ORGANIC REACTWHY IN MIXED AQUEOUS SOLVENTS

A LINK BEIWEEN C S

AND THERMODYNAMICS

ORGANIC REACTIVITY IN MIXED AQUEOUS SOLVENTS

A LINK BETWEEN KINETICS AND THERMODYNAMICS

RIJKSUNIVERSITEIT GRONINGEN

ORGANIC REACTIVITY IN MIXED AQUEOUS SOLVENTS

A LINK BETWEEN KINETICS AND THERMODYNAMICS

PROEFSCHRIFT

ter verlu-ijging van het doctoraat in de Wiskunde en Natuurwetenschappen aan de Rijksuniversiteit Groningen op gezag van de Rector Magnificus Dr.S.K.Kuipers in het openbaar te verdedigen op vrijdag 29 november 1991 des namiddags te 4.00 uur

door Wilfried Blokzijl geboren op 30 maart 1964 te 's-Gravenhage

Promotores: Prof.Dr. J.B.F.N. Engberts Prof.Dr. M.J. Blandamer

Aan mijn ouders

VOORWOORD
Water heeft door de eeuwen heen niet alleen componisten, dichters, schrijvers en schilders maar ook wetenschappers weten te inspireren. Terwijl voor velen de uiterlijke schoonheid van water, in a1 zijn verschijningsvormen, een belangrijke bron van inspiratie was, raakten wetenschappers ook gefascineerd door de moleculaire eigenschappen van de "matrix en de moeder van het leven" (A. Szent-Gyorgy). Ook ik ben de afgelopen jaren ge'infecteerd door het "watervirus". Het complexe gedrag van een ogenschijnlijk zo eenvoudige en alledaagse verbinding als water vormde voor rnij een continue uitdaging die, naar ik hoop, duidelijk wordt bij het lezen van dit proefschrift. De afgelopen vier jaar waarin dit onderzoek is uitgevoerd zijn voor rnij bijzonder prettige jaren geweest. Dit is voor een belangrijk deel te danken aan de goede sfeer op het organisch chemisch laboratorium. Vooral de bewoners en de exbewoners van het "waterlab" hebben altijd bijgedragen tot een goed wetenschappelijk werkklimaat en de zo broodnodige gezelligheid. Mijn grote waardering en dank gaat uit naar Jan Engberts, die binnen zijn werkgroep een sfeer heeft weten te creeren waardoor het mogelijk was op een creatieve manier te "stoeien" met de chemie. De vrijheid en het vertrouwen die hij mij heeft gegeven zijn voor rnij altijd heel belangrijk geweest. Met plezier denk ik terug aan de vele discussies en gesprekken die ik, op het lab of bij andere gelegenheden, met hem heb gevoerd. Deze discussies en gesprekken gingen, vanzelfsprekend, vaak over de chemie, maar ook over de filosofische kanten van het leven. Although the contacts with Mike Blandamer were, due to geographical limitations, less frequent, they were at least as intense. I will not easily forget our two-days sessions. Within these two days, his enthousiasm always convinced me that chemistry is fun. The pile of letters and computer output that crossed the North Sea as well as the frequent telephone calls made doing research time and again "super". I want to express my heartfelt thanks to both Jan and Mike for many discussions during the realisation of the preprint. Mijn dank gaat ook uit naar de leden van de leescommissie, Pr0f.Dr.H.J.C. Berendsen, Prof.Dr.A.M.van Leusen en Prof.Dr.G.Somsen voor hun vlotte correctie van de preprint en hun waardevolle suggesties. Bij dit onderzoek was de technische hulp van Marten de Rapper, wiens blik soms a1 voldoende was om weigerachtige UV-apparatuur tot de orde te roepen, onmisbaar. Anno Wagenaar en Wim Kruizinga wil ik bedanken voor hun geduld toen ik steeds maar weer in H,O, en niet in D,O mijn kinetische NMR-experimenten wilde uitvoeren. Marinus Suijkerbuijk was altijd weer bereid rnij met raad en daad bij te staan bij de experimenten met en rond het GC-apparaat. Bij de experimenten met de gasfaseosmometer, die helaas niet altijd de venvachte resultaten gaven, kon ik altijd weer rekenen op de hulp van Jan Ebels. Klaas Hovius wil ik bedanken voor zijn synthetische hulp. During three months, Francesca Bortolozzo brought some italian atmosphere into our laboratory. She did "of course" an important part of the synthetic and kinetic work, described in Chapter 7. Jan Willem Wijnen haalde (alle) onvermijdelijke typefouten uit de preprint. Tenslotte wil ik Willem Kuil bedanken voor het tekenen van de talloze grafieken en figuren. Hier kan nog steeds geen computer tegenop.

CONTENTS Chapter 1
1.1 1.2

Solvent effects on organic reactions in aqueous mixtures An introduction

1.3 1.4 1.5 1.6 1.7

Solvent effects in chemistry 1 1.1.1 Some words about the historical background ............ 1 1.1.2 Water; menstruum universale? ....................... 1 A survey of approaches to the analysis of solvent effects ............ 2 Correlations with physical properties of the solvent ........ 4 1.2.1 Correlations with emperical and semi-emperical solvent 1.2.2 parameters ..................................... 4 Analysis of solvent effects using solubility and transfer 1.2.3 parameters ..................................... 6 Theoretical treatments of solute-solvent interactions in 1.2.4 relation to solvent effects ........................... 7 Interactions and reactivity in water and in mixed aqueous solvents .......................................... 8 Hydrophobic effects; definitions and the state of the art ............ 8 The need for a quantitative description of solvent effects in mixed (aqueous) solvents. Incentives for this study ...................... 14 Aims of this study ......................................... 15 Survey of the contents ...................................... 15

..................................

Chapter 2
2.1 2.2

A quantitative analysis of solvent effects in mixed solvents Development of a theoretical model

2.3

2.4

2.5

Introduction ............................................. 19 Solvent effects in dilute mixed solvents ......................... 21 2.2.1 Kiiietics of a reaction; transition state theory ............ 21 Chemical potential and activity coefficients of solutes in 2.2.2 dilute mixed solvents .............................. 22 Dependence of excess properties on composition 23 2.2.3 Solute-solute interactions; additivity approach ............ 27 2.2.4 Modifications of the general theoretical model for quantitative analysis of solvent effects in mixed solvents ............................ 29 2.3.1 Solvolysis reactions ............................... 29 2.3.2 Chemical equilibria ............................... 30 2.3.3 Bimolecular reactions .............................. 31 Solvent effects on partial molar enthalpies and entropies 2.3.4 of activation .................................... 33 Solvent effects in mixed binary solvents ......................... 35 Kinetics of reactions; transition state theory ............. 35 2.4.1 Dependences of the standard chemical potential of solutes on 2.4.2 the composition in mixed binary solvents ............... 35 A comparison of the approaches used to develop a theoretical model for the analysis of solvent effects in mixed solvents ................... 39

.........

Chapter 3

Quantitative analysis of solvent effects in highly aqueous media Application of SWAG procedures and a critical appraisal of the additivity principle

3.1 3.2 3.3 3.4 3.5 3.6

Introduction .............................................41 Solute-solute interactions in aqueous solution .................... 42 Medium effects of monohydric and polyhydric alcohols on the hydrolysis reaction of l.benzoyl.3.phenyl.l,2, 44riazole ...................... 47 Towards a better understanding of medium effects in dilute aquous solution of monohydric and polyhydric alcohols ................... 56 Concluding remarks ....................................... 60 Experimental section 60

.......................................

Chapter 4
4.1 4.2 4.3 4.4 4.5 4.6

A survey of solvent effects on some organic processes in dilute aqueous solution

Introduction ............................................. 63 Solvent effects on the hydrolysis of p-methoxyphenyl dichloroacetate in aqueous solutions containing urea and alkyl-substituted ureas ....... 64 Solvent effects on a bimolecular Diels-Alder reaction ............... 72 Solvent effects on an intramolecular Diels-Alder reaction ............ 75 Solvent effects on the keto-en01 equilibrium ..................... 76 Experimental section ....................................... 81

Chapter 5

Alkyl substituent effects on the neutral hydrolysis of l.acyl.3.alkyl.1,2, 4.triazoles in highly aqueous media The importance of solvation

5.1 5.2 5.3

5.4 5.5 5.6 5.7

Introduction ............................................. 83 Alkyl substituent effects. Current views 84 Solvent effects and substituent effects on the hydrolysis of l~acyl.(3~substituted).1,2, 4.triazoles ............................ 86 A link between alkyl substituent effects and solvent effects . . . . . . . . . . 94 The molecular origin of alkyl substituent effects ................... 96 Solvophobic acceleration and substituent effects .................. 97 Experimental section ....................................... 98

.........................

Chapter 6
6.1 6.2 6.3

Diels-Alder reactions in mixed aqueous media Enforced hydrophobic interaction

Introduction 101 Diels-Alder reactions in water; mechanistic considerations and recent developments 103 Kinetic studies of Diels-Alder reactions in mixed aqueous solvents; a theoretical approach .................................... 107

............................................ ......................................

6.4 6.5 6.6 6.7 6.8 6.9

Properties of alcohol water mixtures .......................... Diels-Alder reactions in water and in mixed aqueous solvents; experimental results ...................................... A quantitative analysis of solvent effects on Diels-Alder reactions in mixtures of water and monohydric alcohols ................... Hydrophobic effects on Diels-Alder reactions in water ............. Hydrophobic effects on Diels-Alder reactions in mixed aqueous solvents .................................... Experimentalpart

111

111
123 130 133

........................................ 134

Chapter 7
7.1 7.2 7.3 7.4 7.5 7.6 7.7 7.8

A quantitative analysis of solvent effects on intramolecular Diels-Alder reactions in mixed aqueous solvents
139 140 143

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Intramolecular Diels-Alder reactions; an o v e ~ e w ................ Intramolecular Diels-Alder reactions of furan derivatives Synthesis and intramolecular Diels-Alder reactions of N-furfuryl-Nalkyl maleamic acids Solvent effects on the intramolecular Diels-Alder reactions of N-furfuryl-N-alkyl maleamic acids ............................ A quantitative analysis of solvent effects on intramolecular Diels-Alder reactions in mixed aqueous solvents ................. Discussion and conclusions ................................. Experimental section ......................................

........... ......................................145

146 150 152 154

Chapter 8
8.1 8.2 8.3 8.4

Epilogue A link between organic chemistry in aqueous solutions and hydrophobic effects
159 159 161 167

Introduction ............................................ A quantitative approach for the analysis of solvent effects in mixed solvents. Merits and shortcomings ............................ Hydrophobic effects. Hydrophobic interactions and hydrophobic hydration ................................. Organic reactivity in mixed aqueous solvents ....................

References Summary Samenvatting

169 179

183

I . Introduction Solvent gects on organic reactions in aqueous mirtwes

CHAPTER 1
Solvent Effects on Organic Reactions in Aqueous Mixtures. An Introduction
2.1 Solvent effects in chemistry

2.1.1 Some words about the hirtorical background

Reactivity of molecules and ions in solution is largely dictated by the solvent. Comparison of rate constants in the gas-phase and in solution shows that differences of 10" are not uncommon1. In 1862, Berthelot and PCan de Saint-Gilles were the first to notice the considerable influence of the reaction medium on the rates of homogeneous chemical reactions2. In 1890, Menschutkin performed the first detailed study of the reaction of trialkylamines with haloalkanes in twenty-three different solvents and stated that "solvents are by no means inert in chemical reactions and can greatly influence the course of themN3.Since then many papers have appeared in which more or less dramatic solvent effects are reported on a large variety of chemical processes. Changing the solvent can, in extreme cases4, lead to rate variations in the order of lo9. Not only chemical reactions, but also chemical equilibria are sensitive to the solvent. This was shown by ClaisenS and wislicenus6, who were, in 1896, among the first authors to draw attention to the considerable solvent effect on keto-en01 tautomeric equilibria. The fact that the solvent can seriously affect spectroscopic properties of molecules in solution was demonstrated by ~ u n d t 'in 1878. During the past century, solvent effects have also been reported on a number of other chemical phenomena, such as aggregation8, complexation9, ionisationlo, conformation" and isomerisation12. Solvent effects on chemical reactivity have received close attention. A long-lived goal of chemists at large has been to establish methods and to provide tools to understand and predict solvent effects on chemical reactions. Ultimately, this knowledge should enable chemists to choose in a rational way a suitable solvent for a particular chemical transformation.

1.1.2 W a r ; menstruum univer~ale?'~

According to ancient Greek philosophy every solution was called "water" and all liquids, able to dissolve compounds were classified as "divine water". Water was in fact the first liquid to be considered a solvent. Many publications and textbooks claim that water is in every aspect a unique solvent and liquid. Water certainly is the most extensively studied liquid. Properties, models and theories have been discussed in detai114-16. Although living organisms depend in a unique way on water as a solvent for biochemical transformations, synthetic organic chemists do not particularly like water

1. Introduction Solvent effects on organic reactions in aqueous mixtures

Table 1-1:Summary of organic reactions performed in homogeneous and heterogeneous aqueous media. Reaction Intermolecular Diels-Alder reactions Intramolecular Diels-Alder reactions Claisen rearrangements Aldol reactions of silyl-en01 ethers Benzoin condensation Reduction of alkyl halides with tinhydride Allylation of carbonyl compounds using zinc 18f,g 19c 17e 17f 22a,b, 29 References

as a solvent for chemical reactions. This "hydrophobic" attitude stems largely from the fact that water is far from a "Menstruum Universale" for organic compounds. Moreover, water is often highly reactive towards many organic reagents. Nevertheless, the past decade has witnessed a remarkable reappraisal of water as a solvent for organic reaction^'^-^^. This change in attitude is partly due to the pioneering work of res slow"", who, in 1980, reported intriguing solvent effects of water on notoriously solvent-insensitive Diels-Alder reactions. This discovery inspired others to search for other organic reactions that could benefit likewise from water as a solvent. In addition, the need for solvents that can satisfy the high requirements of current environmental legislation, makes water an interesting choice as a solvent for organic reactions. In Table 1-1 a collection of organic reactions is surnrnarised that are not traditionally performed in water, but were found to benefit from the presence of water as a solvent. The molecular origin of these remarkable and sometimes even spectacular solvent effects in aqueous solution remains unclear.

1.2 A survey of approaches to the analysh of solvent

effect^^'-^^

Solvent effects on chemical processes are usually studied in comparison to reactivity in a reference solvent. In some cases comparison is made with gas-phase reactivity. Generally, analysis of solvent effects has been based on an equilibrium solvation model. However, for some organic reactions the reaction rates can be very high and examples are known for which non-equilibrium or dynamic solvation models have to be used to account for reactivity in solution. Consequently, sophisticated theoretical models have been developed. In the past, quantitative and qualitative approaches for the analysis of solvent effects have been developed almost simultaneously.

1. Introduction Solvent effects on organic reactiom in aqueous mixtures

Qualitative descriptions of solvent effects. Almost all qualitative treatments of solvent effects are based on the simple solvation model, developed by Hughes and Ingold in 1935, for explaining solvent effects on substitution and elimination reactionsB. The model considers mainly the change of electrical interactions between solvent and reacting species during the activation process. Solvents are thus classified according to their ability to solvate ions and molecules. A serious shortcoming of this approach is the fact that the solvent is considered as a continuum without defined structure and that specific solvent-solute interactions are completely neglected. In addition, changes in the structure of solvents as a result of changes in solvation during the activation process are neglected. Also entropic contributions to solvent effects are not incorporated into this model. Especially in highly structured solvents such as water, entropic contributions to solvent effects can be significant. Notwithstanding the weak points of these qualitative models, these approaches are simple and are readily applied; they are still very popular in practical chemistry. Quantitative descriptions of solvent effects. The starting point of most quantitative approaches for the interpretation of solvent effects is based on transition state theory (TST). A general quantitative description of solvent effects is given in Equation 1-1,

in which
A pi,= pic(S)

P ~ H )

and

where ApIsO and ApACOare the transfer standard chemical potentials of the initial state and the activated complex, respectively, for transfer from the reference solvent R to solvent S. The rate constants, found in solvents S and R are k and k,, respectively. The success of any quantitative method in describing solvent effects is determined by the accuracy with which the Gibbs energies of solvation can be calculated. A detailed knowledge of solvent-solute interactions is essential for this goal. The wide range of possible solute-solvent interactions, the actual structure of the solvent, the characteristics of the reacting molecules and the activated complex make enormous demands on the theory. (Semi)quantitative approaches can be subdivided into four major categories: (i) (ii) (iii) (iv) Correlations with physical properties of the solvent. Correlations with empirical solvent parameters. Analysis incorporating solubility andfor transfer parameters. Theoretical treatment of solute-solvent interactions.

In separate sections, these approaches will be briefly outlined and critically discussed.

1. Introduction Solvent pffects on organic reactions in aqueous rnirtures

1.2.1 Correlations with physical properties of the solvent

Solvents can be classified according to many physical properties. Some of these have been used for the correlation of solvent effects. Frequently, the justification for these correlations is found in some theoretical model. Most popular are correlations with the relative permittivity, E, of the solvent. Solvent effects are often related to functions like [(~;l)/(&,+l)] or I/&,. The theoretical basis for these correlations has been given by Eyring, Kirkwood, Laidler and ~andskroene?"'. Another well-known method, developed by Scatchard and Hildebrand, is based on the energy of evaporation of a solvent per unit of volume, the cohesive pressure or cohesive energy density, c 3 8 4 . Related to this approach are correlations with the internal pressure of the or aH2 solvent42,-rr. These methods are mainly used to analyse solvent effects on reactions between neutral molecules. Koppel and palm4' emphasised the importance of the polarisability of the solvent, which can be expressed by a function of the refractive index of the solvent, [(n2-1)]/[p2+l)]. In this light, also correlations with the polarisability parameter P are known4 744. Finally, correlations with the viscosity of the solvent have been used to explain solvent effects on diffusion-controlled reactions45746. Correlations of solvent effects with these solvent parameters are often surprisingly good. The inherent weakness of the method is that the parameters measure macroscopic properties. Specific solute-solvent interactions that occur on a microscopic scale are completely neglected. The structural changes of the solvent, accompanying the activation process, are also neglected. Correlations with physical properties of the solvent are mainly associated with enthalpic contributions to the overall solvent effect, which makes the method less suitable for reactions in highly structured solvents such as water.

1.2.2 Correlations with empirical and semi-empirical solvent parameters

The ability of a solvent to solvate molecules or ions, sometimes rather ambiguously described as the solvent polarity, is difficult to express in terms of physical solvent parameters. This problem forced chemists to search for empirical or semi-empirical scales of solvent polarity. The list of solvent polarity scales is long and lengthens every y e a ? . ' The scales are based on linear solvent energy relationships (LSER) and use solvent effects on properly selected model processes. In the literature, correlations involving solvent polarity parameters are frequently encountered. Some polarity scales have found a wider application in the analysis of solvent effects and will be described briefly below. Most popular polarity scales are based on spectroscopic properties. Spectroscopic data describing solvent effects on UVIVIS transitions of a large number of solvatochromic dyes resulted in a long list of solvatochromic polarity parameters4'. One of the oldest parameters, introduced by ~ o s o w e r ~ ' in .~~ 1958, the Z-value, is still frequently used. Dimroth and ~ e i c h h a r d t ~ ' reported ~' probably the best known and most frequently used solvent polarity parameter, the q 3 0 ) value. The standard probe molecule is a pyridinium-N-phenoxide betaine dye (I), which has a T-T' absorption band with intramolecular charge-transfer characteristics.

I. Introduction Solvent effects on organic reactions in aqueous miaures

This parameter can be determined in many solvents and is very solvent sensitive. Another series of parameters, based on spectroscopic transitions was introduced by The ~.~~. Kamlet, Taft and Abraham and is related to specific properties of the s o l ~ e n t ~ a-scale measures the hydrogen bond donor acidity of the solvent, the p-scale the hydrogen bond acceptor basicity of the solvent and the T*-scale the polarisability or dipolarity of the solvent. The a-scale and the p-scale are based on a solvatochromic comparison method, using solvatochromic shifts of 6nitroaniline and N,N-diethyl-4nitroaniline. Similarly, the T-scale is based on electronic T-T* transitions of seven nitroaromatics. The Acceptor Number (AN), introduced by Gutmanns4, is based on the relative 3 1 ~ chemical ~ shift ~ values ~ of triethylphosphane oxide related to those of the 1:l adduct Et,PO-SbC15. The scale classifies the solvent according to Lewis acidity or the Electron Pair Acceptor property of the solvent. A few solvent polarity scales have been related to chemical equilibria. The Donor Number (DN) measures the Lewis basicity of the solvent'0JS~s6. SbCls was used as a reference compound. The parameter is defined by the enthalpy associated with the adduct formation between antimony pentachloride and Electron Pair Donor solvents. The oldest polarity scales find their origin in kinetic measurements. A famous ~ expressed example is the ionising power, as defined by Winstein and ~ r u n w a l d 'and in the solvent Y-scale. This scale has been developed on the basis of the rate constant for the SN1 solvolysis of t-butylchloride in different solvents. Later, this approach was and ' , others6' in order to extend the procedure to modified by winstein*', ~ e n t l e g ~ ' ~ correlations of solvent effects on reactions involving borderline or even pure S,2 mechanisms. Finally, a large number of polarity scales are based upon solubility data6', transfer parameters6z63, partition coefficient^^^>^' and chromatographic These parameters quantify the phobicity or philicity of a selected compound for the solvent. In physical organic applications the Hansch value^^*^^ as well as the S,-value, recently introduced by brah ham^', are frequently encountered. Both parameters measure the solvophobicity of apolar molecules for different solvents. Some empirical parameters based on partition coefficients, solubility, and Gibbs energies of transfer are mainly used in industrial and engineering applications.

I . Introduction Solvent effects on organic reactions in aqueous mixtures

The inherent weakness of these methods is that the solvent polarity scale is based on a selected process and is therefore not universal. Obviously, only satisfactory correlations can be expected for solvent effects on processes closely related to those used to define the polarity scale. In addition, it is not always clear which of the many types of solute-solvent interactions is expressed by a certain parameter. By careful selection of the model process or by sophisticated comparison methods5z5396&70, some polarity scales have been given a relatively well-defined physical meaning. These methods are based on the fact that some classes of compounds interact with the solvent via a predominant and well-defined mode of interaction. A detailed interpretation of solvent effects in terms of specific interactions based on polarity scales is, however, extremely difficult. The choice of a suitable solvent polarity scale to correlate solvent effects on a new process is often simply pragmatic. A practical limitation is that solvent polarity parameters often cannot be measured in all solvents. It has become clear that no single solvent parameter can account for the complex nature of solvent effects on different chemical processes. To circumvent this problem, multiparameter approaches have been advocated. Koppel and palm7', and later Kamlet, Taft and brah ham" and others71772 developed sophisticated multiparameter equations, incorporating two to four empirical, semi-empirical or physical solvent parameters. Selection of a set of independent solvent parameters that incorporates all possible contributions to the overall solute-solvent interaction is rather arbitrary. Every single parameter measures a combination of distinct contributions to the overall solute-solvent interaction for which both the identity and magnitude are not properly defined. An interpretation of the results of a linear regression analysis is difficult and often highly speculative. From a more practical point of view, in particular for predictive applications, the use of linear regression analysis is seriously limited by the fact that for a proper calculation many datapoints are required. Finally, the theoretical background of LSER's and its application as a tool to analyse or even predict kinetic parameters for new chemical transformations, has been seriously criticised7'. Discussion of the fundamental aspects of LSER's is interesting and often philo~ophical~~.

1.2.3 Analysh of solvent effectsusing solubility and transfer parameters

Measurement of transfer parameters and solubilities to analyse solvent effects is directly linked to the general Equation 1-1. Many techniques are available to measure the standard chemical potentials and partial molar enthalpies and entropies of transfer of compounds from one solvent to the ~ t h e r ~ ' Transfer ~~. parameters can be used (i) to develop a solvent polarity scale or (ii) to determine the change in standard Gibbs energy, enthalpy and/or entropy for the transfer of reactants for a particular process. In the latter case, solvent effects can sometimes be analysed in terms of initial state and transition state effects7'. In fact, measurement of transfer parameters is the only way in which solvent effects can be analysed in terms of initial and transition state effects. In some cases, initial state and transition state effects have been subsequently analysed in terms of solvent polarity scales72. In recent and interesting fundamental studies of reactivity in the gas phase, reactants, present in the gas phase, are stepwise "solvated" by a. limited and accurately

1. Introduction Solvent effects on organic reactions in aqueous mirtures

known number of solvent molecules. The reactivity of these molecules in clusters is monitored as a function of the number of solvent molecules present, and shows that even a small number of solvent molecules can bring about a large change in reactivity77-79. Even with a limited number of solvent molecules, reactivity in the gas-phase starts to resemble the reactivity in the condensed phase.

1.2.4 Theoretical treatments o f solute-solvent interactions in rehation to solvent effeca

The past fifteen years have shown an impressive number of sophisticated theoretical studies concerned with reactivity in the condensed phase. However, an unambiguous and universal theory of the liquid state still does not exist. This fact strongly hampers the theoretical treatment of reactivity in solutions. Early theoretical approaches were mainly based on models of Kirkwood and O n ~ a g e P ~ ' ~ . The first simulations of reactivity in solution appeared in the late sixties. The accessibility of large computers and suitable quantum mechanical methods resulted in a whole new area of theoretical studies of solvent effects. The first ab initio calculations of solvent effects were based on reactive com ounds, "solvated by a few solvent molecules; the so-called super-molecule approachr2. These ab initio approaches are still in the early stages of developments3. Other methods use quantum mechanical approaches to calculate the reaction trajectories in the gas-phase, which are subseMolecular dynamics is used for quently transfered into a box of solvent molecules84986. allowing reacting systems to equilibrate with the solvent along the total reaction path. Until recently only rather simple and elementary processes, like the substitution reaction of chloromethane with chloride ion, have been studied in detail. Strikingly, reactivity in water has received most attention. Many theoretical studies are focused on non-equilibrium solvation. Conventional transition state theory is shown to be unsuitable for the treatment of solvent effects in these casess4. Among the topics which can be studied using modern computers are (i) the time-scale of the reaction dynamics, (ii) the extent and importance of energy flow from solvent to reactants and (iii) the involvement of the solvent dynamics in the activation process. Recent calculation^^^ of the solvent effect on a S,Zprocess in water show that water undergoes a substantial reorganisation well before the change in the charge distribution of the reactants. This reorganisation appears to be crucial for the overall reaction. The development of theoretical models to calculate solvent effects is, however, still in its infancy. Most methods do not take account of mechanistic changes that might be induced by the solvent. The chemical processes, treated by theoretical models, are necessarily still simple and elementary. Consequently, the results do not yet attract the interest of experimental chemists. Unfortunately, the complexity of the methods still creates a considerable gap between experimentalists and theoreticians. It is striking that theoretical approaches of solvent effects do not appear in recent reviews and textbooks on solvent effects in chemist$"32972. Apparently, reading theoretical papers gives experimentalists, dealing with solvent effects, a genuine feeling of dissatisfaction, which makes them turn to their familiar solvent polarity scales.

1. Introduction. Solvent effects on organic reactions in aqueous mi-s

1.3 Interactions and reactivity in water and in m h d aqueous solvents

Organic reactivity in water and in mixed aqueous solvents is determined by interactions of water and cosolvents or cosolutes* with the reactant(s) and the activated comp l e ~ Water-solute ~ ~ ~ ~ ~ interactions . reflect the fact that water molecules are small, moderately polarisable and able to form a highly structured hydrogen bonded network. Induced dipole-induced dipole interactions of water with solutes are small, but the dipole moment of water does enable significant dipole-induced dipole and dipole-dipole interactions with solutes. Obviously, an important contribution to the overall solute-solvent interactions in water is hydrogen bonding. Finally, the interaction of water with charged solutes is very strong. Hydrophobicity and hydrophobic hydration play an important role in the solvation of reactants and activated complex in water and in mixed aqueous solvents. Hydrophobic effects are characterised by intriguing thermodynamic properties and are the result of a combination of water-solute and water-water interactions. Traditionally, hydrophobicity and hydrophobic hydration are considered to be a consequence of the preference of water for interaction with other water molecules over interaction with hydrophobic (i.e. apolar) solutes. In a study of solvent effects on reactions in aqueous reaction media, a discussion of hydrophobicity and hydrophobic hydration is essential. Recent views on these hydrophobic effects will be discussed in more detail in Section 1.4. Interactions of reactant(s) and activated complex with cosolvent or cosolute molecules in aqueous solutions involve all forms of dipolar interactions mentioned above, but are strongly mediated by water. The thermodynamics of these interactions strongly depend on the concentration of the solute molecules. In addition, hydrophobic interactions play a significant role in intermolecular interactions between solutes in aqueous media. The interactions between hydrophobic compounds or hydrophobic moieties in aqueous media is thermodynamically very complex, and some recent ideas about hydrophobic interactions will be also outlined in Section 1.4.

1.4 Hydrophobic effects; &finitions and the state o f the ad9-''

Hydrophobic effects; definition. Disagreements exist in the literature concerning the definition of "hydrophobicity", "hydrophobic effects", "hydrophobic hydration" and "hydrophobic interactions". The terms are strongly .related and some are practically synonymous. Many different definitions for these terms can be encountered in the literatures98. In order to prevent unnecessary ambiguities, a definition of these hydrophobic terms will be given below. The term "hydrophobic effects" describes all phenomena related to the dissolution of non-polar solutes in aqueous media and is, as such, a quite general term. The term "hydrophobicity" is ambiguous and many different interpretations are found in the
The definition of organic, irreactive compounds in aqueous media as cosolutes or cosolvents is ambiguous. Following thermodynamic formalism, the term "cosolute" is more appropriate in dilute aqueous media, whereas in binary aqueous mixtures the term "cosolvent"is prefered.

1. Introduction. Solvent effects on organic reactions in aqueous r n h r e s

literature. The extent of "hydrophobicity" can be expressed in an experimentally accessible parameter measuring the solubility of apolar substances in waterg9. This parameter involves (a) the breaking of the solute-solute interactions and (b) refilling of the vacancy in the apolar medium, (c) creation of a cavity in the aqueous medium, (d) onset of solute-water interactions and (e) rearrangement of the water molecules surrounding the solute. Alternatively, "hydrophobicity" can be expressed in terms of a transfer process of an apolar compound from an apolar solvent to water. In this case, process (a) is replaced by breaking of the solute-solvent interactions. The poor interactions between water and apolar solutes make the small but significant solubility of completely apolar compounds in water rather unexpected. "Hydrophobic hydration" has been suggested to account for this "high" solubility of used the term "hydration apolar compounds in water. Privalov and Gi111009'03J08~'09~111 effect" to describe this phenomenon. "Hydrophobic hydration" can be defined as the combination of process (c), (d) and (e). The term "hydrophobic interaction" seems to express the discomfort of chemists in dealing with non-covalent interactions between apolar molecules in aqueous solution that appear to be predominantly entropic in origin. It is necessary to make a clear distinction between "bulk hydrophobic interactions" and "pairwise hydrophobic interaction^"'^^^^^^. Unfortunately, this distinction is seldomly recognised in discussions about hydrophobic interactions. "Bulk hydrophobic interaction" describes the tendency of apolar molecules or moieties to form solvent unseparated clusters and can be conveniently described as the reverse of process (a)-(e). "Pairwise hydrophobic interaction" refers to the potential of average force between two hydrophobic solutes in water, expressed in G(R), the gradient of which determines the force, necessary to bring the two solutes S from an infinite distance to a distance R, as expressed in Equation 1-292.

In this equation, Uss(R) is the solute-solute interaction potential, or the work required to bring about the same process in vacuum. G~'(R)is the contribution of water to the process in aqueous solution. This term is difficult to quantify and sometimes refered to as hydrophobic interactiong2. The contribution of water to the process of "bulk hydrophobic interactions" is even more difficult to establish experimentally. The terms "hydrophobic hydration" and "hydrophobic interactions" are poorly defined in the literature and the lack of a proper definition has resulted in discussions with a strongly semantic f l a v o ~ r ' ~ ~ ~ ' ~ ~ . Hydrophobic effects; the state of the art. The classical description of hydrophobic hydration, as put forward by Frank and Evanslo4 in 1945 is still very popular. This "iceberg model" was later quantified by Nemethy and ~cheraga"' and Frank and wedo6. In 1959, hydrophobic interactions were introduced in a famous paper by ~ a u z r n a n n ~ These ~'. four papers set the stage for later studies on hydrophobic effects. Figure 1-1 shows the almost exponential increase of papers on hydrophobic effects during the past 25 years. It is impossible to review in this thesis all theories and ideas

I . Introduction. Solvent fleets on organic reactions in aqueous m a r e s

about hydrophobic effects which have been presented in the literature during the last decades. Emphasis will therefore be put on some novel and interesting developments. Recently, the study of hydrophobic effects began to prompt a reconsideration of elder t h e ~ r i e s ' ~ ~ An ~ ' ~important ~ ~ ~ ~ ' . reason for this development is the accessibility of large computers and sophisticated models which have allowed detailed theoretical studies of hydrophobic effects. Also experimental and theoretical studies of protein folding gave a new impulse to the discussion about hydrophobic effects. "New views on hydrophobic effects" have been put forward that seriously contrast with classical descriptions112 l19. In this introducto~ychapter as well as in Chapter 8 new views on hydrophobic effects will be compared to the classical theories on hydrophobic effects. Hydrophobic effects are characterised by a number of remarkable features. The most important feature of hydrophobicity, as expressed in the chemical potential of transfer of a hydrophobic, apolar compound from an apolar phase to water, is the unusual and large heat capacity ~hange"~*"~.
number of t i t l e s

year

Figure 1-1:Number of titles of research papers, containing the word "hydrophobic" as a function of the year.

I . Introduction Solvent effects on organic reactions i n aqueous mixtures

Consequently, partial molar enthalpies and entropies of transfer are highly temperature dependent. The thermodynamic parameters, measuring the process of both pairwise and bulk hydrophobic interactions, are also shown to be strongly temperature dependenta. Privalov et a1.'00~103~108911' reported methods to determine the thermodynamic quantities of hydrophobic hydration, which are very difficult to measure experimentally, and showed that hydrophobic hydration is characterised by a similar temperature dependence of enthalpy and entropy terms. The classical model of hydrophobic effects is mainly based on phenomena observed near room temperature. The hydration shell, surrounding apolar molecules or moieties, is found to be highly structured, minimising the interactions between the apolar molecule or moiety and water and optimising water-water interactions. Detailed t h e o r e t i ~ a 1 ~ "and ~~" computer ~~ simulation studies'22-'" confirmed that the structure of the hydrogen bonded network of water is significantly altered in the first solvation shell surrounding apolar molecules or moieties. The transfer of apolar molecules from an apolar solvent to water is, at temperatures near 25"C, characterised by a positive standard Gibbs energy of transfer, a slightly negative enthalpy and a large negative entropy of transfer*. For a broader temperature range, the thermodynamic parameters of transfer exhibit very characteristic values at two important temperatures. At a temperature TH, which is near room temperature, the enthalpy of transfer is nearly zero, whereas the entropy is strongly negative. At a temperature Ts, which is near 160C, the entropy of transfer is zero, and the enthalpy is highly positive"'. Apparently, the structure forming capacity of water is lost at these high temperatures. The strong changes of AH0 and TASOas a function of temperature are shown in Figure 1-2. The standard Gibbs energy of transfer, as indicated in Figure 1-2, can be described by Equation 1-386~100~'01~'M~10808111.
A GwO = AH* - A Cp[(Ts- ;T)

In this equation, AH*,given by AC,(Ts-T,), corresponds to the enthalpy of transfer of an apolar compound from the apolar liquid phase to water at T , and appears to resemble the enthalpy of evaporation of the liquid, apolar solute. The second term on the right, which is always negative, contributes favourably to the standard Gibbs energy of transfer, and is strongly temperature dependent. This term can be identified as due to hydrophobic hydration. Hydrophobic hydration apparently leads to a decrease in the Gibbs energy of transfer. Both the enthalpy and entropy of hydrophobic hydration are strongly temperature dependent, but appear to be strongly compensating, leading to a small, but favourable overall Gibbs energy of transfer. At 160C it is definitely the unfavourable enthalpy of transfer AH*, which causes the large and positive Gibbs energy of transfer of the apolar solutes from an apolar phase to water. This unfavourable term has to be attributed to the disruption of London dispersion interactions between the apolar solute molecules in the pure apolar, liquid state. At room temperature, however, the loss of entropy determines the unfavourable Gibbs energy of transfer. The loss of entropy, associated with the hydrophobic hydration of the apolar molecules, is also accompanied by a similar gain of enthalpy. As a conse-

1. Introduction Solvent effectson organic reactions in aqueous rniriwes

the height of the bamer. Recently, it has been shown that long-range attractive forces are present between large apolar surfaces148.The molecular origin of these hydration forces is also unclear.

1.5 The need for a quantitative description o f solvent egects in mixed (aqueous) solve&. Incenlives fir this study

In Section 1.2 it has been shown that simple quantitative models to analyse solvent effects in pure solvents which are based on empirical solvent polarity parameters, lack general applicability. Moreover, the molecular basis of solvent effects remains unclear because of the undefined physical significance of these polarity scales. Macroscopic solvent parameters, derived from physical properties of the solvent give an unrealistic picture of the reaction medium as a continuum without specific structure. Specific interactions between solvent and reacting species, as well as the importance of the structure of the solvent, are neglected. More sophisticated, theoretical approaches to analyse solvent effects in pure solvents are either inaccessible or difficult to use for practical problems. Many chemical reactions are performed in mixtures of solvents. Particularly solvent mixtures of water and an organic cosolvent are very popular. Quantitative descriptions of solvent effects in mixtures of solvents are even more complicated than those in the pure solvents. It is, for example, an impossible task to determine solvent polarity parameters in every mixture of solvents. Very often therefore, the assumption has been made that these solvent parameters depend linearly on the composition of the mixture. Obviously, the occurence and consequences of preferential solvation are not taken into account. The even more pronounced complexity of specific solutesolvent interactions is also fully neglected. Patterns of organic reactivity in mixed aqueous solutions are particularly interesting. Kinetic data for reactions in water and in mixed aqueous solvents are often intriguing and their interpretation is a real challenge. The study of organic reactivity in water and in aqueous solutions has become more interesting since water has been shown to induce high reaction rate constants as well as high selectivities for a number of organic reactions, both homogeneous and heterogeneous (see Table 1-1).The low solubility of organic reagents can be overcome by addition of organic cosolvents. The consequences of addition of cosolvents for reactivity and selectivity of reactions in water are unknown. A quantitative study of solvent effects on a series of organic reactions in water and in mixed aqueous solvents, based on a general theory for quantitative analysis of solvent effects in mixed solvents, would offer insight into the molecular basis of rate effects on organic reactions in aqueous media. This knowledge should enable organic chemists to select organic reactions, which can benefit from mixed aqueous solvents with a suitable composition with respect to solubility, reactivity and selectivity, in a rational way. However, it is remarkable that no valid quantitative model exists for the analysis of solvent effects on reactions in mixed solvents.

I . Introduction Solvent effects on m a n i c r e a c h in aquew~s mirtwes

1.6 A h o f this st&

The general aim of this study was to develop and test a new, simple, but general theoretical model for the quantitative analysis of solvent effects in mixed solvents. An important objective was to draw together transition state, theory and thermodynamic formalism to describe thermodynamic properties of solutions. Following this approach, three important demands, made with respect to a novel theoretical model for the analysis of solvent effects in mixed solvents, can be met. Kinetic medium effects in mixed solvents can be expressed in terms of thermodynamic parameters. These thermodynamic parameters can be determined by experimental techni(ii) ques other than kinetic measurements. (iii) The theoretical model enables a quantitative analysis of observed solvent effects leading to further insight into the solute-solvent interactions which govern solvent effects in mixed solvents. (i) A major incentive for the development of a new theoretical model for the quantitative analysis of solvent effects on reactions in mixed solvents has been its possible application to study solvent effects on reactions in mixed aqueous media. In fact, this application would involve a rigorous test for the new approach. The second aim of this study was, therefore, to critically appraise the developed theory by studying the kinetics of, first, simple first-order processes and, later, of bimolecular reactions and equilibria, in mixed aqueous solvents. The general strategy involved systematic variation of the nature of the mixed aqueous solvent (i) by varying of the composition of the mixed solvent and (ii) by varying the structure of the cosolvent(s). In addition, different reaction types have been studied and the structure of the reacting molecules has also been changed by changing substituents. The consequences for a quantitative analysis of the observed solvent effects, based on the novel theoretical model, have been examined. Concomittantly, the variation of the structure of the reactants provides an opportunity for a detailed quantitative study of the contribution of solvation effects to the overall substituent effects of alkyl groups on chemical reactivity in mixed aqueous solvents. A challenging goal was found in the application of the developed theory to elucidate the role of water, and eventually of the apolar cosolvents, in the spectacular rate enhancements of some organic reactions in aqueous media.

1 . 7 Survey o f the contents o f thk thesis

Chapter 1 contains a general introduction in the field of qualitative and quantitative methods for the analysis of solvent effects. After a brief historical overview, emphasis is placed on recent methods to analyse solvent effects. In this context conventional methods are critically discussed. The second part of the chapter is devoted to a description of water and mixed aqueous solvents as reaction media. Attention is focussed on intermolecular interactions in aqueous solvents. Hydrophobic effects are

1. Introduction Solvent effects on organic reactions in aqueous mirrures

clearly defined and recent theories on hydrophobic effects are discussed. Finally, the incentives and aims of the study are summarised. A general theoretical model for the analysis of solvent effects in mixed solvents is reported in Chapter 2. In this chapter, kinetic theory and thermodynamic formalism to describe thermodynamic properties of solutions are drawn together. This leads to theoretical expressions which relate solvent effects in mixed solvents to the composition of the solvent. The general model is elaborated by using two alternative thermodynamic descriptions of the reaction medium, dependent on the composition of the medium. Both methods are discussed in detail and are critically compared. Theoretical expressions are derived for solvent effects on a simple unimolecular reaction in mixed solvents. The expressions are modified in order to analyse solvent effects on solvolysis reactions, bimolecular processes and chemical equilibria. Furthermore, the quantitative treatment for the analysis of solvent effects on Gibbs energies of activation is extended to a quantitative analysis of enthalpies and entropies of activation in mixed solvents. A general feature of all theoretical expressions derived is that they describe the dependence of reactivity on the composition of the reaction medium in terms of interactions of the (co)solvent(s) with the reactants and activated complex, respectively. These expressions form the basis of the second part of the thesis, and will be used frequently. In Chapter 3, solvent effects are described on the neutral hydrolysis of l-benzoyl3-phenyl-1,2,4-triazole in mixed aqueous solvents that contain monohydric and polyhydric alcohols. In the introduction, the vast amount of literature on solute-solute interactions in dilute aqueous solutions is briefly reviewed and some important features are discussed. The new theoretical model, as developed in Chapter 2, is critically tested by analysing the solvent effects of 24 different mono- and polyhydric alcohols on the rates of hydrolysis. The theoretical expression was found to describe the experimental data very well. Particular attention has been paid to the applicability of group additivity approaches for the analysis of solute-solute interactions. It was observed that addivity schemes are only valid under strict conditions and deviation from additivity is discussed in detail. In addition, the effect of urea on solvent effects in mixed aqueous solvents was investigated. Urea reduces the solvent effects of apolar cosolvents, whereas urea itself does not induce a significant solvent effect at all. Solvent effects are expressed in terms of Gibbs energy parameters, describing solutesolute interactions in aqueous solutions, and the results are compared with literature data for the analysis of solute-solute interaction in aqueous media. Emphasis has been placed on the participation of the cosolvent in the solvation shell of the reactant and the activated complex, accompanied by hydration shell overlap. In the first part of Chapter 4, solvent effects on the Gibbs energy, the isobaric enthalpy and entropy of activation of the neutral hydrolysis of p-methoxyphenyl dichloroacetate in mixed aqueous solvents containing urea and alkyl-substituted ureas, are quantitatively analysed. The theoretica1 model, developed in Chapter 2, is shown to describe the dependence of the activation parameters on the composition of the reaction medium. However, higher-order enthalpic and entropic interaction terms are more important for the description of solute-solute interactions in dilute aqueous solutions than higher order Gibbs energetic interaction terms. The origin of this phenomenon is discussed in detail in terms of hydrophobic effects. In the second part of Chapter 4, the theory is successfully applied in the quantitative analysis of solvent

1. Introduction Solvent EfJects on organic r e a c h in aqueous m h s

effects on the Diels-Alder reaction of methyl vinyl ketone with cyclopentadiene, the keto-en01 equilibrium of 2,epentanedione and the intramolecular Diels-Alder reaction acid in mixed aqueous solvents. of N-alkyl-N-furfurylmaleamic A thorough study of the medium effect of ethanol and 1-propanol on the neutral hydrolysis of 18 different l-acyl-3-alkyl-1,~4-triazolesis reported in Chapter 5. The dependence of the solvent effect on the alkyl groups in the substrate was examined in detail. The solvent effects depend critically on the substituent. This implies that the substituent effects of alkyl groups are significantly affected by the composition of the solvent. In the introduction conventional quantitative methods for analysing, understanding and predicting alkyl substituent effects in terms of substituent constants are outlined and discussed. The results show that substituent effects of all@ groups on reactions in aqueous media are strongly governed by solvation effects. It is argued that efforts to describe steric, polar and inductive effects of alkyl substituents either include a substantial contribution of solvation effects or are completely useless. The contribution of solvation to the substituent effect of alkyl groups on reaction rates of reactions in aqueous media is explained in terms of "hydrophobic acceleration". The main theme of Chapters 6 and 7 is the analysis of solvent effects on DielsAlder reactions in mixed aqueous solvents, containing monohydric alcohols across the whole mole fraction range. The applicability is tested of the theoretical expression, derived in Chapter 2, for a quantitative anaIysis of solvent effects in binary solvents. Chapter 6 contains a quantitative study of solvent effects on the rate constants and isobaric activation parameters for bimolecular Diels-Alder reactions of cyclopentadiene with alkyl vinyl ketones as well as with 5-substituted-1,4-naphthoquinones in mixed aqueous solvents. In addition, standard Gibbs energies of transfer of the reactants, activated complex and products of the Diels-Alder reaction of alkyl vinyl ketones and cyclopentadiene from 1-propanol to aqueous solutions of 1-propanol have been determined and analysed. Chapter 7 describes the synthesis and the intramolecular Diels-Alder reaction of four N-alkyl-N-furfurylmaleamic acids. The intramolecular Diels-Alder process undergoes a spectacular rate increase in aqueous reaction media. The solvent effects appear to be very similar to solvent effects on the bimolecular process. Emphasis is placed on solvent effects on the stereochemistry of the intramolecular cyclisation. The solvent effects, reported in Chapters 6 and 7, are satisfactorily described by the derived theoretical expressions. The spectacular rate effects of water on DielsAlder reactions are largely preserved in highly aqueous reaction media. Preferential solvation of the reactants appears to diminish the rate effect of water in the presence of higher concentrations of cosolvent molecules. The spectacular rate accelerations are ascribed to the apparent decrease of the hydrophobic surface of the apolar reactants during the activation process. This "hydrophobic acceleration" is a result of "enforced hydrophobic interaction" of the reactants during the activation process. The possibility of induction of a more polar activated complex in highly aqueous media is tentatively suggested. Chapter 8 evaluates the applicability, merits and shortcomings of the proposed theoretical treatment of solvent effects in mixed aqueous solvents. In this thesis, the dependence of solvent effects on reactivity of apolar organic reactants in water on the concentration of cosolvents, has been mainly explained in terms of bulk and pairwise hydrophobic interactions. Based on the quantitative analyses of these solvent effects a

1. Introduction Solvent effects on organic reactions in aqueous mixtwes

novel model is introduced for the description of hydrophobicity and hydrophobic interactions. This model accounts for the most characteristic properties of hydrophobic effects. Finally, the use of water as a solvent for organic reactions is critically discussed. It is shown that organic reactions can benefit from water as a solvent due to the fact that reactive, hydrophobic species in water and highly aqueous media tend to minimise their hydrophobic exposure to water of the hydrophobic groups during the activation process. Reaction types are suggested which might be accelerated in aqueous solutions. A major part of the work described in this thesis either has already been published or will be published in the near f u t ~ r e ' ~ ~ - ~ ~ ~ .

2. A quantitative anaIysis of solvent L'IJecain mired solvents. Development of a theoretical model

CHAPTER 2 A Quantitative Analysis of Solvent Effects in Mixed Solvents, Development of a Theoretical Model
2.1 Introduction

In this chapter the kinetics of reactions in mixed solvents and thermodynamic formalism for the description of thermodynamic properties of a solution are drawn together. Each separate theme is well established in the chemical literature, and will be briefly introduced. According to the new theoretical model, the combination of kinetic and thermodynamic theories leads to a quantitative description of solvent effects in mixed solvents in terms of the composition of the reaction medium and Mixed thermodynamic parameters having well defined physical significan~e'~'~'~~. aqueous solvents can be classified according to their composition: (i) (ii) Dilute mixed solvents, in which one solvent is present in small mole fractions and is considered as a cosolute; Mixed solvents, prepared from comparable amounts of the two components. Dependent on the choice of the reference solvent, one of the solvents is labeled as a cosolvent.

The exact demarcation between these classes of mixed solvents might be somewhat artificial and depends on the characteristics of the reference solvent as well as of the cosolvent(s) or cosolute(s). An important incentive for defining two classes of mixed solvents stems from the fact that reactants andlor activated complex can be preferentially solvated by the cosolvent. Two thermodynamic descriptions of the reaction mixture are needed to develop a theoretical model for a quantitative analysis of solvent effects on reactions in mixed solvents. The basic thoughts underlying the model are similar. In the next part of this introduction a brief outline will be given of the general procedures and strategies that have been followed. Transition state theory constitutes the general basis for the kinetic analysis of solvent effects. Solvent effects in mixed solvents are caused by the interactions of the cosolvent(s)* with the reactant(s) in the initial state and with the activated complex in the transition state. These interactions result in changes of the chemical potential of the reactants as well as of the activated complex which affect the Gibbs energy of activation because reactant(s) and activated complex are solvated to a different extent. In Scheme 2-1, the cosolvent C interacts favourably with the reactant (IS), whereas the interactions between activated complex (AC) and cosolvent are slightly

In this chapter, the term "cosolvent*is generally used and the distinction between "cosolute" and
"cosolvent" is not made.

2 A quantitative ana3,sis of solvent effects in mired solvents. Development of a theoretical model

unfavourable. The resulting increase in the Gibbs energy of activation is manifested by an effective decrease in reaction rate constant.

Scheme 2-1

The next step is to set down a thermodynamic description of the chemical potential of reactants and activated complex in mixed solvents. Two approaches have been followed. The first approach is based on the theory on dilute solutions of McMillan and ~ a ~ eand r is~selected ~ ~ for J the ~ development ~ of a general theoretical expression that describes solvent effects in dilute mixed solvent^^^'^'^^. The total Gibbs energy of the reaction mixture, including reactants, activated complex, products and cosolvent(s) is determined by the chemical potentials of all components in the reaction mixture. The chemical potential of the reactant(s) and the activated complex can be described effectively by a standard chemical potential, the concentration and the activity coefficients of the reactant(s) and activated complex in the solution. Changes of the chemical potential of reactant and activated complex are a consequence of changes in the activity coefficient of these species. Changes in the activity coefficients of solutes in a mixed solvent are caused by interactions of the solutes with the added cosolvents and are defined in the non-ideal part of the chemical potential. The nonideal part of the chemical potential of the solutes contributes to the excess Gibbs energy of the total solution. The excess Gibbs energy can be conveniently expressed in a molality expansion, using virial coefficients. The physical significance of these virial coefficients is given either by the theory of McMillan and Mayer or by lattice models. In dilute solutions of non-electrolytes the second virial coefficient appears to be the most predominant contribution to the overall excess Gibbs energy. The second order virial coefficient is identified as the pairwise Gibbs energy interaction parameter and is subsequently evaluated in terms of the Savage and Wood Additivity of ~ r o u p s " ~ ~ (SWAG) and the Excess Group Additivity (EGA) appro ache^'^". Molality expansions to describe the excess Gibbs energy of the total reaction

2 A quantitative analysis of solvent eflects i n mired solvents. Development of a theoretical model

medium are not suited for solutions that contain large amounts of cosolvent(s). In these binary solvents, an alternative approach is followed. The changes of the chemical potential of the reactant and the activated complex are interpreted in terms of changes in the standard chemical potential of reactant and activated complex. The standard chemical potential of a solute in solution can be quantified using the theory developed by Kirkwood and and later modified by all"^, Ben-Naim1601161 and ~ e w r n a n ' ~ This ~.~~ theory ~ . expresses intermolecular interactions in solution in terms of inte a1 functions that describe the distribution of cosolvents around solute molecules164,lg . The general procedures outlined above will be followed leading to theoretical expressions for solvent effects on unimolecular first-order reactions. In this chapter, these general procedures will be modified in order to broaden the field of application to solvent effects on chemical equiIibria, bimolecular reactions and solvolysis reactions in mixed solvents. This approach will be extended to the analysis of solvent effects on partial molar enthalpies and entropies of activation. In the last section of this chapter, the two different methods used to develop the theoretical model are compared and critically appraised.

