem Sci ) 1987,98,435 7 M I Davis and G Douheret Thermochim Acta 1991 188,229 8 F Franks and D J G Ives Quart Rev 1966,20 1 9 M J Costigan L J Hodges K N Marsh R H Stokes and C W Tuxford Aust J Chem 1980,33,2103 20 M I Davis Thermochim Acta 1990 157 295 2 1 (a)F H Stillinger Science 1980,209,45 1 (b)W L Jorgenson J Chem Phys 1982,77,4156 22 G C Benson and 0 Kiyohara J Solution Chem 1980,9,791 23 C Dethlefsen P G Sorenson and Aa Hvidt J Solution Chem 1984 13 191 24 M I Davis Thermochim Acta 1983 63 67 25 G Douheret A Pal and M I Davis J Chem Therm ,1990,22,99 26 G Douheret C Salgado M I Davis and J Loya Thermochim Acta 1992 207 3 13 27 P Schuster G Zundel and C Sandorfy ‘The Hydrogen Bond’ North American Amsterdam 1976 28 M I Davis and M E Hernandez submitted for publication 29 S Wieczorek J Chem Thermodjn 1992,24 129 30 C Tanford ‘The Hydrophobic Effect’ John Wiley New York First Edition 1973. Second Edition 1980

ISSN:0306-0012    

DOI:10.1039/CS9932200127

出版商:RSC    

年代:1993

数据来源: RSC