2.2 Solvent effects in dilute mixed solvents

2.2.1 Kinetics o f a reaction; transition state theory

A theoretical model will be developed for a quantitative analysis of solvent effects in dilute mixed solvents by considering a simple unimolecular first-order reaction. This reaction involves the conversion of reactant X to products through an activated complex AC'.

X * ACc

products

A link between a thermodynamic description of the reaction medium, containing reactants X, activated complex AC', cosolvent(s) C and, eventually, products P, and kinetic data is forged using transition state theory. Recognising the important assumptions and inherent weaknesses of this theory, the reactant in the initial state is assumed to be in equilibrium with the activated complex in the transition state166. Therefore

In terms of a standard equilibrium constant of activation, ' K ' , the molality of AC' can be expressed in terms of the molality of reactant X and the activity coefficients of reactant and activated complex.

2. A quantitative anabsis of solvent efects in mired solvents. Development of a theoretical model

In an ideal solution, if the total molality of the solutes is given by m,, in the limit(q+O)y,=y,=l. In this case, neither the reactants nor the activated complex do interact with surrounding solutes. For dilute solutions, it can be assumed that w m .=cJc,, where c is expressed in concentration units, mol dm-3. Therefore, in dilute solutions, according to the law of mass action, the rate constant for this reaction, k, is given by

Following transition state theory, in ideal solutions k can be represented by

In non-ideal solutions, m,+O, reactants and activated complex interact with the cosolvent(s) and k becomes

Hence

Equation 2-6 is now a key relationship which relates kinetic data obtained for reactions in the reference solvent, with kinetic data in non-ideal, mixed solvents containing m, mol kg-' of cosolvent(s) via the activity coefficients of the reactant X and the activated complex AC'.

2.2.2 Chemical potential and activity coemients of solutes in d i l u t e mined solvents

In order to evaluate Equation 2-6 further, the activities of reactants and activated complex must be carefully defined. In the treatment described below the reaction medium contains 1 kg of solvent, m, mole of reactant, m, mole of activated complex, m, mole of product(s) and m, mole of cosolvent(s). The chemical potential of the solutes X as well as AC* can be given by Equation 2-7.

The factor RTlny, measures the deviation of the chemical potential from ideal behaviour. m, is, by definition, 1 mol kg-' and p,' is the standard chemical potential of the solute, i.e. that of a hypothetically ideal solution containing a solute concentrati-

2. A quantitative anabsir of solvent effects in mixed solvents. Development of a theoretical model

on of 1 mol kg-'. The chemical potential of the solvent, p,, is defined as in Equation 2-8

where p,*(l) is the chemical potential of the pure solvent and 4 is the practical osmotic coefficient for which limit(m,+O) +=I. M, is the molar mass of the solvent and m, the molality of the solute S. The total Gibbs energy of the reaction medium, G(sln), in which m,=O, is given by Equation 2-9.

The total Gibbs energy of the reaction medium contains an ideal part G(s1n;id) and a non-ideal part, ~~(sln;nonid), which is defined as the excess Gibbs energy for which

GE(sln;nonid)=RTm,(l-4) + RTmJny,

R%,lny,

Rlh,,hy, 2-10

The activity coefficients of reactants and activated complex are closely related to the excess Gibbs energy of the total reaction mixture according to

This can be shown by taking the derivative of Equation 2-10 with respect to q and by using the Gibbs-Duhem relation for the total reaction mixture

According to the Gibbs-Duhem relation (Equation 2-12) changes in the chemical potential of the solvent affect the chemical potential of the reactants and the activated complex in the reaction mixture. The equations become somewhat more elaborate if m, moles of cosolvent(s) C are present in the mixture. In dilute mixed solvents the cosolvent(s) can be considered thermodynamically as cosolute(s). Therefore, the total Gibbs energy of the reaction medium as well as the Gibbs-Duhem equation can be simply extended with the chemical potential p, of the cosolvent(s). The activity coefficients of the reactants as well as of the activated complex can still be expressed in terms of the excess Gibbs energy of the reaction medium, as shown in Equation 211.

2.2.3 Dependence o f excess pmperties on wmposirion

The excess Gibbs energy of the total reaction mixture, as shown in Equation 2-11, can

2 A quantitative anaQsis of solvent effects in mixed solvents. Development of a theoretical model

only be obtained from experimental data and is characteristic for a given solution. Thermodynamic excess properties of solutions are frequently expressed in expansion series of concentration, density or activity. These virial expansions find a theoretical basis in statistical thermodynamics, as shown by, for example, McMillan and ~ a ~ e r ' ~ and Wood et a ~ . ' ~ According ~. to the theory of McMillan-Mayer (MM-theory), the osmotic pressure, II, can be expressed in a density expansion:143

Virial expansions of the form given in Equation 2-13 are often modified and widely used to represent all kinds of thermodynamic data of solutions. Particularly wellknown are the molality expansions to express the practical osmotic coefficient of a solution,

Also the activity of the solvent, y,, can be expressed in terms of the mole fraction of the solute, x,, as shown in Equation 2-15'~~.

A molality expansion for the excess Gibbs energy of a solution containing only solute-j can be derived from the MM-expansion series, as shown in Equation 2 - 1 6 ' ~ ~ ~ .

Here, gjj and g...are the pairwise and triplet Gibbs energy interaction parameters and P mj is the molality of solute-j. The virial coefficients can be obtained from experimental data, and are sometimes refered to as the Lewis-Randall coefficients16'j. Physical chemists have tried to develop methods to describe the magnitude and sign of these virial coefficients using a theoretical model. In this way, the virial coefficients have been ascribed some physical significance. The magnitude and the sign of the virial coefficients are related to factors such as (i) solute size, (ii) solute-solvent association and (iii) solute-solute interaction^'^^. Concentration expansions of excess thermodynamic properties of solutions have been frequently applied to estimate solute-solute interactions in solutions145.In particular solute-solute interactions in aqueous solution have received much a t t e n t i ~ n ' ~ ' * ' ~ These ~~'~~ models . have been based on (i) lattice or quasi-lattice theories, (ii) distribution theories, (iii) equilibrium theories and (iv) statistical thermodynamics. These approaches have clearly shown that a oversimplified molecular picture of the virial coefficients can be misleading. Kauzmann et al.143developed different approaches for calculating the second and third virial coefficient, based on lattice theories as well as on the MM-theory. In a famous paper143 lattice theories were used to show that solvent-solute interactions decrease the magnitude of the virial coefficients. Moreover, the magnitude of the virial coefficients appears to be clearly related to the size of the solute. The use of

2 A quantitative unabsis of solvent effects in mired solvents. Development of a theoretical malel

Ni,Nj as well as Nj,Ni) and the species in the subscripts indicate all particles involved in the cluster integral, shown in Equation 2-18. In case reactants X, activated complex AC (symbol z), product(s) P and cosolvent(s) C are present, Equation 2-19 yields a summation over all the possible interactions between the solutes present in pairs, triplets and higher order. As shown in Equation 2-11, Iny, is the derivative of the excess Gibbs energy with respect to the molality of x. In the absence of cosolvent(s), the activity of reactants is given by

In the presence of cosolvent(s) this relation is rather elaborate and contains a large number of cross interaction terms. This complicated expression can also be derived for the activated complex and can, together with Equation 2-20, be incorparated into Equation 2-6. A number of important assumptions and simplifications can be made. If the experimental conditions are such that the concentration of reactants, activated complex and therefore products is very low, the medium can, in the absence of cosolvents, be considered ideal. This means that pairwise and higher-order interactions of the reactant, activated complex or product involving other molecules of reactant, activated complex or product can be safely neglected. Moreover, the almost negligible contribution of these terms to the overall excess Gibbs energy in the pure solvent will effectively cancel similar contributions of these interactions in the presence of cosolvent. Combination of Equation 2-20 with Equation 2-6 yields after simplification,

This important equation relates rate constants for reaction in pure solvents to those in solvents that contain m, moles of a cosolvent C in terms of the composition of the mixed solvent and Gibbs energy interaction parameters, describing pairwise and one higher-order interactions between cosolvent C and reactant and activated complex respectively. In dilute reaction media, where the molality of the cosolvent is also low, the dependence of the excess Gibbs energy as given in Equation 2-20, is effectively described by the pairwise terms. A similar result emerges where the theory uses sitesite pair correlation functions98r145. In the presence of small amounts of cosolvent(s), Equation 2-21 can be further simplified to

Inspection of Equation 2-22 shows that ln[k(mc)/k(mc=O;id)] should be a linear

2 A quantitative anabsis of solvent effects in mired solvents. Development of a theoretical model

function of m,. The slope of the linear plot yields the difference in pairwise Gibbs energy interaction parameters, [&-&,I. It is interesting to note that when the molality of reactants is not low, reactant and products can act as cosolvents for the reaction and thereby affect the rate of the reaction.

2.2.4 Solute-solute interactions; additivity apprwches

The pairwise, but also higher-order Gibbs energy interaction parameters are determined by the nature of the solvent, but more profoundly by the nature of the solute. Solute-solute interactions reflect the interactions between the functional groups, which form the solute molecule and are able to interact with their surroundings. ~ a n g m u i r ' ~ ~ noted that simple additivity principles might be used to reasonably predict pairwise energies or enthalpies of interactions. His "principle of surface action" states that "it is reasonable to assume that the field of force about any particular group or radical in a large organic molecule is characteristic of the group and, as a first approximation, is independent of the nature of the rest of the molecule". Since then, additivity approaches are regularly encountered in the literature where they are applied to account for various thermodynamic proper tie^'^^"^^-'^^. In 1976, Savage and Wood introduced a concept of additivity of groups based on empirical equation^"^^. This Savage-Wood Additivity of Groups (SWAG) model is based on three major assumptions: Each functional group-i in solute P interacts with each group-j in solute Q. Each group-i-group-j interaction makes a characteristic contribution to the solute-P-solute-Q interaction parameter. (iii) Each group interaction parameter is independent of all other functional groups and their relative position or stereochemistry in solute-P or solute-Q. (i) (ii) These assumptions lead to the following expression, relating pairwise Gibbs energy group interaction parameters to pairwise Gibbs energy solute interaction parameters.

Here n : and njQ are the number of groups-i and groups-j in solute-P and solute-Q respectively. Gij is the pairwise Gibbs energy group interaction parameter. More recently, Bloemendal and oms sen^'^' developed a similar additivity model for enthalpies in which they take account of the fact that when solute and solvent are identical, gPQ is zero. This model is called the Excess Group Additivity (EGA) approach. The pairwise Gibbs energy solute interaction parameter is related to pairwise Gibbs energy group interactions by

2. A quantitative analysis of solvent effects in mixed solvents. Development of a theoretical model

where n v and njv denote, respectively, the number of groups-i and groups-j in the solvent molecules. Group additivity approaches are generally considered to be first ap roximations and, in their use, several questionable assumptions are made1"16911.d However, attempts have been made to provide empirical additivity schemes a firm theoretical basis. The pairwise Gibbs energy interaction parameter is connected to the MMsecond virial coefficient by the integral function of [exp(-W,&T)], as shown in Equations 2-17 and 2-18, in which W,,,. is the potential of mean force between the l ~ woodlS6% j submitted that, if WpQ is expressed in pairs of molecules. ~ i l l e ~and terms of additive group contributions, Wij must be very small compared to the thermal energy, kT,thereby allowing expansion of the exponential factor into a linear sum of interaction terms.

For some extreme cases, this assumption is not valid. In order for Equation 2-25 to be valid, nearest neighbour modifications, steric hindrance and specific solvation effects must be absent. ~ille~'@ has J pointed to another important theoretical concern in the additivity of Gibbs energy group interaction parameters. Inspection of Equation 2-17 shows that the pairwise interaction term contains a considerable contribution from the partial molar volume of the solute. Partial molar volumes of non-electrolytes in solution are approximately linear functions of the number and types of groups present in the molecule179. The empirical equation, formulating the additivity of group interaction shows a quadratic dependence on the number of the groups present-(see Equation 223 and 2-24)156b. Recently, Marcus and Bloemendal used a quasi-lattice model to provide the empirical SWAG and EGA schemes with a theoretical Recently, Wood ~~ a more sophisticated model, based on site-site interactiand ~ h o m p s o n 'developed ons. The model is too complex to be suited for a thermodynamic description of dilute mixed reaction media. Notwithstanding serious criticism, the simple procedure of predicting signs and magnitudes of solute-solute interactions based on additivity schemes is very useful. Since 1976, thermodynamic data for solutions have been reported by Wood et a1.lS6, Lille et a].'@, Somsen et a1.lS7, Barone et a1.181*182, Borghesani et and others169p1d16711s111851. These studies have provided interaction parameters and a better understanding of the nature of solute-solute interactions in amidic, aqueous and alcoholic solvents. In a pragmatic sense, a breakdown of the applicability of additivity schemes may be used to signal atypical effects. Some interesting data, reported in the literature for solute-solute interactions in aqueous solvents, will be discussed in Chapter 3. The Equations 2-23 and 2-24 can be both incorporated into Equation 2-22. Therefore, solvent effects on the kinetics of a simple first-order process in dilute mixed solvents can be described by

2 A quantitative anahsir of solvent Meets in mired solvents, Development of a theoretical model

In dilute mixed solvents, the plot of ln[k/k,,] versus the molality of the cosolvent(s) is predicted to be linear, and the slope of this plot is given by a sum of pairwise Gibbs energy group interaction parameters, which can, in principle, be retrieved from the literature.

2.3 M o d i f i c a t i o nof the general theoretical model for quantitative analysk of solvent efleds i n mired solvents
2.3.1 SolvolysB reactions

For many reactions in condensed media, the solvent is also a reactant. The case is considered in which reactant X is solvolysed to products through an activated complex AC*, after attack by N solvent molecules S.
X +

N . S * AC*

products

The chemical potentials of reactant X and of the solvent are important. The chemical potential of the solvent is also affected by the presence of solutes or cosolvent(s) (see Equation 2-8). In terms of transition state theory, the standard Gibbs energy of activation, A'G', is given by

Here, q=q+m,+m,+m,. By following a procedure similar to that used in section 2.2.1 for a unimolecular first-order process, it can be shown that

2 A quantitative anabsir of solvent ef/ects in mired solvents. Datelopmenr of a theoretical model

Therefore, by analogy to Equation 2-6, the rate of solvolysis of compound X in the solvent S in ideal solutions is related to the rate of solvolysis in the presence of mc moles of cosolvent(s) in terms of the molality of the cosolvents in the reaction mixture, the activity coefficients of reactant and activated complex and the practical osmotic coefficient of the solvent.

Equation 2-31 can be combined with Equation 2-20 to yield

which provides a quantitative description of solvent effects on the rate of solvolysis reactions in dilute mixed solvents if the reactants are present in small amounts. The pairwise Gibbs energy interaction term can be evaluated by using Equations 2-23 and 2-24. The term Ng5M,mc accounts for the solvent effect on the reactivity of the solvent. Generally this contribution is small, but will depend largely on the molar mass of the i l l not deviate markedly from 1 after solvent. In practice, the osmotic coefficient w addition of small amounts of cosolvent(s). It is also possible to express the osmotic coefficient in a molality expansion (cf. Equation 2-14). As shown by Cassel and the osmotic coefficient can be expressed in terms of if q+m,+m,am, pairwise and higher-order interactions between the cosolvent molecules in the reaction medium.

Experimentally, the osmotic coefficient can be determined from the excess Gibbs energy of mixing of the mixed solvent, according to

2.3.2 Chemical equilibria

Also equilibrium constants characterising chemical equilibria between compound A and compound B depent on the solvent. The equilibrium condition results in the following equation;

2 A quantitative ana&sis of solvent effects in mixed solvents. Development of a theoretical model

In the absence of cosolvents C and in an ideal reaction medium where compounds A and B are only present in trace amounts, y , = y,=l. Addition of cosolvent(s) C affects the activity coefficient of the compounds A and B. The standard equilibrium constant , and y, KO(mc) cannot be affected by the presence of the cosolvent C. Since y depend on m, the equilibrium quotient Q7which is defined as m$m, does depend on m,. Therefore Q(m,) is related to Q(m,=O;id) by

Combination of Equation 2-36 with Equation 2-20 relates the solvent effect of cosolvent C on the equilibrium quotient Q to the composition of the medium in terms of pairwise Gibbs energy interaction parameters involving interactions between A and C and B and C, respectively. Prerequisite is that compounds A and B are present in small concentrations.

Subsequently, the interaction term between brackets can be evaluated, by using Equation 2-23 or 2-24, in terms of pairwise group interaction parameters.

2.3.3 Bimolecular processes

The analysis described so far finds application in understanding the kinetics of a simple first-order or pseudo-first-order process in mixed solvents. The theoretical model for solvent effects on a birnolecular process is complicated. A bimolecular reaction is represented as involving the conversion of reactants A and B to products through an activated complex AC*.

A + B * AC*
According to transition state theory

products

The molality of the activated complex can be represented by

2 A quantitative anabsis of solvent effects in mired solvents. DeveIopment of a theoretical model

where ' K ' represents the equilibrium constant for the activation process, here defined in a molality scale; m,= 1 mol kg*'. According to the law of mass action

where k2 is the second-order rate constant, expressed in dm3 mol-' s-'. According to transition state theory, the second-order rate constant can be expressed as

The conversion from molality scale to concentration scale by simple substitution of the concentration scale by the molality scale in Equation 2-39, is impossible. Therefore the molality scale must be first converted into a concentration scale in which W, is the mass of the solvent, and V is the volume of the total reaction mixture, including cosolvent(s). Hence

where the equilibrium constant is defined on a molar scale. The second order rate constant in this equation is obtained by dividing the pseudo-first-order rate constant, obtained experimentally, by the molarity of the reactant which is in excess. Now, according to procedures described in Section 2.2, it can be shown that

Hence, the solvent effects on second-order rate constants can be expressed in terms of activity coefficients of both reactants and transition state, if proper scale conversion factors are used. Equation 2-43 and Equation 2-20 can be combined, yielding

' as well as W,and W,' are not necessarily the It is important to note that V and V same in magnitude. In Equation 2-44 the solvent effect of cosolvent(s) C on the pseudo-first-order rate constant of the bimolecular reaction is quantified, using a reaction mixture that contains a fixed molality of the reactant in excess to which a certain amount m, of cosolvent is added. The factor between brackets can be analysed in terms of pairwise group interaction parameters, according to Equations 2-23 and 2-24.

2 A quantitative analysis of solvent qects i n mired solvents. Development o f a theoretical model

2.3.4 Solvent eseects on partial m o l a r enthalpies and entropies of activation

In the foregoing sections, the theoretical model has been developed for a description of the dependence of Gibbs energies of activation on the composition of the reaction medium. In a study of solvent effects, changes of the entropy and the enthalpy of activation contain valuable additional information about the molecular origin of the solvent effects8'@. The theoretical model, as developed in Section 2.2, can be extended to a quantitative analysis of solvent effects on the enthalpy and entropy of activation. The partial molar enthalpy of a solute-j in solution is related to the chemical potential, pj at constant pressure.

Hence

where HjO is the standard partial molar enthalpy, which is equal to Hj at infinite dilution. When only solute-j is present in the solution and mi is very small, the pairwise Gibbs energy interaction parameter, describing the solute-j- solute-j interactions, is given by

The pairwise enthalpy interaction parameter, hjj, can be derived from the pairwise Gibbs energy interaction parameter. If the solution is dilute in solute-j

which means that

Hence

Therefore, the partial molar enthalpy of solute-j in dilute solutions, is a linear function of the molality of j. The slope yields the pairwise enthalpy interaction parameter.

2 A quantitative analysis of solvent effects in mired solvents. Development of a theoretical model

Focusing on a unimolecular reaction in a dilute mixed solvent in the presence of cosolvent(s) C, Equation 2-22 yields

The partial molar enthalpy of activation,A*H; for the process is defined by

By taking the derivative of Equation 2-51 with respect to T and using Equations 2-48 and 2-52 it can be shown that
2 A 'H"(mJ - A *H" (m, =O;id)= 7[h,c-hx]mc

2-53

mo
in which the solvent effect on the enthalpy of activation is a linear function of the composition of the reaction medium. The slope is determined by the change of pairwise enthalpy interactions of the reactant X with the cosolvent(s) during the activation process. It is also possible to use Equation 2-21 and incorporate triplet enthalpy interaction terms which yields

pairwise enthalpy interaction parameters can be According to Wood et evaluated in terms of additivity schemes as well.

Here Hi, is the pairwise enthalpy group interaction parameter and nip and n? are the number of groups-i in solute-P and the number of groups-j in solute-Q, respectively. As has been shown by Cassel and the excess enthalpy of a solution can also be expressed in terms of a molality expansion.

The painvise and higher-order enthalpy interaction parameters are determined experimentally, and using Equation 2-54 pairwise enthalpy group interaction parameters can be determined. ~ i l l e y ' ~states '~ that additivity approaches for enthalpic virial coefficients have a much firmer theoretical basis than for Gibbs ener virial coefficients. Recently, Wood et a1. 1569187, Barone et a1?811182, Somsen et al.'gY'?IB, Lilley et

2 A quantitative anaiysis of solvent effects in mixed solvents. Development of a theoretical model

a1.16' and Borghsani et al.ls3 have reported the analysis and determination of pairwise and higher-order enthalpy interaction parameters from dilution experiments as well as from freezing point depression experiments. In chapter 3 some of these results, obtained in aqueous solutions, will be discussed in detail. the pairwise entropy interaction parameter can be Finally, using gpo=hPQ-TsPQ, derived if both the palrwise Gibbs energy and enthalpy interaction parameters are known.

2.4 Solvent enects in mired binary solvents 2.4.1 Kinetics of reaction. Transition state theory

A simple unimolecular first-order reaction will be considered, as represented schematically in Section 2.2.1. The equilibrium condition, shown in Equation 2-1 as derived from transition state theory, can be developed in an alternative way. In the first part of this chapter, the effect of cosolvents on the chemical potential of reactants and activated complex was evaluated in terms of the activity coefficients. An alternative approach is to relate the Gibbs energy of activation for the unimolecular process in solution to the difference in the standard chemical potentials of the reactant in the initial state, k0and the activated complex in the transition state, p,' at fixed T and P.

The change of the Gibbs energy of activation upon transfering the reaction from a reaction medium with composition x2=0 to a reaction medium with composition x, can be represented by

in which k(x2=O) and k(x2) are the rate constants for the reaction in the respective reaction media.

2.4.2 Dependence ofthe stcmdard chemical potentiul of solutes on the composition in mired binary solvents

The chemical potentials of reactants and activated complex in a mixed solvent are affected by interactions with cosolvents. This non-ideal behaviour can be conveniently

2. A quantitative anabsis of solvent effects in mired solvents. LkveIopment of a theoretical model

expressed in terms of the excess Gibbs energy of the total reaction mixture. In dilute solutions the excess Gibbs energy can be evaluated in a molality expansion using pairwise and higher-order virial coefficients. If the molality of the cosolvent increases, the higher-order interaction terms gain importance. The higher-order terms can not be determined accurately from molality expansions. Equation 2-21 is in fact the optimum we can expect from a quantitative analysis of solvent effects in mixed solvents based on the MM approach. In mixed binary solvents which contain cosolvent(s) across the total mole fraction range, in which x, is the mole fraction of cosolvent, higher-order interactions of the cosolvent with the reactant reflect the occurrence of preferential solvation. A theoretical model for the description of preferential solvation effects is offered by the theory of Kirkwood and Buff. In this theo~y,the tendency of solutes or solvents to attract each other is quantified16'1581'59-16363189-191. The Kirkwood-Buff (KB) theory was originally developed to obtain an insight into the molecular organisation within liquids and solutions. An essential parameter in the theory is the so-called Kirkwood-Buff integral, which has been defined ads8

where gpQ is the radial distribution function or angle-averaged and orientationalaveraged pair correlation function. This function measures the conditional probability of finding a particle P in a volume element around particle Q at a distance of r. This parameter must be carefully distinguished from the pairwise Gibbs energy interaction parameter used in the first part of this chapter. The parameter is zero at r values below the contact distance of P and Q and a maximum at the most probable distance. The function approaches unity asymptotically with increasing distance r. The KBintegral GpQis now a measure of the "affinity" of P to Q. The product of the KBintegral with the number density of P is the overall excess or deficiency of particles P in the volume around Q. For short-range intermolecular forces these integrals may be developed in power series in the concentrations of the solutes, the coefficients of which involve distribution functions of sets of n molecules at infinite dilution in the solvent. The resulting expansion is in fact identical to the MM-expansion as shown in * as in Equation 2-18. Furthermore, it can be shown Equation 2-13, with B ~ defined that, for homotactic interaction^'^'

Accordingly, in the limit of very dilute solutions, the KB-integral equals the second osmotic coefficient, which emphasises the relation between the KB-theory and the MM-theory. In the KB-theory all the virial coefficients of the MM-approach are collected in one integral function. Consequently, no conclusions can be drawn from the values of GpQ about the microscopic nature of the interactions that lead to specific solvation effects, because this information is contained in gPQand is lost in the integration process. Since the publication of the KB-theory several authors have reported GpQfuncti-

2 A quantitative ana3)sis of solvent effects in mired solvents. Development of a theoretical model

The theory has been used to probe interactions in binary solvents. More recently ~ a 1 1 and '~~ later Ben-Naim160*161 and ~ e w m a n ' ~ G suggested '~~ a way of using an Inverse KB-theory to treat the problem of preferential solvation of a solute in a binary solvent mixture. This MB-theory offers a quantitative description of solvent effects in solvent m i ~ t u r e s ' ~ ~ " ~ ~ .

0ns189,191

A reaction mixture is considered, containing reactant X, activated complex AC*, products P, solvent 1 and cosolvent 2. The mole fraction of the cosolvent is x , . According to all"^ it can be shown that, if the reaction mixture is dilute in reactant and consequently in activated complex as well as in product, the standard chemical potential of the reactant X is

where

%I
and

CIGX~

Nx2 = c2GX2

2-62

Here, c, is the number density of (co)solvent n, gm is the radial distribution function N is the overaIl excess or deficiency of (co)solof (co)solvent n around reactant X. , vent n in the solvation shell of the reactant. Gm is the KB-integral. p, is the chemical potential of the (co)solvent n. A prerequisite of the theory is that the solute-solute interactions are negligible. Using the Gibbs-Duhem equation for the solvent at constant T,

the standard chemical potential of the reactant is related to the activity coefficient of the cosolvent through KB-integral functions

The activity coefficient can be expressed in terms of the molar excess Gibbs energy of the reaction medium, G ~according , to

From Equation 2-66 it follows that

2. A quantitative ana&sis of solvent effects in mixed solvents. Development of a theoretical model

Combination of Equation 2-65 and 2-67 yields

, ' is given by {(I-x2)Ml + x2M2}/p and p is the density of the medium and where V MI and M, are the molar masses of the solvent and cosolvent, respectively. Equation 2-68 can be rewritten as follows:

where

The difference in affinity of reactant X for solvent 1 and cosolvent 2 is apparently related to the dependence of the standard chemical potential of the reactant on the composition of the reaction medium as well as to the non-ideality of the reaction medium itself. A similar procedure can be followed with respect to the standard chemical potential of the activated complex. Combination of the expressions for the chemical potentials of reactant and activated complex, as given in Equation 2-69, with the expression, derived from transition state theory (Equation 2-58), yields Equation 2-71'~~,

which is used to define a quantity A,. A, is the change in the relative affinities for solvent 1 and cosolvent 2 of the reactant upon going from the initial state to the transition state. Rewriting Equation 2-71 in terms of rate constants yields

in which the derivative of the solvent effect with respect to the mole fraction of cosolvent is a function of (i) a quantity Q, which expresses the non-ideality of the reaction medium, (ii) V , ' , the molar volume of the reaction medium and (iii) 4, which measures the change in preferential solvation of the reactant during the activation process. This theoretical expression for a quantitative treatment of solvent effects in mixed solvents has obvious similarities to the expression derived in the first part of this chapter (Equation 2-21).

2 A quantitative analysis of solvent effects in mired solvents. Development of a theoretical model

It is more tedious to separate the affinity of reactant for solvent 1 and cosolvent 2.

all"^, Ben-Naim161 and ~ e w r n a n 'showed ~~ that, for this purpose, a knowledge of

the partial molar volumes of the solutes is required, and derived that

, is the limiting isothermal compressibility of the medium, V2 is the partial where K molar volume of the cosolvent and V," is the partial molar volume of the solute at infinite dilution. By combining Equations 2-71 and 2-73 a quantity 4, is .defined

which is the change in the affinity of reactant X for solvent 1 upon going from the initial state to the transition state of the reaction. Knowing Akl, AD can be calculated which measures the same parameter, but now with respect to the cosolvent. Rewriting Equation 2-74 yields

Here AV'(x,) is the volume of activation of the reaction at the mole fraction x , of cosolvent, at infiite dilution of reactants. This quantitative approach needs a proper knowledge of the dependence of the non-ideal thermodynamic properties of the reaction medium and of the dependence of the rate constant on the composition of the medium. These dependences have to be fitted to mathematical functions in order to be able to obtain first- and secondorder derivatives. In addition, volumes of activation of the reaction, the molar volume of the solvents and of the cosolvents are required. The treatment can be extended to second-order reactions, schematically represented as shown in Section 2.3.3. The affinity parameters 4 and Akl and A , aie then [GI,-GI,-G,,] and [G2,-G,defined, respectively, as [GI,-G,,]-[GI,-GJ-rG,,-G,,I,
G2~1*

2.5 A comparison of the approaches used to develop a theoretical mo&i for the anaijwis of solvent effects in mired solvents

As has been shown in Section 2.4, the Kirkwood-Buff theory and the McMillan-Mayer theory are related. In the limit of infinite dilution, both theories are even fully identical. The underlying concepts for the development of a theoretical model for a quantitative analysis of solvent effects in mixed solvents, based on the two alternative statistical thermodynamic theories, are very similar. The physical parameters, encountered in the final expressions, describe essentially the interactions of reactants and activated complex with the cosolvent(s).

2. A quantitative analysis of solvent effects in mired solvents. Development of a theoretical model

The theoretical model based on the MM-theory, as shown in Section 2.2, is only valid in mixed solvents in which one of the solvents is in large excess. The exact limits have to be assessed for each solvent mixture. Linking transition state theory with the KB-theory, as shown in Section 2.4, is in principle valid across the whole mole fraction range of cosolvent. The final theoretical expressions, based on the MM-theory and extended with group additivity schemes, allow a quantitative and very subtle analysis of solvent effects in dilute mixed solvents and provide detailed information about the interaction of reactants and activated complex with cosolvents in solution. In fact, the theory provides a quantitative measure for "molecular recognition" processes in these media. Application of the equations, based on the KB-theory, leads to far less detailed information about the microscopic origin of interactions in the reaction medium. Instead, an analysis based on the KB-theory explains solvent effects in mixed binary solvents in terms of bulk interactions that form the basis for preferential solvation of reactants and activated complex in binary solvents. Interpretation of solvent effects in terms of affinities of reactants and activated complex for solvents that make up the mixed reaction medium offers better insight into the local organisation and structure of the reaction mixture, but the approach does not lead to an interpretation of subtle solvent effects in dilute mixed solvents. The models provide complementary information. They give eventually a complete picture of the subtle pairwise interactions between cosolvents and reacting species in dilute mixed solvents which lead, either cooperatively or additively, to bulk interactions and preferential solvation of reactants and activated complex in binary non-dilute solvents. From a more practical point of view, the equations based on the MM-theory need input of kinetic data, obtained in mixed solvents with known composition. The equations, based on the KB-theory, are more complex and require, in addition to kinetic input, also a considerable amount of thermodynamic data of the reaction medium. Fortunately, for many binary reaction media these parameters are known. Application of both approaches for quantitative analysis of solvent effects in mixed aqueous solvents will be presented in the second part of this thesis.

3. A &aI

appraisal of the additivity princble

CHAPTER 3 Quantitative Analysis of Solvent Effects in Highly Aqueous Media. Application of the SWAG Procedures and a Critical Appraisal of the Additivity Principle

Rate constants for chemical reactions in water often change dramatically when organic cosolvents are addedg7@.Chemical reactivity, involving solutes in aqueous solution, is governed by specific, noncovalent interactions between inert cosolvents and the reacting species. These types of interactions are also responsible for the stabilisation of proteins19', biological membranes193, micelles and other aggregates194in aqueous solution. In Chapter 2, we introduced a treatment of solvent effects on reactions in dilute mixed solvents using transition state theory, which provided a link between phenomenological rate constants and parameters, calculated from equilibrium properties of mixed solvents. The dependence of the rate constant on the nature of the added cosolvent and the composition of the solution was analysed in terms of pairwise Gibbs energy interaction parameters, involving interactions of the cosolvent with reactant and activated complex, respectively. In this chapter, a study is presented of the medium effects of 25 monohydric and polyhydric alcohols on the neutral (1).The hydrolysis of the activated amide bond in 1-benzoyl-3-phenyl-l,2,4-triazole mechanism of this reaction is shown in Scheme 3-1.

Scheme 3-1

3. A critical appraisal o f the additivity principle

The hydrolysis proceeds via a dipolar activated complex, in which two water molecules Schematically, ~'~. the reaction can be summarised as follows: are i n ~ o l v e d ' ~ ~
X + N.H20

* [X-N.H20]* products

The general treatment, developed in Chapter 2, can be adopted conveniently for the treatment of solvent effects on a hydrolysis reaction in dilute aqueous media. In the case of solvolysis reactions, the reactivity of the solvolytic agent, in this case water, is also incorporated151. The observed rate effects, expressed in pseudo-first-order rate constants, are quantitatively analysed in terms of Gibbs energy interaction parameters, and subsequently in pairwise Gibbs energy group interaction parameters. The consequences of gradual changes of the structure of the cosolvent are monitored. This large set of kinetic data affords a challenging test for the merits of the additivity rinciples of Savage and Wood SWAG)'^^^, and Somsen and Bloemendal (EGA)' 7', which have been introduced in Chapter 2. Therefore, we have examined the procedures, described by Wood et a1.156bin formulating solute-solute interactions in terms of pairwise Gibbs energy group interaction parameters and hence their success in accounting for trends in kinetic data. In addition, solvent effects of n-propanol on the rate of hydrolysis of 1 have also been determined in urea-water mixtures. Specific, noncovalent interactions between solutes in aqueous solutions have been frequently analysed in terms of pairwise and higher-order interaction parameters. As shown in Chapter 2, the pairwise Gibbs energy interaction parameters are directly related to the potential of average force, acting between the solute particles, via the second osmotic virial coefficient. The magnitude and the sign of the pairwise Gibbs energy interaction parameters give a valid estimate of the force, acting between two solute molecules in water. However, few Gibbs energy interaction parameters have been reported. Enthalpy interaction parameters, however, have been frequently reported. The present results yield detailed information about pairwise Gibbs energy interaction parameters, involving monohydric and polyhydric alcohols and reacting species. The simple additivity schemes are found to be valid for the analysis of interactions involving apolar hydrocarbon moieties. The interactions involving the polar hydroxy groups are complex, and can be applied only under strict conditions in terms of additivity schemes. The present results enable a detailed analysis of the small, but significant deviations from additivity.

P.

3.2 Solute-solute interactions in aqueous solution

The last decade has witnessed a renewed interest in the details of solute-solute interactions in aqueous solution145.This is partly due to the development of refined solution theories and computer simulation techniques. In addition, considerable effort has been directed towards the measurement of thermodynamic properties of aqueous solutions. Both solution theories and computer simulations have been used to calculate potentials of mean force between two solute particles or the related second osmotic virial coefficients. Most theoretical studies have been focused on the calculati-

3. A critical appraisal o f the addin'v@vprinciple

on of interactions between apolar groups or molecules in aqueous media. In contrast, experimental studies show a preference for the evaluation of interactions between more polar compounds. These interactions are more difficult to assess by theoretical calculations. '~~ models to analyse In a classic paper, Kauzmann, Knight and ~ o z a k developed virial coefficients using the dilute solution theory of McMillan and Mayer and wellknown lattice and quasi-lattice theories. Kauzmann, Knight and Kozak showed that attractive forces between a large series of solute molecules increase with increasing size of the alkyl groups as well as with increasing number of functional groups, capable of participating in hydrogen bond formation. Repulsive forces were estimated by calculating the repulsive forces between hard spheres in a solution. Later, Friedman et a1.'21J44 used a r-9-repulsion term to calculate repulsive forces, in which they used estimates of the radius r, based on the partial molar volume of the solutes. Furthermore, Friedman et a1.14 introduced an additional hydration shell overlap term, representing the attractive and/or repulsive forces arising outside the core region of the solute (also called the Gurney overlap term). The latter term appeared to be predominantly attractive for apolar solutes. A different approach was reported by ~ , used a site-site interaction model to calculate solute-solute Franks et a ~ . ' ~who interactions in the gas-phase, and, in a subsequent step, introduced spherical hydration spheres (Gurney cospheres) to model the difference between average solute-solute forces in the gas-phase and in water. In the model described by Friedman, all attractive forces are present in this Gurney overlap term. Notwithstanding basic differences between both approaches, the main conclusions are similar. Water appeared to decrease the attractive forces between small hydrocarbons like methane and ethane, as compared to those present in the gas-phase. Recently, Wood et a1.14' confirmed these conclusions by using a refined site-site interaction model to calculate potentials of average force between solutes in water. Furthermore, in this study it was shown that water tends to increase the attractive forces between large hydrocarbon molecules, with carbon chains becoming as large as 6-8 carbon atomsg1. This fact was already elegantly decribed in an important review of Abraham198. Many models for the calculation of solute-solute interactions have been advanced to reproduce solvation data of apolar compounds in water. As pointed out in chapter 2, solvation data are only valid for the analysis of bulk solute-solute interactions in aqueous solutions. These models, when used to estimate pair interactions, overestimate the attractive forces. Recently, Scheragalg9reported a refined model, based on the calculation of solvation energies to correct pair potentials between solutes in va~uum'~ The ~ . solvation contributions are in fact similar to the Gurney cosphere terms, used by Friedman14 and frank^'^^. Due to the addition of solvation Gibbs energies to the vacuum interactions, attractive forces between apolar groups are always larger than in the gas-phase. This conclusion is in conflict with recent experimental data about pairwise hydrophobic interactions, but it is a convincing demonstration of the opposite effect of water on pairwise and bulk hydrophobic interactions. p an ford^ and Ka~wnann'~' used solvation data to correlate solute-solute interactions in aqueous media with accessible surface areas. These models also overestimate attractive solutesolute potentials. An elegant approach to model solute-solute interactions, presented by Ben-Naimg2, is based on the solubility of apolar gases in water. According to this model, which has been developed to account for interactions at small distances, water

3. A critical appraisal of the additivity principle

increases the attractive interactions between apolar groups at short distances r, but this effect changes sign near contact distance. A very intriguing point, raised by BenNaim and later reappreciated by wood14', is that the overall attractive force between many apolar groups or molecules is smaller than the sum of all individual pair potentials if these groups or molecules are in close contact. During the last fifteen years, a few integral theories have been advanced to calculate the potentials of mean force between two solutes in aqueous solution. In 1977, Pratt and Chandle?' calculated the shape of the pair potential as a function of the distance between two methane molecules. The pair potential, describing the force acting between two methane molecules, oscillates as a function of the distance r, persisting over several water molecular diameters135w, The contact configuration appears to be disfavoured over the solvent-separated configuration. ~ e t t i t t ' ~ ~ ' ~ ~ , ~anaka'~~ and ~'~ others'36 ' have calculated the solute-solute interactions between more complex molecules. The conclusions are, however, contradictory. In general, computer simulations of solute-solute interactions in aqueous media lead to a description of the pair potential as a function of the distance between the solutes. Only in very few cases, a virial coefficient has been actually calculated. One of the earliest simulations was performed by eve ridge'^^^'^^ and Panagali12'. The pair potential between two hard spheres appeared131 to be an oscillating function of the distance r. Later, potential energy-distance curves were reported for interactions between rare gases 029203, and even larger molecules, like 2-methyl-2-propanol1= and methanolm. The exact shape of the potential curves depends very much on the model that has been applied to describe water2". Generally, contact configurations appear to be energetically more favourable, but solvent-separated configurations are more probable because of the larger volume of the solvent-separated well. However, increasing the size of the solute leads to a rapid increase in contact interactions, whereas the volume does not increase as rapidly. Theoretical calculations of solute-solute interactions, involving more complex and polar molecules, are not common. In fact, for the analysis of interactions between molecules in aqueous solutions, encountered in practical chemistry, we have to rely almost completely on qualitative and empirical approaches that have been developed to account for thermodynamic properties of these solutions. Experimental studies on thermodynamic properties of aqueous solutions have resulted in the production of an enormous number of pairwise and higher order interaction parameters (see Table 31). Most frequently, pairwise enthalpic interaction parameters have been determined near room temperature. Unfortunately, enthalpies of solvation of compounds in water and enthalpic interaction parameters are strongly temperature dependent'78. In addition, as shown by Wood et al., enthalpy interaction parameters are largely compensated by entro y interaction parameters, often leading to very small Gibbs 81a,206 . Therefore, the significance and structural implications of energy parameter~"~g-~* the sign and magnitude of enthalpic interaction parameters is questionable. However, the determination of Gibbs energy interaction parameters is more elaborate. For pragmatic reasons, empirical and qualitative analysis of solute-solute interactions in aqueous solutions is generally based on enthalpic interaction parameters. According to an empirical classification of experimental data, solute-solute interactions between non-electrolytes have been distinguished in three types (see below), produced by hydrophobic and hydrophilic spe~ies'~'g~'~~J. The demarcation

'

3. A critical appraisal of the additivity principle

between hydrophobic and hydrophilic molecules can be vague. Hydrophobic molecules are often considered to be structure-making. This is an old and questionable classification and reflects the tendency of water to form a highly structured solvation shell surrounding the apolar parts of the molecule. The disadvantage of this kind of classification becomes clear when hydrophilic compound~.~have to be subdivided into hydrophilic structure-making and hydrophilic structure-breaking solutes.
Table 3-1. Literature References for Pairwise Gibbs Energy and Enthalpy Interaction Parameters of Aqueous Solutions.

Solute(s) Monohydric and Dihydric Alcohols Polyhydric Alcohols Carbohydrates Urea +Derivatives Amides Ethers Amino Acids Carboxylic Acids Esters Urea+ Alcohols Urea +Carbohydrates Urea +Amides Amides+Carbohydrates Amides +Polyols Amides +Ethers Alcohols Carbohydrates

&and% 182ij,k,183a-c,156a 181f,k,n,182, 183a-c,156b,f,g 181m,n,r,182b,f,156b,206 181a,c-e,v,156a,b 181w,168a,c-e,hj-k,156b,f,k 156fj 182e,168a-1 156e 1561 181h,i,1,156b 1810,182b,f7156b 18lgj,p,t9y,182g,h, 168b,g,i 181q,u,156b,h,i 156h,i 156ij 181nj

G , and G,

181f,183a-c,156c,f,g 206 181a,c,156a,c 156c,d,f,k,168a,c-ej-k 156fj 168a-1 156e 1561

181g

156j

3. A critical appraisal of the additiviiy principle

Carbohydrates are generally classified as structure-making hydrophilic molecules, which is attributed to the fact that hydration of these molecules involves extensive hydrogen bonding to water. Urea and related compounds belong to the categoly of hydrophilic, structure-breaking solutes. For many solutes, like glycol and glycerol, the classification is ambiguous. Recognising its weak points, a rough gradation of solutes is very convenient for a subdivision of solute-solute interactions in terms of pairwise interaction parameters. First. a distinction has to be made between homotactic and heterotactic interactions. subdivision of homotactic interactions is relatively straightforward. At room temperaturelSx (i) (ii) (iii) Hydrophobic structure makers Hydrophilic structure makers Hydrophilic structure breakers G,<O; G,>O; G,c O;

Tk>&>O I-pTS,>O
Q<TS,cO.

For heterotactic interactions the situation is more complicated. Barone et a1.181d have defined an empirical relationship to calculate the magnitude of pairwise enthalpic interactions for solutes X and Y, belonging to the same class of compounds.

Where solutes do not belong to the same class of compounds, or if the classification of solutes was ambiguous, these simple rules are not obeyed. Many aspects of heterotactic interactions have, as yet, to be clarified. Heterotactic interactions have been qualitatively analysed in terms of the hydrophobic and hydrophilic domains of the solutes. Trends in pairwise heterotactic interaction parameters arise mainly from the radically different ways in which water molecules are organised in the hydration shell of the solutes. The destructive overlap of local hydration spheres has been shown to be extremely important. In Table 3-1, references are given to studies, in which interaction parameters for aqueous solutions of different classes of solutes are reported. and Somsen and ~ l o e r n e n d a l have ~ ~ ~ 'introduced empirical Wood and equations, based on additivity of painvise group interactions (see Section 2.2.3). This approach has been frequently appIied to analyse thermodynamic properties of aqueous solution^'^^. In Table 3-2 pairwise group interaction parameters are summarised, taken from a recent compilation. Using these painvise group interaction parameters, heterotactic, as well as homotactic interaction parameters can be calculated. The results are certainly encouraging, but deviations from additivity have been described and discussed extensively181a9'68JJ56f~~t167. It has become clear that solute-solute interactions can be very specific. Interactions between polar groups, such as hydroxyl groups, are dependent on the specific orientation of the groups in the molecule, and can dominate less orientationdependent interactions between hydrocarbon groups, as has been shown for interactions between dihydric alcohols in water. Additivity approaches have been criticised remorselessly178.However, application of additivity schemes, recognising the underlying gross assumptions, has provided chemists with a simple but valuable tool for analysing and understanding solute-solute interactions in aqueous media156b.

3. A critical appraisal o f the additi*

prhcipIe

Table 3-2. Gibbs Energies and Enthalpies of Pairwise Functional Group Interactions, Gij and in Aqueous Solution, at 25OC8.

Yj

Group i-j CH2-CH2 OH-OH OH-CH2 CONH-CONH CONH-OH COW-CH2 CHOH-CHOH CHOH-CONH CHOH-CH,

Gi

t,J kg rn01-~

H~ jb, J kg mor2

'Taken from ref.l56f,h. 'standard errors are very large, often 20-40%.

3.3 Medium effects o f monohydric and polyhydric alcohoh on the hydmrysk reaction o f I-

benzo9-3-phenyZ-l,2,QM

The quasi-SWAG approach. The solvent effects of alcohols on the rate of neutral hydrolysis cf 1 can be understood quantitatively in terms of Equation 2-32 (see Section 2.3.1). In dilute aqueous solutions of alcohols, ln[k(mc)/k(mc=O)]is a linear function of the molality of the cosolvent. The slope of the plot yields the difference sum of pairwise Gibbs energy interaction parameters describing the interactions of the cosolvent with the reactant X and activated complex, respectively, [&-g,,]. The factor NM,4mc describes the effect of the cosolvent on the reactivity of water, that is a reactant as well. This equation can be re-expressed using a quasi-group additivity approach.

In this equation, G, and G,, represent the pairwise Gibbs function interactions of, respectively, reactant and activated complex with groups i in the cosolvent. The term between brackets in the right-hand side of Equation 3-2, multiplied by (lfm,)', is defined as G(C) and will be refered to as the solvent effect of cosolvent C on the hydrolysis of 1 . The contribution of each group i to the overall value of G(C) can be calculated from a comparison of solvent effects of structurally related alcohols. It has been shown by Friedman121and Wood'56bthat a methyl group can be represented by three CH groups, according to

3. A critical appraisal of the additivity principle

Actually, only those groups are considered which can interact with surrounding molecules. The quaternary carbon atom in t-butanol is not included in a group additivity approach. If complete additivity is observed, rate effects of alcohols on the hydrolysis of 1 should be governed by two parameters, the contribution of the OH group, expressed as G(OH), and the contribution of the CH groups, expressed as G(CH). In terms of the EGA approach, Equation 3-2 can be written as follows

Here, the functional groups present in the solvent (nv) are taken into account and n v is the number of groups i in the solvent. Alcohol-water mixtures. For a long time ,physical chemists have been intrigued by the fascinating properties of water-alcohol mixtures. In particular at low concentrations of alcohol, the properties of alcohol-water mixtures are extraordinary complex. Refined solution theories, computer simulation techniques and experimental data have resulted in a far better understanding of the structure and dynamics of aqueous solutions of alcoholsm7.The properties of alcohol-water mixtures are usually described by considering separately several distinct concentration ranges. The demarcation between these concentration ranges depends on the characteristics of the particular alcohol. The properties of aqueous solutions of alcohols can change dramatically within a narrow concentration range. In this chapter we are mainly concerned with the very dilute concentration range. In Chapter 6, structural aspects of more concentrated aqueous solutions of alcohols will be discussed. Infinitely dilute solutions are only interesting with respect to the solvation behaviour of a particular alcohol. In dilute solutions, intermolecular alcohol-alcohol, but also alcohol-reactant and alcoholactivated complex interactions start to occur. This gives rise to a concentration dependence of thermodynamic properties and thus to the occurence of solvent effects. By definition, triplet- or higher-order interactions are not observed in dilute aqueous solvents. Complication of the analysis of solvent effects, due to the onset of preferential solvation is therefore very unlikely. We have measured pseudo-first-order rate constants for hydrolysis of 1in a series of dilute (0-1.5 mol kg-') alcohol-water mixtures. In these media, alcoholysis may compete with hydrolysis and therefore the nature of the reaction products (carboxylic acid versus carboxylic ester; see the Experimental Section) was carefully examined. However, the presence of the most reactive alcohol, i.e. methanol (1 mol kg-'), only led to a 5% yield of methyl benzoate. Methyl benzoate was not formed by esterification of benzoic acid with the alcohol. Therefore, in an analysis of medium effects, the contribution of alcoholysis can be safely neglected. Of course, alcohol could also function as a general base in the hydrolysis of 1, but in view of the large excess of water molecules, this process will not contribute significantly to the overall reaction rate. Monohydric Alcohols. The medium effect of a series of monohydric alcohols on the hydrolysis of 1, expressed as ln[k(m,)/k(m,=O;id)], is plotted against the molality of

3. A critical apprairal o f the a d d i t i v i t y principle

the added alcohol in Figure 3-1. AU functions show, in particular at low molalities, perfect linearity. All alcohols induce a decrease of the reaction rate constant. Except for methanol, this is expressed in the negative value of G(C) (Table 3-3), which is a consequence of the less favourable sum of pairwise Gibbs energy group interaction parameters between the added alcohol and the activated complex than between the alcohol and the reactant.

0.5

0.L -

0.3

0.2 -

0.5

1 .O

1.5

rnc,rnol.kg

-1

2.0

Figure 3-1. Plots of -ln[k(m,)/k(m,=O)] vs molality of monohydric alcohols for the , methanol; , ethanol; A , 1-propanol; V , 2neutral hydrolysis of 1 at 25 "C; propanol; , 1-butanol; A , 2-methyl-2-propanol; 0 , 2-butanol; V , 2-methyl-lpropanol; +, cyclopentanol.

3. A critical appraisa! of the additivity principle

This pattern clearly reflects the loss of hydrophobicity of the substrate during the activation process and the hydrophobicity of the alcohols. The decrease of the reactivity of water, due to the presence of alcohols, contributes only very moderately to the overalI medium effect. Only in case of methanol, the rate effect appears to be reversed because of the effect of methanol on the activity of water. If the value of G(C) is zero, the slope of the plot of ln[k(m~/k(m,=O)]versus m, is determined fully by the reduced reactivity of water. This slope amounts to -.036, which is veIy small. G(CH2), and consequently G(CH), can be derived from the increments between the slopes of the lines in Figure 3-1, for example

In this way it is found that G(CH)=-68 J kg m01-~.Using the SWAG approach, G(0H) can now be derived. For example

Clearly, the apolar methylene group retards the hydrolyand G(OH)=226 J kg m01-~. sis reaction, whereas the polar hydroxyl functionality exerts a rate increasing effect. This agrees very well with the notion that the substrate partly looses its apolar character during the activation process. The calculated values of G(C), using derived values of G(CH) and G(OH), are shown in Table 3-3, and agree with the experimental values. The SWAG approach predicts a unique value of G(CH) for all isomers of propanol and butanol. In terms of the counting of functional and interacting groups, even cyclopentanol should induce a similar medium effect as all isomers of butanol. Although the medium effects, induced by the isomers, are significantly different, the differences between the calculated and the experimental value of G(C) are small. This indicates that the configuration of the alkyl group is not an important factor in determining the magnitude of the overall medium effect of monohydric alcohols. When the EGA approach is used, the contribution of the polar hydroxy groups of the alcohol should be compared to the interaction of reactants and activated complex with the hydroxy groups of the water molecules. The contribution of the methylene groups can be calculated from the increments between the slopes of the lines in Figure 3-1, similar to the procedures used for the SWAG approach. In terms of the EGA approach, the hydroxy group of the added alcohol appears to increase the rate constant more than two hydroxy groups of water: the value of G(OH)= -226 J kg mol2

Dihydric Alcohols. Dihydric alcohols also induce a decrease in the rate of hydrolysis . In Figure 3-2 the medium effect, expressed as ln[k(mc)/k(mc=O;id)], of vicinal of 1 dihydric alcohols is plotted as a function of the molality. For other dihydric alcohols similar plots were obtained and excellent linear behaviour was found. Again, the rate retardation is governed by the negative value of G(C) (Table 3-4), indicating that dihydric alcohols exhibit a predominantly hydrophobic behaviour.

3. A critical appraisal o f the a d d i t i v i t y principle

Table 3-3. Neutral Hydrolysis of 1 in Aqueous Solutions of Monohydric Alcohols at 25OC. Application of Quasi-SWAG and Quasi-EGA Approach.

G(c)-,~ J kg mol"
methanol ethanol 1-propanol Zpropanol 1-butanold 2-butanold 2-methyl-1-propanol 2-methyl-2-propanol cyclopentanol

G(c)==~~=, J kg mol"

'Concentration 0-2M. '~x~erimental value. 'Calculated value, using the SWAG approach; G(CH)=68 J kg rn01-~ and G(OH)=226 J kg rn01-~,and using the EGA approach; G(CH)=-68 J kg rn01-~ and G(OH)=-226 J kg rnoU2. d~ncentration 0-1M.

Following the SWAG approach, and using the values of G(CH) and G(0H) as derived for the monohydric alcohols, the G(C) parameters can also be calculated. In most cases, the discrepancy between calculated and experimental values is dramatic. Generally, dihydric alcohols exhibit more pronounced hydrophobic character than anticipated. Only when both hydroxyl groups are not in close proximity (e.g. in 1,4butanediol), is agreement between calculated and experimental value satisfactory. Obviously, it is not possible to express medium effects of monohydric and vicinal dihydric alcohols in terms of two unique painvise Gibbs energy interaction parameters, G(CH) and G(0H). This result is a clear demonstration of the breakdown of the SWAG approach. From the increments that emerge from Figure 3-2, an alternative set of parameters can be derived, describing the medium effect of vicinal dihydric and G(OH)=62 J kg mol". Apparently, alcohols. In this case, G(CH)=-42 J kg rn01-~ G(CH), derived for vicinal alcohols, and consequently also G(OH), are considerably smaller than those for monohydric alcohols. The data suggest that 1,3-diols show behaviour between that of 1,Zdiols and 1,n-diols, where n>3. The medium effects of 1,n-diols where n > 3 can be conveniently analysed with G(CH) and G(0H) parameters derived for monohydric alcohols. In terms of the EGA approach, the dihydric alcohols contain an equal number of hydroxy groups as the water molecule. Therefore, the medium effect has to be totally ascribed to the apolar methylene moieties of the alcohols. Inspection of Table 3-4 shows immediately that isomeric dihydric alcohols exert a quite different medium effect and that the EGA procedure is bound to fail. Using the EGA approach and the G(CH) parameter derived for monohydric alcohols, the hydrophobicity of the dihydric alcohols is largely overestimated (see Table 3-4). Using the alternative parameter, as derived above for vicinal dihydric alcohols, the hydrophobicity is still fairly overestimated.

3- A critical appraisal of the additivity principle

Figure 3-2. Plots of -In[k(m,)/k(m,=O)] vs molality of dihydric alcohols for the neutral hydrolysis of 1 at 25OC; A , glycol; , 1,2-propanediol; 0, 1,2-butanediol; 0, 2,3butanediol; , trans-1,2-cyclopentanediol; W , ck-1,2-cyclopentanediol.The solid lines represent calculated rate constants (see text).

Obviously, proximity effects play a dominant role in determining the contribution of the hydroxy groups to the overall medium effect. The nature of this proximity effect is complex, and depends not only on the relative position of the hydroxy groups, but also on the stereochemistry of the hydroxy groups. This is clearly demonstrated by the significant difference in the medium effect of cis- and tram-1,2-cyclopentanediol. Somewhat unexpectedly, the cyclic pans diol exhibits more apolar character than the cis isomer. Polyhydric Alcohols. Even notoriously hydrophilic polyols, as listed in Table 3-5, are found to induce a remarkable decrease of the rate of hydrolysis of compound 1. Plots

3. A critical appraisal of the additivity principle

of ln[k(mc)/k(mc=O)]versus the molality of the cosolvents show perfect linearity. The experimental values of G(C) are all negative. The calculated values of G(C), using the derived values of G(CH) and G(0H) for monohydric as well as for vicinal dihydric alcohols, show poor agreement with the experimental G(C) values (Table 3-5). Again, the proximity of hydroxy groups on the carbon backbone gives rise to problems for an analysis in terms of the SWAG approach. The polyhydric alcohols appear to behave as surprisingly more hydrophobic solutes than anticipated. The increments between the G(C) values of the polyhydric alcohols, corresponding with the addition of a CHOH-group, are not constant at all. This, again, is a demonstration of the breakdown of additivity schemes, which has to be attributed to the proximity of the polar hydroxyl groups. The EGA approach fails completely in accounting for the observed medium effects. l-Propanol-Urea-Water Mixtures. In Figure 3-3 the rate effects of l-propanol, expressed as In[k(m,mu)/k(mc=mu=O)], for the hydrolysis of 1 is plotted as a function of the molality of l-propanol, at different molalities of urea. Addition of l-propanol to water induces a decrease of the reaction rate (Figure 3-1). In separate experiments it was shown that the addition of urea (0-10 M) does not lead to a significant rate effect and that urea can be considered as an apparently inert cosolvent. This appearance is deceptive, as shown by the fact that in the presence of more than 0.8 mol kg-' of 1propanol, urea enhances the rate constant.
Table 3-4. Neutral Hydrolysis of 1in Aqueous Solutions of Dihydric Alcohols at 25C. Application of Quasi-SWAG and Quasi-EGA Approach.

2,3-butanediol 1,3-butanediol 1,4-butanediol 1,5-pentanediol 1,6-hexanediolg cis-1,2-cyclopentanediol mans-1,2-cyclopentanediol

'Concentration 0-2M. '~x~erirnental value. CCalculatedvalue, using the SWAG approach; G(CH)=68 J kg m01-~ and G(OH)=226 J kg m01-~. dCalculatedvalue, using the EGA a roach; G(CH)=-68 J kg rnol-?. 'Calculated value, using the SWAG approach; G(CH)=-42 J k mol- and G(OH)=62 J kg rn01-~'Calculated value, using the EGA approach: G(CH)=42 I kg mol-' anantration 0-lM.

3. A critical appraisal of the additivityprinciple

Table 3-5. Neutral Hydrolysis of 1 in Aqueous Solutions of Polyhydric Alcohols at 25C. Application of Quasi-SWAG Approach.

hcol glycerol erythritol arabitol sorbitol mannitol

-40 (6) -91 (6) -170 (10) -229 (14) -172 (14) -159 (14)

180 238 496 654 812 812

-44 -24
-4

16 36 36

aConcentration 0-2M. kxperimental value. CCalculatedvalue, using the SWAG approach; G(CH)=68 J kg rn01'~and G(OH)=226 J kg m01-~. dCalc~lated value, using the SWAG approach; G(CH)=-42 J kg mol" and G(OH)=62 J kg m01-~'

This effect is probably a result of reduction of the rate-decreasing effect of l-propanol. In the presence of 10 mol kg-' of urea, 1-propanol even causes an effective rate acceleration.
The SWAG-approach. According to equations 2-26 and 2-27, the kinetic data can also be analysed in terms of an extended painvise group additivity approach. A valid description of the interactions, involving the activated complex, is a matter of concern. Inspection of the mechanism of the hydrolysis reaction, as given in Scheme 3-1, shows that the solvation changes during the activation process are predominantly determined by interactions involving three polarised OH groups. Here, in a first approximation, the polarisation of the carbonyl group in the activated complex is neglected. Thus,

where the alcohol is designated by ROH. Therefore

In case of complete additivity, the solvent effects of alcohols should be solely deterand Go,mined by two pairwise Gibbs energy group interaction parameters, Go,-,,
CH.

Monohydric alcohols. The SWAG approach is only applied to the study of the solvent effects of monohydric alcohols on the neutral hydrolysis of 1 . The factor [L-g,J can be calculated, according to Equation 3-2, using the Gibbs energy interaction parameters given in Table 3-2. Table 3-6 compares the experimental and calculated value of G(C)-

3. A critical appraisal of the a d d i t i v i t y principle

Figure 3-3. Plots of -ln[k(m,mu)/k(mc=mu=O)] vs molality of 1-propanol for neutral hydrolysis of 1 at 25OC in the presence of different molalities of urea (mu); 0 , m =O; ,mu=1 mol kg-'; A ,mu=2 mol kg-'; v ,mu=3 mol kg-'; ,mu=4 mol kg-Y ; H ,mu=5 mol kg-'; A ,mu=10 mol kg-'.

3. A critical appraisal of the additivity principle

The trends are encouraging, especially in the light of the gross assumptions that had to be made. Better agreement is obtained when the pairwise group interaction parameters are adjusted in order to allow for the considerable polarisation and the increased hydrogen bonding interactions of the water molecules in the activated 23 J kg mor2 and complex. Satisfactory values of G(C) are obtained using GCH-OH= Go,-,,= -78 J kg m ~ l(see - ~ Table 3-6).

3.4 Towards a better understanding of medium efects in dilute aqueous solution o f monohydric and polyhydric alcohoLF

The results described in Section 3.3 emphasise the overwhelming importance of solvation and hydration shell overlap in determining kinetic medium effects in highly aqueous reaction media. The observed medium effects are primarily the result of a reordering of the solvation shells during the activation process. Thus, the hydration characteristics of the cosolvent are of paramount importance. As shown in Section 3.2, interactions between non-electrolytes have been classified according to their hydrophobic or hydrophilic behaviour. Monohydric alcohols. Monohydric alcohols are generally classified as hydrophobic molecules. Application of a quasi-additivity approach shows that kinetic medium effects of monohydric alcohols are dominated by a rate decreasing contribution of the apolar CH groups. The contribution is additive and only slightly dependent on the configuration of the alkyl group.

Table 3-6. Neutral Hydrolysis of 1 in Aqueous Solutions of Monohydric Alcohols at 25C. Application of SWAG Approach.

methanol ethanol 1-propanol 2-propanol 1-butanole 2-butanole 2-methyl-1-propanol 2-methyl-2-propanol cyclopentanol

27 (2) -120 (6) -258 (6) -231 (6) -474 (39) -405 (12) -425 (24) -392 (14) -379 (18)

-49.5 -130.5 -211.5 -211.5 -292.5 -292.5 -292.5 -292.5 -292.5

25 -113 -251 -251 -389 -389 -389 -389 -389

aConcentration 0-2M. '~x~erimental value. CCalculatedvalue, using the SWAG approach; G(CH2OH)=27 J kg rn01-~and G(0H-OH)=-24J kg rn01-~.d~alculated value, using the SWAG approach; G(CH2-OH)=46 J kg rn01.~ and G(0H-OH)=-78J kg m01.~. 'Concentration 0-1M. 56

3. A critical appraisal of the additivity princkle

An important reason for the additivity of CH interactions is that the hydration of alkyl groups is governed by water-water interactions that are mainly dependent on the volume of the hydrated hydrophobic group. Therefore, the interactions, involving apolar, hydrated moieties, are not strongly orientational dependent. The solvent effect of the apolar alkyl group is partly counteracted by the hydrophilic OH group. Generally, the contribution of OH groups to the overall solvent effect of monohydric, but also of polyhydric alcohols, can be considered the result of three contributions:
(i) (ii) (iii) Direct OH interactions with reactant or activated complex (which can be mediated by the solvent), the shielding of the apolar groups for interactions with the reacting species and the shielding of apolar groups for interactions with water.

When the number of hydroxy groups, present in the cosolvent, is reduced to zero, the hydrocarbon cosolvent is not soluble enough in water to be able to determine any rate effect. However, the effect of CH groups of alkanes is expected to be significantly larger than -68 J kg mo1-', found for monohydric alcohols. If, on the other hand, only OH groups are present, the effect of the cosolvent will be caused by direct interactions of the O H groups with the reacting species, displacing interactions with water. Consequently, the solvent effect will be very small. This fact has been recognised by Somsen and ~loemendal"", who advocated the application of the EGA concept. Unfortunately, the results obtained by using the quasi-EGA approach for the treatment of kinetic data are unsatisfactory. Most likely, this is due to the fact that the size and volume of the water molecule and the cosolvent molecule are quite different. Therefore, subtraction of two OH interactions from the overall number of OH groups of the cosolvent is quite arbitrary. Direct interaction between the OH group and the reacting species is a result of the difference between the interaction of the OH groups of water and the OH group(s) of the alcohol with the reactant and the activated complex. Therefore, the solvent effect of the OH groups, due to direct interactions, is small.
Dihydric alcohols. Except for glycol, most dihydric alcohols have been classified as relatively hydrophobic, structure-making compounds. However, introduction of a second hydroxy group gives rise to a remarkable differentiation between solvent effects of isomeric diols. Temperatures of maximum density, ultrasonic relaxation and microcalorimetric data show that the solvation behaviour of isomeric diols is substantially different183c12081209. Inspection of the data in Table 3-4 shows that also the solvent effects of l,2-butanediol, 1,Zbutanediol and 1,Cbutanediol are very different. Apparently, 1,2-butanediol exhibits the most hydrophobic behaviour. In terms of the quasigroup additivity approach, this is a consequence of two opposing effects.

(i)
(ii)

The hydrophobic contribution of the apolar CH groups decreases with the addition of a second OH group in close proximity to the first OH group. The counteracting contribution of the O H groups to the effect of apolar methylene groups, decreases with addition of a second OH group in close proximity to the first OH group.

3. A critical appraisal o f the a d d i t i v i t y princple

The contribution of (ii) is more pronounced. Consequently, the overall hydrophobicity of vicinal and, to less extent, of 1,3-dihydric alcohols is amplified as compared to the effect that has been anticipated, based on data obtained for monohydric alcohols. Introduction of a second OH group outside the direct hydration sphere of the first OH group, does not hamper the additivity approach. Apparently, the effects of two hydrophilic OH groups on the overall hydration characteristics of dihydric alcohols can be mutually obstructive. Consequently, the overall characteristics of the hydration shell and subsequent interactions, involving dihydric alcohols, cannot be treated by simple additivity approaches. This effect cannot be simply attributed to the occurence of intramolecular hydrogen bonding, becauce ch-1,2-cyclopentanediol induces a significantly larger rate effect than its trans analogue. The molecular origin of this proximity effect becomes clear, when the medium effects of polyhydric alcohols are studied.
Polyhydric alcohols. Polyols are intriguing compounds, which occur in living organisms ~~~*~~~. as protecting agents of proteins against extreme ~ o n d i t i o n s ' ~ Homotactic interactions between linear or cyclic polyols as well as between carbohydrates, which are classified as hydrophilic, structure-making solutes, are repuhive (myo-inositol is a remarkable exception). The strongly positive pairwise Gibbs energy interaction parameters are remarkable, and reflect the difficulties when attempts are made to crystallise carbohydrates from aqueous solution. This contrasts with the attractive homotactic interactions between monohydric and dihydric alcohols. Heterotactic interactions, involving polyhydric alcohols, are thermodynamically very complex. The results show that solvent effects of polyols are dominated by the rate retarding contribution of the CH groups. Furthermore, introduction of additional OH groups diminishes even more the applicability of additivity schemes. Glycerol, for example, appears to induce a larger decrease of the rate constant than 1,3-propanediol. More striking is the observation that erythritol decreases the rate constant to a similar extent as does l,Zbutanediol, and even more than 1,Cbutanediol. Although polyols are by no means hydrophobic, they nevertheless show a stronger preference for the reactant than for the activated complex. Most polyols enhance the solubility of apolar compounds in water. Generally, the interaction between apolar compounds and polyols is endothermic. Barone et al.lm explained this fact in terms of destructive overlap of hydrophilic and hydrophobic hydration shells. A tentative explanation of the favorable interaction of polyols with apolar compounds in terms of Gibbs energy, is based on the fact that the solvation of the polyols is governed by the hydroxy groups. The apolar CH groups do not necessarily need to be extensively hydrated by hydrophobic solvation shells and are therefore directly available for interactions with apolar cosolutes. In other words, the apolar groups are deshielded for interactions with water by the hydroxyl groups. The hydration characteristics of polyols are predominantly determined by the hydroxyl and polyols can be regarded as alkylated water. In contrast, the solvation of monohydric and dihydric alcohols is dominated by hydrophobic hydration. Therefore, the classification of monohydric and dihydric alcohols as soluble alkanes is more appropriate. The result is that the hydrocarbon groups of both apolar, hydrophobic monohydric and dihydric alcohols, as well as of polar, hydrophilic polyhydric alcohols, are available for hydrophobic interactions with surrounding cosolutes. The exact stereochemistry of polyhydric alcohols is extremely important. The

3. A &a1

appraisal o f the adddive princ@le

better the hydroxyl groups are positioned to fit into the water structure, the better the CH groups are shielded for interactions with water. This fact is elegantly demonstrated in a recent study by Galema et al.212214,who showed that the relative positions of the OH groups in carbohydrate molecules govern the solvent effects of this class of compounds on the hydrolysis of 1.

Aqueous solutions of urea and 1-propanol. Aqueous solutions of urea have been extensively studied in relation to the protein denaturating properties of these solutiBarone, ons. The exact mechanism of denaturation is not well understood257F92151216. Lilley and coworkers have examined enthalpies of interaction between urea and a range of other non-electrolytes (see Table 3-1). Little is known about the Gibbs energy of these interactions. Recent studies have shown that the interaction of urea with other non-electrolytes is due to cornpetion for water18%. The magnitude and the sign of the enthalpic and entropic interaction parameters depend critically on the enthalpic and entropic state of the water molecules in the solvation shells of urea and the cosolute. The effect of urea on pairwise interaction parameters involving nonelectrolytes, is not well documented in the literature. Following the formalism of the McMillan-Mayer theory, the second virial coefficient has to remain unchanged when an additional solute or cosolvent is introduced into the solution. The cluster expansion provides a quantitative measure for the effect of an additional cosolvent molecule R on the interaction of molecules P and Q by using the third virial coefficient g,,. So, the magnitude of gPQ does not respond to the presence of a second solute. If the addition of a second cosolvent does not affect the pair interactions in the solution and the triplet contribution to g is also negligible, the lines in Figure 3-3 would be parallel. Clearly, urea and ?Ipropanol mutually affect the pair interactions w i t h reactant and activated complex in the aqueous reaction medium. It is striking that the lines in Figure 3-3 intersect within a very narrow region of the plot. This phenomenon is forced by the fact that the slope of a plot of ln[k(mcl,m,)/ k(mcl=m,=O)] against mcl is zero or very small across a molality range of O<mcl<mcl* at a certain molality of C2, defined as mao. The physical significance of this plot is that the rate constant for the hydrolysis of 1 in the aqueous solution with m=m,' is apparently not affected when cosolvent C1 is added. If C1 and C2 are 1-propanol and urea, respectively, m , ; is ~ 7 . 5 m01.k~-'and mcl'-0.8 m01.k~-'.Now all , across a molality range of lines, describing 1n[k(m,,,mc2)/k(mcl=mc2=O)] against m O<mc,<mCl', intersect within a small region in the plot near m , ; . This means that at 0.8 mol.kg-' of 1-propanol, the addition of urea does not lead to a significant rate effect across a molality range of O< mu<7.5 m01.k~-'The slope of ln[k/k,,] against the molality of cosolvent can only be close to zero if (i) the cosolvent behaves ideally, or (ii) if the reactants and activated complex are equally solvated, or (iii) if the addition of the cosolvent disturbs the interaction of other cosolvents with the reacting species. Most likely, urea reduces hydrophobic interactions of 1-propanol with apolar solutes in the aqueous media and decreases the probability of finding molecules of 1-propanol in the immediate surroundings of other apolar solutes. The molecular origin of this effect is complex. It is generally believed that urea reduces the capacity of water to form hydrophobic hydration shells, but the contribution of this effect to the overall solvent effect of 1-propanol in aqueous solutions of urea is not clear yet.

3 . A &a1

appraisal of the additivity princ@le

3 . 5 Concluding remarks
Clearly, the present results show that solvent effects of added monohydric and polyhydric alcohols on the neutral hydrolysis of 1in aqueous solution are governed by the participation of the cosolvent molecules in the solvation shell of the reactant as well as of the activated complex. The participation involves overlap of the hydration shells of the cosolvent with hydration shells of the reacting species and the activated complex. The present analysis quantifies the change of the "solvation" of the reactant by the cosolvent during the activation process in terms of Gibbs energies of painvise interactions. All cosolvents studied exhibit a remarkable preference for interaction with the more apolar reactant, leading to a stabilisation with respect to the activated complex and a subsequent decrease of the reactivity. Quantitative analysis of the solvent effects shows, that solvent effects of both hydrophobic as well as hydrophilic cosolvents are dominated by the contribution of alkyl moieties in the cosolvent molecule. The magnitude of the solvent effect is governed by (i) the number of methylene moieties, (ii) the size of the molecule, and (iii) the number and position of the hydroxy groups. When the number and relative positions of the hydrophilic OH groups in the cosolvent, i.e the hydrophilic frame work, is kept constant, solvent effects can be analysed by applying additivity approaches. The magnitude of the contribution of the apolar CH groups to the overall solvent effect is modulated by the OH groups. The latter regulate the availability of the CH groups for interaction with apolar domains of the reactant and activated complex. Whether the pairwise hydrophobic interaction of the apolar CH groups with the reactant as well as with the activated complex occurs via direct London dispersion interactions or mediated by water molecules is a matter of speculation. Undeniably, a considerable number of water molecules have to be expelled from the respective hydrations shells. The structural reorganisation of the hydration shells during the interaction is accompanied by a large entropy and enthalpy effect. However, the relative importance of the contribution of hydration shell overlap in terms of Gibbs energy to the overall Gibbs energy of interaction is still a matter of debate (see Chapter 8).

3 . 6 Experimental section Materials. 1-Benzoyl-3-phenyl-l,2,4-triazole was prepared according to a standard procedure195 (mp 7 9 - 8 0 " ~ ; ~ i t 78.6-79.8"C). .l~~ Glycol and glycerol were distilled in vacuo before use. Erythritol was crystallised twice from ethanol. tram- and circyclopentane-1,2-diol were prepared from cyclopentene according to the methods of Owen and Smith2"(mm-~lopentane-1,2-dio1;mp 51C;Lit?" 50C, cir-cyclopentane1,Zdiol;mp 116-117C;Lit. l7 117C). 1,3-Propanediol was purified as follows: 1,3propanediol (1.0 mol) was added to acetone (1.3 mol) and benzene (50 ml). A few crystals of p-toluenesulfonic acid were added and the mixture was refluxed under removal of water. After 24 h, benzene and the excess of acetone were removed in vacuo, and the trimethylene acetal was purified by distillation; hp 125OC (72 mm). The acetal (10.4 g, 0.09 mol) was hydrolysed by addition of 10 ml of 0.5 N HCl. Acetone

3. A &a1

appraisal o f the additiviry pincple

and water were removed in vacuo, and 1,3-propanediol was distilled twice; the yield was 6.0g, 0.079 mol; bp 120C (15 mm) [lit.bp 108C (11 m m ) ~ ~ ~ ' . All other alcohols were used as supplied. Demineralised water was distilled twice in an all-quartz distillation unit. Urea (Merck) was used as supplied. A l l solutions were made up by weight and contained ca. 3x10' rnol kg-' HCI to suppress catalysis by hydroxide ions. The pH was measured using a Coming 130 pHmeter.
Product analysis. Reaction products for the solvolysis of 1 in alcohol-water mixtures were analysed qauntitatively. Reactions were performed in the presence of 0.50, 1.00 ~ all ~ cases . and 1.50 rnol kg4 of alcohol, with substrate concentrations of ca. 5 x 1 ~ In the pH was kept at ca. 4.5 to suppress hydroxide ion catalysis. After complete reaction, the products were analysed by gas-chromatography (Hewlett-Packard 5890 instrument, equipped with a 15 m wide-bore HP fused silica column) and by GC-MS (Ribermag R-10-10c instrument). The relative amounts of benzoic acid and the corresponding ester were determined. The analysis used known mixtures of authentic esters and benzoic acid for calibration purposes. Only in cases of alcohols containing primary OH-groups significant amounts of ester were found. Independent experiments showed that the esters were not formed by esterification of the benzoic acid after . For different alcohols (1.5 rnol kg"), the percentage of ester complete hydrolysis of 1 formation were as follows: methanol, 5+1%; ethanol 321%; 1-propanol, 2*1%; 2propanol, =I%; glycol, 321%; 1-butanol, ~ 1 % 2-butanol ; and 2-methyl-2-propanol, <0.4%. The yield of ester in the presence of 0-1.5 rnol kg-' of alcohol depends linearly on the concentration of the alcohol. Alcoholysis contributed in all cases less than 5% to the overall reaction rate.

Kinetic Measurements. Pseudo-first-order rate constants were determined by following the change of absorbance at 273 nm. About 3 pL of a stock solution of 1 in acetonitrile (=5x1U2 rnol dm-3) was added to the reaction medium (= 2.5 mL) in a quartz cell, placed in a th~rmostated cell compartment (25.O"C) of a Perkin-Elmer A5 spectrophotometer. The substrate is only slightly soluble in water, and care was taken that the absorption did not exceed 0.5. The reaction was followed for about 10 halflives, and excellent first-order kinetics were observed. Pseudo-first-order rate constants were calculated using a data station PE A5, connected to the spectrophotometer, and were reproducible to within 2%. Reaction rates at each molality of cosolvent were determined at least three times. For the calculation of G(CH), data measured for at least six molalities were used.

4. Solvent effects on some organic processes in aqueous solution

CHAPTER 4

A Survey of Solvent Effects on Some Orgdnic Processes in Dilute Aqueous Solution

Rate constants and derived activation parameters for chemical reactions involving solutes in a ueous solutions are sensitive to the presence of organic cosolvents in these media'788. In Chapter 2, a general method was described for a quantitative analysis of solvent effects in mixed solvents which can be used to analyse solvent effects on organic reactions in mixed aqueous solvents. This general theory was developed for solvent effects on rate constants and isobaric activation parameters of a simple first-order reaction. In subsequent sections, the scope of the theory was broadened and equations were derived for a quantitative analysis of solvent effects on bimolecular reactions, solvolysis reactions and equilibrium quotients of chemical equilibria. The dependence of rate constants and derived activation parameters on the concentration of chemically inert cosolvents can be understood in terms of interactions of these cosolvents with reactant(s) and activated complex. Pairwise and higher-order interaction parameters were shown to relate rate constants, equilibrium quotients and activation parameters to the molality of the added cosolvent. These parameters were subsequently analysed in terms of pairwise group interaction parameters. In chapter 3 it was shown how this additivity principle can be used to identify the magnitude and sign of group contributions to solvent effects of organic cosolvents on reactivity in aqueous s~lution'~'.Additivity schemes were put to a critical test by assessing the merits and limitations of additivity rules for the analysis of solvent effects on a hydrolysis reaction in dilute mixed aqueous solutions, containing mono- and polyhydric alcohols148.The results set the stage for a quantitative understanding of solvent effects of organic cosolvents on the rate constants for organic reactions in aqueous solutions. In this chapter, the applicability of the theory, which is based on the McMillan-Mayer solution theory, is tested for (i) the quantitative description of solvent effects on A*HO and TA*SO of reactions in dilute aqueous solutions, (ii) the description of solvent effects on rate constants and isobaric activation parameters of organic reactions in concentrated (>1.5 mol kg-') aqueous solutions in order to assess the relative importance of pairwise and higher-order interaction parameters and (iii) the quantitative analysis of solvent effects on a unimolecular process, a bimolecular reaction and a chemical equilibrium in aqueous solution. In Section 4.2, solvent effects are described on the pseudo-first-order rate constant, A*HO, and TA*SOfor the neutral hydrolysis of p-methoxyphenyl dichloroacetate in aqueous solutions, containing urea and alkyl-substituted ureas. Attempts are made to describe solvent effects in dilute and also in more concentrated aqueous solutions (0-6 mol k g ' ) in terms of pairwise and higher-order interaction parameters. This type of analysis has reasonable success in accounting for trends in pseudo-first-order rate

4. Solvent effectson some organic processes in aqueous solution

constants in these aqueous mixtures. The quantitative analysis of solvent effects on activation parameters is shown to be valid in dilute solutions (c1.5 M), but appears to be more troublesome for higher molalities of the cosolvents. In the second part of this chapter, the theory is tested with reference to the analysis of solvent effects on a intermolecular as well as an intramolecular Diels-Alder reaction in aqueous media. In Section 4.3, solvent effects are reported of a variety of cosolvent molecules on the second-order rate constant for the bimolecular Diels-Alder reaction of cyclopentadiene with methyl vinyl ketone in aqueous media. Solvent effects of monohydric alcohols on the first-order rate constant for the intramolecular Dielsacid in aqueous solutions are reported Alder reaction of N-furfuryl-N-methylmaleamic in Section 4.4. Finally, in Section 4.5 solvent effects are described on the equilibrium quotient, representing the keto*enol equilibrium of 2,Cpentanedione in aqueous solutions, also containing monohydric alcohols. The solvent effects on these processes are generally rather small. More pronounced solvent effects are found as the mole fraction of the cosolvents is increased. The theory, based on the McMillan-Mayer solution theory is, however, less appropriate for the analysis of solvent effects in binary solvents in which the components of the solvent are present in comparable amounts. The analysis of solvent effects on intermolecular and intramolecular DielsAlder reactions in mixed aqueous solvents, containing comparable amounts of water and organic cosolvent, is reported in Chapter 6 and Chapter 7, respectively. Some relevant aspects of intermolecular and intramolecular Diels-Alder reactions with respect to reactivity in aqueous media are discussed in these chapters as well.

4.2 Solvent effects on the hydrolysis of p-methoxyphenyl dichlomzcetate in aqueous solutions containing urea and &I-substituted ureas

In this section, solvent effects are reported of urea (U), N,N'-dimethylurea (DMU), N,N'-diethylurea (DEU) and tetramethylurea (TMU)on the pseudo-first-order rate constant and isobaric activation parameters for the neutral hydrolysis of p-methoxyphenyl dichloroacetate (1) in aqueous solution. The reaction mechanism for the neutral hydrolysis of 1 is shown in Scheme 4-2. The hydrolysis proceeds via a dipolar activated complex containing two water molecules with three protons in flight219-=. The reaction is surnmarised in Scheme 4-1, where X is the substrate ester and N represents the order of the reaction with respect to water.
X + N H20

* [X.NH20]* -. products

Scheme 4-1
Ester 1is hydrolysed in a reaction medium consisting of water, q and m, mol kg-' of reactant and activated complex, respectively, a small concentration of HCl in order to prevent base-catalysed hydrolysis and m, mol kg-' of the cosolvent. Solvent effects of non-electrolytes on the pseudo-first-order rate constant for neutral hydrolysis of 1 can be quantitatively analysed by using Equation 2-32.

4. Solvent effects on some organic processes i n aqueous solution

- TS
0 II CIZCHC-0C6HLOCH3 -p
+

2 H20

6 0

products

Scheme 4-2

In dilute aqueous solutions the solvent effect can be effectively described by assuming pairwise interactions between cosolvent and reacting species. Following the procedures descnied in Chapter 3 (Section 3.2), Equation 2-32 can be rewritten in the following form;

and

The factor G(C), which equals C(i)G(i), represents the difference in the pairwise Gibbs energy of interaction of the cosolvent molecule C with reactant and activated complex, respectively. According to the quasi-SWAG approach, the solvent effects of the cosolvent C can be considered as a sum of group contribution of groups i that form the cosolvent molecule, expressed as G(i). In Chapter 3 it was shown that plots of ln[k(mc)/k(mc=O)]versus molality of very hydrophobic molecules, such as 1-butanol exhibit clear curvature at molalities below 1.5 mol kg-'. Generally, pairwise interaction parameters are not sufficient to describe solvent effects on reactions in aqueous solutions if either the cosolvent is strongly hydrophobic or the cosolvent concentration is high. Since primary alcohols are reactive towards the substrate that was studied in Chapter 3, a detailed analysis of solvent effects of these cosolvents at higher cosolvent concentration was bound to be inconclusive with respect to the importance of higher-order interaction parameters. Urea and alkyl-substituted ureas do not react with activated ester 1and solvent effects have been studied in aqueous solutions, containing high concentrations of cosolvent. A quantitative description of solvent effects on the rate constant for a hydrolysis reaction in concentrated aqueous solutions is given in terms of Equation 4-3. The contribution of higher-order terms is taken into account and the n-th order contributions are symbolised by Gn(C) which equals ZGn(i) (for pairwise terms the superscript 2 is omitted). The difference, expressed in triplet Gibbs energy interaction parameters, between the interaction of the activated complex with two molecules of cosolvent and the interaction of the reactant with two molecules of cosolvent is, for example,

4. Solvent fleets on some organic processes in aqueous solution

measured by G3(C). G~(c) also contains the effect of cosolvent on the pairwise interaction of cosolvent with the reactant and activated complex (see Section 3.4).

The reactivity of water is also affected by the addition of the cosolvent molecules. The magnitude of this contribution is expressed by the term NM,4m,. The dependence of 4 on the molality of urea and some alkyl-substituted ureas is accurately knownlg2". In Figure 4-1, the solvent effect of urea and the substituted ureas, expressed as NMl~mc, is plotted as a function of the molality of the cosolvents, assuming that G(C) and higher-order terms are negligibly small. All cosolvents induce a significant but similar decrease in the reaction rate constant. The dependences of NMl+mc on the molality mc are characteristic of each cosolvent and the solvent effects are significantly different. The differences are, however, negligible for the analysis of solvent effects. The isobaric enthalpy and entropy of activation for reactions in mixed solvents, containing a small concentration of cosolvent molecules, depend on the molality of the cosolvent. Assuming that the dependence of the osmotic coefficient on the temperature does not contribute significantly to the overaIl change of the activation parameters, Equation 2-53 can be rewritten to yield Equation 4-4.
2 ~ ' H " ( m , = 0 ;id) - A'HO(mJ= -&-h,J
"'0

mc

4-4

The formalism, used for the description of solvent effects on the Gibbs energy of activation for reactions in mixed aqueous solvents, is also applied to the description of solvent effects on the enthalpy of activation. In addition, higher-order interaction terms are introduced.

Pairwise entropy interaction parameters can be derived using the equation gpQ=hpQTSPQResults and discussion. In Figure 4-2, the effects of urea and the substituted ureas on the rate constants, expressed as ln[k(m,)/k(mc=O)], are plotted as a function of the molality of the cosolvents. Urea induces a small but significant acceleration of the reaction, whereas alkyl-substituted ureas decrease the rate constant. Comparison of the slopes in Figures 4-1 and 4-2 confirms that the trend in the rate constants cannot be accounted for on the basis of the reduced reactivity of water. The predominant contribution to the solvent effect must be attributed to pairwise and higher-order interactions of the cosolvent with the reacting substrate. The significant curvature of all plots indicates that higher-order Gibbs energy interaction parameters must be taken into account. The curvature is particularly pronounced for the most hydrophobic cosolvents (TMU and DEU). The experimental results were analysed by fitting the data points to a polynomial.

4. Solvent effects on some organic processes h aqueous solution

The data could be described accurately (h0.997) by assuming pairwise and triplet interactions. The solvent effects at molalities ~1.5mol kg" were also analysed by linear regression. In both cases the functions were forced to cross (0,O). The fitted curves and lines are shown in Figure 4-2. The pairwise and higher-order Gibbs energy interaction terms G(C) and Gn(C)are summarised in Table 4-1. In Figure 4-3 TA'SO and A'HO for the neutral hydrolysis of 1 at 2SC are plotted as a function of the molality of the alkyl-substituted ureas. In aqueous solutions of urea the changes of the activation parameters did not exceed the experimental error. The dependence of entropy and enthalpy of activation on the molality of the cosolvents exhibits well-known compensating behaviou$'@. The decrease in rate constant for the hydrolysis is determined by a dominant entropy term. Both entropies and enthalpies pass through extrema. For DMU, DEU and TMU these molalities are ca. 4, ca. 3 and ca. 2 mol kg-', respectively.

Figure 41. Kinetic analysis of the neutral hydrolysis of p-methoxyphenyl dichloroacetate in aqueous solutions. Dependence of -NM,$m, (N=2) on the molality of urea, ;DMU, 4 ;DEU, A .

4. Solvent effects on some organic processes in aqueous solution

Figure 4-2. Kinetics of the neutral hydrolyis of p-methoxyphenyl dichloroacetate in aqueous solutions containing urea and alkyl-substituted ureas at 2SC. Dependence of ln{k(mc)/k(mc=O)] on molality of urea, ;DMU, A ;DEU, ;TMU, H . Solid lines are curves, calculated by non-linear regression. Dashed lines are calculated by linear regression of data below 1.5 M (see text).

The dependence of the enthalpy of activation on the molality of the cosolvents cannot be described accurately by the theoretical model. Even higher-order interaction terms do not account for the observed trends. A satisfactory description can be obtained for solvent effects on the activation parameters at molalities below the extrema. The experimental errors in the determined activation parameters are much more pronounced than for the rate constants, and a detailed quantitative analysis is not significant. Nonetheless, the experimental data were analysed by linear regression and the results are summarised in Table 4-1. The solvent effects of urea and alkyl-substituted ureas are the result of participation of the cosolvent molecule in the solvation shell of reactant and activated complex, respectively.

4. Solvent efJects on some organic processes in aqueous solution

Figure 4-3. Solvent effects on isobaric activation parameters for the neutral hydrolysis of p-methoxyphenyl dichloroacetate in aqueous solutions at 25OC. Dependence of A*HO (open symbols) and -TA*SO (closed symbols) on the molality of DMU, A ;DEU, and TMU, . The values for A*HOhave been placed upwards 10 kJ mol* for clarity.
Interaction of the reactant and activated complex with the cosolvent is accompanied by the destructive overlap of the hydration shells of both solutes. Solute-solute interactions involving urea and alkyl-substituted ureas have been frequently studied (see Table 3-1). Pairwise enthalpy and Gibbs energy interaction parameters for homotactic interactions have been determined accurately and appear to be attractive (see, for example ref.l8lc,d,e,v). For DMU, DEU and TMU G , is negative and H ,is ,are negative. Also heterotactic interactions, involving positive. For U, both G, and H urea have been studied in great detailm.

4. Solvent fleets on some organic processes in aqueous solution

Table 4-1. Solvent Effects on Isobaric Activation Parameters for the Neutral Hydroly-

sis of p-Methoxyphenyl Dichloroacetate in Aqueous Solutions at 25OC. Cosolvent Urea DMU DEU TMU G(C)? J kg m01-~ 98 (2) -353 (15) -642 (22) -921 (22) G(C),b G3(C),b J kg m01-~ J kg m01'~ 102 -360 -650 -967
-4

H(C): TS(C)P J kg m01-~ J kg mor2

C C

18 46 43

507 (96) 1579 (91) 3605(410)

860 (111) 2221 (113) 4526(432)

aDetermined by linear regression. b~etermined by non-linear regression. 'Determined by linear from G(C) and H(C). CVariation of activation parameten did not regression below 1.5 M. *~alculated exceed the experimental error.

Pairwise enthalpic interaction parameters vary in magnitude and sign, depending rone et~a1.1g2' have used this . Ba on the hydrophobic nature of the other s o l ~ t e ' ~ ~~ *'~ ~ fact to determine the relative hydrophobicity of polyhydric alcohols. Unfortunately, only a few Gibbs energy interaction parameters are known. In the presence of 7 M of urea the solubility of methane and ethane in water is decreased, whereas the solubility of propane and butane is increased224.The effects are, even at these high concentrations of urea, very small. In a famous paper, Roseman and ~ e n c k showed s~ that urea increases the solubility of naphthalene and uric acid in water considerably. Wood et a1.156f9hhave estimated that the pairwise group interaction parameter G(CH?CONH) amounts to 1 J kg m ~ l -which ~ , is an extremely small value. Even less expenmental data are available for heterotactic interactions, involving alkyl-substituted ureas. In the light of the relevance of amide-amide interactions for the structure of peptides and proteins, interactions of the alkyl-substituted ureas with other amides Interactions between two amide functions have received some are intriguing. Although the pairwise Gibbs energy interaction parameter, describing the homotactic CONH-CONH interaction, indicates an extremely strong attractive binding (see Table 3-2), only weak association has been detected between amides in dilute aqueous s o l ~ t i o n ~ -Experimental ~'. evidence suggests the formation of solvent separated pairs, in which water mediates the hydrogen bonding interactions between the amide Inspection of the results in Table 4-1 reveals that the magnitude of solvent effects of the alkyl-substituted ureas on the rate constant for the hydrolysis of 1 are high in comparison to the solvent effects of monohydric alcohols on the rate constant for (see Table 3-3). Recently, Kerstholt et hydrolysis of 1-benzoyl-3-phenyl-l,2,4-triazole a1.2289U9 showed that solvent effects of monohydric alcohols on the rate of hydrolysis of 1are also significantly larger than those on the hydrolysis of the activated amide. The trend in the G(C) was shown to be similar and the solvent effects clearly reflect the loss of hydrophobicity of the substrates during the activation process. Apparently, the activated ester looses more hydrophobic character during the activation process than the activated amide, studied in Chapter 3. The third-order term G3(C) is quite small and cannot be determined accurately. The calculated triplet terms indicate that the

4. Solvent flects on some organic processes in aqueous solution

importance of higher-order interactions increases with increasing hydrophobicity of the cosolvent, as anticipated. According to literature procedures, a molecule of urea is equivalent to 1.5 CONH g r o ~ ~ sA ' ~ molecule ~ ~ ~ . of DMU is written, for the present purposes, as 2 CH,+ lCONH=BCH+CONH. DEU is described as 2CH,+ 2CH2+1CONH= 10CH+ CONH. Similarly, TMU can be represented by 12CH+CONH. This representation of TMU assumes that the N-CO-N grouping is equivalent to one CONH group. These are unhappy assumptions. Nonetheless, they have been frequently and successfully used for the analysis of thermodynamic data in aqueous solution (see references in Table 31). Following the literature procedures for the subdivision of urea and substituted ureas into groups, estimates of G(CH) and G(C0NH) cannot be determined unambiguously. A better result can be obtained by comparing the G(C) values of U, DMU and TMU. Substitution of a hydrogen by a CH, group leads to a decrease in G(C) of ca.-240 J kg m01-~.Comparison of the G(C) values of DMU and DEU shows that increasing the size of the alkyl substituent with a CH, group leads to an additional decrease of ca.-150 J kg mor2. These exercises give an idea of the contributions of the various groups of the cosolvent molecule to the overall medium effect. But the data set is too small to make a reliable estimate of the exact magnitude of the group contributions. Generally, apolar CH groups induce a rate decreasing effect, whereas polar CONH groups increase the rate constant. Solvent effects of urea on rate constants as well as on activation parameters for hydrolysis reactions are small. Apparently, the reactant and the activated complex are indifferent to the presence of urea and the distribution of urea near reactant or activated complex is not dramatically different from the distribution in the bulk. The kinetic solvent effects of the alkyl-substituted ureas are much more pronounced and dominated by hydrophobic interactions. In fact, the contributions of pairwise interactions to the solvent effect have to be derived from the limiting slope of plots of ln[k&,] versus m,. Apparently, pairwise interactions dominate the solvent effect on rate constants over an extended concentration range. The contribution of triplet interactions at low concentrations of cosolvent are surprisingly small. This is revealed by the fact that the pairwise interaction term can be accurately determined by linear regression of the kinetic data, obtained at small concentration (<1.5 mol kg-') of cosolvent. Most striking is the observation that the complex dependence of the activation parameters on the molality of the alkyl-substituted ureas is not reflected in the Gibbs energy of activation. This is caused by the fact that the large and compensating changes in the enthalpy and entropy of activation are a consequence of the destructive overlap of the h drophobic hydration shells, surrounding cosolvents, reactant and activated c 0 m ~ l e x 8 Prerequisite ~~. for these large entropy and enthalpy effects is that apolar solutes are fully solvated by a structured hydration shell. At a certain molality of the cosolvent, dependent on the size and apolar character of the solute, the cosolvent molecules start to compete for the water molecules to form a hydrophobic hydration shell. At this concentration, all water molecules are part of a large, extended hydrophobic hydration shell and water looses its typically aqueous character. Beyond this concentration, destructive overlap of hydration shells does not lead to large entropy and enthalpy changes. Although the solubility of apolar solutes increases due to the addition of apolar cosolvents, water gradually looses the capacity to

4. Solvent effects on some organic processes in aqueous solution

prevent the formation of microheterogeneties in the solution. At these cosolvent concentrations, bulk hydrophobic interactions lead to preferential solvation of the solute in apolar microdomains. Since the factors, determining the dependence of entropy and enthalpy of activation on the molality of the cosolvent are altered beyond what is sometimes refered to as the magic mole fraction, the quantitative analysis of solvent effects on entropy and enthalpy of activation can only be applied below this mole fraction. These microdomains are, of course, very short-lived and the contact of the solutes with water is still extensive. Conclusion. The solvent effects of the alkyl-substituted ureas on the rate constant for the neutral hydrolysis of 1 are the result of changes in the Gibbs energy of pairwise and higher-order interactions of the cosolvents with the reacting species during the activation process. Pairwise and higher-order interactions of cosolvents with reactants and activated complex are accompanied by destructive overlap of the hydration shells. The dependence of the isobaric activation parameters A'HO and TA'SO on the molality of the cosolvent in dilute aqueous solutions (below the "magic mole fraction") is strongly determined by the destructive overlap of the hydration shells. Since the entropy and enthalpy changes, resulting from the destructive overlap of the hydration shells; are strongly compensating, the importance of destructive hydration shell overlap for the solvent effect, which is a Gibbs energy effect, is difficult to assess. On the other hand, the occurence of direct solute-cosolvent interactions in the dilute aqueous solutions cannot be unambiguously confirmed either. In more concentrated aqueous solutions, the importance of preferential solvation is beyond doubt (see for example Chapters 6 and 7). Although the interactions between the reactants and the activated complex on one hand and the cosolvent on the other hand, are clearly hydrophobic in nature, the molecular origin of the favourable Gibbs energy term is difficult to establish. In Chapter 8, the molecular origin of pairwise and bulk hydrophobic interactions in dilute and more concentrated aqueous solutions will be discussed in more detail.

4.3 Solvent effects on a bimolecular Diels-Alder reaction

Since its discovery in 192sU0,the Diels-Alder reaction has been a valuable method for the synthesis of polycyclic products and the generation of other important synthons in organic chemistry. For organic chemists, Diels-Alder reactions provide an opportunity for regioselective and stereospecific introduction of multiple centers of configuration. The mechanism of Diels-Alder reactions generally involves an isopolar activated complex. The activated complex strongly resembles the reactants and small changes in the polarisability of the reactant(s) during the activation process result in corresponding small changes in solute-solvent dispersion interactions. Consequently, solvent effects on Diels-Alder reactions are expected to be small and not very interesting. Detailed kinetic investigations of Diels-Alder reactions in large series of solvents are rare. Generally, rate constants for Diels-Alder reactions upon going from nonpolar to polar solvents vary by a factor between 3 and 1 5 ~ ~ The ~ . last decade, however, witnessed a renewed interest in the study of solvent effects on Diels-Alder reactions"a-

4. Solvent effects on some organic processes in aqueous solution

reactions were shown to undergo dramatic rate enhancements in aqueous solution. The molecular origin of these rate enhancements is not clear. In Chapter 6, solvent effects on Diels-Alder reactions in binary alcoholwater mixtures will be discussed in detail. Mechanistic aspects are presented which can explain the curious rate effects observed in water and aqueous media. In this section, solvent effects are presented of a series of cosolvents on the second-order rate constant for the Diels-Alder reactions of methyl vinyl ketone (MVK)with cyclopentadiene (CPD) (Scheme 4-3) in dilute aqueous solutions.

d,l&c,l%,b,zOasPppJ7,28

. Diels-Alder

Scheme 4-3
In Chapter 2, it was shown that solvent effects on bimolecular reactions in mixed solvents can be analysed using Equation 2-44. Following this approach, solvent effects on a bimolecular Diels-Alder reaction in mixed aqueous solutions are described by Equation 4-5.

Herein G(C) is defined as [gAc+g,,-g,c]/m,2 were A and B are the diene and the dienophile, symbol t represents the activated complex and C is the cosolvent. Therefore, G(C) quantifies the difference between the pairwise Gibbs energy interaction parameters for interactions of the cosolvent with the reactants and with the activated complex, respectively. As explained in Section 2.3.3, the unit in which a rate constant is expressed, is a matter of concern. In Equation 4-5, W and W', and V and V' are the masses of water present in the reaction mixture, and the volumes of the total reaction mixture, respectively. The rate constants are expressed in dm3 mol-' s-l. Results and discussion. In Figure 4-4, the solvent effects of a series of cosolvents on the second-order rate constant of the Diels-Alder reactions of MVK and CPD, expressed as ln[{k(m,).V/W}/{k(mc=O).V'/w'}], are given as a function of the molality of the cosolvents. For clarity, the data-points for the solvent effects of methanol, ethanol and 1-propanol are not shown in the plot. All plots are perfectly linear and the solvent effects can be accounted for by assuming dominant painvise interactions between the cosolvent and the reactants and activated complex, respectively. In Table 4-2 the derived values of G(C) are given. Apparently, all monohydric alcohols induce a similar and small decrease in the rate constant. Urea hardly affects the rate constant, whereas polyhydric alcohols and carbohydrates produce a significant increase in the second-order rate constant. 1,4-Dioxane induces the most pronounced ratedecreasing effect.

4. Solvent effects on some organic processes in aqueous solution

m,, moi kg-'

Figure 4-4. Kinetics of the Diels-Alder reaction of methyl vinyl ketone with cyclopentadiene in aqueous solutions at 25C. Dependence of In[k(mc)/k(mc=O)]on the molality of glucose, ;galactose, A ;glycerol, ;urea, ;2-methyl-2-propanol, and l,Cdioxane, A

At first sight, the increase of the second-order rate constant due to the presence of carbohydrates might seem surprising. However, the rate increase can be related to changes in the volume of the reaction medium due to addition of cosolvents. The dependence of the total volume of the reaction mixture on the molality of solutes is rather complex and depends strongly on the nature of the solute231. Addition of carbohydrates to water leads to a considerable decrease in the volume of the total reaction medium whereas addition of monohydric alcohols leads to an increase in the volume of the reaction medium. Quantitative analysis of the solvent effect is based on the conversion of reactants per second in one kg of water, present in the total reaction medium (k,V/W). The partial molar volume of water in the mixture is therefore of paramount importance. The positive G(C) values must be largely attributed to the effect of solutes on the partial molar volume of water.

4. Solvent effects on some organic processes in aqueous solution

Table 4-2. Solvent Effects on the Diels-Alder Reaction of Methyl Vinyl Ketone with Cyclopentadiene in Aqueous Solutions at 25C.

Cosolvent Methanol Ethanol 1-Propanol 2-Methyl-2-propanol Urea 1,CDioxane Glycerol Glucose Galactose

G(C), J kg rn01-~

Based on these results, it is difficult to assess whether the hydrophobicities of the reactants and the activated complex are more or less similar, which is what one expects for reactions involving an isopolar activated complex, or whether the polarity is changed during the activation process. In Chapter 6, solvent effects will be reviewed over a more extended concentration of monohydric alcohols. Based on these results, an explanation will be given for the small and even rate increasing solvent effect of rnonohydric and polyhydric alcohols.

4.4 Solvent effeects on an intramolecular DieIs-Alder reaction

The first exam les of intramolecular Diels-Alder reactions stem from publications of Alder in 1 9 5 3 4 Subsequently, it was shown that intramolecular Diels-Alder reactions are very useful in the synthesis of polycyclic compounds234.Chapter 7 will be completely devoted to the results of a study of intramolecular Diels-Alder reactions of Nalkyl-N-furfu~ylmaleamic acids (2) in mixed aqueous solvents. In Chapter 7, general aspects of structure and reactivity of compounds, able to undergo intramolecular Diels-Alder reactions will be reviewed. In addition, mechanistic aspects of the intramolecular Diels-Alder reactions of compounds 2 will be discussed in detail. Solvent effects on intramolecular Diels-Alder reactions are usually very small and resemble solvent effects on their intermolecular analogues232.Also for intramolecular Diels-Alder reactions, small changes in the dipole-moment of the reactant during the activation process are the main cause for solvent effects235.Detailed studies of solvent effects on intramolecular Diels-Alder reactions are rare. Most attention has been placed on the re 'oselectivity, the stereochemistry and the relation between structure and reactiviegb? Regioselectivity and stereospecitkity can be very different for intramolecular Diels-Alder processes and their intermolecular analogues. Our interest in intramolecular Diels-Alder reactions is prompted by the question whether intramolecular Diels-Alder reactions experience rate enhancements similar to those observed for their intermolecular analogues in water and in aqueous solutions (see Chapter 6).

4. Solvent effects on some organic processes in aqueous solution

In this section, solvent effects are reported on the first-order rate constant for the intramolecular Diels-Alder reaction of N-methyl-N-furfurylmaleamic acid in aqueous solutions, containing monohydric alcohols (Scheme 4-4).

Scheme 4-4
Solvent effects on the first-order rate constant for the intramolecular Diels-Alder reaction of 2 in aqueous solution can be analysed using Equation 2-22. Following the procedures advanced in Chapter 3, the factor [g,,-g,,]/m? is defined as G(C). Group contributions of groups i, that form the cosolvent molecule, to the overall solvent effect are expressed in G(i), and the solvent effects can be calculated by Equation 4-6.

Results and discussion. In Figure 4-5, solvent effects on the first-order rate constant for the intramolecular Diels-Alder reaction of 2, expressed as ln[k(m,)/k(m,=O)], are plotted as a function of the molality of a series of monohydric alcohols. All plots are linear. The solvent effects can be effectively described by pairwise interactions of the cosolvents with the reactant and the activated complex, respectively. In Table 4-3, derived values of G(C) are collected. All cosolvents induce a decrease in the rate constant. This trend is due to the rate decreasing contribution of the apolar CH groups, as revealed by a negative G(CH) (-22 J kg mo1-'). The polar OH groups induce an increase in the reaction rate constant (G(OH)= 24 J kg m01'~).The solvent effects are rather modest. Nonetheless, the activated complex appears to be significantly less hydrophobic than the reactant.

4.5 Solvent eflects on the k e t o ~ = ~ equilibrium ~ol

In general, 8-diketones, both pure and in solution, exist as a mixture of three forms: (i) the ck-enolic form, (ii) the diketo form and (iii) the trans-enolic form. For 2,4pentanedione (PD) these forms are shown in Scheme 4-5. The mans-enolic form is only observed in special cases237and generally the ketoeenol equilibrium constant K, can be defined as

4. Solvent efJec~ on some organic processes in aqueous solution

Figure 4-5. Kinetics of the intramolecular Diels-Alder reaction of N-methyl-N-furfurylmaleamic acid in mixed aqueous solutions at 25OC. Dependence of -ln[k(m,)/k(m,=O)] on molality of methanol, ;ethanol, ;l-propanol, A and 2-methyl-2-propanol, w .Solid lines represent the line derived from the experimental data (see text).

Table 4-3. Solvent Effects on the IntramoIecular Diels-Alder Reaction of N-Methyl-NFurfurvlmaleamic Acid in Aaueous Solutions at 2SC.

Cosolvent Methanol Ethanol 1-Propanol 2-Methyl-2-propanol

G(C),B J kg m01-~ -45 (4) -78 (5) -124 (5) -174 (6)

G(CITb J kg m01-~ -42 (11) -86 (13) -130 (15) -174 (17)

aDerived from experimental results. b~alculated, assuming G(CH)=-22 J kg molq2 and G(OH)=24 J kg mor2.

4. Solvent fleets on some organic processes in aqueous solution

K, = [cis-enol]
[htol

ii

iii

Scheme 4-5
Ketopenol equilibria are very sensitive to the nature of the s ~ l v e n t ~ ~As ~ - early ~~'. as 1896, Claisens described that "there are compounds capable of existence in the form -C(OH)=CH-CO- as well as in the form -CO-CH2-CO-; it depends on the nature of the substituents, the temperature and, for dissolved compounds, also on the nature of the solvent which of the two forms will be more stable". Later, Wislicenus6 concluded, based on a study of the ketoeenol tautomerism in eight solvents, that the keto form predominates in alcoholic solutions, the en01 form in chloroform and benzene. Based on these results, ~tobbe"' classified solvents into two groups, which reflect a division into protic and aprotic solvents. Since these initial reports, solvent effects have been studied on a large variety of compounds that can exist in two tautomeric forms3'. Ketolenol ratios in the gas-phase were shown to approach the values found in apolar, aprotic solvents23s. The cis-enolic form is favoured by solvents of low polarity, whereas polar solvents displace the ratio towards the diketo form. Ketolenol ratios in large series of solvents have been frequently correlated with solvent polarity scales, physical properties of the solvent and solubility data of the en01 and the diketone. ~ l m s l e y showed ~ ~ ~ that ketolenol ratios can be roughly correlated with the polarity of the solvent. The best correlation was found for A+B (a scale advocated by Swain242,which expresses the basicity and the acidity of the solvent). Correlations with ET(30), E and .rr' are satisfactory. The internal hydrogen-bonding of the en01 tautomer, and in particular that of the 2,4-pentanedione, has fascinated chemists for many years244. In order to explain observed solvent effects on the ketoenol equilibrium, direct hydrogen-bonding between the en01 and the solvent has often been suggested to compete with the intramolecular hydrogen-bonding in the cis-enolic form238.Recently, it was shown that the intramolecular hydrogen-bonding of the enolic form remains intact in both aprotic and protic solvents442.There is no good evidence that the solvents favour either side of the equilibrium by hydrogen-bonding more strongly with either the en01 or keto tautomer. Consequently, solvent effects have been rationalised by the fact that the diketo form is usually more dipolar than the cisenolic form. This conclusion might seem somewhat surprising, but intramolecular hydrogen-bonding helps to reduce the

4. Solvent effectc on some organic processes in aqueous solution

dipole-dipole repulsion of the carbonyl groups, which is quite unfavourable in the diketo form. Since the initial work of Reevesa5, 'H-NMR analysis has been the prefered method for analysing ketopenol equilibria. In some cases, where the life-time of one of the species makes detection by NMR-spectroscopic methods impossible, IRspectroscopy must be A major problem in the determination of ketooenol equilibrium constants is the fact that the concentration-quotient (which is the ratio of the concentrations of the enolic form and the keto form) depends on the concentratiThe ~ ~ effect ~. of dilution on the ketopenol equilibrium has on of the f l - d i l c e t ~ n e ~ but has not been reported in detail until 1987. Elmsley et al.a2 been re~ognised'~~, determined the dependence of the ketoeenol equilibrium quotient of PD on the mole fraction of PD in different solvents. The authors showed that the equilibrium constants should be determined by extrapolation to infinite dilution. The dependence of the equilibrium quotient on the mole fraction of PD appeared to be sensitive to the solvent, both in magnitude as well as in sign. Especially in water, the dependence on the mole fraction of PD is very large, hence accounting for the fact that reported equilibrium constants for the keto-en01 equilibrium of PD in water show a large scattePya2. The problem arising from the difference between the equilibrium constant and the equilibrium concentration-quotient has been anticipated in Section 2.3.2. Solvent effects on chemical equilibria in mixed solvents are explained in terms of the solvent effect on the equilibrium quotient. Equation 2-37 can be used to describe the solvent effect on the ketoeenol equilibrium quotient of PD in aqueous solutions. Again, [gAcg,c]/m: is defined as G(C). The en01 is represented by B, and the di-ketone by A. Following the procedures, described in Section 4.4, the contribution of groups i to the overall solvent effect, expressed in G(i), can be derived:

G(C) and G(i) are defined as the difference in the pairwise Gibbs energy of interaction of cosolvent molecule C or groups i with the diketone and the enol, respectively. Usually, solvent effects on keto-en01 equilibria have been analysed in terms of macroscopic properties of the solvent, like dielectric constant or polarity238. This type of analysis stresses the importance of specific intermolecular interactions for the solvent effects on ketopenol equilibria in aqueous solution. reported that the dependence of the Results and discussion. Elmsley et a ~ . " ~ percentage of en01 in water on the mole fraction of PD is +267a2 (the slope of the versus the molality of PD is ca. 0.3, see below). The plot of in[Q(mpD)/Q(mpD=O)] equilibrium quotient for the keto-no1 equilibrium of PD in aqueous solutions is defined as [enol]/[keto] and was determined by 'H-NMR at a mole fraction of PD of 5x10'. This solution can be considered as thermodynamically ideal and the equilibrium quotient, determined in this dilute aqueous solution, can be considered as the equilibrium constant. At 25"C, the measured equilibrium constant was 0.193+0.008 (Literature data: 0.148242,0 . 1 9 ~ and ~~ 0.23'~~).Addition of monohydric alcohols affects the equilibrium quotient.

4. Sohent qects on some organic processes in aqueous solution

Figure 4-6. Ketoenol equilibrium of 2,Cpentanedione in aqueous solutions at 25OC. Dependence of ln[Q(m,)/Q(mc=O)] on the molality of methanol, ;ethanol, ;lpropanol, A and 2-methyl-2-propanol, .

In Figure 4-6, the solvent effects of the monohydric alcohols on the equilibrium quotient of PD, expressed as ln[Q(m3/Q(mc=O)], are plotted as a function of the molality of the alcohols. AIl plots are linear. The solvent effects are small, and addition of apolar cosolvents favours the enol. In Table 4-4, derived values of G(C) are given. It is found that G(CH)=37 J kg mor2 and G(OH)=-108 J kg mol". Interestingly, the dependence of the equilibrium quotient on the mole fraction of PD itself can also be interpreted as a cosolvent effect. Frequently, the dependence of the equilibrium quotient on the concentration of tautomeric compounds has been interpreted in terms of association or stackinP2. However, it seems more likely that the dependence of the equilibrium quotient on the concentration of the tautomeric compound is a consequence of pairwise and higher-order homotactic interactions in the aqueous medium.

4. Solvent effects on some organic processes in aqueous solution

Table 4-4. Solvent Effects on the Equilibrium Quotient ([enol]/[keto]) of the Keto+no1 Equilibrium of 2,CPentanedione in Mixed Aqueous Solutions at 2SC. Cosolvent Methanol Ethanol I-Propanol 2-Methyl-2-propanol
rnolm2.

G(C),B J kg mor2

G(C),b

J kg m01-~
3 (16) 77 (20) 151 (23) 225 (26)

8 (4) 69 (8) 156 (6) 228 (14)

'Derived from experimental data. b~alculated, assuming G(CH)= 37 J kg m01-~and G(OH)=-108 J kg

From the data of ~ l m s l e p who ~ , did not consider a logarithmic dependence of Q on the mole fraction, G(C) can be estimated as 315 J kg mor2, which is in reasonable agreement with the solvent effects of the monohydric alcohols. In summary, solvent effects on the ketoenol equilibrium of 2,4-pentanedione in aqueous solutions can be accurately described assuming pairwise interactions of the added cosolvents with the diketone and cis-enol, respectively. Apparently, painvise interactions of the monohydric alcohols as well as of both tautomers of PD with the cis-en01 are more favourable than with the diketone, indicating that the cis-en01 is the more hydrophobic isomer.

4.6 Experimental section

Materials. p-Methoxyphenyl dichloroacetate was prepared according to standard procedures248(mp 62-63OC, 1it.~~~62.2-62.80C). Methyl vinyl ketone was freshly distilled before use. Cyclopentadiene was prepared from its dimer immediately before use. Nmethyl-N-furfurylmaleamic acid was synthesised as described in Chapter 7. 2,4Pentanedione was purified by distillation. Urea (Baker) was used as supplied. N,N'-dimethylurea and N,N'-diethylurea were purified by rec~ystallisationfrom ethanol and petroleum ether, respectively. Tetramethylurea was distilled in vacuo. All monohydric alcohols and 1,Cdioxane were of the highest purity available and used without further purification. Demineralised water was distilled twice in an all-quartz distillation unit. All solutions were made up by weight. pH-measurements were performed by using a Coming 130 pH-meter. Kinetic measurements. Pseudo-first-order rate constants for the hydrolysis of pmethoxyphenyl dichloroacetate were determined by following the change in absorbance at 288 nm (formation of p-methoxyphenol) as a function of time. About 8x10~~ cm3 of a stock solution of the substrate (about 5x10-~ mol dm" in acetonitrile) was added to a reaction medium in quartz cells placed in the thermostated cell compart-

4. Solvent @ecB on some organic processes in aqueous solution

ment of a Perkin Elmer A5 spectrophotometer. The reaction medium contained mol dm" of HCI to suppress catalysis of the hydrolysis by hydroxide ions. The reaction was followed for about 10 half-lives. Excellent first-order kinetics were observed. Rate constants were calculated using a data station PEA5 connected to the spectrophotometer and were reproducible to within 2%. Rate constants were measured at at least five different temperatures. Plots of l n [ m vs 1 / T were perfectly linear, and the slopes provided enthalpies of activation by linear regression analysis. The entropies of activation were calculated using the Gibbs energy of activation at 2SC and the enthalpy of activation. Second-order rate constants for the Diels-Alder reaction of methyl vinyl ketone with cyclopentadiene were determined as described in detail in Chapter 6. First-order rate constants for the intramolecular Diels-Alder reaction of N-methyl-Nfurfuryhalearnic acid were determined as described in detail in Chapter 7. The equilibrium quotient for the ketooenol equilibrium of 2,bpentanedione (PD) was determined by 'H-NMR (Varian VXR-300 (300 MHz) instrument). The concentration of PD was 5x10" mol kg-'. The NMR tube was equipped with an external standard and H,O could be used as solvent. The equilibrium quotient was determined by integrating the signals at 2.97 ppm (keto) and at 1.77 ppm (enol). The integration was performed after 250 pulses. Each measurement was repeated several times, and the equilibrium quotients were reproducible to within 2 %.

5. Alkyl substinrent fleets. The importance o f solvation

Alkyl Substituent Effects on the Neutral Hydrolysis of 1-Acyl-(3-substituted)-1,2,4-triazolesin Highly Aqueous Media. The Importance of Solvation.

As shown in Chapters 3 and 4, non-covalent interactions between chemically inert cosolvents and reacting substrates in highly aqueous solutions can dramatically affect rate constants of chemical reactions in these media. The quantitative treatment of solvent effects in dilute aqueous media, put forward in Chapter 2, was critically tested in the previous chapters. It was shown that solvent effects are governed by the participation of the chemically inert cosolvents in the solvation shell of the reactant as well as of the activated complex. The different ways in which reactant and activated complex are solvated, causes the observed solvent effects. Careful application of additivity schemes showed that solvent effects are usually dominated by interactions of apolar, hydrophobic parts of the cosolvent. Analysis of solvent effects in dilute aqueous media was used in a retrospective way to identify the contributions of different groups in cosolvents on solvent effects. In dilute aqueous solutions, pairwise interactions predominate solvent effects, unless cosolvent molecules form microphases in the aqueous solution. In this chapter, a quantitative study is reported of alkyl substituent effects and solvent effects on the pseudo-first-order rate constants for the neutral hydrolysis of eighteen 1-acyl-(3-substituted)-1,2,4-triazoles(la-j, 2a-b, 3a-c, 4a-b and 5) in highly aqueous solutions, containing ethanol and 1-propanol. Furthermore, kinetic medium effects and substituent effects on the water-catalysed hydrolysis are completed with the determination of rate constants for the acid-catalysed hydrolysis, solvent deuterium isotope effects for the neutral hydrolysis and spectroscopic data of the substrates.

II / R,-C-N
'

N>C/R~

I
C N ~

la, R,=Me, R2=H lb, R,=Et, R2=H lc, R,=n-Pr, R2=H Id, R, =i-Pr, RZ=H le, Rl =i-Bu, R2=H lf, R,=s-Bu, R2=H Ig, R,=t-Bu, R2=H lh, R,=n-Pent, R2=H li, Rl=3-Pent, R2=H lj, Rl=Ph, R2=H

Scheme 5.1

5. All# subsbent effects. The importance o f solvarion

We will examine the extent to which the variation of the molecular structure of substrates affects the solvent effects of cosolvents and associated group contributions in highly aqueous media. To this end, the variation of the structure of the cosolvents is limited. In contrast, the structure of the reactive substrate is varied. We restrict our attention to the study of substituent effects of apolar alkyl groups. Substituent effects of alkyl groups are difficult to analyse quantitativelyg. Several previous approaches are summarised in Section 5.2. The through-bond effects of alkyl groups on the reactivity of the substrates is expected to be very small compared to those of other substituents (see Section 5.2). The consequences of systematic variations of the alkyl substituents R, and R2 on both solvent and substituent effects, have to be explained mainly in terms of steric and solvation effects. A clear link between substituent effects and solvent effects is found. The importance is emphasised of solvation effects on the magnitude and sign of alkyl substituent effects. Interestingly, an increased hydrophobicity of the reacting substrates enhances the reactivity in water.

5.2 Alkyl substhunt efeects. Cumnt views

Substituent effects of alkyl groups on rate constants and equilibrium constants in - ~ ~ ~ . alkyl substituent effects are solution have been studied e ~ t e n s i v e l ~ Usually analysed using Linear Free Energy Relationships (LFER). The abundance of scales of substituent constants shows a strong resemblance to the wealth of solvent polarity scales. Sometimes theoretical calculations provide the empirical scales with a physical basis252-258. Substituent effects are generally governed by (i) polar, (ii) steric and (iii) solvation effects259. In principle, polar effects can be separated into field, inductive and resonance effects. Steric effects and solvation effects are through-space effects. Field effects are transmitted either through space or through the solvent molecules (if present), in contrast to inductive and conjugative effects, which are generally transmitted directly through the bonds of the molecular chain. Evaluation of substituent effects of alkyl groups is a complex task. Identification of the individual contributions to the overall substituent effect is difficult and has prompted controversies in the literature. An important question concerns the ambiguous evidence for inductive substituent effects of alkyl The first and still most frequently used scale for quantifying polar effects uses a*, as defined by the Taft-Ingold equation261.

Thus, the polar effect of the substituent R (or R') is evaluated relative to methyl by comparison of the rate constants for the base(B)- and acid(A)-catalysed hydrolysis of esters RC02R7. Based on 13C-NMR data, Fliszar et a1.262 have shown, that a' values are related to the charge distributions in the alkyl groups. Later, Levitt and w i d i n e 9 defined the a, scale, mainly based on ionisation potentials for molecules in the gasphase. This scale was used to measure inductive substituent effects. In fact, the a, scale is directly correlated to the u* scale. Substituent effects of alkyl groups were

5. Alkyf substituent fleets. The importance o f solvation

shown to depend on the structure and the electrical characteristics of the molecule. Levitt and W i d i ~ ~ grationalised ~'~ the magnitude and sign of the alkyl substituent constants by (i) charge delocalisation, (ii) through-the-bond and (iii) electric field models. In all models, the inductive effect has been attributed to the polarisation of Alkyl groups were heteroatornic bonds, caused by differences in ele~trone~ativiq". treated as electron donating substituents. Following this approach, alkyl substituent effects have been frequently correlated with the electronegativity of the alkyl group, expressed in xRm. Houk et alez3 have criticised the interpretation of correlations of ionisation potentials in terms of alkyl inductive effects and prefered explanations in terms of dominant hyperconjugative effects. In the gas-phase, alkyl groups are able to stabilise adjacent charges, either negative or positive. Taftm has explained this effect, induced by a neighbouring charge, in terms of the polarisability of alkyl groups, expressed in a substituent constant P. Recently, Hansonm has expressed not only polar and inductive substituent constants, (a* and a,, respectively), but also the group electronegativities (x,), and the polarisability (P) of alkyl substituents in terms of connectivity characteristics. Essentially, the number of the carbon atoms and the pattern of their bonding determine the magnitude of alkyl substituent constants. Both a* and a, were shown to be linear functions of the electronegativity of the alkyl groups. Probably, the polarisibility has a hyperconjugative element. Notwithstanding the experimental evidence cited above, the existence of inductive effects of alkyl substituents has been questioned by several a ~ t h o r s ~ ~In ~ -addition, %~. the significance of a' as a measure for polar substituent effects of alkyl groups has been frequently disputed. For example, inductive effects of alkyl groups are hardly manifested in reactivity. Physical manifestations of alleged inductive substituent effects of alkyl groups are mainly found in NMR data, ionisation potentials and thermodynamic parameters. Generally, in order to obtain a reliable polar or inductive substituent constant, many sets of kinetic data are used. The controversy about the existence of inductive effects of alkyl groups and the significance of polar alkyl substituents constants is mainly caused by the fact that (i) a* and a, are much smaller for alkyl groups than for other substituents, (ii) many more data are available for alkyl groups than for other groups which leads to a divergence of the averaged substituent constant and (iii) the localised field and/or inductive effect of alkyl groups is very small. Consequently, substituent constants for alkyl groups are inevitably flawed with large errors. ~ i t c h i e ~ ~ ,e ~ a and ? Chartonx5 have even argued that the small differences between a* of alkyl groups are a result of artefacts and that differences in inductive effects of alkyl groups, expressed in a,, do not exist at all. These authors claim that the Taft a* constants for alkyl groups probably measure steric effects. A few attempts have been made to separate and quantify steric effects of alkyl groups. In fact, it was again Taft, who was the first to define a steric substituent , ,based on Equation 5-2261. constant, E

Thus, the steric effect of substituent R is evaluated relative to methyl by measuring the rate constants for the acid-catalysed hydrolysis of esters RCOOEt. Subsequently,

5. AIkyl substituent effects. The importance of solvation

Taft analysed the substituent effects of alkyl groups in terms of multiple parameter equations, of the form given in Equation 5-3, in which both polar and steric effects are incorporated.

log k = a + p,E,

+ p,a,

5-3

DeTar et a1.266 showed that E , is correlated with both a*and a,,which makes a joint correlation with E, and either a* or a, of diminished significance. Conflicting claims , .268, exist about steric constants, and alternative steric substituent constants, such as E E,C 269, and 0"' have been introduced to replace the E, scale. In addition, Chart~n~ successfully ~' analysed the steric contribution of a w l substituents in terms of a branching equation. The contribution of solvation effects to substituent effects has always been treated as a The general attitude towards the contribution of solvation effects to the overall substituent effects of alkyl groups has been characterised by pragrnatism. Solvation effects are assumed to be effectively the same for different substrates. Alternatively, it has been assumed that they tend to parallel steric effects. In either case it was thought possible to obtain, at least, relative values of steric effects. The validity of this assumption is not borne out by theoretical or experimental evidence. Recent1 , Charton reported attempts to quantify the role of solvation in substituent effects2 In this chapter, the importance of solvation will be analysed for the overall substituent effect of alkyl groups on a simple hydrolysis reaction.

7 '.

5.3 Solvent effects and substituent gects on the hydrdysir triazohs

I-acyl-(3-sUbs~d)-1,2,4-

Theoretical background. In water-rich media, the pH-independent hydrolysis of l-acyl(3-substituted)-1,2,4-triazolesproceeds via a general-base catalysed rocess, in which the water molecules act as a nucleophile and a general base19 '%, respectively; Scheme 5-2.

Scheme 5-2
86

5. AIkyl substiruetat fleets. The importance o f solvation

Depending on the effectiveness of acid- and base-catalysed hydrolysis, the rate constants me pH-independent in a particular pH range (ca.3-6). However, this range is critically determined by the substituents R, and R, in the substrate. The acidcatalysed hydrolysis shows the characteristics of specific-acid catalysis and proceeds via reversible protonation at N-4in the triazole ring, followed by rate-determining attack of water at the carbonyl moiev3(Scheme 5-3).

Scheme 5-3
For the neutral hydrolysis, the reaction medium consists of water, m, and m, mol kg-' of reactant and activated complex, respectively, a small concentration of HCl in order to prevent base catalysis, and m, mol kg-' of a cosolvent C (ethanol or l-propanol). In all experiments, the substrate concentration is very small (ca.10" mol kg-'). Product analyses revealed only minor amounts of products formed by alcoholysis (see Experimental Section). Therefore, solvent effects are a consequence of the interaction of reactant and activated complex with the added cosolvent molecule. Previously, we derived Equation 2-32 , which relates the pseudo-first-order rate constant for solvolysis reactions in dilute mixed solvents to the rate constant in the pure solvent. This equation can be applied to describe the solvent effect of ethanol and 1-propanol on the neutral hydrolysis of compounds 1-5. In this chapter we use the formalism introduced in Chapter 3. Kinetic data are quantitatively described by

Herein, the factor [CG,-CG,,] is defined as G(C), where G, and GSirepresent the pairwise Gibbs energy interaction parameters of groups i in the cosolvent with reactant and activated complex, respectively. G(C) can be subdivided into group contributions G(i), following a quasi-SWAG approach (see Section 3.3). In this study, solvent effects of ethanol and 1-propanol are expressed as G(Et0H) and G(Pr0H). In chapter 3 it was shown that the group contribution of the CH groups and the OH group of monohydric alcohols can be determined from a comparison of the solvent effect of ethanol and propanol. The solvent effects of ethanol and 1-propanol on the neutral hydrolysis of a series of 1-acyl-(3-substituted)-1,2,4-triazolesare subsequently

5. AIhyl substintent effects. The importance o f solvation

Table 5-1. Neutral and Acid-Catalysed Hydrolysis of la-j. Rate Constants and Solvent Deuterium Isotope Effects at 25C.

la lb lc Id le

If
lg lh li li

methyl ethyl n-propyl i-propyl i-butyl S-butyl t-butyl n-pentyl 3-pentyl vhenvl

203

9.4

2.62

'Rate constant for neutral hydrolysis in water containing ca.10~m01.k~-' of HC1. k a t e constant for neutral hydrolysis in aqueous solution, containing 5 mol kg-' of 2-propanol. '%ate constant for acidcatalysed hydrolysis. d ~ oneutral r hydrolysis.

Table 5-2. Neutral and Acid-Catalysed Hydrolysis of 1-Acyl-(3-substituted)-1,2,4triazoles. Rate constants and Solvent Deuterium Isoto~e Effects at 25OC.

l a methyl
2a methyl

H
t-butyl C1

2b Ib 3a 3b
3c

lg
4a

4b lj 5

methyl ethyl ethyl ethyl ethyl t-butyl t-butyl t-butyl phenyl phenyl

H
methyl t-butyl phenyl

H
methyl t-butyl

H
phenyl

216 59.7 692 290 132 82.8 212 341 168 106 203 127

20 56 29 67 80 9.5 68 213 241 9.4 2.8

3.00 3.06 2.92 3.13 3.19 3.15 3.24 3.09 3.04

C

2.62 2.95

aRate constant for neutral hydrolysis. b ~ a t e constant for acid-catalysed hydrolysis. CBecause of the high sensitivity for acid catalysis, an accurate value of the SDIE for the neutral hydrolysis is difficult to obtain.

5. Alkyl substituent effects. The importance of solvation

Table 53. Spectroscopic Data for 1-Acyl-(3-substituted)-1,2,4-triazoles.

1 3 cchemical shift,'
R l
l a methyl
2a methyl

R 2
H
t-butyl C 1

cm-'

vc=c,,a

UV 1 nm

C=O

8 9 I'pm

C3

C5

2b lb 3a 3b 3c lc Id le If lg
4a

4b lh li l j 5

methyl ethyl ethyl ethyl ethyl n-propyl i-propyl i-butyl s-butyl t-butyl t-butyl t-butyl n-pentyl 3-pentyl phenyl phenyl

H
methyl t-butyl phenyl

H H H H H
methyl t-butyl

H H H
phenyl

1754 1749 1765 1752 1747 1749 1746 1751 1751 1750 1750 1735 1729 1726 1748 1748 1716 1706

218 226 225 221 226 226 258,270d 217 218 218 219 220 227 229 219 219 250
273

167.9 152.8 1 4 3 . 1 173.6 168.2 143.2 171.4 171.3 173.4 171.7 152.6 162.6 171.7 163.0 1 4 3 . 2 1 4 3 . 5 1 4 3 . 2 143.8 1 4 3 . 5

174.5 1 5 2 . 8

1 7 5 . 1 1 5 2 . 1 145.0 1 7 5 . 1 162.0 145.4 1 7 5 . 3 172.7 1 4 5 . 0 173.5 1 5 2 . 6 143.2 164.2 1 5 2 . 8 145.7 164.3 163.8 1 4 6 . 3

aIn CC14 b ~ water. n ' I n CDCl,. d~avelengthfor monitoring the kinetics of the hydrolysis.

If R,is an alkyl group, the solvent deuterium isotope effect (SDIE) increases with , is an alkyl group, the SDIE is less increasing size of the alkyl substituents. When R sensitive to the size of the substituents. It is general1 assumed that hydrophobic The significant difference effects are more pronounced in D20than in Hz& e and compounds If and l g indicates, however, between the SDIE for compounds l that interpretation of these SDIE's is extremely difficult. Substituent effects on spectroscopic parameters. Carbonyl stretching wavenumbers, ~ shifts ~ for ~ the carbonyl carbon and wavelengths of UV maxima and 1 3 ~ chemical the triazole carbon atoms C3 and C5, are collected in Table 5-3.The carbonyl absorption is only slightly dependent on the nature of the alkyl substituents R, and R,. Only for substrates in which R,is a t-butyl group, or in which R,and/or R2are phenyl 3 Cchemical shift of the groups, does the C=O frequency deviate markedly. The 1 , . carbonyl carbon is also sensitive to variation of the alkyl substituents R, and R Essentially, introduction of alkyl moieties causes a downfield shift. The chemical shift of the C3 carbon changes drastically when the hydrogen on C3 is substituted by an alkyl group. This pattern is not unexpected. The chemical shift of the C5 carbon is, however, almost indifferent to substituent effects.

5. Alkyl substituent effects. The importance of solvation

Remarkable, therefore, is the small but significant change in the chemical shift of the C5 carbon in the substrates in which R, is a t-butyl group, or phenyl group. UV spectra are hardly sensitive to changes of the alkyl substituents. Extension of the Telectron delocalisation by the introduction of phenyl groups induces significant shifts of the W maxima. The mutual dependence of substituent effects and solvent effects. In Figures 5-1 and 5-2, ln[k(mc)/k(mc=O)] is plotted as a function of the molality of ethanol and 1propanol, respectively. Here, k is the pseudo-first-order rate constant for neutral hydrolysis of compounds la-lj. Both ethanol and 1-propanol induce a significant decrease of the reaction rate constant. Similar plots can be made for the solvent effect of ethanol and 1-propanol on the neutral hydrolysis of substrates 2-5. All plots show perfect linearity. As shown in Tables 5-4 and 5-5, all solvent effects are dominated by the negative G(C).

0.5

1.O

1.5

m,. mol.kg

-1

Figure 5-1. Plots of -In[k(mc)/k(mc=O) vs molality of ethanol for the neutral hydrolysis of 1-acyl-1,2,4-triazoles (R1COTrR2, R2=H) at 2SC; Rl=Me, 0 ;R,=Et, ;R,=iProp, A ;Rl=n-Prop, V ;R,=Ph, ;R,=i-But, ;Rl=t-But, A ;R,=3-Pent, 0 .

5. Alkyl substituent effects. The importance o f solvation

Figure 5-2. See legend to Figure 5-1, with cosolvent 1-propanol instead of ethanol.

Generally, the reduced activity of water, due to the presence of cosolvents, contributes only moderately to the overall rate effect. The negative G ( C ) is predominantly caused by the rate decreasing contribution of the CH groups, expressed by the negative value of G(CH) (see Tables 5-4 and 5-5). This pattern is consistent with a loss of hydrophobic character of the substrate during the activation process. G(Et0H) and G(Pr0H) depend critically on the nature of the substrate. It is striking that not only the alkyl group R,, but also the alkyl group R , which is quite far removed from the actual reaction centre, exerts a pronounced influence on the observed solvent effects. This means that we have to account for differences in the

5. AIkyl substituent Meets.

IRe importance of

solvation

Gibbs energy of pairwise interactions of the cosolvents with the substituents in the reactant and in the activated complex. An important result of the strikingly different solvent effect of apolar cosolvents on the rate of neutral hydrolysis of related compounds is shown in Table 5-1. Here, the rate constant for neutral hydrolysis of compounds la-j is given, as determined in the presence of 5 mol kg-' of 2-propanol. The sequence, observed for the reactivity of the substrates in pure water, is completely changed. Table 5-4. Solvent Effects of Ethanol and 1-Propanol on the Neutral Hydrolysis of laj at 25OC. Application of a Quasi-SWAG Approach.

R~
la lb lc Id le lg methyl ethyl n-propyl i-propyl i-butyl t-butyl l i 3-pentyl lj phenyl

G(EtOH), J kg mor -52 (2) -103 (2) -107 (1) -155 (2) -133 (2) -185 (1) -202 (2) -83 (2)

G(PrOH), J kg rn01'~ -97 (2) -158 (3) -182 (3) -237 (3) -226 (3) -289 (2) -324 (5) -172 (3)

G(CH), G(OH), J kg rn01-~ J kg mol" -22 (2) -28 (3) -38 (2) -41 (3) -47 (3) -52 (2) -61 (4) -45 (3) 59 37 82 50 102 75 103 142

Table 5-5. Solvent Effects of Ethanol and 1-Propanol on the Neutral Hydrolysis of 1Acyl-(3-substituted)-1,2,4-triazolesat 25OC. Application of a Quasi-SWAG Approach.

la 2a lb 3a 3b
3c

lg 4a 4b
lj

methyl methyl ethyl ethyl ethyl ethyl t-butyl t-butyl t-butyl phenyl phenyl

H
t-butyl

H
methyl t-b~tyl phenyl

H
methyl t-buvl H phenyl

5. Alkyl substituent effeccrs. The importance o f solvation

5.4 A link between alkyl subsfhent effects and solvent effects

As shown in Chapter 3, the solvent effect of monohydric alcohols is attributed to the participation of the cosolvent in the solvation shell of the reactant as well as of the activated complex. The magnitude and sign of the observed solvent effect directly reflect the relative loss or gain of hydrophobicity of the substrate during the activation process. The kinetic data show that the loss of hydrophobicity of related substrates 15, undergoing neutral hydrolysis, depends critically on the nature of substituents. Clearly, the solvation of the substituent in the initial state of the reaction, is quite different from the solvation in the transition state of the reaction. A quantitative analysis of solvent effects in terms of a quasi-SWAG approach quantifies this difference in terms of G(C), G(CH) and G(0H). It is remarkable that G(CH), obtained for different substrates, shows a perfect correlation with the hydrophobicity of the substituent. To illustrate this fact, G(CH) is (Cfi) of the alkyl plotted as a function of Rekker's hydrophobic fragmental constants274 substituent R, (see Figure 5-3). A similar dependence, though less pronounced, can be , . In obtained with respect to G(CH) and the hydrophobicity of the substituent R Figure 5-4 we have constructed a plot, using as the horizental axis the value of Cfi for R, and as vertical axis the value of Cfi for R , . The contribution of the CH groups to the overall solvent effect of ethanol and 1-propanol, expressed as G(CH), is indicated for the particular substrates. In the lower left-hand comer, the most polar substrates are localised, whereas the most apolar substrates are found in the upper right-hand corner.

0 0.5

1.0

1 .S

2.0

2.5
Ifi

3 .O

Figure 53. Plot of -G(C) vs the sum of Rekker's hydrophobic fragmental constant of R , . The substrates are the 1-acyl-1,&4-triazoles (R,COTr);R,=Me, 0 ;R,=Et, ;R, =i-Prop, A ;R, =n-Prop, ;R,=Ph, V ;R, =i-But, rn ;R, =t-But, ;R,=IPent, A .

5. AlkyI substituent effects. The importance o f solvatwn

If. R,

Figure 5-4. Diagram showing the sum of the Rekker's hydrophobic fragmental constants for R, and R2 on the horizontal and vertical axis, respectively. Substrates are 1-acyl-(3-substituted)-1,2,4-triazoles (R,COTrR2). The numbers, given near the symbols are the solvent effects of the apolar CH groups on the rate constant for neutral hydrolysis of the substrates at 25"C, expresses as G(CH):la, 0 ;lb, ;lc, A ;Id, a ;le,A ;lg, ;li, 4 ;lj, V ;2a, r ;3a, 0 ;3b, 4 ;3c, D ;4a, 4 ;4b, 0 ;5, b .

It appears that the loss of hydrophobic surface of the substrate during the activation process is strongly dependent on a combination of (i) the size of alkyl substituents R1 and Ra (ii) the position of the alkyl substituent with respect to the reaction centre and (iii) the overall hydrophobicity of the substrate. The effect of alkyl groups appears to be synergic. Therefore, substrates with an overall similar hydrophobicity can exhibit a remarkably different behaviour towards solvent effects (compare, for example, compounds 2a and 4a). For substrates with a comparable hydrophobicity and a comparable positioning of the substituents, still subtle differences in solvent

5. Alkyl subsrituent @em. lke importance o f solvation

effects of apolar cosolvents are observed. The solvent effect of ethanol and l-propano1 on the neutral hydrolysis of substrates l c and Id, and of l e and l g are significantly different. The loss of hydrophobicity during the activation process of substrates, in which the alkyl substituent R, is branched at the /3 carbon, is smaller than for substrates in which the alkyl group is branched at the a carbon. One can speculate whether this is a result of an increase in the distance to the reactive centre. Clearly, the introduction of alkyl substituents destabilises the reactant with respect to the activated complex. Consequently, the substrate is more sensitive towards hydrolysis. Addition of an apolar cosolvent stabilises the reactant with respect to the activated complex and thereby reduces the reactivity of the substrate. Therefore, the presence of apolar cosolvents neutralises or even reverses the contribution of solvation to the overall substituent effect of alkyl groups. This proposal is confirmed experimentally in Table 5-1, where the presence of 5 mol kg-' of 2-propanol is shown to completely reverse the substituent effect of the alkyl groups. Interestingly, the pK, of RCOOH in water also increases in the series methyl, ethyl, i-propyl and r-butyl, which has been a major point of discussion with respect to alkyl substituent effects266. The trend is understandable in the light of the solvation contribution to the overall alkyl substituent effect.

5.5 The molecuIar origin o f alkyl substbent effects

Introduction of alkyl substituents leads, in the absence of polar or steric effects, to a decrease of the Gibbs energy of activation for reactions in water or highly aqueous solutions. A prerequisite is, that the substrate loses apolar character during the activation process. The loss of hydrophobic surface of compounds, dissolved in water, is, in terms of Gibbs energy, a favorable process. Inspection of the substituent effects of alkyl groups, summarised in Tables 5-1 and 5-2, shows that alkyl substituent effects on rates of neutral and acid-catalysed hydrolysis of compounds 1-5 are not solely a result of solvent effects. In addition, steric effects and, possibly, polar effects of alkyl groups, though the contribution of the latter will be very small, have to be taken into account. These contributions are responsible for the observations that (i) the kinetic and spectroscopic behaviour of substrates in which R, is a phenyl or a t-butyl group deviates from the kinetic and spectroscopic behaviour of the other substrates studied, (ii) the effect of R2 substituents on the rate constants for water and acid-catalysed hydrolysis is opposite and (iii) the difference between substituent effects of substituents R, is significant i f the alkyl groups have a similar h drophobicity . Following suggestions, made by Pinkus et aL2" 6, who studied the carbonyl stretching frequencies and dipole moments of a series of ketones and esters, the abnormally low stretching frequencies of compounds in which R, is t-butyl are most likely the result of an out-of-plane twisting of the t-butyl group due to steric interference with the triazole ring. The remarkable chemical shift of the C5 carbon in the ring, observed for these compounds, supports this explanation. This steric interference is absent for the other alkyl groups. Recent calculations277 have indeed confirmed that the triazole ring and the C=O bond are not coplanar if the R, group is either a tbutyl group, or a phenyl group.

5. Alkyl substiiuent effects. The importance o f solvation

Table 5-6. p&-Values of substituted 1,2,4-Triazoles and 1,2,4-Triazolium Ions in Water at 20C. 1,2,4-Triazole Derivative p c phb

a1,2,4-Triamle. bl,2,4-~riazolium ion. In Table 5-6, the pK, is listed for a series of substituted triazoles and triazolium ions. According to Kroger et al.278,the pK, correlates satisfactorily with a , , , . As shown by the pK, of the triazolium ions, protonation of the triazole ring is facilitated by introduction of an alkyl group on C3. This fact explains the increased sensitivity of the acyltriazoles towards acid-catalysed hydrolysis, when alkyl substituents are introduced on the C3 position. The importance of solvation for this substituent effect is difficult to assess. As shown by the p& of the triazoles, the leaving ability of the triazole ring is reduced by introduction of alkyl substituents on C3. This might explain the decreased rate constant for neutral hydrolysis upon introduction of alkyl groups on C3. The molecular origin of the alkyl substituent effect remains obscure. The abnormal behaviour of substrates in which phenyl groups are present is due to the possibility of extensive charge delocalisation. The pK, of the 3-phenyl-1,2,4triazolium ion is very low (Table 5-6), which explains the extremely low sensitivity towards acid-catalysed hydrolysis. The difference between the substituent effects of isomeric alkyl groups is largely attributed to steric effects and polar effects. Molecular models show that only branching at the /? carbon as well as on the y carbon of R, leads to considerable shielding of the carbonyl carbon, whereas this effect is absent for alkyl groups branched at the a carbon. Therefore, attack of a nucleophilic agent is significantly hindered. Confirmation of this fact by application of a branching equation does not prove unambiguously that only steric effects are responsible for the observed substituent effects. It has been argued that branching at the a carbon leads to a reduced possibility for hyperconjugationm, which might also contribute to the observed substituent effect.

5.6 Solvophbic acceleration and s u b s h n t effects

The present study shows that a quantitative treatment of solvent effects on simple hydrolysis reactions in highly aqueous reaction media in terms of a quasi-SWAG approach, offers a useful basis with which to assess the contribution of the solvent to alkyl substituent effects on reactivity in water-rich binary solvents. The important

5. Alkyl substituent effects. The Theportance o f solvation

conclusion is that steric and polar substituent effects are strongly modulated by the solvation of the reactant and the activated complex. The relative extent of substituent effects can be changed dramatically and in some cases substituent effects can even be reversed by changing the solvent composition. In terms of a Hammett-type analysis this means that (i) the solvation effect is expressed in the magnitude of the dependence of the measured quantity on the substituent and (ii) any correlations with substituent constants is largely governed by the dominant contributions of steric and solvation effects. This fact prompts three important conclusions: (0 (ii) In aqueous media the determination of substituent constants for alkyl substituents that measure purely steric and/or polar effects is impossible. Substituent constants for alkyl groups, or hydrophobic substituents in general, obtained for reactions in either water or highly aqueous solutions, are strongly determined by solvation effects. An observed correlation of substituent effects of alkyl groups for processes in aqueous solutions with known substituent constants means that the solvation effects parallel the steric and/or polar effects of alkyl groups.

(iii)

Chemical reactivity of hydrophobic reactants in water is strongly enhanced by the fact that the water "forces" the reactant to reduce its hydrophobic surface during the activation process. Tentatively, this is a consequence of easy fluctuations of the hydrogen-bonded structure of water together with the ability of the activated complex to adapt itself to the aqueous environment. In the absence of steric and polar effects, increasing the hydrophobicity of the reactant by increasing the size of the alkyl substituents leads to increased reactivity in aqueous solution. The contribution of solvation to the overall substituent effect of hydrophobic groups can therefore be called solvophobic or hydrophobic acceleration.

5.7 l3qerimental section

Materials. Ethanol (p.a.), 1-propanol (p.a.) and 2-propanol were supplied by Merck and were used without further purification. Demineralised water was distilled in an all-quartz distillation unit. All solutions were made up by weight and contained 4x10" mol dm" of HCl (to suppress catalysis by hydroxide ions) for the measurements of rate constants for neutral hydrolysis and appropriate concentrations of HCl for the measurement of rate constants for the acid-catalysed hydrolysis. Solvent deuterium isotope effects were measured in D,O solutions that contained 4 x 1 ~ 'mol dm-3 DCl. Generally, the pH range in which the actual reaction rate constant for hydrolysis is independent of the pH, is very small for those compounds that undergo very effective acid-catalysed hydrolysis. Therefore, measurements were performed between pH 4 and pH 4.5. The pH of the binary mixtures was always carefully checked before and after the reaction using a Corning 130 pH-meter. In case of compound 4b no reliable SDIE could be determined because of the extreme sensitivity towards acid-catalysis. The 1-acyl-(3-substituted)-1,2,4-triazolesla-j, 2b and 5 have been synthesised as reported previously1%. The new substrates were prepared from the corresponding acyl

5. AIkyl substinrent effects. The importance o f solvatwn

chloride and 3-substituted 1,2,4-triazole, according to a standard pr~cedure"~. Solid 1acyl-(3-substituted)-124-triazoles were recrystalised twice from n-hexane and liquid substrates were distilled in vacuo. The 3-t-butyl-1,2,4-triazole, 3-methyl-1,2,4-triazole, and 3-phenyl-1,2,4-triazole were prepared with use of standard procedures. NMR spectra were recorded on a Varian VXR-300 instrument with TMS as an internal standard.

l-Methanoy1-3-S-butyl-lJ,4-tri8u)le(2a): bp 100-10lC (1 mm Hg); 'H-NMR (CDCI,) 8 1.34 (s,9H), 2.64 (s,3H), 8.75 (s,lH) ppm; ',C-NMR (CDC1,) S 21.99 (p), 28.78 (p), 32.82 (q), 143.15 (q), 173.63 (q) ppm. 1-Ethanoyl-3-methyl-l,2,4-triamle(3a): mp 71-73OC; 'H-NMR (CDCI,) S 1.25 (t73H),2.38 (s,3H), 3.10 (q,2H), 8.81 (s,lH) ppm; 13C-NMR (CDCI,) S 7.66 (p), 13.73 (p), 28.01 (s), 143.46 (t), 162.58 (q), 171.29 (q) ppm. 1-Ethanoyl3-5-butyl-1,2,4-triazole(3b): bp 75-77C; 'H-NMR (CDCI,) 6 1.22 (t,3H), 1.32 (s,9H), 3.05 (q,2H), 8.75 (s,lH) ppm; 13C-~MR (CDCI,) S 7.65 (p), 27.96 (p), 128.73 (p), 132.81 (q), 143.23 (t), 171.67 (q), 173.41 (q) ppm. l-Ethanoyl3-phenyl-l,2,4-triazole(3c): mp 60-62OC; 'H-NMR (CDCI,) S 1.33 (s,3H), 3.18 (q,2H), 7.45 (m,2H), 8.92 (s,lH) ppm; ',C-NMR (CDCI,) S 7.77 (p), 28.11 (s), 126.79, 128.56, 129.37, 130.2, 143.84 (t), 163.46 (q). 171.70 (q) ppm. l~-Butanoyl-3-rnethy1-1,2,4-tri8~)le (43): bp 50-51C (0.7 mm Hg); 'H-NMR (CDCI,) S 1.38 (s,9H), 2.34 (s,3H), 8.72 (s,lH) ppm; 13C-NMR (CDCI,) S 13.83 (p), 26.68 (p), 41.08 (q), 145.37 (t), 161.99 (q), 175.14 (q) ppm. l~-Butanoyl-3t-buty1-1,2,4triamle (4b): bp 180-182C (8 mm Hg); 'H-NMR (CDCI,) 6 1.30 (s,9H), 1.40 (s,9H), 8.70 (s,lH) ppm; 13C-NM~ (CDCI,) S 26.71 (p), 28.75 (p), 32.70 (q), 41.06 (q), 145.01 (t), 172.66 (t), 172.66 (q), 175.29 (q) ppm.
Product analysis. The reaction products of the solvolysis of the conpounds la, lb, lg, l j , 4a and 5 were analysed quantitatively. Hereto, reactions were performed in the presence of 0.5, 1.0 and 1.5 mol kg-' of ethanol and 1-propanol, respectively. The substrate concentration in these experiments was always about 5x10" mol dm-3, the pH was between 4 and 4.5. The pH was checked before and after the reaction. After completion of the reaction, the products of the reaction were analysed by gas chromatography (Hewlett-Packard 5890 instrument, equipped with a 15 m wide-bore fused silica column), and, if necessary, by GC-MS (Ribermag R-10-10c) with the authentic triazoles, acids and esters as reference materials. The relative amount of alcoholysis was determined by calibration. In every case, traces of ethyl and propyl ester could be observed. Independent experiments showed that esters were not formed by esterification of the acid formed after hydrolysis. The yield of the esters depends linearly on the concentration of the alcohol. Although slight differences were found between the amount of ester for the different substrates studied, the amount of ethyl ester formed at 1.5 mol kg-' ethanol never exceeded 3+1%, and the amount of propyl ester was even smaller, that is 2*1%. Kinetic analysis therefore shows that the occurrence of alcoholysis does not hamper our quantitative analysis of solvent effects. Kinetic Measurements. The pseudo-first-order rate constants were determined by following the change of absorbance at appropriate wavelengths; see Table 5-3. About 3-5pL of a concentrated stock solution of the substrate in acetonitrile (ca. 5x10-' mol

5. Alkyl substituent effects. The importance of solvation

dm3) was added to the reaction medium (ca. 2.5 cm3) in a quartz UV cell (1 cm) that was placed in a thermostated cell compartment of a spectrophotometer (Perkin Elmer 15, or a Perkin Elmer 1 2 ) , both equipped with a data station. Care was taken to avoid inhomogeneous reaction mixtures in view of the low solubility of the apolar substrates. Therefore, initial concentrations of the very apolar substrates were kept as low as possible. For compound 4b some solubility problems troubled the kinetic analysis. In this case, a delay time was applied after the injection of the stock solution before collection of the data points. In addition, the solution was vigorously stirred with an external microstirrer. The reactions were followed for about 10 half-lives, and excellent pseudo-first-order kinetics were observed in all cases. For every kinetic run, about 100-300 datapoints were collected. Rate constants were calculated using a fitting program as well as by applying the end-value approach. Rate constants, at each molality of cosolvent, were determined at least three times, and were reproducible to within 1%.For the calculation of G(C) from data obtained at, at least, six molalities, a least-squares procedure was used, and the line was forced to go through (0,O). The second-order rate constants for the acid-catalysed hydrolysis were calculated by a least-squares analysis from pseudo-first-order rate constants obtained at six different pH values.

6. Diels-Alder reactions in aqueous media. Enforced hydrophobic interaction

CHAPTER 6
Diels-Alder Reactions in Mixed Aqueous Media. Enforced Hydrophobic Interaction

For a wide range of solvent systems, both rate constants and stereospecificities of Diels-Alder reactions are little or only moderately sensitive to changes in the nature of the reaction mediumz3'. In 1980, Breslow et a1.17" showed that, quite surprisingly at that time, Diels-Alder reactions are dramatically accelerated in aqueous solutions. In the past ten years, a number of studies have revealed curious rate effects on different types of cycloaddition reactions in aqueous solutions17ad~1Sae~19a~bbma-C~21a~bb23 Recently, Grieco et a1.280 reported similar rate enhancements for a Diels-Alder reaction in diethyl ether containing 5 M lithium perchlorate. Not only Diels-Alder reactions are accelerated in aqueous solutions, but also Claisen rearrangernent~'~"~, the benzoin c~ndensation'~", and aldol reactions of silyl en01 ethers '%(see Table 1-1). A common aspect of these reactions is that the reactants are apolar, neutral species. Grieco et a1.18 showed that these accelerations are not just curiosities but that water can be a highly useful solvent for organic transformations of synthetic interest. The high reactivity in aqueous solutions facilitates reaction at low temperatures to prevent decomposition and elimination processes which might occur at high reaction temperatures. In some cases products were obtained which hitherto were inaccessible using organic solvents. In addition, for several processes the stereospecificity and regioselectivity are significantly chan ed when the reaction is performed in aqueous solutions23. Recently, Waldmann et a15 studied ehiral catalysis of Diels-Alder reactions in mixed aqueous solutions. Unfortunately, these authors were not able to perform the reactions without the presence of a considerable amount of organic cosolvents. In all cases the yield of the reactions was significantly higher than in organic solvents. However, a substantial improvement of the stereoselectivity was not observed. An important disadvantage of water is that the solubility of organic substrates is often small. Although the rate constants do not necessarily decrease dramatically when microheterogenities are present in the reaction medium'7ac, an acceptable high solubility of the reactants is a prerequisite for synthetic applications. This can be achieved by (i) employing organic cosolvents to increase the solubility of the apolar reagents18', or (ii) introducing hydrophilic substituents in the substrate molecule^'^^^^^. Lubineau et a1.19b showed that introduction of carbohydrate substituents in the diene in order to increase the low solubility immediately leads to much smaller rate accelerations of these Diels-Alder reaction in aqueous solutions. It is not known, however, how the reactivity of organic reactions in aqueous media depends on the concentration of the cosolvent(s). From the previous kinetic and synthetic studies in aqueous solutions, three crucial questions emerge:

6. Diels-Alder reactions in aqueous media Enforced hydrophobic interaction

(i)

What is the molecular origin of the curious rate effects on Diels-Alder reactions and related transformations in aqueous solutions? (ii) How does the reactivity of the above processes in aqueous solutions respond to addition of organic cosolvent(s) or cosolutes? (iii) Which organic reactions can be accelerated in aqueous media and how do substituents affect these rate accelerations?

A proper understanding of these factors should enable chemists to select organic reactions and appropriate aqueous solvents in order to benefit from the properties of water as a solvent for organic reactions. In this chapter, an attempt is made to answer the questions (i-iii). The chapter describes a quantitative analysis of solvent effects on the intermolecular Diels-Alder reactions of cyclopentadiene (1) with alkyl vinyl ketones (2a,b) and 5-substituted-1,4-naphthoquinones(3a-c) (Scheme 6-1) in water and mixed aqueous solvents containing monohydric alcohols.

Scheme 6-1 Diels-Alder reactions, involving the cloaddition of dienes to quinones have been cornerstones in the synthesis of steroids", cortisonesm2and many natural productsm. Consequently, much information has been accumulated concerning the regioselectivity and reactivity of Diels-Alder reactions involving quinones, naphthoquinones and anthraq~inones~'~. The Diels-Alder reactions of cyclopentadiene with alkyl vinyl ketones or related acrylic acids and derivatives are also well-known. The latter reactions have been frequently used as models in studies of the effect of external factors on reactivity, stereoselectivity and regioselectivity of Diels-Alder processes20ac*2332859286. In Section 6.2, some general mechanistic aspects of Diels-Alder reactions will be described. Subsequently, these mechanistic aspects are associated with external factors which govern the reactivity, stereospecificity and regioselectivity of Diels-Alder reactions in solution. The second-order rate constants for the cycloaddition of dienophiles 2a,b and 3a-c with diene 1 in aqueous solutions are, respectively, about 200 and 5800 times larger

6. Dieh-Alder reactions in aqueous media. Enforced hydrophobic interaction

than those in n-hexane. In comparison to the known rate effects on Diels-Alder reactions in aqueous solutions, the rate effect of water on the Diels-Alder reaction of cyclopentadiene with the substituted naphthoquinones is extremely high. In the past eleven years, several attempts have been made to explain the molecular origin of the high reactivity in aqueous solutions. In Section 6.2, these explanations will be reviewed and critically discussed. The kinetics of Diels-Alder reactions in mixed aqueous solvents across the mole fraction range, OSx,S1 where x2 is the mole fraction of the cosolvent, cannot be analysed by assuming pairwise interactions of the cosolvent with reactants and activated complex, respectively. This chapter will delineate how the quantitative treatment, developed from the model of Kirkwood and Buff, can be applied to account for the dependence of kinetic parameters on the composition of the mixed aqueous solvent. This analysis is supported by the application of the Kirkwood-Buff treatment for the quantitative analysis of transfer parameters for the transfer of 1, 2a and 2b from 1-propanol to mixed aqueous solvents, containing 1-propanol. The applicability of the theory, developed in Chapter 2, for a quantitative analysis of solvent effects on reactions in mixed solvents in which both solvents are present in comparable amounts, is critically tested. Furthermore, solvent effects are reported on isobaric activation parameters (A*HOand TA*SO)of Diels-Alder reactions in mixed aqueous solvents. The exceptional rate accelerations in aqueous solutions are primarily interpreted in terms of "enforced hydrophobic interaction" between diene and dienophile.

6.2 DieLF-AIder reactions in water. Mechanistic considerations and recent developments

Mechanistic aspects of Diels-Alder reactions. Since the discovery of the Diels-Alder reaction in 1928~~", the mechanism of the reaction has been studied extensively2. Knowledge of the mechanism of Diels-Alder reactions is very important for understanding the influence of external factors on Diels-Alder reactions. Studies of the mechanisms of Diels-Alder reactions are still strongly dominated by the WoodwardHoffmann rules. Theoretical predictions, ab initio calculations, approximation methods and frontier molecular orbital theory all contribute to our knowledge of the mechanism of Diels-Alder reactions232.Studies of the mechanism of intermolecular DielsAlder reactions have been characterised by attempts to interpret all Diels-Alder reactions in terms of one universally valid mechanism. Notwithstanding the fact that many [4+2]-cycloadditions involve a symmetry-allowed synchronous bond formation, a two-step mechanism cannot be ignored in all cases. Simultaneous bond formation may be severely hindered by steric or electronic effects and consequently the otherwise energetically less favourable two-step mechanism can become a serious competitor for the one-step process. An important matter of concern in discussions about the mechanism of DielsAlder reactions is the occurrence of a charge-tranfer (m) intermediate287. The occurrence of charge-transfer intermediates is plausible for the cycloaddition of very electron-rich dienes to very electron-poor dienophiles. Some evidence, however,

ti DieLr-Alder reactions in aqueous media. Enforced hydrophobic interaction

suggests that charge-transfer is an important step in every cycloaddition reaction287. Claims for the intermediacy of a true CT-intermediate are controversial. A current debate centres on the possibility that the activated complex takes advantage of CTinteractions, even without prior formation of a true CI'-complex, via a process called "electron-transfer activation". Although the activation process must involve the transfer of electrons to the region where new bonds have to be formed, the extent of electron transfer and concomittant formation of radical ions is unclear. Development of some partial charges during the activation process has been suggested by frontier molecular orbital (FMO) ~ a l c u l a t i o n s Diels-Alder ~ ~ ~ ~ ~ . reactions are characterised by small, but significant p-values, indicative of significant electronic substituent effects232. FMO and ab initio calculations have been very useful in the analysis of the rate constant and regioselectivity of Diels-Alder reactionsmtm. The rate constant and the site of reactivity for reactions of an electrophilic molecule with a nucleophile is dominated by the interaction of the LUMO of the electrophile with the HOMO of the nucleophile. If the orbitals which are involved in the reaction step are brought closer in energy, the overlap is more efficient and the rate constant is enhanced. has been frequently used as a mechanistic tool289The volume of activation, A*VO, 291. In extensive review articles, LeNoble et collected volumes of activation for a large series of Diels-Alder reactions. Generally, the volumes of activation are negative. The activated complex resembles the products with respect to spacial requirements, but in some cases, the volume of activation is even more negative than the overall reaction volume. The mechanistic interpretation is twofold. First, the activated complex is considered to be highly ordered, and formed by simultaneous bond formation; secondary orbital overlap causes an additional decrease of the activation volume. Second, part of the volume of activation must be attributed to the contribution of the solvent. The formation of partial charges during the activation process leads to a decrease of the volume of activation in polar solvents, due to what has been called "electr~striction"~~. The effect of external factors on Diels-Alder reactions. A number of external factors, other than temperature, have been identified which may affect the rate constant, regioselectivity and stereospecificity of Diels-Alder reactions in solution. A useful and of effective contribution to the question of mechanism was provided by the discove~y the catalysis of Diels-Alder reactions b Lewis-acids which may accelerate the rate of Lewis-acids affect the orbital coefficients of addition by several powers of ten the reactants by interacting with substituents of the dienophile. Recently, Kelly et a1.295 showed that compounds able to form extensive hydrogen bonds to the dienophile, also catalyse Diels-Alder reactions in aprotic media. A less spectacular but very useful extension of the cycloaddition reactions is provided by the fact that Diels-Alder reactions are generally accelerated by applying external pressure2%. The rate effects can be quite dramatic if the volume of activation is strongly negative. The stereospecificity and regioselectivity of the cycloaddition can be significantly altered by changes in pressure if the activated complexes leading to the regio- and stereoisomers are characterised by different volumes of activation286.For Diels-Alder reactions involving cyclopentadiene, the endo-exo product ratio is determined by secondary orbital overlap. An increase in pressure favours one of the two isomeric products297.Recently, it has been reported that also antibodies can accelerate Diels-Alder reactions"'.

mm-L.

6 DieIs-Alder reactions in aqueous media. Enfmed hyirophobic interaction

Many Generally, solvent effects on Diels-Alder reactions are negligibly cycloadditions proceed as fast in the vapour phase as in nonpolar solvents. In a few incidental cases larger solvent effects (accelerations of a factor of 35 on going from an apolar to the most polar solvent) are These solvent effects are attributed to extreme changes in the dipole moment of the reacting species during the activation process. Only in rare cases polar or zwitterionic intermediates are responsible for exceptionally large solvent effectsz7. Desimoni et used the FMO approach to l i n k solvent effects on Diels-Alder reactions with catalysis by Lewis acids. Solvent effects on the cycloaddition of 2,3-dimethylbutadiene to 5-substituted-1,4naphthoquinones were explained by the donor-acceptor interactions between the dienophiles and the solvents. This was confirmed by a correlation between the acceptor numbers of the solvent (see Section 1.2.2) and the rate constant for the Diels-Alder reactions.
Recent developments. Solvent effects on Diels-Alder reactions have been studied frequentl?, though extensive and detailed analyses of solvent effects on Diels-Alder reactions in large series of organic solvents are rare. In 1962, Berson et a1.=' studied solvent effects on the stereoselectivity of Diels-Alder reactions of cyclopentadiene with several dienophiles by means of linear free energy relationships and introduced the 0 solvent polarity scale based on the endo/exo product ratio. In fact, this study reported the first attempt to perform Diels-Alder reactions in water and mixed aqueous solvents: "We made some attempts to conduct experiments in aqueous systems ..............., but the low solubility of cyclopentadiene in the aqueous solvents ....... led us to abandon these." Twenty years later, Breslow et a1.17' reported a first kinetic study of Diels-Alder reactions of cyclopentadiene with several dienophiles in aqueous solutions. According to Breslow, "hydrophobic packing" of the two reactants is the most likely explanation for the large rate effects of water on Diels-Alder reactions". This idea is mainly based on a series of kinetic experiments on Diels-Alder reactions in aqueous solution in the presence of "salting in" and "salting out" materials. The enhancement or breaking of the water structure by the added cosolutes is used as a criterion for the promotion or inhibition of "hydrophobic packing". Breslow et al.", and later also Schneider et a ~ . supported ~, this view by performing these Diels-Alder reactions in the presence of P-cyclodextrine. Comparable rate effects were observed and complexation of the reactants in the hydrophobic interior pocket of the polysaccharide was suggested to replace the "hydrophobic packing" in the aqueous solution. The ideas of Breslow, recently enunciated in a review paper17g, seem to be widely a~cepted~~~ As ' ' .will be shown in Section 6.7, the molecular basis of this "hydrophobic packing" is rather ambiguous. The main problem is concerned with the configuration of the reactants in the initial state of the reaction. Alternative explanations were provided by Griecol' and Lubineau19 and coworkers. Grieco suggested the occurrence of micellar catalysis for some Diels-Alder reactions, in which large concentrations of a surfactant-like diene were The exceptional rate effects were attributed to hydrophobic binding of the apolar dienophile to the micellar aggregate of the diene. For the Diels-Alder reactions of unactivated iminium salts with a series of dienesl& and the effective intramolecular Claisen rearrangementslsg in aqueous solution, this explanation is not appropriate. Unfortuna-

6. Diels-Alder reactions in aqueous media. Enforced hydrophobic interaciion

tely, Grieco et a1.18 did not report any details about the kinetics of their synthetically very useful organic processes in aqueous solution. Nonetheless, the high yields in spite of ambient pressure, low reaction temperatures and short reaction times are indicative of high rate constants. Also the rate acceleration of the aldol condensation of silyl en01 ethers, reported by Lubineau et al.lk, excluded micellar catalysis. Both authors suggested a possible link between the effects of external pressure on Diels-Alder reactions (see above), performed in conventional organic solvents and the lar e rate enhancements in aqueous medialg1? They refer to publications of Daclc4', who stressed the relation between the activation volume A ' V of a chemical reaction and the cohesive energy density (CED) of the solvent. Generally, reactions with a negative volume of activation are accelerated by solvents with a high CED. The CED of the solvent can be expressed in terms of the Hildebrand solubility parameter (a,*) (see section 1.2.2). Application of the Hildebrand solubility parameter to the analysis of solvent effects on chemical reactions is restricted to reactions of neutral molecules involving a neutral activated complex. However, Dack argued that the ideas can also be applied for reactions involving ionic reactants or a charged activated complex42. The effect of CED of the solvent on rate constants is small. For many organic solvents the CED, which is also called the cohesive pressure, is related to the internal pressure k ~ ~ internal pressure acts on the of the solvent. According to a citation of ~ a c "solvent rate of nonpolar reactions in the same direction as external pressure". The suggestion that the internal pressure of the solvent is the explanation for the exceptional solvent effects on Diels-Alder reactions in aqueous media is supported by the exceptionally large rate acceleration of Diels-Alder reactions in diethyl ether in the presence of 5M of lithium perchlorate (known for its high internal pressure)280.However, the CED is a measure for the enthalpy of evaporation per volume of solvent, whereas the internal pressure is defined as the change in internal energy of a solvent as it undergoes a very small isothermal expansion. The internal pressure results from forces of attraction between solvent molecules exceeding the forces of repulsion (mainly dispersion and dipole-dipole interactions). For highly structured liquids, such as water, the CED deviates strongly from the internal solvent pressure. Compared to organic solvents, the internal pressure of water is even exceptionally small. In fact at 277 K, the internal ~, and Breslowl'g did not pressure of water is close to zero. ~ u b i n e a u ' ~Grieco"' discriminate between internal pressure and the CED and erroneously concluded that water, having an extremely large CED, exerts a high internal pressure on organic reactions in this medium. The exceptionally high reactivity of Diels-Alder reactions in aqueous media has also been related to the solubility of the reactants in aqueous media2092311%. Reaction rate constants and stereospecificities of Diels-Alder reactions in mixed aqueous solvents have been correlated successfully with the solvophobicity parameter Sp6' and q ( 3 0 ) ~ ~ ~The ' ~ . correlation of the reaction rate constants of Diels-Alder reactions ' ~ in fact, also with the CED of the solvent, emphasised by ~ r i e c o "and ~ u b i n e a u was, indicative of the im ortance of the solubility of the reactants. Schneider et a 1 performed a detailed study of the solvent effects on the DielsAlder reactions of cyclopentadiene with a series of esters of acrylic acid in aqueous solutions as a function of the hydrophobicity of the dienophiles. Following the ideas of Breslow, Schneider implicitly assumed complexation of the reactants in the aqueous solution. The anticipated correlation between the hydrophobiclty of the dienophiles,

6 DieLF-Alder reactions in aqueous media. Enforced hyirophobic interaction

expressed in the Hansch T-increments, and the rate constants for the Diels-Alder reactions in water was, however, not observed'. This forced Schneider to conclude that hydrophobic complexation can be in fact counterproductive due to the fact that this might result either in homotactic complexation of the reactants or in complexes, having the 'bong" geometry for reaction to occur. Recently, photochemical Diels-Alder reactions have been studied in aqueous s o l ~ t i o n s ~ In ' ~ some ~ ~ . cases the product distribution and the regioselectivity of the process were very different in aqueous solutions as compared with that in organic solvents. This was related to the enhanced photo-efficiency in aqueous media. The interpretation of solvent effects on photochemical reactions is very complex, but the effects have been frequently attributed to stacking of the reactants in aqueous media. Finally, Diels-Alder reactions have been studied in "organised aqueous media, (ii) ~ ~micellar ~ ~ ~ , systemsw1 and (iii) clay emulsion^^^. such as (i) r n i c r ~ e r n u l s i o n s ~ These microheterogeneous systems share the following two common features: (i) they provide a hydrophobic pocket in which hydrophobic molecules can be assembled and (ii) they offer an interface between the aqueous exterior and a hydrocarbon-like interior wherein the reactants can be aligned. Frequently, Diels-Alder reactions are significantly accelerated in these systems. Remarkably, the stereospecificity of DielsAlder reactions in micellar systems and microemulsions and the stereospecificity found in polar environments, are similar. Rate effects have been attributedw1 to (i) cage effects (the ability of microdomains to hold two reactive species together long enough for reaction to occur), (ii) pre-orientation of the reactants (which resembles stacking), (iii) microviscosity and (iv) the local concentration effect. Some organic reactions, involving reagents that are only sparingly soluble in water, have also been performed in heterogeneous aqueous media. The rate constant still was considerably higher than in organic solvent^"^^.

6.3 Kinetic studies of DieIs-Alder reactions in m k d aqueous solvents; theoretical apprwch

Theoq. Transfer parameters for 1,2a, 2b and the product of the reaction of 1with 2a have been determined by measuring vapour pressures. We consider a solution of a solute-j at a temperature T. The solvent, or mixed solvent, is liquid-1. The solute is volatile so that at equilibrium the partial pressure of the solute-j above the solution is pj. It is now assumed that the vapour forms a perfect gas. For the gas phase, the chemical potential of the substance-j is given by:

Here, p0 is the standard pressure and Ccj"(g) is the chemical potential of the compound-j as a perfect gas at pressure pO.For the compound-j in solution, the chemical

* w e were not able to reproduce the rate constants reported by Schneider et al. for the DielsAlder reactions of cyclopentadiene with esters of acrylic acid (see Experimental Section).

6 .Diels-Alder reactions in aqueous media. Enforced hydrophobic interaction

potential is related to molality m, using Equation 6-2:

By definition, limit(mj+O)yj=l.O. At equilibrium, pj(g)=pj(sln,l). Therefore

In dilute solutions of compound-j on a molality scale, according to Henry's Law, we have

where Hj is Henry's constant. So, if mj is very small (yj=l.O),

A similar equation can be derived for a solvent-2 I f the solutions are dilute (mj+O), the standard chemical potential of transfer of solute-j from solvent-1 to solvent-2, where solute-j has unit molality, is given by Equation 6-6.

Values of H. were calculated from the slope at infinite dilution of the plot of pj as a function of the molality of j. flsln;2)-lr,D(sln;l) is the transfer parameter. The vapour pressure of a solute is very sensitive to stacking or clustering of the solute. According to Equation 6-4, a plot of pj against mj is linear for an ideal solution in which yj is unity. Any significant curvature (and deviation of yj from 1) is indicative of pairwise and higher order interactions between the solutes, possibly leading to association. In Chapter 2, a quantitative treatment was developed, based on the theory of Kirkwood and Buff, which accounts for the dependence of rate constants on the composition of a mixed solvent across a mole fraction 0 s x 2 S l , where x2 is the mole fraction of the cosolvent. The analysis of the kinetic data is based on the standard chemical potentials of the reactants and the activated complex in a mixed solvent, containing solvents 1 and 2. According to Equation 2-71, a parameter 4 can be 4 is determined by four Kirkwood-Buff integral functions, Gxl, G , G,, and G , , . These integral functions measure the probability of finding a solvent molecule 1 or 2 in a volume element around reactants X and activated complex (symbol +), respectively. Therefore, 4 measures the change in the relative affinities

6.DieIs-Alder reactions in aqueous media. Enforced hydrophobic interaction

of the reactant, upon going from the initial state to the transition state, for the solvents present in the mixed solvent. In fact, as shown in Section 2.4, six KirkwoodBuff integral functions are required to determine the magnitude of 4 for bimolecular reactions. To avoid unnecessary complications in the formulae, X represents both diene and dienophile. The quantitative analysis of solvent effects on Diels-Alder reactions in mixed aqueous solutions then involves the calculation of 4 as a function of x,. For the calculation of 4 three parameters are required that have to be derived from experimental data. 4 is related to the dependence of the rate constant on the composition of the mixed solvent, as expressed by the differential d{ln[k(x2)/k (x,=O)])/dx,. The determination of this first-order derivative involves fitting the dependence of ln[k(x2)k(x2=0)] on the composition of the solvent. It was found to be impossible to formulate a unique mathematical function which accounts for the observed trend. Since the first-order derivative must be a continuous function, the data were approximated by cubic ~plines~'~(see experimental section). Also important is the property Q, which represents the non-ideal behanour of the mixed solvent. Q is a function of the composition of the mixed solvent (see Equation 2-70). The key step in the determination of Q involves the calculation of the secondorder differential d2GmE/d$. Error propagation analysis shows that the calculation of Q is the largest source of error in the evaluation of 4, especially when Q approaches zero (see also Section 6.6)l9'. The excess molar Gibbs energies of mixing were calculated on the basis of activity coefficients or vapour pressures which have been reported , ' on the mole fraction in the literature (see experimental part). The dependence of G x, was fitted to the orthogonal polynomial, given in Equation ~ 7 ~ .

where

This orthogonal polynomial is particularly well suited1630435 for analysis of the excess molar Gibbs energy of a binary aqueous mixture where GmEis large and positive. Subsequently, the second-order derivative with respect to the mole fraction of cosolvent was computed. The third parameter, the dependence of the partial molar volume on the composition of the mixed aqueous solvent, was calculated on the basis of densities which were retrieved from the literature. The change of affinity of the reactants upon going from the initial state to the

6 Diels-Alder reactions in aqueous media Enforced hyirophobic interaction

6.4 Properties of alcohd-water mirtures.

For many years, aqueous solutions of low-molecular weight alcohols have been of interest to many scientists and technologists. The complete miscibility of alcohols with water and low costs make mixtures of water and alcohols an attractive class of solvents for industrial applications. Although chemists have been intrigued by the remarkable properties of these mixtures in the highly aqueous concentration range, their properties have usually been determined over the whole concentration range (see, for an extensive review, ref.15). High precision microcalorimetry, nuclear magnetic relaxation studies, scattering methods and new theories of liquids and solutions have contributed to a better understanding of the structure and dynamics of aqueous solutions of alcohols. Properties of alcohol-water mixtures are discussed considering separately several concentration ranges. The demarcation between these concentration ranges is determined by more or less abrupt changes in the properties of alcohol-water mixtures. The following concentration ranges have successively been distinguished (i) infinitely dilute aqueous solutions and dilute aqueous solutions, (ii) water-like mixed aqueous solutions, and (iii) alcohol-like mixed aqueous soIutions. In Section 3.2, the properties have been discussed of infinitely dilute and dilute solutions. The thermodynamic properties of these solutions can be accurately described by concentration expansions, involving pairwise and higher-order terms. The alcohol molecules are fully solvated by water molecules. At higher concentration of alcohol, partial molar quantities, such as partial molar volumes and heat capacities, exhibit extrema. These extrema are very sensitive to temperature and pressure and depend on the nature of the alcohol molecule. Light scattering data, ultrasound measurements, kinetics and SAXS indicate the onset of aggregation-like processes (defined by Franks as typically aqueou~'~). The increased affinity of alcohol molecules for other alcohol molecules is elegantly shown by extrema in the Kirkwood-Buff integral functions representing the mutual affinity of alcohol molecules. Beyond these extrema the mixed aqueous solvent is characterised by short-lived water-rich and l quantities exhibit a second, less pronounced alcohol-rich clusters. The ~ a r t i a molar change at still higher alcohol content. Beyond this concentration, the properties of the mixed aqueous solvent start to resemble those of the pure alcohol.

6.5 Dieb-Alder readions in water and in mired aqueous solvents; erperimental r e &

The dependences of the rate constants for Diels-Alder reactions on the composition of the binary water-alcohol mixtures are characterised by two critical mole fractions of cosolvent. These mole fractions almost coincide with the critical mole fractions that characterise the thermodynamic properties of the alcohol-water mixtures (see Section 6.4). Following the subdivision of mixed aqueous solvents, containing monohydric alcohols, described in Section 6.4, the dependences are reported of (i) rate constants, (ii) A'HO and TA'SO and (iii) standard chemical potentials of reactants and activated complexes on the composition of the solvent. Four separate concentration ranges are identified.

6 Dielr-Alder reactions in aqueous media. Enforced hydrophobic interaction

Water and organic solvents. Second-order rate constants and isobaric activation parameters for the intermolecular Diels-Alder reactions of diene 1 with dienophiles 2a, 2b and 3 a c in water and in 1-propanol, are listed in Table 6-1. The rate constants for the Diels-Alder reactions of 3a-c with cyclopentadiene in 1-propanol decrease in the followin order: 3cc3ac3b. In aqueous solutions, however, we find 3aSb<3c. Houk et a 1 5 used frontier molecular orbital approaches to calculate substituent effects on rate constants and regioselectivities of Diels-Alder reactions, involving a large series of quinones. Electron-donating substituents in the quinones raise the orbital coefficients and decrease the rate constants, whereas electron-accepting substituents lower the orbital coefficients and lead to an increase of the rate constant. When the hydroxy substituent of compound 3b (also known as juglone) adopts the syn-conformation, the hydroxy group binds to the peri carbonyl via intramolecular hydrogen bonding. In this case, the hydroxy substituent acts as an acceptor. If the hydroxy group is constrained to the anti-conformation, intramolecular hydrogen bonding with the peri carbonyl is impossible and the hydroxy group acts as a donor. l l solvents284.The FMO calculations predict The syn-conformation is more stable in a rate constants in the order of 3cc3a<3b. This order was only found in aqueous solutions. Table 6-1. Second-order Rate Constants and Isobaric Activation Parameters for the Diels-Alder Reactions of Cyclopentadiene (1) with Dienophiles 2a, 2b, and 3a-c in Aqueous Solutions and Organic Solvents, at 25OC.

Da
2a

Solvent Methanol Ethanol 1-Propanol Water Water+S~S~ Water +SDSC water+CI'ABb Water+CTABC water +CHPd I-Propanol Water 1Propanol Water 1Propanol Water 1-Propanol Water

k2

A*GO,

A *HO,

dm3 mol-' s-'

kJ morl

kJ mol-'

-TA 'So, kJ mol-I

2b 3a 3b
3c

Dienophile. 50 mmol in 1 kg of water. pyrrolidone in 1 kg of water.

a

100 mmol in 1 kg of water

750 mmol N-cyclohexyl-

6.Diek-AUer reactions in aqueous media. Enforced hydrophobic interaction

Desimoni et a1.299found the following order of rate constants of Diels-Alder reactions of 5-substituted-naphthoquinones with 2,3-dimethylbutadiene in organic solvents: 3a<3c<3b. In summary, the order of reactivity is sensitive to the solvent and cannot be predicted accurately by FA40 calculations. The rate constants for all Diels-Alder reactions studied are spectacularly enhanced in aqueous solutions. The magnitude of the rate acceleration is dependent on the nature of the dienophile. The Gibbs energies of activation, found in 1-propanol, are reduced by 10-14 kJ mol-' in aqueous solutions. The magnitude of this reduction is dependent on the size andlor hydrophobicity of the dienophiles. The substituent effects are moderate but significant. Both entropies and enthalpies of activation are more favourable for reactions in water than in 1-propanol and contribute cooperatively to the rate acceleration. Generally though, the entropy term is the dominating factor.

Figure 6-1. Gibbs energy of activation for the Diels-Alder reaction of 1with 3c in nhexane (I), tetrachloromethane (2), benzene (3), 1,4-dioxane (4), THF (5), chloroform (6), dichloromethane ( 7 ) , acetone (8), DMSO (9), acetonitrile (lo), 2-propanol (ll), ethanol (12), N-methylacetarnide (13), N-methylformarnide (14), methanol (IS), glycol (16), trifluoroethanol (17) and water (18) as a function of q 3 0 ) at 25C.

6 DM-Alder reactions in aqueous media. Enforced hydrophobic interaction

Figure 6-2. Gibbs energy of activation for the Diels-Alder reaction of 1with 3c in nhexane (I), THE: (2), benzene (3), tetrachloromethane (4), 14-dioxane ( 9 , acetone (6), acetonitrile (7), DMSO (81, dichloromethane (9), chloroform. (lo), N-methylformarnide ( l l ) , Zpropanol (12), 1-propanol (13), ethanol (14), methanol (15), trifluoroethanol (16) and water (17) as a function of the acceptor number (AN).
The second-order rate constants for all Diels-Alder reactions studied increase with increasing polarity of the reaction medium. The Gibbs energy of activation (A*GO)for the Diels-Alder reaction of 1 with 3c correlates with w 3 0 ) for a series of organic solvents (see Figure 6-1). The data point for water deviates strongly from the observed trend. Following procedures advanced by Desimoni et a1.299, these Gibbs energies of activation were also correlated with acceptor numbers (AN) of the solvents. Although the correspondence with the results obtained by Desimoni et al. is very good for the organic solvents, the data point for water again deviates strongly

6 DieIs-Alder reactions in aqueous media.Enforced hydrophobic interaction

from the observed trend (Figure 6-2). Generally, the reaction rate constant is much larger in water than anticipated on the basis of correlations with solvent polarity parameters. The correlation of the Gibbs energy of activation with the Hildebrand solubility parameter is poor, although this correlation seems to account for the exceptionally large rate constant, observed in water (Figure 6-3). A number of experiments were performed to assess aggregation of the reactants in aqueous solution. In Figure 6-4, the vapour pressure of 1,present in the vapour above a solution of 1in water, is plotted as a function of the molality of 1 (in fact, the peak area of the GC-peak is plotted, see Experimental Section). Only at high concentration (near the solubility limit), does the plot begin to deviate significantly from linearity. Apparently, 1 does not aggregate in aqueous solution at concentrations used in our kinetic experiments.

Figure 6-3. Gibbs energy of activation for the Diels-Alder reaction of 1 with 3c in nhexane (I), tetrachloromethane (2), benzene (3), THF (4), chloroform (S), acetone (6), dichloromethane ( 7 ) , 1,Cdioxane (8), 2-propanol (9), trifluoroethanol (lo), acetonitrile (ll), ethanol (12), DMSO (13), methanol (14), glycol (IS), and water (16) as a function of aH2at 2SC.

4 Diels-Alder reactions in aqueous media. Enforced hydrophobic interaction

Peak A r e a x 1 0 " ~

0.00

0.02

0.04

0.06
-1

0.08

mcpd.mol kg

Figure 6-4. Plots of the peak area of cyclopentadiene, obtained after injection of a standard volume of vapour, withdrawn from the vapour above an aqueous solution of cyclopentadiene as a function of the molality of cyclopentadiene, at 2SC; solution of cyclopentadiene in pure water, w and solution of cyclopentadiene in aqueous solution, containing 10 % (wh)of ethanol, .
Addition of 10 % w h of ethanol completely inhibits aggregation at these concentrations (see Figure 6-4). Similar experiments, performed with aqueous solutions of alkyl vinyl ketones clearly showed that these dienophiles do not aggregate in water. The low volatility of the substituted-naphthoquinones precludes determination of vapour pressures. The concentration of the 5-substituted-naphthoquinones, used for the mol kg-'. Even stacking of molecules, characterised kinetic experiments, was 10~-10~' by an extensive T-electron system, at these concentrations is highly improbable. Finally, the second-order rate constants for the Diels-Alder reactions were measured using different excess concentrations of diene or dienophile (see Experimental Section). The second-order rate constants were reproducible within the error limits which indicates that the reactants do not form aggregates in the aqueous solution.
Dilute mixed aqueous solutions. In Section 4.3, the solvent effect of monohydric alcohols on the rate constant of the Diels-Alder reaction of 1 with 2a has been discussed in detail.

6 DieIs-Alder reactions in aqueous media. Enforced hydrophobic interaction

k2,dm mol

- 1 - 1

Figure 6-5. Second-order rate constants for the Diels-Alder reaction of 1 with 3a in mixed aqueous solutions as a function of the mole fraction of water at 25C; , ethanol; ,1-propanol; A ,2-methyl-2-propanol.

In this concentration range pairwise intermolecular interactions govern the medium effect of organic cosolvents and solvent effects of monohydric alcohols are very small. It was concluded that the interactions of the activated complex and reactants with the added cosolvents are very similar and that the activated complex resembles the initial state. The more pronounced solvent effect of 1,4-dioxane is difficult to reconcile with this explanation. In dilute solutions (x1>0.95), the second-order rate constants, expressed in dm3 mol-' s-', and Gibbs energies of activation do not change dramatically (see Figures 6-5, 6-6, 6-7 and 6-8). Remarkably, the rate constants for the Diels-Alder reactions of 1 with 3s-c even increase slightly upon addition of 2-methyl-Zpropanol and 1-propanol. This remarkable effect was not observed for the Diels-Alder reactions of 1 with 2a-b. The standard Gibbs energies for the transfer of diene 1 and for the product of reaction of 1with 2a from 1-propanol to dilute mixed aqueous solutions of 1-propanol are highly unfavourable (see Figures 6-9 and 6-10). The transfer of the other reac-

6. Diels-Alder reactions in aqueous media. Enforced hydrophobic interaction

tants, 2a and 2b is significantly less unfavourable. Remarkably, the standard chemical potentials of the activated complexes of the reactions of 1with both 2a and 2b in 1propanol are almost identical to the standard chemical potentials in water and dilute aqueous solutions of 1-propanol. The standard chemical potentials of reactants as well as activated complexes do not change dramatically as a function of the mole fraction of cosolvent in dilute aqueous solution. The dependence of A*HO and TA*SO (at 298.15 K) for the reaction of 1with 3b on the mole fraction of 1-propanol in dilute aqueous solutions is dramatic (see Figure 611). The changes of A'HO and TA*SOacross the mole fraction range of 0<x2<0.07 are spectacular and amount to 24 kJ mol-'. The enthalpy of activation decreases sharply upon addition of 1-propanol. Simultaneously, the entropy of activation exhibits an equally dramatic decrease. The effects almost fully compensate each other. In the absence of an organic cosolvent, the rate acceleration of the Diels-Alder reaction of 1 with 3a is almost completely due to a more favourable entropy of activation. It is interesting to note that a similar acceleration can be achieved in a mixed aqueous solution of 1-propanol, x2=0.07, which is due entirely to a more favourable enthalpy of activation.

Figure 6-6. Gibbs energy of activation for the Diels-Alder reaction of 1 with 2a in mixed aqueous containing methanol, A ;ethanol, ;l-propanol, o ;2-methyl-2propanol, 0 as a function of the mole fraction of water at 2SC.

4 Diek-Alder reactions in aqueous media. Enforced hydrophobic interaction

The dependence of the endolexo ratio of the products on the molality of cosolvents for the reaction of 1with 2a and 2b in dilute mixed aqueous solvents is similar (see Figure 6-12 for the endolexo product-ratio of the reaction of 1 with 2a). The endo product, which is already favoured in organic solvents, is almost exclusively formed in aqueous solutions. The endolexo ratio decreases smoothly with the addition of monohydric alcohols. In Table 6-1, second-order rate constants are given for the Diels-Alder reaction of 1 with 2a in the presence of a cationic (cetyItrimethylammonium bromide) and an anionic (sodium dodecylsulphate) surfactant. The critical micellar concentrations are 0.95 mM and 8.3 mM, respectively. In addition, the second-order rate constants are given for the Diels-Alder reaction of 1with 2a in the presence of N-cyclohexyl yrrolidone. The latter cosolvent forms microdomains in dilute aqueous solutions%'. All additives induce a sharp decrease of the rate constant.

Figure 6-7. Gibbs energy of activation for the Diels-Alder reactions of 1 with 2a in mixtures of water and ethanol ( 0 ) and 1-propanol (0)and of 1with 2b in mixtures of water and ethanol (m) and 1-propanol (e)as a function of the mole fraction of water, at 2SC. The Gibbs energy of activation in mixtures of 1-propanol and water has been displaced upwards by 4 kJ mol-' for clarity.

6. Dieb-Alder reactions in aqueous media. Enforced hydrophobic interactbn

Figure 6-8. Gibbs energy of activation for the Diels-Alder reaction of 1 with 3a (circles), 3b (squares), 3c (triangles) in mixtures of water and ethanol (unfilled symbols), and 1-propanol (filled symbols) as a function of the mole fraction of water, at 2SC. The Gibbs energy of activation in mixtures of 1-propano1 and water has been displaced upwards by 4 kJ mol-' for clarity.

Water-like aqueous solutions. At certain mole fractions of cosolvent the rate constants of the Diels-Alder reactions decrease rapidly (the Gibbs energies of activation increase markedly). These mole fractions are dependent on the hydrophobicity of the cosolvent as well as on that of the reactants. For aqueous solutions of methanol, ethanol, 1-propanol and 2-methyl-2-propanol, these mole fractions are ca.0.15, ca.0.08, ca.0.04 and ca.0.04, respectively. These mole fractions separate the dilute aqueous solution from the water-like mixed aqueous solution. The rapid increase of the Gibbs energy of activation occurs at a lower mole fraction of cosolvent if the hydrophobicity of the reactants is enhanced (see Figures 6-7 and 6-8). The increase of the Gibbs energy of activation is more pronounced if the cosolvent is more hydrophobic (see Figure 6-6). A second critical mole fraction, which is less pronounced, marks the end of the sharp increase of the Gibbs energy of activation. The first critical mole fraction, observed for the solvent effect on the Diels-Alder reaction of 1with 3b in aqueous solutions of 1-propanol coincides with the extrema of the isobaric entropy and enthalpy of activation. Beyond the first critical mole fraction

6 DieLF-Alder reactions in aqueous media. Enforced hydrophobic interaction

A*HOand TA*SOincrease simultaneously. The contributions do not compensate each other. The decrease of the rate constant is dominated by the unfavourable enthalpic contribution (see Figure 6-11). The endolexo product ratios for the reactions of 1 with 2a and 2b continue to decrease smoothly as a function of the mole fraction of cosolvent. The shapes of the curves do exhibit a clear change near the second critical mole fraction (see Figure 612). The standard chemical potentials of 1and the product of the reaction of 1with 2a decrease sharply as a function of the mole fraction of 1-propanol (see Figures 6-9 and 6-10). The dienophiles 2a and 2b are very soluble in water, and the standard chemical potentials exhibit a more or less linear dependence on the mole fraction of l-propanol.

Figure 6-9. Standard Gibbs energies of transfer for reactants, activated complex and products of the Diels-Alder reactions of 1with 2a from 1-propanol to mixtures of 1propanol and water as a function of the mole fraction of water at 25OC;2a, A ;I, rn ;initial stata=(1+2a), a ;activated complex, 0 ;product, 0 .

6. Diels-Alder reactions in aqueous media. Enforced hydrophobic interaction

0 .O

0.2

0.4
x ( HZO)

0.6

0.8

1 .O

Figure 6-10. Standard Gibbs energies of transfer for reactants and activated complex for the Diels-Alder reaction of 1 with 2b from 1-propanol to aqueous mixtures of 1;initial state propanol as a function of the mole fraction of water;2b, A 1 (1+2b), ;activated complex, 0 . Interestingly, the standard chemical potential of 2a is even lower in a mixed aqueous solution of 1-propanol than in the pure solvents. Most striking, however, is the fact that the standard chemical potentials of the activated complexes for the reactions of 1 with 2a and 2b are hardly affected by addition of 1-propanol. Alcohol-like mixed aqueous solutions. Beyond the second critical mole fraction, kinetic parameters and stereospecificities of Diels-Alder reactions and standard chemical potentials of reactants, products and activated complexes show only a modest dependence on the solvent composition. Two observations are particularly interesting. First, the rate acceleration of the Diels-Alder reaction of 1 with 3b in mixed aqueous solutions is, except in dilute aqueous solutions, governed by a more favourable enthalpy of activation. Second, the

6. Diels-Alder reactions in aqueous media. Enforced hydrophobic interaction

solvent effects on the Diels-Alder reactions of 1 with 2a and 2b in mixed aqueous solvents appear to be virtually caused by the destabilisation of the reactants.

6.6 A quallritative analysis of solvent effects on Dieh-Alder reactions in mixtures o f water

and monohydric nlcohoIs

Treatment of the data and sources of error. The experimental quantities necessary to obtain the affinity parameters 4, and & have been outlined in Section 6.3 and the crucial role of Q was emphasised. The uncertainty in Q affects the affinity parameters with respect to sign and magnitude.

Figure 6-11. Isobaric activation parameters for the Diels-Alder reaction of 1with 2a as a function of the mole fraction of water in mixtures of water and 1-propanol; A*GO (+), A*HO( 0 ) and -TA'SO ( 0 ) at 25OC. The values of A'GO have been displaced downwards by 40 kJ mol-' for clarity.

6. Diels-Alder reactions in aqueous media. Enforced hydrophobic interaction

[endo] / [exo]

Figure 6-12. Endo/exo product ratio for the Diels-Alder reaction of 1 and 2a in and 2-methyl-2mixtures of water and methanol, A ;ethanol, ;I-propanol, 0 propanol, 0 as a function of the mole fraction of water, at 25OC.

The impact of errors in Q is highlighted when this quantity approaches zero. If the solvent is thermodynamically stable, Q must fulfill the thermodynamic stability condition (Q>O for 0 1 ~ ~ 1 1 )If . Q is zero, phase separation will occur. Furthermore, lim(x,-+1)Q=lim(x2~l)Q=l. The most important source of errors results from the fitting of the molar Gibbs energies of mixing, which have been either retrieved from literature or calculated from published vapour pressure data. Unfortunately, excess properties of mixed aqueous solutions of some monohydric alcohols have either not been determined at 25OC or are rather inaccurate. This puts severe demands on the fitting procedure. We have used Equation 6-9 to fit the dependence of the excess molar Gibbs energies of mixing on the composition of the mixed aqueous solvent. The results are shown in Figure 6-13. If the mole fraction of the monohydric aIcohols increases, the occurrence of small, short-lived alcohol-rich and water-rich domains becomes increasingly important. The non-ideal behaviour of the mixed solvents is most pronounced at the mole fraction were Q is closest to zero: x2=0.4 and x2=0.2 for aqueous mixtures of ethanol and 1-propanol, respectively. For 1-propanol the condition of thermodynamic stability was not fulfilled, even though the mixture is completely miscible.

4 DieIs-Alder reactions in aqueous media Enfmed hydrophobic interaction

. .

0.0

0.2

0.4
x (H,O)

0.6

0.8

1 .O

Figure 6-13. Q (see text) for mixed aqueous solutions of 1-propanol, as a function of the mole fraction of water, at 25C.

and ethanol,

Since the fitting procedure itself gave satisfactory results, the thermodyrnamically inconsistent result must be attributed to errors in the thermodynamic parameters. Matteoli et a1.191 used fitting procedures which impose the condition of thermodynamic stability. We chose not to manipulate the vapour pressure data to obtain positive values of Q across the whole mole fraction range. The partial molar volumes VmOand V, contribute significantly to the affinity parameters. We used volumetric data having an accuracy of at least k0.5 cm3 morl over the whole concentration range. The errors in V , and VmO propagate in the values of A, and A, proportionally to 1/Q and xJQ, respectively. This shows that the accuracy of the volume parameters becomes important when Q approaches zero. The dependence of the kinetic data on the composition of the mixed solvent was approximated by cubic splines (see experimental part). The uncertainty in the firstorder derivative aA*G0(x2)/ax, contributes significantly to the error in the affinity parameters. Although the absolute values of the first-order derivatives are subject to a considerable error, the trends in the kinetic data are reflected accurately by the fitted spline. Similar considerations apply to the analysis of the transfer parameters. An additional source of error is due to the fact that the volumes of activation are

4 Die&-Alder reactions in aqueous media. Enforced hyirophobic interaction

not known for the reactions studied in this chapter. Volumes of activation for DielsAlder reactions in water have, to our knowledge, not been determined. The error in the estimated volumes of activation may lead to a considerable error in the affinity parameters only when 4, and A, are small. In addition, the volume of activation of organic reactions in aqueous solutions depends on the composition of the mediumu1. Generally, the effect of these errors is negligibly small if the affinity parameters are large. Results and discussion. The kinetic solvent effects are understood in terms of preferential solvation of both the reactants and the activated complex by one or the other of the components of the solvent mixture. The term "preferential solvation" means that the composition of the cosphere of the reactant and activated complex A, and 4 are determined by differs from that of the bulk solvent. The values of 4,, Kirkwood-Buff integral functions GIs and G, which are determined by the composition of the solvent in close proximity of the solute S. In Figure 6-14, values of 4, Akl and k , are plotted as a function of the mole fraction of water for the Diels-Alder reactions of 1 with 2a and 2b in ethanol-water mixtures. The solvent dependences of the affinity parameters for the Diels-Alder reaction of 1 with 3b in mixed aqueous solutions of ethanol, are shown in Figure 6-15. For the cycloadditions of 1with 3a and 3c, the solvent dependences of Ak, 4. and Ala are very similar. Analysis of solvent effects on the rate constant of Diels-Alder reactions in aqueous solutions of 1propanol is seriously hampered by the fact that Q becomes zero near x2=0.2. This thermodynamic inconsistency leads to asymptotic results for the affinity parameters ( & . a , or -00 for x2+0.2). An example is given in Figure 6-16, where the affinity parameters are plotted for the Diels-Alder reaction of 1 with 2a in aqueous solutions of ethanol and 1-propanol. In a l l plots, 4 increases dramatically with increasing x2, which points to a marked difference between the composition of the solvation shells and the bulk solvent. It is difficult to interpret the solvent dependence of 4. In particular for a bimolecular process 4 is determined by a series of integral functions. Further insight into the factors, affecting the solvent dependence of the rate constant on x2 is provided by 4, and &. Recalling that and A, are given by [GI,-GI,-GI,] and [G,,-G,-G,], respectively, where A and B are the diene and the dienophile, the solvent dependence of 4 can be interpreted in terms of a decreasing preference of the reactants for the cosolvent upon forming the activated complex. More information was obtained by analysing the standard chemical potentials of 1, 2a, 2b and the corresponding activated complexes in aqueous mixtures of 1-propanol. In Figure 6-17, the dependence is analysed of the standard chemical potentials of (i) 1, (ii) 2a, (iii) the initial state (1+2a) and (iv) the corresponding activated complex on the composition of the aqueous solution of 1-propanol by plotting [GIs-G,] as a function of the mole fraction of water. Unfortunately, the standard chemical potentials could not be analysed near x2=0.2. Clearly, however, reactants as well as activated complex are preferentially solvated by 1-propanol when the mixed aqueous solvent approaches x2=0.2. An important point to note is that the magnitude of the Kirkwood-Buff integral function, G,, is strongly determined by the tendency of the cosolvents to aggregate. p2G, (p2 is the number density of the cosolvent) represents the average excess (or

6. Diels-Alder reactions in aqueous media Enforced hydrophobic interaction

deficiency) number of cosolvent molecules in the whole space around molecule S with , does not only reflect the presence of the cosolvent respect to the bulk. Therefore, G in the first solvation shell but in the whole cosphere. When the mixed solvent is characterised by the presence of short-lived alcohol-rich and water-rich domains, the composition of the cosphere of the apolar solute is largely determined by clustering of the organic cosolvent. Hence, if Q of a mixed aqueous solvent approaches zero, and the solute S has a slight preference for an apolar environment, [GI,-G,] becomes strongly negative (see Figure 6-17). Nevertheless, the solute is not solely solvated by the organic cosolvent and still experiences the presence of water. The activated complexes of the Diels-Alder reactions of 1 with 2a and 2b exhibit chameleon-behaviour and are indifferent to the composition of the mixed aqueous solvent (see Figure 6-9 and 6-10).

Figure 6-14. Affinity parameters 4, 4, and A, for the Diels-Alder reactions of 1 and of 1with 2b, 0 in mixed aqueous solutions of ethanol as a function with 2a, of the mole fraction of water, at 25OC.

6 DieLr-Alder reactions in aqueous media. Enfotced hydrophobic interaction

Figure 6-15. Affinity parameters 4 , 0 and A, for the Diels-Alder reaction of 1with 3b in mixed aqueous solutions of ethanol at 25C.
Although the activated complex is preferentially solvated by 1-propanol in the waterlike aqueous solution of 1-propanol its standard chemical potential is hardly affected (see Figure 6-16). Consequently, the solvent dependence of [GI,-G2,] is much smaller Hence, the increase in the Gibbs energy of activation for the than that of [Gl~-GZIs]. Diels-Alder reaction of 1with 2a in the water-like aqueous solutions of 1-propanol is determined by preferential solvation of the reactants by the organic cosolvent. Interestingly, the dependences of the affinity parameters 4, 4, and k, on the composition of the aqueous solution of ethanol show extrema which are less pronounced if more hydrophobic dienophiles are involved in the Diels-Alder reaction (see Figure 6-14). This can be explained by the fact that the affinity parameters are determined by the quotient [i3A*G0(xz)l&z]lQ. Preferential solvation of the reactants can lead to a pronounced increase in the Gibbs energy of activation before Q has reached its lowest value. Subsequently, the affinity of the organic cosolvents for the reactants is more pronounced in dilute aqueous solutions if the reactants are more hydrophobic (see Figure 6-14). It is not inconceivable that the apolar reactants even induce the formation of microdomains.

6. Dieh-Alder reactions in aqueous media. Enforced hydrophobic interaction

Figure 6-16. Affinity parameters 4 (closed symbols) and A, (open symbols) for the Diels-Alder reaction of 1with 2a in mixed aqueous solutions, containing ethanol, and 1-propanol W as a function of the mole fraction of water, at 25OC.

Unfortunately, the applied method for quantitative analysis of solvent effects on DielsAlder reactions is rather inaccurate in dilute aqueous solutions. The most serious problem in extracting the affinity parameters is that one needs several sets of thermodynamic and kinetic data to obtain these parameters. The accuracy of the analysis can be significantly improved if Kirkwood-Buff integrals could be determined directly from an experiment. A considerable improvement might be provided by a method, reported recently, to directly measure the property Q332.An interesting development is reported by Bot et who advance methods to determine Kirkwood-Buff integrals on the basis of Rayleigh-Brillouin light scattering data. The cubic spline approximation of the dependence of the Gibbs energy of activation on the composition of the medium can be improved by the determination of more data points. Finally, it is desirable to determine the dependence of the volume of activation of the reaction on the compositon of the mixed aqueous solvent.

4 --Alder

reactions in aqueous media. Enfwed hydrophobic interactton

Nevertheless, the simplified approach, presented in this section, has considerable potential in explaining solvent effects on organic processes in mixed aqueous solvents.

6.7 Hydmphobic e

on DieIs-Alder readions in water

Water is a bad solvent for apolar solutes. Yet, water is able to dissolve to some extent the apolar reactants, involved in many Diels-Alder reactions by forming an extensive hydrophobic hydration shell (see also Section 8.3). If the interactions between the reactants are extremely favourable, water is not capable to prevent aggregation of these compounds.

Figure 6-17. Analysis of standard Gibbs energies of transfer of 1, 2a, initial state and activated complex from 1-propanol to aqueous mixtures of 1-propanol at 25OC. [G,,G,,] as a function of the mole fraction of water where S is 1, A ;2a, 0 ;initial state (1+2a), R and activated complex, w , respectively.

6. Diet%-Alderreacriom in aqueous media. Enfoced hydrophobic interaction

Stacking of organic solutes in aqueous solutions is a well-known phenomenon and has been described extensively for organic dyes and nucleosides. These molecules are Generally, association of solutes in organic characterised by extended .rr-sy~tems~~'. solutions as well as in water is counteracted by an unfavourable entropy factor. In water, hydrophobic hydration shells also counteract aggregation at room temperature (see Section 8.3). The present results (Section 6.5) provided no indication of homotactic aggregation or stacking of the reactants at the concentrations used for the kinetic experiments. Heterotactic stacking of the reactants is also highly unlikely. Apparently, the pairwise hydrophobic interactions do not lead to aggregation of the reactants, because the Gibbs energy of this process is unfavourable. This immediately explains the ambiguity of the term "hydrophobic packing" as the explanation of the rate enhancement of Diels-Alder reactions in water. An additional argument against "hydrophobic packing" of the reactants is provided by the fact that intramolecular Diels-Alder reactions are accelerated by a similar large factor in water and aqueous solutions (see Chapter 7). A further point of concern is the fact that pairwise hydrophobic interactions do not necessarily lead to a complex that possesses the right geometry to undergo a cycloaddition reaction. Inhibition of the Diels-Alder reaction by hydrophobic interactions, leading to homotactic complexes or heterotactic complexes of the wrong geometry for cycloaddition, as suggested by Schneider et a ~ . is ~~, not confirmed by the present results (Section 6.5). In spite of the fact that water is capable of preventing aggregation of apolar solutes when their concentration is sufficiently low, the apolar solutes do sense each others presence by pairwise and higher-order interactions. Pairwise and higher order hydrophobic interactions between the apolar reacting species might lead to a concentration dependence of the rate constant (a similar dependence was observed for the keto-en01 equilibrium quotient of 2,4-pentanedione). The frequency of these hydrophobic "encounters" is proportional to the concentration of the reactants and the molar Gibbs energy, associated with these interactions. However, the concentrations of the diene and the dienophiles, used for the kinetic experiments, are low. Consequently, pairwise interactions and, for certain, higher-order interactions are rare and the concentration dependence of the rate constant can be neglected. A bimolecular reaction necessarily involves the formation of a direct, solventunseparated complex of diene and dienophile during the activation process. When two relatively apolar reactants are involved in this process and the medium is highly aqueous, we might speak of an "enforced hydrophobic interaction". The term "enforced" is introduced here (i) to emphasise the fact that the complexation of the diene and the dienophile is not necessarily favourable in terms of Gibbs energy and (ii) to reflect the forced geometry of the association process. In order to account for the exceptionally large rate acceleration of Diels-Alder reactions in aqueous solutions, the "enforced hydrophobic interaction" of diene and dienophile has to be more favourable (or more appropriately, less unfavourable) in water than in organic solvents where we might speak of an "enforced solvophobic interaction". The Gibbs energy, associated with "enforced hydrophobic (solvophobic) interactions'' is given by1!

6 Diels-Alder reactions i n aqueous me&

Enforced hydrophobic interaction

The gradient of AG(R) provides the average force between the solutes S. The first term represents the direct diene-dienophile pair potential, which is the work required in the process of bringing diene and dienophile from an infinite separation to a distance R in vacuum. The second term takes account of the solvent effect on the same associative process. The first term is similar in every solvent. The exceptionally large acceleration of Diels-Alder reactions in water must therefore be fully attributed to the second term. This solvent effect has to be explained in terms of the transfer of the reactants and activated complex from an organic solvent to water. The standard Gibbs energy of transfer of 1, 2a and 2b from 1-propanol to water is highly unfavourable (Section 6.5, Figures 6-9 and 6-10). The standard chemical potentials of the activated complexes are almost independent of the composition of the mixed aqueous solvent and are similar in monohydric alcohols and in water. Apparently, the "enforced hydrophobic interaction", involving (i) the enforced pairwise interaction between the reactants (the complex might, tentatively, be identified as the CT-complex), (ii) the volume contraction during the bond formation step and (iii) an increase in polarisability, dramatically reduces the hydrophobic surface of the reactants. The chameleon-like behaviour of the activated complex in mixed aqueous solvents (see Figures 6-9 and 6-10) suggests that water induces a more polar activated complex that can be optimally accomodated in the aqueous medium. Recently, Jorgensen et a1.352 reported computer simulations of the Diels-Alder reaction of 1 with 2a. This study supports the idea that enhanced polarisation of the activated complex in aqueous solution contributes strongly to the acceleration of aqueous Diels-Alder reactions. This process has important implications for the isobaric activation parameters of Diels-Alder reactions in water. The "enforced hydrophobic interaction" of the reactants is accompanied by destructive overlap of the hydrophobic hydration shells of the reactants. This phenomenon clearly distinguishes water from organic solvents and leads, at room temperature, to a favourable entropic contribution, which is almost fully canceled by an equally unfavourable enthalpy term (see Section 8.3). Consequently, destructive overlap of hydrophobic hydration shells causes both T A ' S O and A'HO to be equally increased in water with respect to organic solvents. The solvent effect, associated with the desctructive overlap of hydrophobic hydration shells is negligibly small. The actual solvent effect is due to the fact that the London dispersion interactions which stabilise the initial state (both reactants) in organic solvents, are absent in water. During the activation process the hydrophobic surface of the reactants is dramatically reduced. In water, the interactions between water and the activated complex might be enhanced by induced polarity of the activated complex. Consequently, interactions of the activated complex with water and organic solvents are similar. Hence, the A'HO is more favourable (or, more correctly, less unfavourable) in water than in organic solvents. In summary, the enthalpy of activation of DielsAlder reactions in aqueous solutions is affected by destructive overlap of the hydrophobic hydration shells of the reactants as well as by the lack of interactions between the reactants and water. Both contributions are counteractive. Whether the enthalpies of activation of Diels-Alder reactions in organic solvents are more or less favourable than in water depends on the magnitude of both contributions. Our results (see Table 6-1) suggest that the contribution of the unfavourable enthalpy term, due to the destruction of hydrophobic hydration shells, is surpassed by the favourable enthalpy contribution. The entropy of activation of Diels-Alder reactions in aqueous solutions

6 Diels-Alder reactions in aqueous media. Enforced hydrophobic interaction

differs from that in organic solvents mainly due to overlap of hydrophobic hydration shells and is therefore more favourable in water than in organic solvents (see Table 61). The increased preference for the endo-product of Diels-Alder reactions of 1 with 2a and 2b can be explained by the fact that formation of the endo-product is characterised by a more negative volume of activation and a concomittantly more pronounced reduction of the hydrophobic surface area. Recently, Diederich et a1.= showed that complexation of organic compounds in large, apolar guest-molecules was more effective in water than in organic solvents. As for the isobaric activation parameters of Diels-Alder reactions, the enthalpy as well as the entropy of complexation were found to be more favourable in water than in organic solvents. In this case, the enthalpy term was shown to dominate the overall solvent effect. Apparently, the favourable enthalpy contribution is so large that it not only outweighs the favourable entropic contribution inspite of the competing unfavourable enthalpic term. This is not inconceivable considering the large decrease of solute-solvent interactions. Interestingly, a superficial analysis of the isobaric activation parameters suggests a major role of hydrophobic hydration in the exceptional acceleration of Diels-Alder reactions in aqueous solutions. A more sophisticated analysis reveals that the molecular origin of the exceptional rate acceleration is provided by the enthalpically less favourable interactions between the reactants and water, which are relieved during the activation process. The large rate constants for Diels-Alder reactions in aqueous solutions are an example of "hydrophobic acceleration" (see chapter 5). Section 8.3 will be devoted to a detailed discussion of hydrophobic effects and hydrophobic interactions in particular.

6.8 Hydrophobic effects on Diels-Alder reactions in mired aqueous solutions

Solvent effects on the rate constants of Diels-Alder reactions in mixed aqueous solvents reflect the interactions of cosolvents with the diene, the dienophile and the activated complex, respectively. In dilute aqueous solvents (Section 6.5), pairwise hydrophobic interactions govern the interactions of the cosolvent with the reacting species and the activated complex. The extreme dependence of the entropy and enthalpy of activation for the Diels-Alder reaction of 1with 3b on the molality of 1propanol (Figure 6-11) is governed by two effects. First, at the so-called "magic mole fraction" of cosolvent, which depends on the size and hydrophobicity of the cosolvent, all water molecules are part of an extended hydrophobic hydration shell (see Section 4.2). Beyond this mole fraction the overlap of the hydrophobic hydration shells of the reactants, accompanying the activation process, cannot lead to release of "hydrophobic will be dramatihydration water" to the bulk solution. Consequently, A'HO and TA*SO cally decreased at the magic mole fraction. A second contribution is due to the fact that the hydrophobic surface of the reactants decreases significantly during the activation process. Pairwise interactions between the organic cosolvent and the reactants involve destructive overlap of the hydrophobic hydration shells. Since the activated complex is less hydrophobic, the pairwise interactions also result in a gradual decrease of A*HO and TA*SO (see also Section 4.3).

6. Diels-Alder reactions in aqueous media. Enforced hydrophobic interaction

We anticipated a reduction of the Gibbs energy of activation due to these pairwise interactions. However, the present results show that in some cases Diels-Alder reactions are even accelerated due to the addition of organic cosolvents (see, e.g. Figure 6-5). The interpretation of this effect must be that addition of 1-propanol also has a competing favourable effect on the "enforced hydrophobic interaction". Indeed, ~ e n - ~ a showed i m ~ ~that pairwise hydrophobic interactions are more favourable in the presence of small concentrations of an organic cosolvent. Consequently, the overall solvent effect on Diels-Alder reactions in dilute aqueous solutions will be small and it depends on the magnitude of the competing contributions whether addition of organic cosolvent will increase or decrease the rate constant. The onset of preferential solvation of the reactants in water-like mixed aqueous solvents (Section 6.6) cancels the driving-force for the "hydrophobic acceleration" of Diels-Alder reactions in dilute aqueous solvents and the Gibbs energy of activation increases dramatically. A'HO and TA'SO gradually approach values, found in the pure alcohol. The complete disappearance of aqueous phases inaugurates the alcohol-like mixed aqueous solvents. Aqueous solutions of surfactants and N-cyclohexylpyrrolidone are, already at very low concentrations, characterised by the presence of micelles and small clusters, respectively. Although the diene and the dienophile are preferentially solvated in these phases and the local concentration of the reactants is considerably enhanced, the reduced rate constant in these apolar environments reverses this concentration effect. Hence, rate constants of Diels-Alder reactions decrease sharply due to addition of surfactants above the CMC or organic cosolvents that tend to aggregate at low concentration.

6 . 9 Experimental part

Materiat. 1,6Naphthoquinone (3s) and 5-hydroxy-1,4-naphthoquinone (3b) were commercially available (Aldrich) and were recrystallised from methanol; mp 121122OC and 162-163OC, respectively (litm 123C and 163OC, respectively). 5-Methoxy1,4,-naphthoquinone (3c) was prepared from 3 4 as reported in the literature299,and was recrystallised from methanol; mp 185-186C (lit.m mp 187'C). Methyl vinyl ketone (2a, bp 81C) and ethyl vinyl ketone (2b, bp 102-103C) (Janssen) were freshly distilled before use. Cyclopentadiene (1) was prepared from its dimer immediately before use. Dimineralised water was distilled twice in an all-quartz distillation unit. All cosolvents were of the highest purity available.

Kinetic measurements. Pseudo-first-order rate constants were determined by following the change in absorbance at appropriate wavelengths in a quartz UV cell (1 cm) that was placed in a thermostated cell compartment of a Perkin Elmer A2 spectrophotometer, equipped with a standard personal computer. For reactions in highly aqueous media with half-lives longer than about ten minutes, the cuvettes were carefully sealed to prevent evaporation of cyclopentadiene. Evaporation of cyclopentadiene may otherwise seriously hamper the kinetic measurements and can lead to dramatic errors

6. Die&-Alder reactions in aqueous media. Enforced hydrophobic interaction

in the determined rate constants, because the evaporation process is also approximately first-order. A recent example of the danger of measuring evaporation of cyclopentadiene instead of a rate constant for the Diels-Alder process are the exceptionally high rate constants of Diels-Alder reactions of acrylic acids with cyclopentadiene, reported by Schneider et a ~ . ~ . Reactions of 1 with 2a and 2b were followed by monitoring the decrease of absorbance of cyclopentadiene at 250-255 nm (dependent on the composition of the reaction mixture), using an excess of the alkyl vinyl ketone. About 5-8 pL of a concentrated stock solution of cyclopentadiene in 1-propanol was added to the reaction medium (3 cm3) containing a known concentration of alkyl vinyl ketone. Typical concentrations, used for the determination of rate constants were 1 x 1 0 ~ mol dms of cyclopentadiene and 3.5x10-~ mol dm" of alkyl vinyl ketone. Reactions of dienophiles 3a-c with 1 were followed by monitoring the decrease of absorbance of the dienophiles at, respectively, 333, 426 and 400 nm. An excess of 1 was used. Especially in highly aqueous solutions it was appropriate to add 1, dissolved in a very concentrated stock solution in the cosolvent, just before the measurement, followed by addition of 5-8 pL of a stock solution of the dienophile. Stock solutions were made in the cosolvent. After addition of the two reactants, the solution was stirred briefly with an external micro-stirrer. Typical concentrations, used in the kinetic runs were 6-8x10'~mol dm" of 1and 5x10" mol dm" of 3a-c. All reactions were followed for at least 5 half-lives and were perfectly first-order. A total of 200-600 data points were collected and pseudo-first-order rate constants were calculated by fitting the data points to an exponential function. Rate constants were reproducible to within 1%, except for those Diels-Alder reactions of cyclopentadiene with naphthoquinones in highly aqueous solutions, which were reproducible to within 3%. Isobaric activation parameters were calculated with the Eyring equation, for kinetic data obtained at at least six different temperatures. To prevent evaporation of 1, rate constants were measured in the temperature range 10-35C. Plots of ln[k/T] against 1 E were perfectly linear, and A*HOwas determined by least-squares regression. Finally, pseudo-first-order rate constants of all Diels-Alder reactions were also determined by employing different excess concentrations of 2a, 2b and 1, respectively. Second-order rate constants were reproducible to within 3%.
Product analysis. The Diels-Alder reaction of 1 with 2a and 2b were performed in aqueous solutions on a preparative scale. 0.01 moles of 1and 0.01 moles of 2a and 2b were dissolved in 1 kg of water and stirred during 15 minutes at room temperature. The aqueous solution was extracted with ether (2x75 ml). After evaporation of the ether, the norbornene products was obtained almost quantitatively. The reaction products were analysed by GC (Hewlett-Packard 5890 instrument, equipped with a 15 m wide bore HP1 fused silica column), GC-MS (Ribermag R-10-10C) and NMR (Varian VXR-300 (300 MHz) instrument). The NMR spectra matched spectra, that were reported in the literature. The endolexo ratios of the products were determined by NMR and GC, following procedures, reported by Segushi et a~.~". The retention times of the exo and endo products of the reaction of 1with 2b were 4.6 min. and 5.0 min., respectively, at 150C. The retention times of the exo and endo products of the reaction of 1with 2a were 4.5 min. and 5.0 rnin., respectively, at 120C. The endolexo product ratios were also determined by NMR. The characteristic signals for the

6. Diels-Alder reactions in aqueous media. Enforced hydrophobic interaction

products of the reaction of 1 with 2a are: 'H-NMR (CDCI,) 2.00 ppm (CH, endo) and 2.10 ppm (CH,, exo), and for the reaction of 1 with 2b: 'H-NMR (CDCl,) 1.01 ppm (CH, endo) and 1.06 (CH, exo). The results obtained by GC and by NMR were identical. No side reactions were observed. Transfer parameters. Vapour pressures of 1,2a, 2b and the product of the reaction of 1 with 2a in mixed aqueous solutions, containing 1-propanol, were determined by analysing the composition of the vapour above the solution. The concentration of the compound in the vapour phase is linearly related to the vapour pressure. The composition was determined by GLC (Hewlett-Packard 5890 instrument, equipped with a 15 m wide bore HP1 fused silica column). Solutions were prepared by weight. To ensure that there was a large volume of vapour above the solution, special wide flasks, sealed with self-sealing septum caps were used.
(i) Determination of homotactic association of the dime and the dienophiles. The flasks, containing one compound in increasing concentrations, were simultaneously placed in a thermostated water bath at 25 -cO.lC and allowed to equilibrate for an hour. To prevent condensation of the solvent (particularly of 1-propanol) on the inside of the septum cap, the flasks were submersed as far as possible in the water bath and the water was covered with an insulating sheet. Atmospheric pressure inside each flask was maintained by piercing a very narrow hypodermic needle through the septum cap. Vapour samples were taken using a sealed gas-syringe, that was kept at 2SC. After withdrawal of the vapour, the samples were expanded quickly to prevent condensation and immediately analysed by chromatography. The vapour above each solution was chromatographed three times. After each run, the syringe was cleaned carefully. The peak area was determined by integration. Peak areas were reproducible to within 3%. In a typical experiment, 200 pL of vapour was chromatographed. The peak area was plotted as a function of the molality of the compounds in water and in mixed aqueous solutions, containing 10 and 20 mole percent of ethanol. The applied concentration of the compounds in solution was selected on the basis of the volatility of the compound. Typical concentration ranges for solutions of 2a, 2b and 1were lo45x10-~mol kg". For cyclopentadiene in aqueous solution small deviations from linearity could be observed only at higher concentrations (>0.04 mol kg-'). However, linear plots were obtained in mixed aqueous solvents containing small amounts of ethanol. For the alkyl vinyl ketones no deviations from linearity were observed. (ii) Determination of the standard Gibbs energies of transfer. Appropriate concentrations were choosen to determine peak areas as a function of the composition of the mixed aqueous solvent. Concentrations were selected as such that (i) the peak areas, obtained for each run throughout the whole composition range, were similar, and (ii) the solutions could be treated as ideal. In this way, the dependence of the GLC-signal on the concentration of the compound did not affect the measurement. In addition, clustering or stacking of the solute did not hamper the analysis. In a typical experiment, ten solutions with increasing mole fractions of 1-propanol were analysed simultaneously. The peak areas were determined as described above. The peak areas were divided by the concentration of the solute, defined in mol dm". The data were then used to calculate the transfer parameters, as described in-section 6.3, for the

6 &Is-Alder reactions i n aqueous media. Enforced hydrophobic interaction

transfer of the solute from 1-propanol to mixtures of water and 1-propanol. Quantitative analysis. Excess Gibbs energies of mixing of 1-pro an01 and water were calculated based on data and methods reported by Dawe et a1.k The G~~ data are fitted as a function of temperature and mole fraction. This fitting method was used to estimate the excess Gibbs energies of mixing at 25OC. Excess molar Gibbs energies of mixing of ethanol and water were calculated using the vapour pressure measurements reported by ~ o b s o n ~ The ' ~ . dependence of the excess molar Gibbs energy of mixing on the composition of the medium was fitted by applying Equation 6-7 (see also ref.303). Q was calculated by evaluating the second-order derivative. Volumetric parameters for the aqueous mixtures of 1-pro an01 and water were calculated by using the density data reported by Dawe et al.' and Benson et al?lO. Volumetric parameters for the aqueous mixtures of ethanol and water were calculated, using the data reported by Benson et al.310. The volumes of activation for the reactions of cyclopentadiene and alkyl vinyl ketones were estimated, using volumes of activation reported by Segushi et al.311 for the Diels-Alder reaction of cyclopentadiene with acrylic acid:A'Vo= -30 cm3 mol-'. The volumes of activation for the Diels-Alder reaction of cyclopentadiene with the 5substituted-naphthoquinones were estimated, using volumes of activation reported by LeNoble et ~ I . ~ ~ ' : A * V O =cm3 - ~ ~mol-'. The dependence of ln[A*G0(x2)]and $(x,)-pr(x2=1) on x2 was fitted using a spline appro~imation~'~. A spline approximation is based on the generalisation of a 1 into subintervals with function f (x,) by partitioning the mole fraction range O l x , ~ common endpoints, called nodes (O=a,,<al, ........<a,=l). In each subinterval, the function f(x) is defined by a polynomial. Therefore, instead of approximating f(x) by one single function (as was tried, without satisfactory results), f(x) is approximated by a series of n polynomials. The functions f(x) thus obtained are called splines. Different splines can be defined. We have approximated the dependence of the kinetic data on the composition of mixed solvent by cubic splines. A cubic spline f(x) on the interval 01x2Sl is, by definition, a continuous function which has continuous first-order and second-order derivatives everywhere in that interval and is represented by a series of polynomials that, in each subinterval, do not exceed a degree of three. Generally, the data were accurate enough to be used as nodes, forcing the cubic spline to cross all data points. The occurrence of unrealistic ripples in the fitted spline were avoided by deleting strongly deviating data points. The approximation as well as the calculation of the first-order derivative were performed using a commercially available program.

7. Solvent eflects on intramolecular DieIs-Alder reactions in aqueous solutions

number of intermolecular Diels-Alder reactions. A model was developed which accounts for the exceptional reactivity of intermolecular Diels-Alder reactions in aqueous media. In summary, the acceleration of Diels-Alder reactions in aqueous solutions was ascribed to "enforced hydrophobic interactions" between the diene and the dienophile. The term "enforced hydrophobic interaction" emphasises the fact that diene and the dienophile do not aggregate spontaneously in water. Instead, the diene and the dienophile form an "enforced complex" with a well-defined geometry for the cycloaddition to occur. It was argued that water is able to prevent spontaneous aggregation of the diene and the dienophile by hydrophobic hydration of the reactants (see also Section 8.3). During the last decades, IMDA reactions of furan derivatives have received considerable attention. In the first part of this chapter, an overview will be given of mechanistic aspects of IMDA reactions in general and IMDA reaction of furan derivatives in particular, as far as they are relevant for the interpretation of the solvent effects. The stereochemical aspects of IMDA reactions will be emphasised. The second part of this chapter is devoted to the description of solvent effects on the IMDA reaction of compounds 2a-d. Particular attention will be paid to the stereochemical aspects of the cyclisation and the effect of the hydrophobicity of the alkyl substituent. Solvent effects on the IMDA reactions in mixed aqueous solutions containing ethanol and 1-propanol, respectively, are analysed quantitatively by applying the theory that has been developed in Chapter 2 and successfully applied in the analysis of solvent effects on bimolecular Diels-Alder reactions in Chapter 6.

7.2 Intramolecular DieLF-Alder reactions; an overview

The memorable publication of Diels and Alder230,in 1928, constituted the beginning of the success story of the Diels-Alder reaction. For synthetic organic chemists the Diels-Alder reaction presents almost unrivalled opportunities for regioselective and stereospecific introduction of multiple centers of configuration. Many publications have established the utility of Diels-Alder reactions in the synthesis of natural products and valuable synthons. The intramolecular version of the Diels-Alder reaction is over twenty years younger than its bimolecular counterpart. Alder reported the first example of an IMDA reaction233. In 1963, Brieger deliberately sought to apply the IMDA reaction to natural product synthesis3''. Since then, interest in IMDA reactions increased almost exponentially. An extensive list of publications (see below), in which IMDA reactions are used to construct polycyclic ring systems, showed that IMDA reactions are a preparatively useful extension of the bimolecular Diels-Alder reactions. In this chapter, we are concerned with [4+2] cycloadditions. IMDA reactions are characterised by the fact that the diene and the dienophile are constrained in the same molecule. When the reacting substrates are themselves cyclic and/or have ring substituents, complex molecules can be formed in one single step (Scheme 7-2). The enhanced reactivity in comparison to their bimolecular analogues, the regioselectivity and the stereospecificity account for the widespread use of internal cycloadditions in the synthesis of natural products. An indication of the activity in the field of IMDA reactions is the frequency with

7 .Solvent effects on intramolecular DieIs-Alder reactions in aqueous solulions

which review articles and monographs309 have been published recently: 1974~'~, 1976~'~, 1977315,l98O3l6, 1984317218,1987319,and 199d9".

ortho

meta

Scheme 7-2
IMDA reactions can be classified into a number of major reaction types, based on the structure of the diene moiety: (i) acyclic dienes, (ii) endocyclic dienes, (iii) aromatic dienes (iv) o-quinodimethanes and (v) aficyclic systems. Examples of these reaction types are schematically shown in Scheme 7-3.

Scheme 7-3
Examples of IMDA reactions of a clic dienes are widespread and are found in the synthesis of many natural products3 This reaction type is used to construct 5,6-, 6,6-, 6,7-, 6,12- and even 6,14-ring systems. IMDA reactions of endocyclic dienes involve the intramolecular cycloaddition of 1,3-dienes which are partly or wholly incorporated in a ring. The chemistry of aromatic dienes is dominated by styrene and its congeners. The in situ generation of ortho-quinodimethanes which are trapped intramolecularly by a dienophile was developed by Oppolzer and ~ e l l e ?and ~ has been studied extensively, particularly for the synthesis of steroids321.Reports of IMDA reactions of alicyclic systems are still rare316. IMDA reactions are generally limited to additions in which the diene and the dienophile are linked by a bridge of three to four atoms. The longer and more flexible the bridge becomes, the more IMDA reactions resemble the analogous bimolecular reactions. IMDA reactions are exothermic. The entropy of activation is still negative,

r'.

% Solvent efJects on inhamofeeular&&-Alder r e a c h in aqueous sohtiorrs

though considerably less negative than that of its birnolecular counterpart. As a result of this "entropic assistance", the Gibbs energy of activation of IMDA reactions is 2230 kJ mol-* more favourable than for analogous bimolecular reactions. Although IMDA reactions occur generally in a better yield and require milder conditions, no IMDA reactions have been observed for systems which do not react intermolecularly. The strain, accommodated during the activation process of many IMDA reactions can be considerable and is dependent on the size of the ring. This strain primarily governs the success of an IMDA reaction, though formation of highly strained ring systems is not uncommon. Products of IMDA reactions are, however, frequently sensitive to retro Diels-Alder processes and therefore many IMDA reactions are reversible3''. The volume of activation for IMDA reactions is negative, and though it is enerally less negative than for the intermolecular Diels-Alder reactions, Isaacs et al?'showed that also IMDA reactions can be considerably accelerated by applying external pressure323. Ring fusion is determined by a complex interplay of conformational, steric and electronic effects, which may vary independentl?l9. The regioselectivity or direction of IMDA reactions is determined by the number of bridge atoms and the conformation of the diene. As shown schematically in Scheme 7-2, the intramolecular addition can lead to addition in a "ortho" or "meta" sense3". The preference for the ortho or meta addition is mainly governed by the length of the bridge between diene and dienophile. In addition, the stereospecificity of the IMDA reactions is of considerable importance. Craig has reviewed stereochemical aspects of IMDA reactions319. IMDA reactions and intermolecular Diels-Alder interactions involve synchronous bond formation with characteristic stereospecificity (see ref. 232 for a review on these mechanistic aspects of intermolecular Diels-Alder reaction). IMDA reactions exhibit supplementary features which differentiate between the endo- and ao-mode of addition.

endo trans-diene

cis-fused

H
exo trans-fused

cis-diene exo

cis-fused

Scheme 7-4

7. Solvent effects on intramolecular DieIs-Alder reactions in aqueous s o l u ~

The terms endo and exo refer to the orientation of the dienophile activating group with respect to the diene function. The conformation of the diene is extremely important for the stereochemistry of the IMDA reaction. The conformation of the diene can be cis or frans (see Scheme 7-4). For the cis-diene, the dienophile chain is forced into an exo-mode of attack since the endo-mode of attack is highly strained. This requirement leads to a ck-fused adduct. For the trans-diene, which is thermally more stable, both orientations are possible. For IMDA reactions secondary orbital overlap is not decisive and the endo-rule, which is generally valid for the description of the stereospecificity of intermolecular Diels-Alder reactions, fails. Steric effects cause a moderate but significant preference for the ao-conformation, leading to a pans-fused product (Scheme 7-4). By modification of the bridge, however, both transand cis-fused products can be obtained. The regioselectivity and stereospecificity of IMDA reactions rely on kinetically-controlled reactions. The major products might well be thermodynamically less favourable. To avoid loss of stereospecificity and regioselectivity, IMDA reactions are carried out at the lowest possible temperature. Finally, the stereospecificity can, as for intermolecular Diels-Alder reactions, be significantly affected by adding Lewis acids, which act as catalysts318.

7.3 Intramolecular DieIs-Al&r reactions o f furan derivatives

The substrate molecules studied in this chapter belong to the class of endocyclic transdienes. As shown in Section 7.2, the IMDA reaction of these compounds can lead to both trans- and ck-fused products. Since the length of the bridge between the diene and the dienophile is short, only "ortho" products are expected. In 1966, Bilovic et reported the intramolecular cyclisation of the furan derivative 2a (see Scheme 71). Ever since, many other cases have been reported of cyclisations involving a variety In Scheme 7-5, reported variations in the of furfuryl derivatives2351236S22325-3279. substrate structure are ;horn schematically. Bimolecular Diels-Alder reactions of furan and furan derivatives with olefins that are activated by only one electronwithdrawing substituent, take place slowly with low yields of usually mixtures of isomers, and are often reversible3". Unfortunately, many IMDA reactions involving furans are reversiblew3279328. Nevertheless, the scope of IMDA reactions of furan derivatives has been studied in detail (see Scheme 7-9, and a considerable number of compounds were shown to undergo IMDA reactions at ambient pressure. The success of IMDA reactions of these furan derivatives is strongly determined by steric effects. The intramolecular cycloaddition of the furan derivatives occurs only when the substrates can adopt a conformation in which the furan and the dienophile are geometrically disposed for T-overlap.

Scheme 7-5

Z Solvent fleets on intramolecular Die,%-Alderreactions in aqueous solutions

The presence of an ester function in the bridge between the diene and the dienophile (X in Scheme 7-5) forces the molecule into a favourable s-tram-conformation about the ester linkage (compare with Scheme 7-6, where X=NR-CO). Recently, Jung et a1.235showed that, in order to cyclise, the substrate has to rotate about the ester bond to give the s-cir-conformer. IMDA reactions of these compounds are therefore very slow, or not observed at A similar effect is possible when secondary and tertiary amide functions are present in the bridge (see Scheme 7-6). Substitution of alkyl groups for hydrogen atoms in the bridge that links the two reactive centers accelerates the IMDA reaction of these compounds. This gem-alkyl effect, also called the 'Thorpe-Ingold effect", favours the reactive rotamep. Compounds with an aliphatic bridge readily undergo intramolecular cyclisation. As shown by Parker et the presence of (i) hetero atoms, such as oxygen, (ii) a carbonyl amide in the bridge, do not hamper the IMDA reaction either. group or (iii) a tertia~y Increasing the length of the bridge decreases the reactivity, but 5-, 6- and 7-membered r i n g s can be easily formed. Recently, Hamilton et a1.236 showed that the IMDA reaction of a furan derivative where n=l, m=O, X=NR-CO, R,=R,=COOH and R,=R,=H is dramatically accelerated by synthetic receptors that bind via hydrogen bonding to the carboxyl groups. The substrate is then forced into a very reactive s-ch-position. In some cases, the stereospecificity and regioselectivity of IMDA reactions of furan derivatives has been studied. Compounds with an aliphatic bridge or with a carbonyl function in the bridge lead to a mixture of ch- and trans-fused products. The IMDA reaction of compound 2a was shown to yield exclusively the trans-fused product via an exo-mode of cyclisation (see Scheme 7-I)~". Solvent effects on IMDA reactions have not received much attention, since their effects were generally believed to be small and uninteresting. Recently, however, Jung et al.235 showed that the solvent effect on the IMDA reaction of furan derivatives k , being as large as 3200). This solvent effect has where X=-OCO- is substantial ( been attributed to the fact that the substrates prefer the s-trans conformer, whereas the activated complex necessarily adopts the s-cir conformation. Due to the overlap of dipoles in the s-ch conformation, the dipole moment of the activated complex is significantly larger than that of the starting material and therefore polar solvents accelerate the cyclisation. Interestingly therefore, the reverse process does not experience a major solvent effect. Solvent effects on IMDA reactions are predominantly caused by changes in the dipolarity of the substrate during the activation process.

s-cis

s-trans

Scheme 7-6

Z Sobent effects on intramolecular Diet%-Alderreactions in aqueous solutions

Since many IMDA reactions are characterised by substantial structural changes on going from starting material to the product, solvent effects can be quite significant. Williams et al." reported some details of an IMDA reaction in aqueous solution. It was suggested that the aqueous solution promotes coiling of the lipophilic chain, linking the reactive groups. Consequently, the regioselectivity of the IMDA reaction was significantly altered. The authors did not report solvent effects on the reaction rate of the IMDA reaction.

7.4 Synthesis and intrnmolecular Dick-Alder re&n

o f N-furfwyl-N-alkylmaleamic a c i d s

N-Furfuryl-N-alQlmaleamic acids (2s-d) were prepared by coupling the appropriate N-furfuryl-N-al$l amines to maleic anhydride at room temperature (see Scheme 7-1). Upon standing, a mixture of the amine and maleic anhydride deposited the acids 2. The maleamic acids immediately convert into the cyclised products, the 2-alkyl-3-0x05 , 7 a - e n d o o x o - 3 a , 4 , 5 , 7 a - t e t r a h y d r o - 4 - i s o ~ cacids (3a-d). The intermediate products 2a-d could not be isolated in pure form. Progress of the IMDA reaction of 2a-d was followed by dissolving the substrates in CDCl, and monitoring the conversion by 'H-NMR. The subsequent spectra, taken at selected time intervals, confirmed the occurence of 2a-d as intermediates and revealed a smooth conversion of the compounds into the cyclised products 3a-d (see Experimental Section). Acids 2a-d exist in both the s-trans and the s-cis conformation. In order to cyclise, the s-tram substrates must also adopt the s-cir conformation (Scheme 7-6). Rotation about the tertiary amide bond is sufficiently slow to allow the determination of the equilibrium constant by NMR (see Experimental Section). Equilibrium constants for the s-trans-s-cis equilibrium of compounds 2a-d in CDC13 are given in Table 7-1.

trans-fused p r o d u c t Scheme 7-7

cis- fused

product

Apparently, the s-trans and the s-cis conformers are present in almost equal amounts. Increasing the size of the alkyl substituent on nitrogen favours the s-cir conformation. Conversion of the s-trans conformer into the reactive s-cir conformer is not rate determining since the rate of disappearance of both conformers was equal. The reaction products were exclusive1 formed by addition in the "ortho" sense, as had been established by Bilovic et al?' for a substrate in which R=phenyl. The stereospecificity of the IMDA reaction of the substrates has been determined by

Z Solvent gects on intramolecular Diek-Alder reactions in aqueous solutions

assigning the configuration of the protons H, H, and He. In principle, the relative stereochemistry of the protons Hd and H ,is determined by the ch-conformation of maleic anhydride. Nonetheless, Bilovic et a1.324 showed that the endo- and exo-acid (Scheme 7-7) can easily interconvert. Fischer et al? reported methods for the characterisation of the ring stereochemistry of IMDA adducts of furan derivatives by 'H-NMR. The assignment was based on the coupling constants of the hydrogen atoms H, Hd and He. The differentiation between the endo and exo H, proton was derived from the coupling constant with Hc (J(Hd,-H,)=4-5 Hz and J(Hd,d,-Hc)=O Hz).In all cases the conformation of H,, was purely endo, which means that only exo-acids were obtained. The stereochemistry of He was assigned by the coupling constant for Hz and J(Hd,,,,-H,,)=3.5 Hz). For all the coupling with Hd (J(Hdcndo-Hecndo)=9 products, He was exclusively endo. The IMDA reactions of 2 a d therefore involve cyclisation via the exo-mode of addition, leading to the trans-fused adducts. With respect to the stereospecificitity of these IMDA reactions, it is interesting to note that the protons Hi and H, of the reaction products 3b-c are diastereotopic, which was shown by 'H-NMR (see experimental part). Apparently, the nitrogen is a chiral centre and the stereochemistry is determined by the conformation of the ring. Retro Diels-Alder reactions are well-documented235. Many IMDA processes involving furan derivatives are reversible and it was necessary to determine whether the reverse process was occuring in this case. The IMDA reactions of 2a-d in CDCl, were monitored for 24 hours. In all cases, the conversion was uantitative and no remaining maleamic acids could be detected. Although Jung et alqu have shown that this is not an absolute proof for the absence of any reverse reaction, our data confirm that the reverse process is at least extremely slow at room temperature. Consequently, the rate constant for the IMDA reactions can be determined by application of simple first-order kinetic analysis.

7.5 Solvent effects on the lMDA reaction o f N-fi&ryl-N-alkylmaleamic acids

The IMDA reaction of 2a was followed by 'H-NMR in a series of solvents. In Table 7-1 the equilibrium constant for the s-ck-s-trans equilibrium, found in these solvents, is reported. Solvent effects on the conformation of the substrate are moderate. Unfortunately, the equilibrium constant in D,O is difficult to determine accurately because conversion to the cyclised product is extremely fast. Nevertheless, it can be safely concluded that the equilibrium constant in aqueous solution is not dramatically different from those in organic solvents. Jung et al.= reported a considerable solvent effect on the conformational equilibrium of furan derivatives with an ester function in the bridge (see above). The conformational equilibrium of tertiary amide lacks, however, strong dipole effects that controle the conformational equilibrium of esters, which explains the small solvent effect on the conformational equilibrium of the maleamic acid 2a. Solvent effects on the conformation equilibrium of the other substrates are not expected to be dramatically different. The first-order rate constants and Gibbs energies of activation for the IMDA reactions of 2a-d where the solvents are water, ethanol and 1-propanol are given in Table 7-2.

7 .Solvent @ec@ on intramolecular &Lr-Alder reactions in aqueous solutions

Table 7-1. Equilibrium Constants ([s-cis]/[s-nuns]) for the Conformational Equilibrium of N-Furfuryl-N-alkylmaleamic Acids 2a-d and Percentages of s-tram and s-cis Conformers in a Series of Organic Solvents, at 25OC.

Compound

Solvent CDCl, Benzene-d6 DMSO-d6 Methanol-d Acetone-d6 D2O CDCl, CDCI, CDCl,

Determined by 'H-NMR at a concentration of ca.10-~mol accurately because of rapid conversion into 3a.

dm-,.

Could not be determined

Table 7-2. Rate Constants and Gibbs Energies of Activation for the IMDA Reactions of 2a-d in a Series of Organic Solvents and in Water at 25OC.

Compound

Solvent Ethanol 1-Propanol Hexane 1,4-Dioxane Dichloromethane Acetonitrile Trifluoroethanol Water Ethanol 1-Propanol Water Ethanol 1-Propanol Water Ethanol 1-Propanol Water

Rate Constant, sl

A *Go,

kJ mol"

Z Solvent effects on intramolecular &Is-Alder reactions in aqueous solutions

In addition, the rate constants and Gibbs energies of activation are reported for the IMDA reaction of 2a in a series of other organic solvents. The rate constants for
reaction in monohydric alcohols increase in the order 2a>2b>2c12d, and parallels the increased preference for the reactive s-czk conformer. The solvent effect on the IMDA reaction of 2a is moderate, and shows no correlation with the polarity of the soIvents. The exceptionally low reaction rate constants for the IMDA reaction in 1 , 4 dioxane and dichloromethane suggest some relation with the internal pressure T of the solvent, which is extremely high for these solvents. This trend seems to disagree with the negative volume of activation, that is generally found for IMDA reactions. The data set of solvent effects on IMDA reactions, given by Jung et a].=, is too smail to confirm any correlation with properties of solvents. The reaction rate constants for the IMDA reactions in aqueous solution are extremely large. The acceleration of the IMDA reaction is most pronounced for compounds 2a and 2b.

0.0

0.2

0.4

0.6

0.8

1 ,O

x(HZO)
Figure 7-1. Gibbs energies of activation for the IMDA reactions of 2a-d in mixed aqueous solutions, containing ethanol as a unction of the mole fraction of water at 25C; 2a, W ;2b, A ;2c, ;2d, a

7 .Solvent effectson intramolecular DieIs-Alder reactions in aqueous solutions

Figure 7-2. Gibbs energies of activation for the IMDA reactions of 2a-d in mixed aqueous solutions, containing 1-propanol, as a function of the mole fraction of water ;2b,A ;2c,+ ;2d, . at 25C; 2a,

Interestingly, compound 2d is completely soluble in water at the concentration required for the determination of the rate constant for the IMDA reaction. At these - ~ kg-') the substrate does not form micelles. In Figures 7-1 concentrations ( 1 0 ~ - 1 0mol and 7-2 the Gibbs energies of activation for the IMDA reactions of 2a-d at 25OC are plotted as a function of the mole fraction of water in mixed aqueous solutions, containing ethanol and 1-propanol, respectively. Two critical mole fractions are distinguished, separating three ranges of solvent composition. The general trends resemble the patterns, observed for intermolecular Diels-Alder reactions in mixed aqueous solution (see Chapter 6). However, the critical mole fractions are less pronounced for the IMDA reactions. In the dilute aqueous solution, the Gibbs energy of activation is only moderately affected by addition of cosolvents. In Section 4.4, the solvent effects on the IMDA reactions of 2a in dilute aqueous solutions were analysed quantitatively. The Gibbs

7. Solvent effects on intramolecular Diels-Alder reactions in aqueous solutions

energy of activation decreases more rapidly after the first critical mole fraction. These water-like mixed aqueous solutions (see Chapter 6) are limited by a second critical mole fraction. In the alcohol-like mixed aqueous media the Gibbs energies of activation decrease steadily as a function of the mole fraction of cosolvent, though in most cases less pronounced than in the water-like aqueous solutions. As for the intermolecular Diels-Alder reactions, the critical mole fractions are strongly dependent on the hydrophobicity of the substrate. The first critical mole fraction for solvent effects on the IMDA reaction of 2d in mixed aqueous solutions is, for example, difficult to detect, whereas the rate constant for the IMDA reaction of 2a is hardly affected over a large mole fraction range of cosolvent.

7.6 A quantitative analysis of solvent effects on intramolecular Diels-Alder reactions in mired aqueous solvents

Solvent effects on the IMDA reactions of 2a-d in mixed aqueous solutions, containing ethanol and 1-propanol, respectively, over the total mole fraction range were analysed by applying the Kirkwood-Buff theory (see Chapter 2). According to Equation 2-71, parameters A, were defined, which measures the relative affinity of the substrate, upon going from the initial state into the transition state, for the solvents present in the solvent mixture. 4 is governed by the thermodynamic non-ideality of the mixed solvent, quantified by the second derivative of the dependence of the molar Gibbs energy of mixing of the components on the composition of the solution. 4 is furthermore determined by the magnitude of the solvent effect as a function of the composition of the medium, expressed as the first derivative of the plot of ln[k(x,)/k(x,=O)] versus the mole fraction x , . Finally, 4 is related to the molar volume of the mixed solvent. In Section 6.3 the methodology is described in detail, showing how these subsequent parameters can be derived or calculated. The solvent effects on the IMDA reactions were analysed using a similar procedure. An extension of the theory enables the evaluation of solvent effects on the reactants and the activated complex of reactions in mixed solvents. According to Equation 2-75, parameters 4, and A , quantify the changing affinity of the reactant during the activation process for solvent 1 and solvent 2, respectively. Calculation of 4, and k, requires similar parameters to those used in the calculation of 4. In addition, the dependence of the partial molar volume of the cosolvent and the volume of activation of the reaction on the composition of the medium have to be determined. As shown in Section 6.3, the partial molar volumes of the cosolvents can be retrieved from the literature. The volume of activation for the IMDA reactions was estimated, based on a study of Isaacs et al.3n. The dependence of the volume of activation on the composition of the medium was neglected (see Section 6.3). In Figure 7-3, the affinity parameters 4, and f& are plotted as a function of the mole fraction of water for the IMDA reaction of 2a in mixed aqueous solvents containing ethanol. In Figure 7-4, a similar plot is given for the affinity parameters 4 describing the solvent effects on the IMDA reactions of compounds 2a-d. Analysis of solvent effects on the IMDA reactions in mixed aqueous solvents, containing 1propanol is seriously hampered by the large error in the affinity parameters near

7 .Solvent effects on intramolecular DieIs-Alder reactions in aqueous solutions

x,=O. The general patterns, observed for the dependence of the affinity parameters on the composition of the medium strongly resemble the dependence for the intermolecular Diels-Alder reactions. The change in preferential solvation of the reactant during the activation process is most pronounced if the mixed aqueous solvent exhibits extreme non-ideal thermodynamic behaviour. Interestingly, the dependence of A, on the composition of the medium is less pronounced if the reactant becomes more hydrophobic (see Figure 7-4). This is a consequence of the reduced sensitivity of the Gibbs energy of activation for the mole fraction of the cosolvent, described by dA'GOldx,. For compounds 2a and 2b, the solvent effects are almost similar, and the increase in hydrophobicity results in an even slightly more pronounced solvent effect. For compounds 2c and 2d, the sensitivity to addition of apolar cosolvents is strongly reduced.

Figure 7-3.Affinity parameters for solvent effects on the rate constant of the IMDA reaction of 2a in mixed aqueous solutions of ethanol as a function of the mole fraction ; h 1 0, ;&, of water at 25OC; 4,

7. Solvent effects on intramolecular Diels-Alder reactions in aqueous solutions

Figure 7-4. Affinity parameters A, for solvent effects on the rate constant of the IMDA reactions of 2a-d in mixed aqueous solutions of ethanol at 25OC; 2a,. ;2b,A ;2c,+ ;2d, .

The affinity parameters Akl and AM show that the solvent effects are a consequence of the preferential solvation of the reactant by the organic cosolvent, which emphasises the close similarities with solvent effects on the intermolecular Diels-Alder reaction in mixed aqueous solvents.

7.7 Solvent effects on intrumlecular DieIs-Alder reactions in mired aqueous solvents; discussion and comI~~ions

Rate constants for IMDA reactions of compounds 2a-d are greatly enhanced in water and dilute aqueous solutions. The exceptionally large reduction of the Gibbs energy of activation of the IMDA in aqueous solution, compared to organic solvents is compara-

7 .Solvent effects on intramolecular DM-Alder reactions in aqueous solutions

ble to the corresponding reductions that have been found for intermolecular DielsAlder reactions. Interestingly, the conformation equilibrium of compounds 2a-d in organic solvents and in aqueous solution shows that the s-cis conformation is not favoured. Apparently, cis-trans isomerisation does not lead to a significant reduction of the the hydrophobic surface which suggests that hydrophobic interactions between the T-electron systems do not contribute significantly to the stability of the s-cb conformer. The exceptionally large rate constant of IMDA reactions of 2a-d in aqueous solution cannot be attributed to an enhanced concentration of s-cb conformer. Apparently, these compounds do not spontaneously "coil" intramolecularly to reduce the interactions of the apolar moieties with water, which has been suggested for some substrates which undergo IMDA reactions in aqueous solutionz5.By analogy to the explanation of the exceptional solvent effect on bimolecular Diels-Alder reactions in highly aqueous solvents, "enforced hydrophobic interactions" also accounts for the exceptional rate enhancements of these IMDA reactions in aqueous media. The "enforced" intramolecular coiling reduces a considerable part of the unfavourable interactions between the reactant and water. Moreover, the partial molar volume of the reactant decreases during the bond formation step and the development of partial charges reduces the hydrophobic surface even further. The rather large difference between the solvent effects on the IMDA reaction of 2a and 2b on one hand and 2c and 2d on the other hand is surprising. With respect to compound 2d it might be argued that the conformation of the reactant in a conventional organic solvent is significantly different from that in aqueous solutions. At the concentrations, used for the kinetic experiments, no aggregation of the reactant occurs. The n-octyl substituent will, however, tend to minimise its apolar surface area in aqueous solutions. This makes a comparison of the reactivity of this compound in aqueous solution with the reactivity in organic solvents more difficult. It is anticipated that the octyl chain "solvates" parts of the reactive moieties of the reactant, reducing the interactions with water. This intramolecular "solvation" is less favourable if the reactive moieties have become more polar during the activation process. For compound 2d, this explanation is not appropriate. The molecular origin of the significantly smaller acceleration of the IMDA reaction of this compound in highly aqueous solutions remains yet unclear. Solvent effects on IMDA reactions in mixed aqueous solutions are governed by interactions between the organic cosolvents and the reactants and activated complex, respectively. In dilute aqueous solutions, pairwise interactions dominate the solvent effects and, consequently, solvent effects are small. As described in Chapter 6, an increase in the concentration of the cosolvent leads to preferential solvation of the reactant and an increased affinity of the cosolvents for homotactic interactions. Due to preferential solvation, the unfavourable interactions of the apolar reactant with water are reduced and the motive for the "hydrophobic acceleration" is undermined. In summary, this chapter shows that rate constants of unimolecular organic reactions of apolar molecules are also dramatically enhanced in aqueous solution, provided that a considerable part of its hydrophobicity is relieved during the activation process.

7.Solvent effects on intramdecular DieLF-Aldm reactions in aqueous solutions

7.8 Experimental section

Materials. Demineralised water was distilled twice in an all-quartz distillation unit. Solvents and cosolvents were of the highest purity available and were used without further purification. Known compounds were purchased from commercial suppliers. NMR spectra were recorded on a Varian VXR-300 istrument, with TMS as internal standard. Melting points were taken on a Mettler microscope connected to a Mettler FP1 melting and boiling point apparatus. The pH of reaction media was determined by using an Orion SA720 pH-meter, equipped with a Ross pH electrode. N-Furfuryl-N-methyl amine (la). Compound l a was prepared using a slightly modified literature procedure329,starting from 18 g (0.19 mol) of furfural and 75 g of an aqueous solution (25%) of methyl amine (0.60 rnol), followed by catalytic hydrogenation (2 g Pd/C, 5%). After extraction of the acidified aqueous layer with ether, the aqueous layer was made alkaline with 20% NaOH and extracted with chloroform. Yield, after distillation, 16.72 g (0.15 mol, 79 %) of la: bp 55OC (19 mm) (Lit. 58OC (24 rnrn)). Spectral data matched reported data324. N-Furfuryl-Nethyl amine (lb). Compound l b was prepared by procedures as for la. 24 g (0.25 mol) of furfural and 100 g of a 25% aqueous solution of ethyl amine (0.25 mol) and 2 g Pd/C 5% yielded, after distillation, 20.0 g (0.16 mol, 64%) of lb: bp 1 W C (54 mm). 'H-NMR (CDCI,) S 1.06 (t,3H), 1.35 (broad,lH), 2.65 (q,2H), 3.75 (s,2H), 6.14 (d,lH), 6.27 (dd,lH), 7.32 (d,lH) ppm. ',C-NMR (CDCI,) S 14.98 (p), 43.15 (s), 45.9 (s), 106.45 (t), 109.84 (t), 141.46 (t), 153.85 (q) ppm. Analysis calculated for CjH,,NO: C:67.17%; H: 8.86%; N:11.19%; found: C:66.60%; H: 8.87%; N:11.00%. N-Furfuryl-N-(n-pmpyl) amine (lc). Compound l c was prepared by refluxing 24 g (0.25 mol) of furfural and 32.5 g (0.55 mol) of n-propylamine in 100 mL of benzene, under removal of water for 1 hour. The subsequent procedure was the same as that for l a and l b using 2 g of Pd/C 5%. Yield, after distillation, 21.2 g (0.153 mol, 61%) of lc: bp 140C (45 mm); 'H-NMR (CDCI,) S 0.88 (t,3H), 1.37 (broad,lH), 1.46 (m,2H), 2.54 (t,2H), 3.73 (s,2H), 6.13 (d,lH), 6.26 (dd, lH), 7.31 (d,lH) ppm. 13CNMR (CDCI,) S 11.53 (p), 22.92 (s), 46.02 (s), 50.84 (s), 106.59 (t), 109.78 (t), 141.4 (t), 153.88 (q) ppm. Analysis calculated for CBH13NO:C: 69.03%; H: 9.41%; N:10.06%; found; C: 68.54%; H: 9.46%; N:10.37%. N-Furfhryl-N-(n-octyl) amine (Id). Compound Id was prepared by a similar procedure as for lc. 7.7 g (0.080 mol) of furfural and 31.2 g (0.241 mol) of n-octyl amine and 1 g Pd/C 5% yielded, after distillation, 8.36 g (0.040 mol, 50 %) of Id: bp ll5OC (3 mm); 'H-NMR (CDCl,) S 0.89 (t,3H), 1.26 (broad,l2H), 1.31 (broad,lH), 1.45 (t,2H), 2.60 (t,2H), 3.76 (s,2H), 6.15 (d,lH), 6.29 (dd,lH), 7.34 (d,lH) ppm. 13CNMR (CDCI,) S 13.98 (p), 22.55, 27.19, 29.14, 29.39, 29.91, 31.71, 46.18 (s), 49.11 (s), 106.49 (t), 109.87 (t), 141.49 (t), 153.98 (q) ppm. Analysis calculated for C,,H,NO: C:74.59%; H:11.07%; N: 6.69%; found; C:74.18%; H:11.45%; N: 6.79%. N-Furfuryl-N-methylmaleamic acid (2a). 100 mg (1 mmol) .of finely powdered

Z Solvent eecects on intramolecular Diels-Alder reactions in aqueous solutions

maleic anhydride was dissolved in 140 mg (1.26 mrnol) of l a on a ceramic plate. The solution deposited compound 2a upon standing. The solid was repeatedly washed with ether.
N-Furfuryl-N-ethylmaleamic acid (2b). This compound was prepared as described for 2a. 100 mg (1 mmol) of maleic anhydride and 156 mg (1.25 mmol) of l b yielded 2b which could not be obtained in analytically pure form. N-Furfuryl-N-(n-propy1)maleamic acid (2c). This compound was prepared as described for 2a. 100 mg (1 mmol) of maleic anhydride and 174 mg (1.25 mmol) of l c yielded 2c which could not be obtained in analytically pure form.

N-Furfuryl-N-(n-octy1)maleamic acid (2d). This compound was prepared by mixing 60 mg (0.5 mmol) of finely powdered maleic anhydride with 105 mg (0.5 mmol) of Id. 2d could not be obtained in analytically pure form.
Solids 2a-d were characterised by dissolving them in CDCI, and immediately recording a 'H NMR spectra. Although a considerable percentage (40%) of the product was already converted into the cyclised product (3a-d), the proton signals confirmed the occurrence of intermediate products 2a-d. The possibility of a bimolecular Diels-Alder reaction between maleic anhydride and the furan derivatives, eventually followed by amide formation, could be excluded. It was extremely difficult to derive the 'H NMR data of the pure compounds, since many signals overlap. In addition, the compounds are equilibrium mixtures of the s-ck and the s-trans form (see below). The coupling of the maleic anhydride and the furan derivative involving the formation of an amide bond was confirmed in all cases by appearance of the characteristic signals of the vinylic protons of the maleamic acid. Also the signals of Hf and H, were extremely useful for monitoring the coupling of the acid and the amine and the subsequent IMDA reaction. By following the IMDA reaction as a function of time, in all cases the disappearance of the characteristic vinylic protons (two doublets) of the maleamic acid and the furan-ring was observed. In addition, the chemical shift of H, and H, changed dramatically. Since the chemical environment of the proton H, and H, is different in the product, two doublets appeared.

2-Alky1-3-oxo-5,7a-end~xo-3a,4,5,7a-tetmhyd~isoind01ine-4-e~0~8rboxylic acids (3a-d). These compounds were prepared by dissolving in ethanol the solid, deposited from the solutions of maleic anhydride and the appropriate N-furfuryl-N-alkyl amines (see above). Compounds 3a and 3b precipitated from the solution, and compounds 3c and 3d were isolated by evaporation of ethanol after 10 hours. The compounds were recrystallised from ethanol (3a and 3b), n-hexane (3c) and dichloromethane/n-hexane 2o/m (3d). Yield of 3a, after recrystallisation, 55 mg (0.28 mmol, 28%); mp.180-182C (Lit.324 184-185C); 'H NMR (CDC13) S 7.79 (d,H,, J(~,nd,-Hd,nd,)=9.16 Hz), 2.87 (d,H,,J=9.16), 2.91 (s,3H), 3.75 (d,H,J(HrHg)=12.09 HZ), 4.01 (d,Hg,J=12.09 Hz), 5.23 (s,H,), 6.40-6.55 (m, Ha and H,) ppm. l3 C NMR (CDCI,) S 30.28 (p), 46.04 (t), 50.01 (t), 51.11 (s), 82.48 (t), 88.8 (q), 134.52 (t), 137.19 (t), 172.47 (q), 172.53 (q) ppm. Yield of 3b, after recrystallisation, 61 mg (0.27 mmol, 27%); mp 172-174C; 'H NMR

7 . Solvent effects on intramolecular Dieh-Alder reactions in aqueous solutions

(CDC13) 6 1.17 (t,3H), 2.90 (d,H,,J(H,,-Hd,)= 9.15 HZ), 2.84 (d, H,), 3.31 (m,H,), 3.51 (m,H,), 3.76 (d,H,J@H& 12.09 Hz), 4.02 (d,H,,J= 12.09 Hz), 532 (s,H,), 6.49 (s,H, and H,) ppm. C N (CDCI,) 6 12.18 (p), 37.91 (s), 46.09 (t), 48.4 (s), 50.39 (t), 82.44 (t), 88.83 (q), 134.58 (t), 137.22 (t), 172.15 (q), 172.58 (q) PPm. Yield of 3c, after recrystallisation, 105 mg (0.45 mmol, 45%); mp 162-165C; 'H NMR (CDCl,) 8 0.88 (t,3H), 2.83 (m,2H), 2.80 (d,&,J(&,d,-Hd,do)=9.15 H Z ) , 2.88 (d, Hd), 3.22 (m, Hi), 3.35 (m,Hh), 3.72 (d,H,J(HrH )=12.09 H Z ) , 3.96 (d,H ,J=12.09 Hz), 5.26 (s,Hc), 6.45 (.,Ha and Hb) ppm. ' C (CDCI,) 6 11.03 (p), $0.25 (s), 44.64 (s), 45.92 (t), 48.85 (s), 50.47 (t), 82.33 (t), 88.8 (q), 134.64 (t), 137.18 (t), 172.46 (49 172.89 (4 PPm. Yield of 3d, after recrystallisation, 40 mg (0.13 mmol, 26%); dec. 167C; 'H NMR (CDCI,) 60.85 (t,3H), 1.22 (m,lOH), 1.55 (m,2H), 2.82 (d, He,J(He,,,-Hd,do)=9.15 HZ), 2.88 (d,Hd,J=9.15 Hz), 3.25 (m,Hi), 3.42 (m,Hh), 3.75 (d,H,J(HrH )=12.08 HZ), 3.99 (d,H ,J=12.08 Hz), 5.40 (d,Hc,J=1.5 Hz), 6.45 ( ~ , H ~ , J ( H ~ - H ~ ) Hz), =?.~ 6.50 ( d d , ~ ~ , ~ ( f i ~ - ~Hz, ~)= J(Hb-Ha)=5.5 1.5 Hz) ppm. 1 3 cNMR (CDCI,) 8 13.94 (p), 22.48 (s), 26.48 (s), 26.88 (s), 28.99 (s), 31.61 (s), 43.18 (s), 47.05 (t), 48.98 (s), 49.5 (t), 82.95 (t), 89.17 (q), 134.16 (t), 137.52 (t), 171.64 (q), 172.71 (q) ppm.

'

Ndk

Determination of the s&-s-trans equilibrium of compounds 2a-d. The chemical shift of almost all protons of compounds 2 a d are strongly sensitive to the conformation of the amide-bond. Most convenient for the determination of the conformation equilibrium are resonance signals for the vinylic protons He and H, and the protons H, and H,. In the s-ck conformation the vinylic protons are affected by the furan-ring which causes a considerable upfield shift. In addition, the difference in the chemical environment of the protons becomes more pronounced, and the coupling constant increases concomittantly. The chemical shift of protons H, and Hg in the s-tram conformation is more upfield than those for the s-cis compound. For compound 2b is found: 'H NMR (CDCI,) G 4.53 (s,H, and H,, s-tram), 4.63 (s,H, and H,, s-cis), 6.63 (d,H,, J(H,-He)= 12.82 Hz, s-ch), 6.82 (d,H,,J= 12.82 Hz, s-ck), 7.35 and 7.38 ppm (dd,He and H,, s-tram). For compounds 2a, 2c, and 2d similar chemical shifts were found. The equilibrium constant was determined after equilibration at 25C for 30 minutes. In D20 it was impossible to allow the solution to equilibrate at 25OC because of the extremely rapid conversion into the cycIised product. In this case, the equilibrium constant was determined immediately after mixing of the reactants l a and maleic anhydride. Kinetic measurements. The IMDA reactions of 2 a d were followed by determining the decrease with time of absorbance at 225-230 nm (the actual wavelength was slightly sensitive to the solvent) in a quartz UV cell (1 cm), that was placed in a thermostated cell compartment of a Perkin Elmer 12 spectrophotometer, equipped with a standard personal computer. 20 mg (0.2 mmol) of finely powdered maleic anhydride was dissolved in an excess (0.25 mmol) of the appropriate N-furfuryl-Nalkyl amine on a ceramic pIate. The solution deposited the maleamic acids 2a-2d upon standing (3-5 minutes). The solid was carefully and repeatedly washed with ether and then dissolved in 1 ml 1-propanol. The IMDA reactions were followed after injection

7. Solvent effects on intramolecular Did-Alder reactions in aqueous solutions

of 8-10 pL of this stock solution into the reaction medium that contained about 1x10-~ M of HCl (protonation of the carboxylic acid). The stock solution was freshly prepared before every measurement. All reactions were followed for at least 5 halflives and were perfectly first-order. 200-600 data points were collected and first-order rate constants were calculated by using a fitting program. Rate constants were reproducible to within 1%. Quantitative analysis of the solvent effects. Procedures were followed as described in Section 6.9. The volumes of activation of the IMDA reactions were estimated on the basis of the volume of activation of the IMDA reaction of N-phenyl-N-furfurylmaleamic acid (-25 cm3 rn01-l)~~~.

8. Epilogue. Organic chemistry and hydrophobic effects

CHAPTER 8
Epilogue. Hydrophobic Effects on Organic Reactions in Aqueous Solutions

In the preceding chapters, we have come a long way from the simple observation that rate constants of organic reactions in mixed solvents, and in mixed aqueous solvents in particular, are sensitive to the composition of the medium. The major aim of this thesis was to establish a theoretical model for the qualitative ideas. In this final chapter, we stand back a little and will draw together some conclusions that have emerged from this work. Furthermore we make some suggestions for further research in this area. In Section 8.2, quantitative treatments of solvent effects on organic processes in mixed aqueous solvents are judged on their merits. These quantitative treatments account for the solvent effects in terms of interactions between the reactants and the activated complex on one hand, and the organic cosolvents on the other hand. In Chapters 6 and 7 we have shown that the solvent effects on intermolecular and intramolecular Diels-Alder reactions in mixed aqueous solvents are also strongly modulated by interactions between the reactants or reactive moieties of a reactant. Intermolecular and intramolecular interactions were shown to be largely governed by hydrophobic effects in highly aqueous mixtures. In Section 8.3 we present a navel model which accounts for the most important features of hydrophobic effects in aqueous solvents presented in this thesis. An important objective of this thesis is to emphasise the practical utility of water and binary aqueous mixtures as solvents for organic processes. In the final section, the role of water in determining rate constants, regioselectivities and stereospecificities of organic processes is stressed. A model is advanced according to which organic processes can be identified that might substantially benefit from aqueous solvents as reaction media.

e * 8.2 A quantitative approach for the analysis of solvent effects in mixed solvents. M short-comings

and

The theoretical models, developed in Chapter 2, are valid for the general analysis of solvent effects in mixed solvents. This thesis has emphasised their utility for mixed aqueous solvents, and this section focusses on this particular aspect. Conventionally, dependences of kinetic parameters on composition of binary aqueous mixtures are examined in terms of the corresponding dependence of pseudoequilibrium parameters, based on transition state theory (e.g. Gibbs energies of

8. Epilogue. Organic chemkby and hydrophobic efJec&

activation), on solvent c o m p ~ s i t i o n ~ "Usually, ~ ~ ~ . these activation parameters are plotted as a function of mole fraction composition of the aqueous mixture. The dependence of rate constants on solvent are understood in terms of the dependence of standard chemical potentials and related standard thermodynamic properties of reactants and activated complex. Previously, this dependence has been analysed by a series of semi-quantitative approaches3'. In this thesis a new approach was developed which considers the dependence on composition of solvent of rate constants in terms of activity coefficients for reactants and activated complex. In this approach the standard chemical potential and related standard properties of reactant(s) and activated complex remain, by definition, unchanged when an organic cosolvent is added to the aqueous solution. The dependence of rate constants on the composition of the solvent is described in terms of the molality of cosolvents. Other concentration scales can be used as Our approach shows some similarities with earlier and ~ ' amm met?, but is more sophisticated. The model is approaches of ~ r d n s t e d ~ developed on the basis of the solution theory of McMillan and ~ a ~ eand r 'incoqno~ rates the additivity schemes advanced by Savage and Wood (SWAG-approach)' b. The theoretical treatment can be used to analyse Gibbs energies of activation for reactions in mixed aqueous solvents across a quite extended mole fraction range. However, the theory is most valuable for the analysis of solvent effects in dilute aqueous solvents where solvent effects are characterised by pairwise interactions. The model has been extended for the analysis of solvent effects on partial molar quantities, such as enthalpies and entropies of activation. This approach is shown to be valid only for the analysis of solvent effects in very dilute aqueous solutions (see Section 4.2). A point of concern is that the underlying assumptions of the SWAG ap roach are too drastic. The model fails to account for nearest-neighbour effects1p8..157h and stereochemical aspects1810~178 of the intermolecular interactions in aqueous solvents. This study, however, clearly shows that this criticism, while valid, should not obscure the considerable potential of treatments based on simple additivity schemes. The theoretical model allows the interpretation of a vast amount of kinetic data in dilute aqueous solutions and has exciting prospects for the analysis of solvent effects in terms of group contributions. The limitations of group additivity schemes were used, in retrospect, to elucidate the details of pairwise interactions in dilute aqueous solutions. The type of anaIysis does full justice to the notion that solvent effects in aqueous solvents result from a delicate balance between competing group contributions and depend on the relative position and stereochemistry of functional groups of the cosolvent. Moreover, the approach satisfies the important goal which is to establish a procedure for a quantitative analysis of solvent effects using parameters that can be determined from the properties of simple solutes in aqueous solution. Since the above method is limited to solvent effects in dilute aqueous solvents, an alternative treatment has been developed, based on the inverse Kirkwood-Buff The structural models, emerging from this kind of analysis offer an interesting insight into solvent effects on kinetics of organic reactions in binary aqueous solvents, containing cosolvents across a mole fraction range of OSx,ll. Solvent effects are understood in terms of preferential solvation of reactants and and A, (Section 2.4.2). activated complex, expressed in affinity parameters 4, b1 The analysis emphasises the microscopic structure of the mixed aqueous solvent and

8. Epilogue. Organic chemirby and hydrophobic effects

signals the presence of short-lived cosolvent-rich phases in the aqueous solution. A serious problem in calculating the affinity parameters is that one needs several sets of thermodynamic data to obtain the integral f u n c t i ~ n s ' ~ ~ ~The '~~ analysis -'~~~ requires, ~. apart from kinetic data, also thermodynamic and volumetric data of the reaction medium, as well as volumes of activation of the reaction. The quantitative approach can, in principle, be very accurate for an analysis of solvent effects in mixed aqueous solvents across the whole mole fraction range. Unfortunately, the required molar Gibbs energies of mixing of the binary aqueous solvent, reported in the literature, are often not very accurate. In particular when the mixed aqueous solvent is thermodynamically highly non-ideal, errors in excess molar Gibbs energies of mixing lead to physically impossible exponential or even asymptotic behaviour of the affinity parameters (see Chapter 6). The accuracy of the analysis of solvent effects in more dilute aqueous solutions is determined by the accuracy with which the dependence of the Gibbs energy of activation on the composition of the mixed solvent can be described mathematically. In addition, the lack of experimental volumes of activation of organic reactions in mixed aqueous solvents may give rise to a small but significant error in the affinity parameters. Recently, methods have been advanced for deriving Kirkwood-Buff integrals directly from experiment343. In addition, sophisticated methods have been reported to obtain the property Q (see Equation 2-70), which is related to the molar Gibbs energy of mixing, in a more reliable rnan~~e?~'. In spite of these sources of error, solvent effects on organic reactions in mixed aqueous solvents can be analysed satisfactorily using estimated volumes of activation and cubic spline approximations of the dependence of the Gibbs energy of activation on the composition of the solvent. In combination with an inverse Kirkwood-Buff analysis of standard chemical potentials of the reactants, the treatment provides a detailed insight into the role of preferential solvation for solvent effects on organic reactions in mixed aqueous solvents. Quantitative analyses of solvent effects in mixed aqueous solvents, based either on the solution theories of McMillan and Mayer or on the inverse Kirkwood-Buff theory, are clearly superior to analyses of solvent effects which require the correlation of kinetic parameters with macroscopic solvent parameters or solvent polarity scales. Furthermore, the SWAG method provides exciting opportunities for using solvent by effects as effective probes for mechanistic studies in physical organic chemist? applying a more fine-grained subdivision of group interaction paramaters. We must, however, be wary of adopting a host of interaction parameters.

8.3 Hydrophobic eflects. Hydrophobic intenadions and hydrophobic hydration

In this section some novel ideas about hydrophobic effects, reported recently in the literature~103~08081199125-338, will be critically appraised. These new views, and those which have emerged from the work presented in this thesis, are utilised to develop a more general view on hydrophobic effects and on hydrophobic interactions in particular. We do not claim to offer an exhaustive treatment of all aspects of hydrophobic effects, but the aim of our model is to account in a qualitative sense for the most important features of hydrophobic effects.

8. Epilogue Organic chemirtry and hyhphobic Mats

Four important themes have governed discussions between adherents of the classic view and the new view on hydrophobic effects: (i) definition of hydrophobic effects, (ii) arguments about the choice of standard states, (iii) underestimation of the temperature dependence of hydrophobic effects and (iv) lack of discrimination between pairwise and bulk hydrophobic interactions (see ref. 344). As argued extensively in Section 1.4, recent discussions about hydrophobic effects have been obscured by the lack of a proper definition of the subject of discussion. This fact led to a debate that has been strongly dominated by semantics. A considerable part of this semantic discussion can be traced back to ar ments about the standard states that have been used to study hydrophobic effects" The argument is centred on the question whether hydrophobic effects have to be related to the transfer of apolar compounds from the liquid phase or from the gas phase to aqueous solutions. Examination of solubility (Gibbs energy) and entropy properties in terms of molecular models is severely complicated by these standard-state choices. Abraham337 convincingly showed that it is misleading to use the liquid state as standard state for developing a molecular model for hydrophobic hydration. The alternative (the gaseous standard state) is disadvantageous as well, since the interpretation of the entropy of solvation is rather ambiguous. Thermodynamic parameters for the hydration of apolar liquids in water have also been erroneously used to develop molecular models for hydrophobic interactions 911'083114. The solvation of liquid hydrocarbons in water largely reflects the disruption of bulk interactions between the apolar species, and does not provide adequate information about pairwise hydrophobic interactions. Bulk hydrophobic interactions are extremely sensitive to the concentration of the solute and cannot be adequately described using solubility data. Privalov and ill'^ advocated an alternative, but artificial, standard state: a compressed gas, having the same entropy and heat capacity as the liquid, but the same enthalpy as the gas. The entropy differences between liquid and gaseous hydrocarbons are very similar and close to 90 J K ' mol-'. This fudged standard state evoked a lot of criticism. This criticism, although well-founded, unfortunately diverted the attention from the interesting and novel ideas of Gil11'0*114119338 and ~ r i v a l o about ~ ~hydrophobic ~ ~ ~ effects. An important merit of the work of ~ u m r y " ~and , later of Privalov and Gill (see references cited above) was that these authors considered hydrophobic effects over a large temperature range. Classic descriptions of hydrophobic effects were mainly concentrated on the exceptional features observed near room temperature. The "new view" on hydrophobic effects, advanced by these authors, has been outlined in Section 1.4. For the sake of argument, the main features of the temperature dependence of hydrophobic effects will be briefly reviewed. The most characteristic property of the solvation of apolar gases and apolar liquids in water is the exceptionally large and positive heat capacity change. This heat capacity change decreases with increasing temperature and approaches zero at elevated temperatures (>lO(PC). Whether the heat capacity change, associated with the solvation of apolar solutes in water, completely disappears at high temperatures is still a matter of concern114. The large and positive heat capacity change implies that the enthalpy of solvation of apolar gases and liquids in water is highly temperaturedependent, and increases with increasing temperature. The entropy of solvation of apolar gases and liquids is strongly temperature-dependent as well. Although the

P'*'.

'

8 .Epilogue. Organic chemistty and hyirophobic effects

solvation of apolar compounds in water is accompanied by a strongly negative entropic contribution, TASO incremes with increasing temperature. The Gibbs energy, associated with the transfer of apolar gases and liquids to water is far less sensitive to temperature changes and increases slightly with increasing temperature (see Figure 12 and Section 1.4). Interestingly, TASO and AH0, associated with the solvation of apolar gases and liquids in water increase simultaneously with increasing temperature. This suggests that the process of dissolution is accompanied by a process occurring close to its This means that a considerable part of the large and equilibrium temperature1159119. negative entropy of solvation is compensation entropy, canceling a favourable enthalpic term. The compensation behaviour is considered to be a consequence of structural reorganisation of water in the immediate surroundings of apolar species, e.g. the formation of a hydrophobic hydration shell. The existence of a fluctuation process in bulk water has been confirmed several times341. Recently, Muller114 developed a theoretical model which accounts for these observations in terms of two rapidly fluctuating conformations of water with equal Gibbs energy. Decreasing the temperaIt should be ture or introduction of apolar solutes favours the "dense" ~tructure~"~. emphasised, however, that the classic "iceberg" concept still provides a convenient qualitative explanation for the large entropy losses and large heat capacity changes, observed upon solvation of a nonpolar solute in water near room t e m p e r a t ~ r e ~ l . ~ . The hydrophobic hydration shells "melt" with increasing temperature. Remarkably, the enthalpy of solvation of liquid hydrocarbons is zero near room temperature, whereas the enthalpy of solvation of gaseous hydrocarbons is zero near 160C. The entropy of f liquid hydrocarbons is also zero near 160C. PrivalovlOB, and solvation o others99-105339~340 suggested that the hydrophobic hydration shell disappears near 160 O C . The validity of this statement is difficult to assess experimentally. The results, presented in this thesis and recent studies of the temperature dependence of hydrophobic effects prompt some important changes in our attitude towards hydrophobic effects. Most important is the notion that the structural changes of the aqueous solution upon introduction of apolar species are not responsible for the low solubility of apolar compounds in water. The emphasis on the entropy anomaly of hydrophobic hydration, which in the classic view of hydrophobic effects was considered to be the main cause for hydrophobic effects, has diverted the attention from the importance of enthalpic contributions. Remarkably, the almost complete disappearance of the negative entropy term for the hydration of apolar compounds at elevated temperatures does not lead to a reduction of the hydrophobicity of apolar compounds since the Gibbs energy of hydration even increases with increasing temperature. This prompts the idea that hydrophobicity is principally determined by an unfavourable enthalpic contribution. But the suggestion of Privalov and c~workers"'~ that the hydrophobicity is entirely determined by the enthalpy of disruption of London dispersion interactions between the apolar molecules is unrealistic. Hydro hobicity is due to the incompatibility of water and apolar compounds. Privalov et al?'underestimate the importance of the fact that water will attempt to maintain its hydrogen bonded network. Therefore, not only the disruption of London dispersion interactions between apolar molecules, but also the unfavourable Gibbs energy, associated with the formation of a cavity in water necessary to accomodate the apolar solute and the weak interactions between the solute and water make water a bad solvent for apolar

8. Epilogue. Organic chemirtiy and hy&ophobic effects

compounds. Enthalpic factors are clearly dominant. At lower temperatures, the formation of more structured hydrophobic hydration shells introduces an unfavourable entropic term. In spite of this fact, the overall Gibbs energy of solvation becomes more favourable. This is due to a dominating favourable contribution of the heat of solvation. In this context, we emphasise the importance of the positive heat capacity change which even increases with decreasing temperature. Hence, the favourable enthalpic contribution becomes even more favourable at lower temperatures. The continuous topological rearrangement of water molecules suggests, however, that the Gibbs energy changes, associated with these fluctuations, are small341.Since addition of hydrophobic solutes does enhance the formation of a more structured hydration shell network, it might be speculated that the formation of a hydrophobic hydration shell enhances the interactions between the apolar solute and water. This might be attributed to bulk-polarisation of the hydrophobic hydration shell induced by rapid proton-hopping (see ref. 347 for a theoretical approach of induced polarisation fluctuations in water). This prompts the remarkable conclusion that the formation of a hydrophobic hydration shell reduces the hydrophobicity of apolar compounds and prevents pairwise hydrophobic association. This view is supported by who showed conclusively recent studies of Watanabe and a1.'02b and Wood et that the tendency of two hydrocarbon molecules, containing more than two carbon atoms, to associate in an aqueous solution is less than their tendency to associate in vacuum. When the formation of a hydrophobic hydration shell is seriously hampered, e.g. by increasing the temperature, the solute-water interactions will decrease, leading to an enhanced hydrophobicity. Hence, at lower temperatures water has a remarkable potential of accomodating apolar compounds by increasing its interactions with the solute. Interestingly, although the solubility of apolar compounds in "cold water might be considered to be exceptionally high, the absolute solubility is lower than at more elevated temperatures. This is due to the fact that the absolute solubility is determined by [AGO(transfer)R](which is [-Rln[solubility]]). Nevertheless, the conclusion is justified that the hydrophobicity is most pronounced at higher temperatures. The reverse applies to hydrophobic hydration. Recent continuum-based calculations of hydration entropies of alkanes by Rashin et al."b and Monte Carlo computer simulations of alkanes in water by Jorgensen et al.346even indicate that the exothermic enthalpies of hydration of gaseous alkanes can be fully attributed to the interactions between water molecules and the nonpolar solutes. According to Rashin et al."b, these interactions are more favourable when the water molecules are "in some configuration around the nonpolar solute". The need to invoke "structure-making" effects in the sense of "enhanced hydrogen-bonding" of water molecules in the hydrophobic hydration shell might therefore be not necessary to explain the exothermic enthalpies for the hydration of nonpolar gases near room temperature. The occurrence of configurations in which the water molecules in the hydrophobic hydration shell interact more favourably with nonpolar solutes can, be interpreted as the formation of some structure. In according to Rashin et a1.345b, contrast, Tobias et Pratt et a1.13' and Rossky et al.349do find enhanced waterwater interactions upon introduction of apolar solutes in water by using moleculardynamics. Generally, gross assumptions were made to simplify the computations of hydration entropies and enthalpies". In addition, the calculated entropic and enthalpic parameters are characterised by large errors, reflecting the statistical

8. Epilogue Organic chemistry and hy&ophobic effects

averaging of molecular-dynamics trajectorie~~~~. These aspects make it difficult to draw any definite conclusions with respect to the thermodynamics of hydrophobic hydration on the basis of theoretical studies. The understanding of hydrophobicity and hydrophobic hydration is a decisive factor in a proper understandig of hydrophobic interactions. As emphasised in Section 1.4, a careful distinction must be made between pairwise and bulk hydrophobic interactions. The quantitative pairwise and bulk interactions in aqueous solution. On this basis it was concluded that the contribution pairwise hydrophobic interactions is generally limited to dilute aqueous solutions. We found that pairwise hydrophobic interactions are characterised by the following features: (i) A destructive overlap of hydrophobic hydration shells. This process is accompanied by the release of water molecules from the hydrophobic hydration shells to the bulk solution. The enthalpy and entropy, associated with this process, are strongly dependent on the temperature. At room temperature both enthalpy and entropy are large and positive. Their magnitudes decrease with increasing temperature. The Gibbs energy of this process (iii) A reduction of the apolar surface exposed to water and a concomitant decrease of water-solute interactions. At room temperature, hydrophobic hydration increases the interactions pairwise hydrophobic interactions pairwise hydrophobic interactions. In the classic view of hydrophobic effects, this driving-force was associated with the destruction hydrophobic interactions and hydrophobic association. Hydrophobic interactions are therefore most pronounced at elevated temperatures, where the formation of a hydrophobic hydration shell is seriously hampered. In spite of the fact that pairwise hydrophobic interactions are characterised by a favourable Gibbs energy term, it was shown that apolar solutes do not necessarily aggregate in aqueous solution. The frequency pairwise hydrophobic "encounters" is proportional to the concentration of the apolar solute as well as to the Gibbs energy, associated with the process (see Chapter 2). Aggregation is counteracted not only by hydrophobic hydration but also by the fact that aggregation is, in general, an entropically unfavourable process. I f the enthalpy, associated with the painvise hydrophobic interaction between the apolar solutes is similar to, or outweighs both competing terms, stacking or aggregation of the apolar solutes can occur. This is well-established for organic dyes and other solutes, characterised by extensively delocalised T-electron systems307.Aggregation can also be induced by destruction of

8. Epilogue. Organic chemistry and hydrophobic effects

the hydrophobic hydration shells. This can be brought about (i) by increasing the temperature (see above) or (ii) by increasing the concentration of the apolar solutes. The effect of temperature is difficult to assess quantitatively. On one hand, increasing the temperature enhances pairwise hydrophobic interactions, but on the other hand the importance of the unfavourable entropic term, associated with aggregation, is simultaneously enhanced. An increase of the concentration of apolar solutes is more effective. At a certain concentration, dependent on the size and hydrophobicity of the solute, all water molecules will be part of an extended hydrophobic hydration shell. Beyond this concentration, incomplete hydrophobic hydration shells are formed which insufficiently counteract aggregation. If, simultaneously, the frequency of hydrophobic "encounters" increases due to the increasing concentration of the apolar solutes, the origin of hydrophobic clusters is a fact. Hydrophobic interaction between many apolar solute particles in a cooperative process, leading to hydrophobic clusters, is defined as bulk hydrophobic interaction. The thermodynamics of bulk hydrophobic interactions are extremely dependent on the concentration of the solutes and the size and morphologies of the solute-rich clusters. On the other hand, bulk hydrophobic interactions are not very sensitive towards temperature. A recent example is given by Nusselder et a1.342, who reported a small temperature dependence of critical micellar concentrations for a series of surfactants. We suggest that the best method to examine bulk hydrophobic interactions is to assess the dependence of partial molar quantities (e-g. chemical potentials) of the solute on the concentration of the solute in water (see Chapter 6). A clear change in the partial molar quantities is indicative for the onset of bulk hydrophobic interactions, leading to preferential solvation of the solute by itself. The driving-force for preferential solvation (bulk hydrophobic interactions) is similar as for pairwise hydrophobic interactions, and is governed by the favourable London dispersion interactions between the solutes and the reduction of the hydrophobic surface, exposed to water. Bulk hydrophobic interactions for saturated hydrocarbons are much more attractive in water than in the gas-phase, in contrast to pairwise hydrophobic interactions. The release of water moIecules from hydrophobic hydration shells to the water-rich clusters leads to significant and largely compensating enthalpy and entropy effects. The importance of these terms gradually looses importance when the concentration of the solutes increases. The thermodynamic conditions under which the formation takes place of aggregated structures with different life-times and morphologies is still a crucial question in the study of bulk hydrophobic interaction^^'^. It is clear that entropy is a powerful factor in determining which structure is thermodynamically preferred. In addition, energy contributions, which however, a very important factor is provided by bounda~y reflect the interactions between water and the aggregated molecules at the surface of the aggregate. The latter is extremely important in preventing macroscopic phase separation and strongly depends on the amphiphilic behaviour and geometric packing properties of the solutes353. The demarcation between pairwise and bulk hydrophobic interactions is somewhat ambiguous. In moderately concentrated aqueous solutions, triplet and higher-order interactions may occur without actually inducing serious aggregation of the apolar solutes. Unfortunately, a thermodynamic theory does not exist which links the properties of dilute and infinitely dilute aqueous solutions characterised by pairwise

8. Epilogue Organic chemistry and hydrophobic effects

hydrophobic interactions, with the properties of mixed aqueous solvents characterised by the presence of water-rich and solute-rich domains. It is therefore a challenging task to develop a theoretical model that accounts for the transition of pairwise to bulk hydrophobic interactions. This model must account for nonadditive or cooperative nature of water-water and solute-solute interactions. Furthermore, although integral equation theoriesm1 and computer s i m u l a t i ~ n s have ' ~ ~ been ~ ~ ~applied ~~ frequently to compute Gibbs energies of pairwise hydrophobic interactions, the decomposition into enthalpic and entropic contri%utions has not been camed out348. In order to verify our new ideas on hydrophobic interactions, further theoretical study is necessaryThe situation becomes even more complicated if the solutes contain hydrophilic substituents. Hydrophilic groups affect the hydrophobic interactions dramatically (see the ' , hydration of hydrophilic moieties is highly Chapter 3). As shown by ~ b r a h a r n ~ ~ favourable in terms of the standard Gibbs energy. The molar Gibbs energies of solvation of gaseous methanol and ethanol in water are, for example, negative. The results, presented in this thesis, show that hydrophilic moieties reduce the hydrophobicity of the solute and reduce the importance of hydrophobic interactions. The presence of more than one hydrophilic moiety gives rise to unexpected phenomena. For example, polyhydric alcohols are quite hydrophobic compounds if interactions are considered with other apolar solutes in aqueous solutions. This might not have been anticipated on the basis of their hydrophilic structure and solubility in water. The hydrophobic appearance was ascribed to the fact that hydration of these compounds is governed by hydrogen-bonding of water molecules to the hydroxy groups. Hydrophobic hydration of the methylene moieties is strongly reduced. Consequently, the hydrocarbon moieties are deshielded and hydrophobic interactions are not counteracted by a hydrophobic hydration shell (see above). Hence, the apolar methylene moieties are accessible for London dispersion interactions with apolar cosolutes.

8.4 Organic reactivity in mixed aqueous solvents

As stated in the introductory chapter, water is not a popular solvent among synthetic organic chemists. That this attitude is not always justified is shown by the fact that water can be an attractive solvent for intermolecular as well as intramolecular DielsAlder reactions with respect to the rate constant as well as the stereospecificity. Grieco et a1.18 have already reported a number of other synthetically important reactions that exhibit an exceptionally large rate enhancement in water. In this section we want to generalise some aspects of organic reactivity in aqueous media. A general approach is developed according to which organic processes can be selected that could benefit from water as a solvent. Exceptionally high reaction rates of organic reactions in aqueous solution are reserved for those reactions involving apolar reactants which lose a considerable part of their hydrophobic surface during the activation process. In the terminology used in this thesis, this anomalous solvent effects is refered to as "hydrophobic acceleration". Organic reactions that are expected to be accelerated in aqueous solutions are pericyclic reactions, many types of rearrangements and complexation processes. Also

8. Epilogur Oganic chcmirny and hyirophobic effects

several photochemical reactions and organic processes involving radicals might be candidates for "hydrophobic accelerationt'. In a few cases hydrophobic acceleration has It is difficult to generalise these already been observed for condensation reactions17G19. incidental observations. Condensation reactions do involve apolar reactants, but the rate determining step can involve ionic intermediates. Whether or not condensation reactions are accelerated in water depends strongly on the stabilisation of these intermediates. The stereospecificities and, tentatively, also the stereoselectivities of organic reactions between apolar reagents can be significantly altered in aqueous solution. Most pronounced effects might be anticipated if the stereochemically different pathways are characterised by a considerably different reduction of the hydrophobic surface during the activation process. Reactants might be designed to meet these requirements. Aggregation of reactants in the aqueous solution must be avoided since hydrophobic acceleration will then be inhibited. In dilute aqueous solutions water is able to prevent aggregation very effectively (see Section 8.2). Exceptions are photochemical processes, which are favourably affected by aggregation of the reactants. A very interesting aspect of hydrophobic accelerations was revealed by intramolecular Diels-Alder reactions which were shown to be accelerated in water as well (Chapter 7). Remarkably, intramolecular aggregation of the diene and the dienophile in aqueous solutions was also not observed. Apparently, water not only prevents intermolecular aggregation but also intramolecular aggregation. More detailed studies of the conformation of apolar compounds in aqueous solutions are necessary to see whether this statement is generally true. The requirement that reactants must be apolar compounds to obtain hydrophobic acceleration is at the same time the limiting factor for organic reactivity in aqueous solutions. Organic reactants are often sparingly soluble in water. The solubility of organic reagents can be enhanced by addition of apolar cosolvents. Generally, organic cosolvents can be added without serious effects on the rate constant across a mole fraction range of 01~~50.1.Preferential solvation of the reactants due to bulk hydrophobic interactions must be avoided since this will immediately lead to a considerable reduction of the rate constant. Unfortunately, water is also reactive towards some organic reagents or particular functional groups. Reaction of organic reagents with water is difficult to prevent (see ref. 17f for a 'fvater-proof' reducing agent). If the desired reaction is very fast, the occurrence of side reactions may be minimised. In many respects, water is a unique solvent. Water is able to dissolve hydrophobic molecules or moieties and to prevent aggregation of these molecules in spite of the favourable interactions between these molecules. In organic chemistry this fact can be used to accelerate reactions as well as to increase the stereospecificity of chemical transformations. Since water is also an ecologically sound solvent, it is desirable to study the applicability of water as a solvent for organic processes in a systematic way. We hope that this thesis will provide the initial impetus to reform the "hydrophobic attitude" of synthetic organic chemists.

Lisi of references

References
Tanaka, K;Mackay, G.I.;Payzant, J.D.;Bohme, D.K CanJ.Chem 197654,1643 Berthelot, M.;Pean de Gaint-Gilles, L. AnnChim.Phys. 1862,65,385 and 1862,66,5 Menschutkin, N. Z.Phys.Chem. 1890,.5,589 Cram, D.J.;Rickborn, B.;Kingsbury, C.k;Haberfield, P. JAmChemSoc. 1%1,83,3678 Claisen, L .LiebhnChem 18%,291,147 Wislicenus, W. LiebAnnChem. 18%,291,147 Kundt, A. ChemZentralbl. 1878,498 Eicke, H-F. Topics in Current Chemistry; Springer Verlag; Berlin; 1980 p.112 (a) Smithrud, D.B.;Diederich, F. J h C h e m S o c . 1990,212,339 @) Izatt, R.M.;Bradshaw, J.S.;Nielsen, S.A.;Lamb, J.D. ChemRev. 1985,85,271 Marcus, Y. Ion Solvation; Wiley; New York; 1985 (a) Jorgensen, W.L. J.Phys.Chem. 19e875304 (b) Abraham, R.J.;Bretschneider, E. Medium Effects on Rotational and Conformational Equilibria; Orville-Thomas, W.J.(ed.); Wiley, New York; 1974, p.481 Ben-Naim, A. J.Phys.Chem. 1975,63,2064 Universal solvent . ( e d . ) ; Plenum; New York; 1972 Water. A Comprehensive Treatise, Franks, F Water Science Reviews, Franks, F.(ed.); Cambridge University Press; 1985 Water and Aqueous Solutions, Ben-Naim, k(ed.); Plenum; New York; 1974 (a) Rideout, D.C;Breslow, R. J4m.Chem.S~. 1980,l02,7816 (b) Breslow, R.;Maitra, U.;Rideout, D. Tetrahedron Lett. 1983,24,1901 (c) Breslow, R.;Maitra, U. Tetrahedron Lett. 1984,25,1239 (d) Breslow, R.;Guo, T. J.Am.ChemSuc. 1988,l10,5613 (e) Kool, E.T.;Breslow, R. J.Am.Chem.Soc.1988 JlO, 15% (0 Light, J; Breslow, R. Tetrahedron Len. 199031,2957 (g) Bresluw, R. Acc.ChemRes. 1991,24,159 (a) Grieco, P.k;Yoshida, K;Garner, P. J.Org.Chem. 1983,48,3139 (b) G r i m , P.k;Garner, P.;Zhen-min, H. Tetrahedron Len. 1983,24,18!37 (c) Larsen, S.D.;Grim, P.A. JAmChem.Soc. 1985J07,1768 (d) Grieco, P.k;Galatsis, P.;Spohn, R.F. Tetrahedron Left. 1986,42,2847 (e) Grieco, P.A.;Larsen, S.;Fobare, W.F. Tetrahedron Len. 1986,27,1975 ( f ) Grieco, P.k;Brandess, E.B.;McCann, S.;Clark, J.d. LOrg.Chem. 1989J4,5849 (g) Brandes, E.B.;Grieco, P.k;Gajewski, J.J. J.Org.Chem. 198954,515 (a) Lubineau, A. J.Org.Chem. 1986,.51,2144 @) Lubineau, A.;Quenau, Y. J.Org,Chem. 1987,52,1002 (c) Lubineau, A;Meyer, E. Tetrahedron 1988,44,6065 (a) Schneider, H.J.;Sangwan, N.K LChem.Soc.;Chem.Comm. 1986,1787 (b) Schneider, HJ.;Sangwan, N.K Angew.Chem. 19879,924 (c) Schneider, H-J.;Sangwan, N.K. J.Chem.Soc.;Perlcin Trans.Z11989,1223 (a) Ahmad-Zadeh Sammiii, A.;Savignac, k de;Rico, I.;Lattes, k Tepahedron 1985,41,3583 @) Lattes, k J.ChimPhys.Phys. -Chem.Biol. 1987,84,1061 (a) Luche, J.L.;Petrier, C.;Einhorn, J. Tetrahedron Len. 1985,26,1449 (b) Petrier, C.;Luche, J.L. J.Org.Chem. 1985J0,912 Cativiela, C.;Garcia, J.I.;Mayoral, J.A.;Avenoza, A;Pregrina, J.M.;Roy, M . k J.Phys.0g.Chem. 1991,4,48 Laszlo, P.;Lucchetti, J. Tetrahedron Lett. 1984,25,2147 Williams, D.R.;Gaston, R.D.;Horton, I.B. Tetrahedron Lett. 1985,26,1391 Keana, J.F.W.;Guzikowski, kP.;Morat, C.;Volwerk, J.J. 1Org.Chem. 1983,48,2661 Braun, R.;Schuster, F.;Sauer, J. Tetrahedron Left. 1986f 7,1285 (a) Waldmann, H. LiebAm 1990,671 @) Waldmann, H.;Drager, M. LiebAnn. 1990,681 Nokami, J.;Otera, J.;Sudo, T.;Okawara, R. Organomet. 1983,2,191 Reichhardt, C. Solvents and Solvent Effects in Organic Chemistry; VCH, Cambridge; New York; 1990 Amis, E.S.;Hinton, J.F. Solvent Effects on Chemical Phenomena; Academic Press; New York;

1973 Abraham, M.H. Progr.Phys..OrgChem. 1974J1.1 Cooper, KA;Dahar, M.L;Hughes. ED.;Ingold, CK;MacNulty, B.J.;Woolf, L.I. J.Chem.Soc. 1948,2043 Glasstone, S.;Laidler, KJ.;Eyring, H. The Theory of Rate Processes; McGraw Hill; New York; 1941. p.400 Kirkwood, J.G.;Westheimer, E LChernPhys. 1938,6,506 Laidler, KJ.;Landskroener, P . k Trans.Far.Soc. 1956,52,200 Kirkwocd, J.G. J.Chem.Phys. 1934.2.351 Hildebrand, J.H.;Prausnitz, J.M.;Soott, R.L Regular and Related Solutions; Van NostrandReinhold; Princeton; 1970 Barton, ARM. Handbook of Solubility Parameters and Other Cohesion Parameters; CRC Press; Boca Raton; 1983 Grieger, R.A;Eckert, CA JsPmChemSoc. 1970,202,7149 Koppel, LA;Palm. V.A. Advances in Linear Free Energy Relationships; Chapman, N.B.; Shorter,J.(eds.); Plenum; New York; 1972 (a) Dack, M.R.J. LChenrEduc. 1974,.51,341 @) Dack, M.R.J. ChemSoc.Rev. 1975,4,211 Ritchie, C.D.; SoluteSolvent Interactions;Coetzee, J.F.;Ritchie, C.D.(eds.); Dekker; New York; 1%9 Chawla, B.;Pollack, S.K;Lebrilla, C.B.;Kamlet, M.J.;Taft, R.W. JAm.Chem.Soc. 1981J03.6924 Dobis, O.;Pearson, J.M.;Schwarc, M. J&Chem.Soc. 196890,278 Kochi, J.K JdmChemSoc. 197092,4395 Reichhardt, C Angew.ChemZntEdEng. 1979J8.98 Kosower, E M . J h C h e m S o c . 1958,80,3253 Kosower, E.M.;Mohammad. MJ. J&ChmSoc. 197193,2713 Dimroth, K;Reichhardt. C;Siepmann, T.;Bohlmann, E Jusfus LiebdnnChem. 1%3,66I,l Reichhardt, C Atre AppLChem. 1982,.54,1867 Kamlet, M.J.;Abboud, J.LM.;Taft, R.W. J h C h m S o c . 19779,6027; 8325 Kamlet, MJ.;Taft, R.W.;Abboud, J.LM.;Abraham, M . H .LOrg.Chem.l983,48,2877 Mayer, U.;Gutmann, V.;Gerger,W. MonatskChem. 1975,206,1235 (a) ibid. 1977J08,489 and 757 Gutmam, V.;Wychera, E .ZnorgNucLChemLeft. 1966,2257 Marcus, Y LSoLChem. 1984J3.599 Grunwald, EWinstein, S. J.Am.Chem.Soc. 1948,70,846 Winstein, S.;Fainberg, AH.;Grunwald, E. J&Chem.Soc. 1957,79,4146 Bentley, T.W.;Schadt, EL;R.Schleyer, P. von J.Am.Chem.Soc. 197698,7667 Bentley, T.W.;Carter, G.E. J.Org.Chem. 1983,48,579 (a) Peterson, P.E.;Vidrine, D.W.;Waller, FJ.;Henrichs, D.M.;Magaha, S.;Stevens,B. JAm. ChemSoc. 19779,7%8 (b) Abraham, M.H.;Grellier, P.L;McGill, R . k J.ChemSoc.; Perkin Trans.II 1988,339 Menger, EM.;Venkataram, U.V. J.Am.ChemSoc. 1986,208,2980 Marcus, Y .J.Phys.Chem. 1987.91,4422 Hansch, C;Lm, A Substituents Constants for Correlation Analyses in Chemistry and Biochemistry; Wiley, New York; 1979 Leo, A LChem. Soc.; Perkin Trans.I1 1983,825 Snyder, LR. J.Chromatogr.Sci. 1978,Z6,223 Kovats, E.;Weiss, P.B. Ber.Bunsenges.Phys.Ckrnm 1965,69,812 Kamlet, M.J.;Taft, R.W. J h C h m S o c . 197698,377 and 2886 Abboud, J.LM.;Kamlet, MJ.;Taft, R.W. A.ogr.P@s.Org.Chem.l981 J3,485 Taft, R.W.;Abboud, J.LM.;Kamlet, M.J.;Abraham, M.H. J.SoLChem.l985J4,153 Koppel, I.k,Palm, V.A. Advances in Free Energy Relationships; Chapman, N.B.; Shorter,J.(eds.); Plenum; New York; 1972; Chapter 2 Recent review: Abraham, M.H.;Grellier, P.L;Abboud, J.LM.;Doherty, R.M.; Taft,R.W. CanJ. Chem. 1988,66,2673

Wold, S.;Sjostrom, M. Acta ChemScmB40 1986.270 Abraham, M.H. Pure ApplChem. 1985,57,1055 Buncel, E;Wilson, H. Acc.ChemRes. 1979J2,42 Buncel, E;Wilson, H. LChemEd. 1980,57,629 Kebarle, P.;Caldwell, G.;Magnera, T.;Sumer, J. &e AppLChem. 198537,339 Riveros, J.M.;Jose, S.M.;Takashima, K Adv.Phys..OrgChem. 1985,21,197 Bohme, D.K;Mackay, G.I. J h C h e m . S w . 1981J03,978 Beveridge, D.L;Kelly, M.M.;Radna, R. J&Chem.Soc. 1974,963769 Birnstock, P.;Hofhnann, H-J.;Kohler, H-J. 2kor.ChimAeta 1976,42,311 Review: Simkin, B.Ya.;Sheikhet. LI. LMoLLiq. 1983 , 2 7,79 Moreau, M.;Turq, P. Chemical Reactivity in Liquids, Plenum; New York; 1988 Gertner, BJ.;Whitnell, R.M.;Wilson, KR.;Hynes, T. J&Chem.Soc. 1991J13,74 Kuznetsov, AM. 3.Phys.Chem. 199094,8664 Rossky, P.J.;Huston, S.E;Zicchi, D.k J.Am.ChemSoc. 1989J11.5680 Blandamer, M.J.;Burgess, J.;Engberts, J.B.F.N. Chem.Soc.Rev. 1986J4,237 Engberts, J.B.F.N. Water. A Comprehensive Treatise, Franks,F.(ed.); Plenum; New York; 1979 Vo1.6 Chapter 4 Franks, E Water. A Comprehensive Treatise; Franks,F.(ed.); Plenum; New York; 1979; Vol.1 Chapter 1 Hydrophobic Interaction; ChemSoc.Faradaj..Symp.1982J 7 Tanford, C. The Hydrophobic Effect; Wiley; New York; 2nd edition; 1980 Ben-Naim, k Hydrophobic Interactions; Plenum; New York; 1980 Hildebrand, J.H. J.Phys.Chem. 1968,72,1841 Nemethy, G.;Scheraga, H.k;Kaumann, W. XPhys.Chem. 1968,72,1842 Cramer, R.D. JAm.ChemSoc. 1977,99,5408 Hildebrand, J.H. Boc.NatLAcadSciUSA. 1979,76,194 Tanford, C. Proc.NatUcadSci USA. 1979,76,4175 Spolar, R.S.;Ha, J.H.;Record, M . T . Roc.Natl.AcadSciUSA. 1989,86,8382 Ha, J.H.;Spolar, R.S.;Record M.T. J.MoLBio1. 1989,2@,801 Murphy, KP.;Privalov, P.L;Gill, S.J. Science 1990,247,559 Baldwin, R.L. Proc.NatLAcadSciUSA. 1986,83,8O!X (a) Dill, K k Science 1990,250,297 @) Watanabe, QAndersen, H.C. J.Phys.Chem. 1986,90,795 Privalov, P.L;GilI, SJ.;Murphy, KP. Science 1990,250,297 Frank, H.S.;Evans, M.W. LChenrPhys. 1945J3,1445 Nemethy, G.;Scheraga, H . k LChemPhys. 1%2,&3382; 3401 Frank, H.S.;Wen, W-Y. Disc.Far.Soc. 1957,24 Kauunann, W. Adv.Protein Chem. 1959J4.1 Privalov, P.L.;Gill, S.J. Adv.Protein Chem 198839,191 Privalov, P.L. AnnRev.Bwphys.Chem. 1989J8,18; 47 Gill, SJ.;Wadso, I. ProcNatLAcadSciUSA. 1976.73.2955 Privalov, P.L.;Gill, S.J. Aue AppLChem. 1989,612,1097 Lumry, R.;Gregory, R.B. The Fluctuating Enzyme; Welsh,C.R.(ed.); Wiley, Interscience; 1986 Makhatadze, G.L;Privalov, P.L I.Chem.Them. 1988,20,405 Muller, N. Acc.ChemRes. 1990,23,23 Mirejovsly, D.;Arnett, E.M. JAmChem.Soc. 1983J05,1112 Ramadan, M.S.;Evans, D.E;Lumry, R.F.;Philson, S. JPhys.Chem. 1985,89,3405 Evans, D.F.;Ninham, B.W. J.Phys.Chem. 1986,90,226 Dec, S.F.;Gill, S.J. LSoLChem. 1985J4,827 Gill, S.J.;Dec, J.F.;Olobon, G.;Wadso, I. I.Phys.Chem. 1985,89,3758 Kang, Y.K;Nemethy, G.;Scheraga, H.A. J.Phys.Chem. 1987.91.4105 and 4109 Friedman, H.L;Krishnan, C . V .3.SoLChem. 1973,2,119 Rossky, P.J.;Karplus, M. JAmChemSoc. 1979J01,1913 Swaminathan, S.;Harrison, S.W.;Beveridge, D.L J&ChcnrSoc. 1978J00,1705 Alagona, G.;Tani, A. LChem.Phys. 1980,72,580

Lzkt of references

Lee, B. Biopobmers 1985,24,813 (a) Pierotti, R.A. J.Phys.Chem. 1%5,69,281 @) Morel-Desrosiers, N.;Morel, J-P. CanLChem. 1981,59,1 Sinanoglu, 0.Proceedings of the International Conference on Molecular Association in Biology; Pullman, B. ( d ) Academic; ; New Yorlr; 1968 Panagali, C.;Rao, M.;Berner, B.J. JChemPhys. 1979,71,2975 and 2982 Clementi, E .LPhys.Chem. 1985,89,4426 Jorgensen, W.L;Swenson, CJ. J h ChemSoc. 1985J 07,1489 Scheraga, H.A. AmNYAcadSci. 1985,439,170 Karplus, M. h N Y AcadSci. 1985,439,107 Warshell, A. Proc.Natl.Acad.SciUSA. 1984,81,444 Struthers, R.S.;Rivier, J.;Hagler, AT. AnnNYAcadSci. 1983,439,227 Pratt, LR.;Chandler, D. LChem.Phys. 1980,73,3430 and 3434 Tani, A. MoLPhys. 1984J1,161 Pettitt, B.M.;Karplus, M.;Rossky, P.J. J.Phys.Chern. 1986,90,6335 Ravishanker, G.;Mizei, M.;Beveridge, D.L. Faraday Symp.ChemSoc. 1982J7,79 Kang, Y.K;Nemethy, G.;Scheraga, H.L. LPhys.Chem. 198891,1382 Tanaka, H.;Nakanishi, K;Touhara, H. JChem.Phys. 1984,81,4065 Tanaka, H . LChemPhys. 1987,86,1512 Pettitt, B.M.;Rossky, P.J. J.Phys.Chem. 1985,83,781 Kozak, J.J.;Knight, W.S.;Kauzmann, W. J.Chem.Phys. 1968,48,675 Friedman, H.L,Krishnan, C.V. AnnNYAcadSci. 1973,204,79 Thompson, P.T.;Davis, C.B.;Wood, R.H. J.Phys.Chem. 198892,6386 Franks, F. LChemSoc.Faraday Trans.1 1977,74.830 Attard, P. J.Phys.Chem. 198993,6441 Blokzijl,W.;EngberrsJ.B.EN.;Blandamer,M.J. J h C h e m S o c . 1990J12,1197 Blokzijl, W.;Blandamer, M.J.;Engberts, J.B.F.N. J.Org.Chem. 1991,56,1832 Blokzijl, W.;Engberts, J.B.F.N.;Blandamer, M . J . J.Am.Chem.Soc. 1986J08,6411 Blokzijl,W.;EngbensJ.B.F.N.;Jager J.;Blandamer,M.J. LPhys.Chem.1987Ql,,ti022 Blokzijl, W.;Blandamer, M.J.;Engberts, J.B.F.N. JAm.Chem.Soc. 1991,213,4241 Blandamer, M.J.;Burgess, J.;Engberts, J.B.F.N.;Blokzijl, W. ChemSmAnnRep. 1991,in press McMillan, W.G.;Mayer, J.E. J.ChemPhys. 1945J3.276 Langmuir, I. Collec.Symp.Monogr. 1925,3,48 (a) Cassel, R.B.;Wood, R.H. LPhys.Chem. 1974,78,2465 (b) Savage, J.J.;Wood, R.H. J.SoLChem. 19762,733 (c) Okamoto, B.Y.;Wood, R.H. J.Chem.Soc.Faraday Trans.1 1978,74,1990 (d) Lilley, T.H.;Wood, R.H. LChern.Soc.Faraday Trans.1 1980,76,901 (e) Harris, A.L.;Thompson, P.T.;Wood, R.H. J.Sol.Chem. 19809,305 ( f ) Tasker, I.R.;Wood, R.H. J.Sol.Chem. 1982,11,729 (g) Spitzer, J.J.;Tasker, LR.;Wood, R.H. LSoLChem. 1984J3.221 (h) Spitzer, J.J.;Suri, S.K;Wood, R.H. LSoLChem. 1985,8,561 (i) Spitzer, J.J.;Suri, S.K;Wood, R.H. LSoLChern. 1985,8,571 (j) Suri, S.K;Spitzer, J.J.;Wood, R.H.;Abel, E.G.;Thompson, P.T. J.SoLChem. 1985J1,781 (k) Grollier, J.P.E.;Spitzer, J.J.;Wood, R.H.;Tasker, LR. J.SoLChem. 1985J4,3M (I) Suri, S.K;Wood, R.H. JSoLChem. 1986J5,705 (a) Rouw, A;Somsen, G. JChemSoc.Faraday Tram.1 1982,73,3397 (b) Bloemendal, M.;Somsen, G. LSoLChem. 1983J2,83 (c) Bloemendal, M.;Somsen, G. /.SoLChem. 1984J3,Bl (d) 1985J07,3426 (e) Bloemendal, M.;Booij, Bloemendal, M.;Somsen, G. J&Chem.Soc. K;Somsen, G. J.ChemSoc.Faraday Trans.1 1985,81,1015 ( f ) Bloemendal, M.;Rouw, kC.;Somsen, G. J.Chern.Soc.Faraday Trans.1 1986,82,53 (g) Pielcarski, H.;Somsen, G. Can.J.Chem. 1986,64,1721 (h) Bloemendal, M.;Sijpkes, AH.;Somsen, G. J.So1.Chem. 1986J5,81 (i) Bloemendal, M.;Somsen, G. J.Sol.Chem. 1987J6,367 (j)Bloemendal, M.;Marcus, Y. AIChE Journal 1987,33,1800 (k) Bloemendal, M.;Somsen, G. J.Chem.Them. 1987J9,l (1) Bloemendal, M.;Marcus, Y.;Booij, M.;Hofstee, R.;Somsen, G. LSoLChem. 1988J 7,15 (a) Kirkwood, J.G.;Buff, F.P. LChemPhys. 1951J9,744 ;The theory is covered in more recent textbooks and publications. See for example @) Debenedetti, P.G. 1Chem.Phys. 1987,87,1256 (c) Matteoli, E. Mansoori, G.A. ( e d s . ) Fluctuation T h e o r y of Mixtures; Taylor and Francis;

List of references New York, 1990 Hall, D.G. LChenrSoc.Faraday Trm.I1972,68,25 Ben-Naim, k CelLBwphys. 1988J2,255 Ben-Naim, A. LChemPhys. 1977.674884 Newman, ICE. J.Chem.Soc.Faraday Trans.I1988,84,1387 Covington, AK;Newman, ICE. LChem.Soc.Fraday Trm.I1988,84,1393 Blandarner, M.J.;Blundell. N.L;Burgess, J.;Cowles, H.J.;Engberts, J.B.F.N.;Horn, I.M.; Warrick, P.jr. J.Am.ChenrSoc.l99OJ12,6854 Blandamer, MJ.;Blundell, N.N.;Burgess, J.;Cowles, HJ.;Horn, LM. I,ChemSoc.Faraday Trans.1 1990,86,277 Glasstone, S.;Laidler, KJ.;Eyring, H. The Theory of Rate Processes; McGraw-Hill; New York; 1941 Wood, R.H.;Lilley, T.H.;Thompson, P.T. LChemSoc.Faraday Trans.1 1978,74,1301 (a) Blackburn, G.M.;Lilley, T.H.;Walmsley, E. LChemSoc.Faraday Trans.1 1980,76,915 (b) Lilley, T.H.;Moses, E.;Tasker, LR. LChemSmFaraday Trans.1 1980,76,906 (c) Blackburn, G.M.;Lilley, T.H.;Walmsley, E. J.Chem.Soc.Faraday Trans.1 1982,78,164 ( d ) Blackburn, G.M.;Lilley, T.H.;Milburn, P.J. LSoLChem. 1984J3.789 (e) Blackbum, G.M.;Lilley, T.H.;Milbum, PJ. ThermochimActa 1985,83,289 ( f ) Blackburn, G.M.;Lilley, T.H.;Milburn, P.J. LCkmSoc.; ChmCommun 1985,299 (g) Blackbum, G.M.;Lilley, T.H.;Milbum, P.J. LChemSoc.Faraday Trans.1 1985,81,2191 (h) Kent, H.E.;Lilley, T.H.;Milburn, P.J.;Bloemendal, M.;Somsen, G. J.SoLCkm. 1985J4,lOl (i) Blackburn, G.M.;Lilley, T.H.;Milburn, P.J. LSoLChem. 1986J5,99 (j) Blackburn, G.M.;Lilley, T.H.;Milbum, P.J. J.Chem.Soc.Faraday Tram.I1986,82,2%5 ( k )Arnold, P.;Lilley, T.H. J.ChemThem1. 1985J7.99 (1) Check, P.J.;Lilley, T.H. J.Chem.Soc.Faraday Trans.11988+?4,1927 Franks, F.;Pedley, M.;Reid, D.S. LChemSo~FaradayTram1 1976,72,359 Veytsman, J.G. J.Chem.Phys. 199094.8499 Friedman, H.L J.SoLChem. 1972J,287 and 413 and 419 Guggenheim, E.A. Trans.Far.Soc. 1960,56,1159 Mayer, J.E. Equilibrium Statistical Mechanics; Pergamon; Oxford; 1%8,chapter 4 Garrod, J.E;Herrington, T.M. J.Phys.Chem. 1%9,73,1877 Schneider, H.;Kresheck, G.C.;Scheraga, H.A. LPhys.Chem. 1965,69,1310 Spink, CS.;Wadso, I. J.Chem. Thenn. 1975,7561 Kiyohara, O.;Perron, G.;Desnoyers, J.E. Can.LChem. 1975,53,2591 Franks, F.;Pedley, M.D. J. Chem.Soc.Faraday Trans.I 1983,79,2249 Leslie, T.E;Lilley, T.H. B i o p o ~ e r s 1985,24,695 Bloemendal, M.;Marcus, Y. LSoLChem. 1989,l8,437 (a) Barone, G.;Castronuovo, G.;Crescenzi, V.;Elia, V.;Rizzo, E. LSoLChem. 1978,7,179 (b) Barone, G.;Castronuovo, G.;Della Volpe, C.;Elia, V.;Grassi, L. LPhys.Chem. 197983,2703 (c) Barone, G.;Castronuovo, G.;Elia, V.;Menna, k J.SoLChem. 1979,8,157 (d) Barone, G.;Cacace, G.;Castronuovo, G.;Elia, V. J.SoLChem. 19809,607 (e) Barone, G.;Cacace, G.;Castronuovo, G.;Elia, V. CanLChem. 1981,59,1257 ( f ) Barone, G.;Bove, B.;Castronuovo, G.;Elia, V. LSoLChem. 1981J0,803 (g) Barone, G.;Cacace, G.;Castronuovo, G.;Elia, V. LChem.Soc.Faraday Trans1 1981,77,1569 (h) Barone, G.;Castronuovo, G.;Elia, V. Adv.Mol.RelaxInt.Proc. 198223,279 (i) Barone, G.;Elia, V.;Rizzo, E. LSoLChem. 1982,l1,687 (j) Cesaro, k,Russo, E.;Barone, G. IntJ.Peptide Protein Res. 1982,20,8 (k) Barone, G.;Cacace, P.;Castronuovo, G.;Elia, V. Carbohydr.Res. 1983J19,l (1) Abate, V.;Barone, G.;Castronuovo, G.;Elia, V.;Rizzo, E . LSoLChem. 1983J2,645 (m) Barone, G.;Cacace, P.;Castronuovo, G.;Elia, V.;Lepore, U. Carbohjdr.Res. 1983J15,15 (n) Barone, G.;Castronuovo, G.;Doucas, D.;Elia, V.;Mattia, C.k LPhys.Chem. 198387,1931 (0) Abate, V.;Barone, G.;Castronuovo, G.;Elia, V.;Savino, V. LChemSoc.Faraday Trans.I1984,80,759 ( p ) Barone, G.;Cacace, P.;Elia, V. J.Chem.Soc.Faraday Trans.1 1984,80,2073 (q) Barone, G.;Castronuovo, G.;Elia, V.;Stassinopoulou, K LChemSoc.Faraday Trand 198480,3095 (r) Barone, G.;Cacace, P.;Castronuovo, G.;Elia, V. LSoLChem. 1984J3,625 (s) Abate, V.;Barone, G.;Castronuovo, G.;Elia, V.;Savino, V. J.Chern.Soc.Faraday Trans.1 1984,80,759 ( t ) Barone, G.;Cacace, P.;Elia, V.;Cesaro, A. J.Chem.Soc.Faraday Trans.1

List of references

1984,80,759 (u) Barone, G.;Castronuovo, G.;Elia, V.;Stassinopoulou, K;Della Gatta, G. JChemSoc.Faraday Trans. 1984,80,U)95 (v) Barone, G.;Castronuovo, G.;Elia, V.;Muscetta, M. 3.SoLChem. 1986J5,129 (w) Della Gatta, G.;Barone, G.;Elia, V. J.SoLChem. 1986J5,157 (x) Barone, G.;Castronuovo, G.;Elia, V.;Tosto, M.T. /.SoLChem. 1986J5,199 (y) Christinziano, P;Lelj, F,;Amodeo, P.;Barone, V, ChemPhys.Lef2. 1987,240,401 (z) Barone, G.;Castronuovo, G.;Del Vecchio, P.;Elia, V. J.ThrrmoLAna1. 1988,34,431 (a) Barone, G.;Castronuovo, G.;Del Vecchio, P.;Elia. V. Ll'hennaLAnal. 198834,479 (b) Barone, G.;Castronuovo, G.;Del Vecchio, P.;Elia, V.;Tosto, M.T./.SoLChem. 1988J7,925 (c) Wunburger. S.;Sartorio, R.;Guar.ino, G.;Nii, M. J.ChemSoc.Flyaday Trans.1 198834,2279 (d) Barone, G.;Castronuovo, G.;Del Vecchio, P.;Elia, V. LChemSoc.Faraday 7b.s. I 1988,84,1919 (e) Barone, G.;Castronuovo, G.;Del Vecchio, P.;Elia, V.;Puzziello, P. LSoLChem. 1989J8,1105 (f) Cascella, C.;Castronuovo, G.;Elia, V.;Sartorio, R.;Wunburger, S. J.ChemSoc.Faraduy Trw.I 1989,85,3289 (g) Barone, G.;Castronuovo, G.;Del Vecchio, P.;Giancola, C. J.ChemSoc.Faraday Trm.1 1989,85,2087 (h) Christinziano, P.;L.elj, F.;Amodeo, P.;Barone, G.;Barone, V. J.ChemSoc.Faraday Trans.1 1989,85,621 (i) Andini, S.;Castronuovo, G.;Elia, V.;Fasano, L. J.Chem.Soc.Faraday Trans.1990,86,3567 (j) Cascella, C;Castronuovo, G.;Elia, V.;Sartorio, R.;Wunburger, S. J.ChemSoc.Faraday Trans.1990,86,85 (k) Wurzburger, S.;Sartorio, L.;Elia, V.;Cascella, C. J.Chem.Soc.Faraday Trans.1 199086,3891 (1) Castronuovo, G.;Elia, V.;Magliulo, M. CanLChem. in press (m) Castronuovo, G.;Elia, V.;Giancola, C.;Puzziello, S. LSoLChem in press (n) Barone,G.;Rizzo,E.;Volpe,V. LChemEngData 1976,21,59 (a) Borghesani, G.;Pedriali, R.;Pulidori, F.;Scaroni, I. J-SoLChem. 1986J5,397 @) Borghesani, G.;Remelli, M.;Pulidori, E Thermo.ChimActa 1988J37.165 (c) Borghesani, G.;Pedriali, R.;Pulidori, E JSoLChem. 1989J8.289 Jolicoeur, C.;Riedl, B.;Desrochers, O.;Lenelin, L.L;Zamojska, R.;Emea, 0. J.SoLChem. 1986, 15,109 Herrington, T.M.;Mole, E.L. LChemSoc.Faraday Trans.I1982,78,213 Zielenkiewicz, P. ThennochimActa 1984,79371 Okamoto, B.Y.;Wood, R.H.;Desnoyers, J.E.;Perron, G.;Deiorme, L. /.SoLChem. 1981J0,1981 Heuvelsland, kW.J.M.;Visser, C. de;Sornsen, G. LChernSoc.Faraday Trans.I1981,77,719 (a) Donkersloot, M.C.A. J.SoLChem. 19798,293 @) Patil, KJ. LSoLChem. 1981J0,315 (c) Tadashi, K J.Phys.Chem. 1984,88,1248 (d) Nishikawa, K;Kodera, Y.;Iijima, T. LPhys.Chem 198791,3649 (e) Nishikawa, IC;Hayashi, H.;Iijima, T. I.Phys.Chem 198993,6559 (9Hayashi, H.;Nishikawa, K;Iijima, T. J.PhysSChem1990$4,8334 (a) Marcus, Y. LChem.Soc.Faroday T r m . 1989,85,3019 (b) Marcus. Y. Aue and AppLChem. 1990,62,899 (c) Marcus, Y. LChemSoc Faraday Trans. 199187,1843 . LChem.Phys. (a) Matteoli, E.;Lepori, L J.Phys.Chem. 198286,1994 @) Mattwli, E.;Lepori, L 1984,80,2856 (c) Lepori, L.;Matteoli, E. LPhys.Chem. 1988Q26997 (d) Hamad, E.Z;Mansoori, G.k;Matteoli, E.;Lepori, L. Z.Phys.Chem 1989J62.27 Scheraga, H.A. Acc.ChemRes. 1979J2,7 Fendler, J.H. Membrane Mimetic Chemistry; Wiley; New York; 1982 WennerstrOm, H,Lindman, B. Phys.Rep. 1979,52,1 Kanijn, W.;Engberts, J.B.F.N. Tetrahedron Len 1978,1787 Mooij, H.J.;Engberts, J.B.F.N. RecL Trav.Chim.Puys-&rs 1988J07,185 Clark, kH.;Franks, F.;Pedley, M.D.;Reid, D . S . JAm.Chem.Soc. 1977,73,290 Abraham, M.H. LChemSoc.Fraday Tranr.Z198480,153 Gibson, KD.;Scheraga, H.A. Roc.NatLAcadSci USA. 1%7,58,420 Nozaki, Y.;Tanford, C. LBioLChem 1971,246,2211 Pratt, L.R.;Chandler, D. J.Chem.Phys. 1977,67,3683 Pratt, LR. Ann Rev.Phys.Chem 1985,36,433 Cramer 111, R.D. JAmChem.Soc. 1977,!29,5408 Okazaki, S.;Touhara, H.;Nakanishi, K J.Chem.Phys. 1984,81,890 Finney, J. Water SciRev. 1985J,92 Franks, F. Pure AppLChem. 1987,59,1189 Franks, E;Desnoyers, J.E. Water SciRev. 1985J,171

L i s t of references
McDonald, D.D.;McLean, A;Hyne, J.B. LSoLChem. 1978,7,63 Nashikawa, S.;Mashima, M. J.ChemSoc.Fraday Trans.1 1982,78,1249 Dipaola, G.;Belleau. B. CanJ..Chem. 198533,3452 Dipaola, G.;Belleau, B. CanJ.Ckm 197735,3825 Galema, S.k;Blandamer, MJ.;Engberts, J.B.F.N. JhChcnrSoc. 1990J12,%65 Galema, S.k;Hgiland, H. J.Phys.Chem. 199195,5321 Galema, S.k,Blandamer, MJ.;Engberts, J.B.EN. LOrg.Chem submitted Shibata, k,Yamashita, S.;Yamashita, T. BuILChemSocJpn. 1984,57,662 Herskovitz, T.T.;Behrens, CF.;Siuta, P.B.;Pandolfel, E.R. BiochimBiophysdcta 1988,493,192 Owen, J.N.;Smith, P.N. LChemSoc. 1952,4026 Olsen, S . ZNatuVorsch. 1946J.679 Engbersen, J.EJ.;Engberts, J.B.F.N. J.Am.ChemSoc. 197496,1231 Holterman, H.AJ.;Engberts, J.B.F.N. J&Soc.Soc. 1980J02.4256 Holterman, H.AJ.;Engberts. J.B.F.N. J.Org.Chem 1983,48,4025 Kanijn, W.;Engberts, J.B.F.N. unpublished results Roseman, M.;Jencks, W.P. J.Am.ChemSoc. 197597,631 Wetlaufer, D.B.;Malik, S.IC,Stoller, L;Coffin, R.I. J.Am.ChemSoc. 1964,86,509 Klotz, LM.;Franzen, J.S. J h C h m S o c . 1 % 2 , 8 4 , 3 4 6 1 Finer, EG.;Franks, F.;Tait, MJ. JsPmChemSoc. 197294,4424 Frank, H.S.;Franks, E LChemPhys. 1968,48,4746 Kerstholt, R.P.V.;Engberts, J.B.F.N.;Blandamer, M.J. J.Chem.Soc.;ChemComm. 1991,1230 Kertsholt, R.P.V.;private communications Diels, O.;Alder, K Justus LiebAmChem 1928,460,98 Lilley, T.H. "Water Science Reviews 5"; Franks, F. (ed.);University Press;Cambridge;l990;p.161 See for example, (a) Riicker, C.;Lang. D.;Sauer, J.;Friege, H.;Sustmann, R. Chem.Ber. 1980J13,1663 @) Huisgen, R. Pure ApplChem. 1980,52,2283 (c) Sauer, J.;Sustmann, R. Angew.Chem;ZntEdEn& 1980J9,779 Alder, K;Schumacher, M.Fortschr.ChemOrg.Naturstoffe 1953J0,66 Brieger, G.;Bennett, J.N. ChemRev. 1980,80,63 (a) Jung, M.E.;Geway, J. J.Am.Chem.Soc. 1989J11.5469 @) Jung, M.E.;Gervay, J. J h C h e m Soc. 1991J13.224 Hirst, S.C;Hamilton, AD. JAnChenrSoc. 1991J13.382 see for example Noy, R.S.;Gindin, V.A;Ershov, B.k;Kol'tsov, kI.;Zubkov, V.A Org.MagnReson. 1975J09.7 Mills, S.G.;Beak, P.J.Org.Chem. 1985,.50,1216 Almdal, K;Eggert, H.;Hammerich, 0. Acta ChemScan Part B 1986,40,230 Moriyasu, M.;Kato, A.;Hashimoto, Y. LChem.Soc. Perkin Tranr.11 1986,515 Stobbe, H. L i e b h C h e m . 1903,326,347 Elmsley, J.;Freeman, N.J. LMoLSmtct. 1987J61.193 Swain, C.G.;Swain, M.S.;Powell, kL;Alunni, S. JAmChem.Soc. 1983J05,502 Elmsley, J.;Freeman, N.J.;Parker, R.J. LChemSoc. Perkin Tram.11 1986,1479 (a) Reeves, LW. C d . C h e m 1957,35,1351 (b) Reeves, LW.;Schneider, W.G. CanLChem 1958,34793 Drexler, E.J.;Field, KW. J.ChemEd 197693,392 Spencer, T.C.;Holmboe, E.S.;Kirshenbaum, M.R.;Firth, D.W.;Pinto, P.B. CanLChem 1982,60,1178 Engbersen, J.EJ.;Engberts, J.B.F.N. J.Am.ChemSoc. 197597,1563 (a) Shorter, J; Advances in Linear Free Energy Relationships; Chapman,N.B.; Shorter, J.(Eds.); Plenum; New York; 1972 @) Shorter, J. Q.Rev.ChemSoc. 1970J4.433 Adcock, W.;Khor, T.-C. J.Org.Chem 1978,43,1272 Bordwell, EG.;Fried, H.E. Tetrahedron Lett 1977,1121 Brauman, J.I.;Blair, LK JAmChemSoc. 197092,5986 Houk, KN.;McAlduff, E.J.;Mollere, P.D.;Strozier, R.;Chang, Y.-M. J.ChemSoc.ChemComm 1977,141

List of references

MacPhee, A-J.;Dubois, J.-E.Tetrahedron Len. 1976,2471 Hudson, R.F.;Eisenstein, O.;Anh, N.T. Tetrahedron 1975,31,751 De Frees, DJ.;Hehre, W.J.;Sunko, D.E J h C h m S o c . 1979JO1.2323 De Tar, D.F.;Luthma, N.P. J h C h e m S o c . 1980J02,4505 Bingharn, R.C;Schleyer, P.von R. J h C h e m S m 197193.3189 Levitt, LS.;Widing, H.F. Prog.P@s..Olg.Chem. 1 9 1 5 J2,119 Hanson, P. I.ChemSoc.Perkin Trans.ll1984,lOl Taft, R.W.;"Steric Effects in Organic Chemistry"; Newmam. M.S.,Ed.; Wiley, New York, 1956, Chapter 13 Fliszar, S.;Beraldin, M.T. Cun.J.Chem 1979J7,1772 (a) Taft, R.W.;Taagepera, M.;Abboud, J.LM.;Wolf, J.F.;De Frees, D.J.;Hehre, W.J.;Bartmess, J.E.;McIver, R.Jr. J h C h e m S o c . 1978J00,7765 @) De Frees, D.J.;Taagepera, M.;Levi, B.k;Pollack, S.K;Summerhays, KD.;Taft, R.W.;Wolfsberg, M.;Hehre, W.H. JAmChemSoc. 1979J01.5532 Ritchie, CD.;Sager, W.F. Prog.Phys.Org.Chem. 1%4,2,323 Charton, M. JAmChemSoc. 1975,97,3691 De Tar, D.FJ.Org.Chem 1980,45,5166 (a) Charton, M. JAaChemSoc. 1977,!N,5687 @) Charton, M. J.Org.Chem 1979,44,903 (c) Charton, M. Rog.P@s..Olg.Chem1981J3,119 (d) Charton, M. LChemSoc. Perkin Trum.IZ 1983,97 MacPhee, JA;Panaye, k;Dubois, J.-E. J.o7g.Chem.1980,45,1164 Idow, J.P.;Schreck, J.O. J.l.Org.Chem 1978,43,4002 SjOstrOm. M.;Wold, S. J.ChemSoc.Perkin Duns 1 1 1979,1274 Charton, M , private communication Huheey, J.E. LOrg.Chem 1971,36,204 Fox, J.P:Jencks, W.P. JAmChem.Soc. 1974,96,1436 Rekker, R.F. The Hydrophobic Fragmental Constant; Elsevier; Amsterdam, 1977 Pinkus, kG.;Custard, H.CJr. LPhys.Chem 1970,74,1042 Pinkus, kG.;Lin, EY. J.MdStruct. 1975,24,9 Perjessy, A;private communication KrOger, CF.;Freiberg, WZChem 1965,5,381 (a) Mietchen, R.;Krt)ger, C.F. Z C h m 1%7,7,184 @) Hoggarth, H. LChemSoc. 1949,1160 Grieco, P.A.;Nunes, J.J.;Gaul, M.D. JAmChmSoc. 1990J12,4595; see also Forman, M.k;Dailey, W.P. J.Am.Chem.Soc. 1991J13,2791 in which the rate acceleration in by LiC104 in ether i s attributed to Lewis acid catalysis by Lithium ion. See, for example Woodward, R.B.;Sondheimer, F.;Taub, D.;Heusler, K;McLamore, W.M. JAmChem.Soc. 1952,744223 Sarett, LH.;Arth, G.E.;Luker, R.M.;Beyler, B.M.;Poos, G.L;Johns, W.F.;Constantin, J.M. J A m ChemSoc. 1952,74,4974 (a) Anand, N.;Bindra, J.k;Ranganathan, S. Art in Organic Synthesis; Holden-Day; San Francisco, 1970 @) More recent examples are Chigr, M.;Fillion, H.;Rougny, k;Berlion, M.;Riondel, J.;Beniel, H. ChemPhurm.BulL 1990,38,688 (c) Nebois, P.;Barret, R.;Houda, F. Tetrahedron Len. 1990j11.2569 Rozeboom, M.D.;Tegmo-Larsson, I-M.;Houk, KN. J.Org.Chem 1981,46,2338 Berson, JA;Hamlet, Z;Mueller, W . k JAmChmSoc. 1%2,84,297 Sera, k;Segushi, K ; Otsuki, Y.;Manuyama, K BulLChemSocJpn 1975,48,3641+ref.'s Sustmann, R.;Dern, M.;Kasten, R.;Sicking, W. ChenrBer. 1987J20,1315 See for example Birney, D.M.;Houk, KN. JAmChmSoc. 1990J12.4127 McCabe, J.R.;Eckert, C . k Acc.ChemRes. 1974,7,251 and references cited therein (a) Asano, T.;LeNoble, W.J. ChemRev. 1978,78,407 (b) Eldik, R.van;Asano, T.;LeNoble, W.J. Chem.Rev. 1989@,549 Mooijman, F.R.;Engberts, J.B.F.N. J.Org.Chem 1989,54,3993 Ketter, A;Glash. G.;Hemnann, R. LChemRes. 19909,278 Narasaka, K Synthesis 1991,l

LiFt of references

Roush, W.R. Adv. %loaddit. 1 9 9 0 , 2 , 9 1 Kelly, T.R.; Meghani, P.;Ekkundi, V.S. Tetrahedron Lett 199031,3381 (a) Insaacs, N.S. Liquid Phase High Pressure Chemistry, Wiley; New York, 1981 (b) Ben Salem, R.;Jenner, G. Tetrdwmn Lett 1 9 8 6 . 2 7 . 1 5 7 5 (c) Klarner, E-G.;Dogan, B.M.J.;Ermer, O.;von EDoering, W.;Cohen, M.P. Anp.Chem;Int.EdEngL 1986,27,1575 (d) Kltirner, F.-G. Chemie in Unserer Zeit 1 9 8 9 , 2 3 , 5 3 Dauben, W.G.;Gerdes, J.M.;Look, G.C. Synthesb 1986,7332 Nagy, O.B.; C d .C h 1985,63,1382 (a) Corsicocoda, k;Desimoni, G.;Ferrari. E;Righetti. P.P.;Tacconi, G. Teaahedron Len. 1984,40,1611 @) Desimoni, G.;Faita, G.; Righetti, P.P.;Tornaletti, N.;Visigalli, M. J.Chem.Soc. 1 1989,437 (c) Desimoni, G.;Faita, G.;Righetti, P.P.;Toma, L Tetrahedron Perkin Trans 1 1 9 9 0 . 4 6 , 7 9 5 1 Hunt, I.;Johnson, C.D. J. Chem.Soc.Perkin Trans 11 1991,1051 See for an excellent review and references: Ramamurthy, V. Tetrahedron 1986,42,5753 Fargues, R.;Maurette, M.T.;Oliveros, E,;Riviere, M.;Lattes, A JPhotochem 1982J8.101 Mayer, H.;Schuster, F.;Sauer, J. Tetrahedron Len. 1 9 8 6 . 2 7 . 1 2 8 9 Kreyszig, E. Advanced Engineering Mathematics ; Wiley, New York, 1979 Swinton, F.L;Rowlinson, J.S. Liquids and Liquid Mixtures; 3rd '4.;Buttenvorths; London, 1982 Burgess, J.;Hubbard, CD. JAmChemSoc. 1984,206,1717 Diederich, E;Smithrud, D.B. JslmChemSoc. 1990J12,339 Vitagliano, V.;Ortona, O.;Sartorio, R.;Constantino, L J.Chem.Eng.Dafa 1 9 8 5 , 3 0 , 7 @) Jorgensen, W.L;Severance, D.L JAm. C hSoc. 1990,212,4768 Brieger, G. JAmChemSoc. 1963,85,3783 Dobson, H.J.E. J.ChemSoc. 1925,2866 Benson, G.C.;Kiyohara, 0 LSoLChem. 1 9 8 0 , 9 , 7 9 1 (a) Segushi, K;Sera, k;Maruyama, K Tetrahedron Lett. 1973,1585 (b) Segushi, K;Sera, k,Maruyama, K. BullChemSoc.Jpn 1974,47,2242 (c) Segushi, K.;Sera, k;Otsuki, Y.;Maruyama, K BulLChemSocJpn 1975,48,3641 Taber, D.E Intramolecular Diels-Alder Reactions and Alder Ene Reactions; Springer; New York, 1984 Carlson, R.G. Ann.Rep.MedChem 19749,270 Mehta. G. LChemEduc. 1976J3,551 Oppoker, W .Angew-Chem; In&EdEng. 1977J6,lO Brieger, G.; Bennett, J.N. ChmRev. 1 9 8 0 , 8 0 , 6 3 Ciganek, E. Org.Reac&1984J2,l Fallis, AG. CanLChem 1984,62,183 Craig, D. ChemSoc.Rev. 1987J6,187 Oppolzer, W.;Keller, K J h C h e m S o c . 197193,3836 See for example Funk, R.L;Vollhardt, KP.C. Chem.Soc.Rev. 1980,9 (a) Hamood, LM.; Leeming, S.k;Isaacs, N.S.;Jones, G.;Pickard, J.;Thomas, R.M.;Watkin, D. Tetrahedron Len 1988,29,5017 @) Beeke, P.van den;Isaacs, N.S. Tetrahedron Lett. 1982,23,2147 Jenner, G . Angew.Chem.; Znt.EdEng. 1975J4,137 (a) Bilovic, D. Croat ChemActa 1%6$8,293 @) Bilovic, D. CroatChemActa 1968,40,15 Parker, Kk;Adamschuk, M.R. Tetrahedron Lett. 1978J9,1689 Gschwend, H.W.;Hillman, M.J.;Kisis, B.;Rodebaugh, R.K LOrg.Chem 1976,41,101 F i h e r , K;Hiinig, S. J.Org.Chem 1987,52364 Dauben, W.G.;Krabbenhoft, H.O. JAmChemSoc. 197496,3664 Bullock, M.W.;Hand, J.J.;Stokstad, E.LR. J.Am.ChemSoc. 1956,78,1956 Koga, Y.;Siu W.W.Y.;Wong, T.Y.H. J.Phys.Chem. 199094,7700 Haak, J.R. Ph.D. Thesis;Groningen, 1986 Kato, T.;Fuyijama, IQNomura, H. BulLChemSocJpn 1982J5.3368 Blandamer, MJ.;Burgess, J.;Robertson, R.E.;Scott, J.M.W. ChemRev. 1 9 8 2 , 8 2 , 2 5 9 Zucker, L;Hammett, LP. J&.ChemSoc. 1939,61,2791

L i s t of references
Brgnsted, J.N. Z.Phys.Chem 1922,l02,169 Galema, S.A.;Blandamer, M.J.; Engberts, J.B.F.N. J.Org.Chem 198924,1227 Abraham, M.H. I.ChemSoc.Faraday Trans.I1984,80,153 (a) Naghibi, H.;Dec, S.F.; Gill, S.J. J.Phys.Chem 198690,4621 (b) Naghibi, H.;Dec, S.J.;Gill, S.J. I.Phys.Chem 198791,245 (c) Naghibi,H.; Ownby, D.;Dec, S.E;GiIl, S.J. J.ChemEng.Data 1987,32,422 Rettich, T.R.;Handa, Y.P.;Battino, R.;Wilhelm, E. J.Phys.Chem. 1981,85,3230 Shinoda, K. J.Phys.Chem 1977,81,1300 (a) Stillinger, F.H. Science 1980,73,3404 @) Jonas, J. Science 1982,216,1179 (a) Nusselder, J.H.H.; Ph.D. Thesis; Groningen, 1990 @) Nusselder, J.J.H.;Engberts, J.B.F.N. J.Colloid IntdSci. 1991, in press Bot, k;Wegdam, G.H. J.Phys.Chem 199195,4679 Wood, R.H.;Thompson, P.T. Proc.NatLAcadSci. 1990,87,946 (a) Rashin, A.A. J.Phys.Chem. 199094,1725 (b) Rashin, k k ; Bukatin, M.A. J.Chem.Phys. 199195,2942 Jorgensen, W.L; Gao, J.; Ravimohan, C. J.Phys.Chem. 1985,89,3470 Sprik, M. J.Phys.Chem 199195,2283 Tobias, D.J.;Brooks 111, C.L. J.Chem.Phys. 199092,2582 Zichi, D.k;Rossky, P.J. J.Chem.Phys. 1986,84,1712 Jorgensen, W.L.;Buckner, J.K;Boudon, S.;Tirado-Rives, J. LChem.Phys. 1988@,3742 Hilvert, D.;Hi11, KW.;Nared, KD.;Auditor, M-T.M. JAmChemSoc. 1989,l11,9262 Blake, J.F.;Jorgensen, W.L. J.Arn.ChemSoc. 1991,l13,7430 Israelachvili, J.N.;Marcelja, S.; Horn, R.G. Q.Rev.Biophys. 1980J3,121

SUMMARY
The notion that reactivity of molecules and ions in solution' is largely dictated by the solvent was established more than a century ago. Since then, chemists have searched for a better understanding of solvent effects on chemical reactions. To date, many chemists still prefer to analyse solvent effects on the basis of macroscopic solvent parameters (dielectric permittivity, polarisability, etc.). Equally popular are quantitative or semi-quantitative methods, based on microscopic solvent polarity parameters (&(30), Z). The importance of specific interactions between reactants and activated complex on one hand and the solvent on the other hand is difficult to interpret on the basis of solvent polarity parameters. Unfortunately, these methods also fail to account for solvent effects arising from structural changes of the solvent during the activation process. In highly structured solvents, such as water, these structural changes might contribute significantly to the overall solvent effects. These shortcomings are even amplified in the analysis of solvent effects on chemical processes in mixed solvents. This must be attributed to preferential solvation of reactants or.activated complex. In addition, mixed solvents are frequently characterised by the presence of short-lived microheterogeneties, containing an excess of one of the components. Mixed aqueous solvents, which constitute attractive solvents for many different applications, are particularly sensitive to the occurrence of microheterogeneties. Recently, aqueous solutions were introduced as solvents for organic reactions that were conventionally carried out in organic solvents. The rate constants of some organic reactions appeared to be dramatically enhanced compared to those in organic solvents. In some cases, the stereospecificity of the reactions in aqueous solution was also significantly different from that in organic solvents. A major goal of this study was to develop a general model for a quantitative i x e d solvents. The applicability of analysis of solvent effects on chemical processes in m this model was appraised by the analysis of intriguing and often unexpected solvent effects on chemical processes in mixed aqueous solvents. Secondly, the newly developed model was used to elucidate the molecular origin of the remarkable accelerations of Diels-Alder reactions in aqueous solutions. In Chapter 1, the reader is introduced into the field of quantitative and semiquantitative methods that are currently available for the analysis of solvent effects. Futhermore, the recent development of organic chemistry in aqueous solutions is briefly outlined. Hydrophobic effects play a decisive role in solvent effects on organic reactions in (mixed) aqueous solvents. Ideas about the molecular origin of hydrophobic effects have been dominated by classic models of Frank, Evans and Kauzmann. Recently, these models have been severely criticised. In this chapter, a proper definition is provided for hydrophobic effects. Furthermore, some novel views on the molecular origin of hydrophobic effects are summarised. The theoretical models, developed in Chapter 2, set the stage for a quantitative analysis of solvent effects in mixed aqueous solvents, presented in the following chapters. In this chapter transition state theory and thermodynamic formalism to describe thermodynamic properties of a solution are drawn together. Both themes are firmly established in the literature and are only briefly introduced. The novel link between thermodynamic properties of a solution and kinetic parameters results in theoretical expressions which relate rate constants of chemical reactions in mixed solvents with the composition of the mixed solvent. With respect to the quantitative

description of the thermodynamic properties of the mixed solvent, a strict distinction is made between binary solvents (i) in which one of the solvents is present in a large excess and (ii) in which both solvents are present in comparable amounts. In the first case, a theoretical expression is derived on the basis of (i) the concentration dependence of excess thermodynamic properties of the total reaction medium, (ii) the solution theory of McMillan and Mayer and (iii) additivity schemes, advanced by Savage and Wood. Solvent effects are thus explained in terms of pairwise and higherorder group interaction parameters, representing interactions of groups, forming the cosolvent molecule, with those present in the reactants and the activated complex, respectively. In the second case, the possibility of preferential solvation of reactants and/or activated ,complex is taken into account. A theoretical model is developed on the basis of the solution theory of Kirkwood and Buff. Following this approach, solvent effects in mixed solvents are quantified by the affinity of reactants and activated complex, respectively, for the solvent molecules present in the mixture. The methodology, which is developed to account for solvent effects on rate constants of unimolecular reactions in mixed solvents, is subsequently extended for the analysis of solvent effects on rate constants of (i) solvolysis reactions, (ii) bimolecular reactions and (iii) chemical equilibria. In addition, methods are advanced for a quantitative treatment of solvent effects on isobaric entropies and enthalpies of reactions in mixed solvents where the cosolvent is present in small amounts. In Chapter 3, the theoretical expression, derived in Chapter 2, is critically tested for the quantitative analysis of solvent effects of a series of 24 monohydric and polyhydric alcohols on the pseudo-first-order rate constant for the neutral hydrolysis in dilute aqueous solutions ( c 2 mol kg-'). Emphaof 1-benzoyl-3-phenyl-l,2,4-triazole sis is placed on a critical appraisal of the applicability of group additivity schemes. Linear alcohols having the same "hydrophilic framework" (the relative position and number of hydroxy groups is similar) exhibit a remarkable additivity with respect to the contribution of the apolar groups to the overall medium effects. The contribution of the hydroxy groups to the overall solvent effect of polyhydric alcohols cannot be treated by applying additivity schemes. The hydrolysis reaction, which is characterised by an apolar reactant and a polar activated complex, is retarded by the addition of the organic cosolvents. The methylene groups are shown to decrease the rate constant, whereas the hydroxy groups enhance the rate constant. A main conclusion is that the solvent effects largely reflect hydrophobic interactions of the cosolvent molecules with the reactant and the activated complex, respectively. These interactions lead to participation of the organic cosolvent in the solvation shell of the reactants and activated complex. This process involves destructive overlap of hydrophobic hydration shells. Hydroxy groups reduce the overall hydrophobicity of the organic cosolvents. However, hydroxy groups also reduce the importance of hydrophobic hydration of the apolar hydrocarbon moieties for the overall hydration of the cosolvent molecule. It is argued that the absence of these local hydrophobic hydration shells increases the accessibility of the methylene moieties for interactions with apolar reactants and/or activated complexes in the aqueous solution. Hence, polyhydric cosolvents exhibit some "hydrophobic" character notwithstanding the fact that polyhydric alcohols are generally very soluble in water. The insights which are obtained are of direct relevance for molecular recognition processes. Finally, it is shown that urea reduces the solvent effect of monohydric alcohols by interfering with the process of hydrophobic interactions. In Chapter 4 it is shown that pairwise and higher-order Gibbs energy interaction parameters fully account for the solvent effects on the pseudo-first-order rate constant

for the neutral hydrolysis of p-methoxyphenyl dichloroacetate in aqueous solutions of urea and alkyl-substituted ureas, even in case when the concentration of the cosolvents is very high (0-8 mol kg-'). Although the importance of third-order contributions increases with increasing concentration, pairwise interactions largely govern solvent effects on the rate constant of reactions in these aqueous solvents. A different pattern is observed for the quantitative analysis of solvent effects on A2H0 and TA2S0. The results indicate that solvent effects on these activation parameters strongly reflect structural changes of the solvent during the activatioh process. It is argued that a quantitative analysis of the dependence of enthalpies and entropies of activation on the corn osition of the aqueous solvent is only valid in very dilute solutions (<I-1.5 mol kg - ). In the second part of this chapter, equations for a quantitative analysis of solvent effects in mixed solvents, derived in Chapter 2, are successfully applied for the quantitative analysis of solvent effects on, successively, (i) the intramolecular DielsAlder reaction of N-furfuryl-N-methyl maleamic acid, (ii) the bimolecular Diels-Alder reaction of cyclopentadiene with methyl vinyl ketone and (iii) the keto-nol equilibrium of 24-pentanedione in aqueous solutions of monohydric alcohols. Chapter 5 describes a study of alkyl substituent effects on the pseudo-first-order rate constants for the neutral hydrolysis of a series of 18 1-acyl-falkyl(or pheny1)1,2,4-triazoles in dilute aqueous solutions of ethanol and 1-propanol, respectively. Current ideas about the molecular origin of alkyl substituent effects are briefly outlined. Generally, the influence of the solvent on the magnitude of alkyl substituent effects is completely neglected. The present results provide strong evidence that this is fully unjustified for the analysis of alkyl substituent effects on reactions in aqueous solvents. Increasing the hydrophobicity of the reactant by introducing alkyl substituents or by increasing the size of the alkyl substituent destabilises the reactant with respect to the activated complex. Consequently, the reactant becomes more prone to hydrolysis. The concomitant increase of the pseudo-first-order rate constant is defined as "hydrophobic acceleration". In addition, this chapter describes some intriguing spectroscopic data of the reactants, solvent deuterium isotope effects and rate constants for the acid-catalysed hydrolysis reaction in aqueous media. Chapter 6 is concerned with solvent effects on intermolecular Diels-Alder reactions of cyclopentadiene with alkyl vinyl ketones and with S-substituted-naphthoquinones in binary mixtures of water with ethanol and 1-propanol, respectively. Although rate constants of Diels-Alder reactions are known to be only moderately affected by changes in the solvent, the rate constants are considerably enhanced in dilute aqueous solutions. The occurrence of "hydrophobic packing" of the diene and the dienophile, often suggested to be the driving-force for the observed acceleration of Diels-Alder reactions in aqueous solutions, is excluded on the basis of the experimental results. The standard chemical potentials of reactants and activated complexes, as well as the isobaric activation parameters for the Diels-Alder reactions in water and in mixtures of water with 1-propanol reveal that the "hydrophobic acceleration" has to be attributed to "enforced hydrophobic interactions" between the diene and the dienophile during the activation process. Largely responsible for this fact is the reduction of apolar surface of the reactants during the activation process. The term "enforced" is introduced to emphasise that (i) the association of diene and dienophile is, in terms of Gibbs energy, an unfavourable process and that (ii) the geometry of the complex is imposed by the reaction pathway. Tentatively, the suggestion is made that the aqueous solvent induces an enhanced polarity of the activated complex. This chapter also contains a study of the endolexo product distribution of the Diels-Alder

reactions in mixed aqueous solvents. In dilute aqueous solutions the endo products appear to be almost exclusively formed. Addition of organic cosolvents induces a considerable decrease of the rate constants. Remarkably, the largest reduction of the rate constants are observed within a rather limited concentration range of cosolvent. Quantitative analysis of the solvent effects indicates that the reduction of the rate constants coincides with the presence of water-rich and cosolvent-rich microheterogeneties in the mixed aqueous solvent. The reduction of the rate constant can be attributed to preferential solvation of the reactants. The rate constants of intramolecular Diels-Alder reactions are also considerably enhanced in aqueous solutions. In Chapter 7, solvent effects are reported on intramolecular Diels-Alder reactions of a series of N-furfuryl-N-alkyl maleamic acids in mixed aqueous solutions of ethanol and 1-propanol. Solvent effects on the cisltrans isomerisation of the amide bond are shown not to be responsible for the acceleration of the intramolecular ring closure. Remarkably, in spite of a covalent bridge between both hydrophobic moieties, intramolecular association of the diene and the dienophile is not enhanced in aqueous solutions. The solvent effects on the intramolecular DielsAlder reactions are almost fully identical to those on the bimolecular Diels-Alder processes, described in Chapter 6. The "hydrophobic acceleration" of intramolecular Diels-Alder reactions in dilute aqueous solvent is thus ascribed to "enforced hydrophobic interactions" between the covalently bridged diene and dienophile. Chapter 8 evaluates the merits and shortcomings of the quantitative treatment of solvent effects on reactions in mixed aqueous solvents, developed and applied in this thesis. However, the largest part of this chapter is devoted to the introduction of a novel model for hydrophobic effects. An important feature of this model is that the formation of a hydrophobic hydration shell is proposed to increase the interactions between apolar solutes and water. This implies that the formation of a hydrophobic hydration shell increases the solubility of hydrophobic compounds and decreases the strength of hydrophobic interactions. Furthermore, the model stresses the temperature dependence of hydrophobic effects. Since hydrophobic hydration shells "melt" with increasing temperature, both hydrophobic interactions and hydrophobicity increase with increasing temperature. The favourable Gibbs energy, associated with hydrophobic interactions, is considered to be largely determined by enthalpic effects, resulting from the difference between the London dispersion interactions between the hydrophobic solutes on one hand, and the solute-water interactions on the other hand. This chapter is concluded with a general analysis of organic reactivity in aqueous solutions. It is argued that organic reactions, involving apolar reactants and reagents can be subjected to "hydrophobic acceleration" and changes of the stereospecificity of the process. Most pronounced effects are anticipated for reactions which involve a nonpolar activated complex. The solubility of the reactants can be enhanced by addition of apolar cosolvents, although strong preferential solvation of the reactants by the organic cosolvents, must be avoided. An initial impetus is provided for a further systematic study of organic reactivity in aqueous solutions.

A1 meer dan een eeuw geleden groeide het besef dat de reactiviteit van moleculen en ionen in oplossing grotendeels wordt bepaald door het oplosmiddel. Sinds die tijd hebben chemici gezocht naar methoden om deze mediumeffecten beter te begrijpen. Ook nu nog analyseren veel chemici mediumeffecten bij voorkeur aan de hand van parameters, die een maat zijn voor macroscopische eigenschappen van het oplosmiddel (dielectrische permittiviteit, polariseerbaarheid, etc.). Kwantitatieve en semikwantitatieve methoden, gebaseerd op polariteitsparameters (ET(30), Z), mogen zich evenzeer verheugen in een grote populariteit. Bij een dergelijke aanpak wordt het belang van specifieke interacties tussen de reactant en het geactiveerd complex enerzijds en het oplosmiddel anderzijds sterk onderschat en hun bijdrage kan in feite niet worden bepaald. Bovendien zijn deze methoden ontoereikend om stmcturele veranderingen van het oplosmiddel, die in het medium plaatsvinden tijdens het activeringsproces, kwantitatief te analyseren. Zeker in sterk gestructureerde oplosmiddelen, zoals water, kunnen deze effecten een grote invloed hebben op het waargenomen mediumeffect. Deze tekortkomingen worden nog eens versterkt als oplosmiddelparameters worden gebmikt om mediumeffecten op chemische processen in gemengde oplosmiddelen te analyseren. Een belangrijke oorzaak hiervan is het optreden van voorkeursolvatatie van reactanten enlof geactiveerd complex. Bovendien worden gemengde oplosmiddelen vaak gekenmerkt door de aanwezigheid van microheterogeniteiten waarin B C n van de componenten in overmaat voorkomt. Met name gemengde waterige oplosmiddelen, voor vele toepassingen bijzonder aantrekkelijke media, zijn berucht vanwege deze verschijnselen. In het afgelopen decennium is steeds vaker melding gemaakt van het gebruik van waterige oplossingen als media voor organische reacties die traditioneel in organische oplosmiddelen worden uitgevoerd. De snelheidsconstante van een aantal organische reacties in waterige oplosmiddelen bleek opmerkelijk veel groter te zijn dan die in organische oplosmiddelen. In een aantal gevallen werd ook een duidelijk afwijkende stereospecificiteit van de reactie waargenomen. Een belangrijk doe1 van deze studie was het ontwikkelen van een algemeen geldende theorie voor een kwantitatieve analyse van oplosmiddeleffecten op chemische processen in gemengde oplosmiddelen. De toepasbaarheid van de theorie werd getest door het analyseren van de vaak onverwachte en intrigerende oplosmiddeleffecten op chemische processen in gemengde waterige oplosmiddelen. Een tweede doe1 was om de ontwikkelde theorie te gebruiken om, via een kwantitatieve studie van mediumeffecten, de moleculaire basis van de opvallende mediumeffecten op DielsAlder reacties in waterrijke media beter te begrijpen. In hoofdstuk 1 wordt de lezer ingewijd in het terrein van kwantitatieve en semikwantitatieve methoden die momenteel worden gebruikt voor het analyseren van oplosmiddeleffecten. Bovendien worden de recente ontwikkelingen van organische chemie in waterrijke oplosmiddelen kort uiteengezet. Hydrofobe effecten spelen een doorslaggevende rol bij oplosmiddeleffecten in watemjke oplosmiddelen. De visie op hydrofobe effecten is lange tijd bepaald geweest door de inmiddels klassiek geworden modellen van Kawrnann, Frank en Evans. De laatste tijd staan deze modellen echter steeds vaker bloot aan kritiek. In dit hoofdstuk worden hydrofobe effecten nauwkeurig gedefinieerd en wordt een overzicht gegeven van enkele belangrijke nieuwe visies op de moleculaire basis voor het optreden van hydrofobe effecten. De theoretische modellen die worden ontwikkeld in hoofdstuk 2 vormen de basis

voor de kwantitatieve analyses van oplosmiddeleffecten op chemische processen in waterige oplosmiddelen zoals die in de volgende hoofdstukken staan beschreven. In dit hoofdstuk wordt een relatie gelegd tussen de theorie van het geactiveerd complex enerzijds en een thermodynamische benadering van het reactiemengsel anderzijds. Het relateren van de thermodynamics van het reactiemengsel aan kinetische parameters is nieuw. In het algemeen leidt deze link tot vergelijkingen die een verband leggen tussen snelheidsconstanten voor reacties in gemengde oplosmiddelen en de samenstelling van het oplosmiddel. Er wordt met betrekking tot de thermodynamische beschrijving van het reactiemengsel een onderscheid gemaakt tussen binaire oplosmiddelen waarbij (i) b6n van beide oplosmiddeIen in grote overmaat aanwezig is en (ii) waarbij de componenten in vergelijkbare hoeveelheden aanwezig zijn. In het eerste gevd wordt gebruik gemaakt van (i) de concenuatie afhankelijkheid van thermodynamische eigenschappen van het reactiemengsel, (ii) de theorie van McMillan en Mayer en (iii) het additiviteitsprincipe, geintroduceerd door Savage en Wood. Oplosmiddeleffecten worden beschreven in termen van groepsinteractie-parameters voor de interacties tussen de functionele groepen die tesamen het cosolventmolecuul vormen en die, respectievelijk, van de reactanten en het geactiveerd complex. In het tweede geval wordt rekening gehouden met voorkeursolvatatie van reactanten enlof geactiveerd complex op basis van de theorie van Kirkwood en Buff. Het oplosmiddeleffect wordt in dit geval uitgedrukt in de affiniteit van de reactanten en het geactiveerd complex voor de componenten van het gemengde oplosmiddel. De methodologie, die is ontwikkeld voor het kwantitatief beschrijven van oplosmiddeleffecten op een unimoleculaire reactie, wordt vemolgens uitgebreid naar een analyse van oplosmiddeleffecten op (i) bimoleculaire reacties, (ii) solvolyses en (iii) evenwichten. Ook wordt een vergelijking afgeleid die een kwantitatief verband legt tussen de entropie en de enthalpie van activering voor een reactie, en de samenstelling van het verdunde gemengde oplosmiddel. In hoofdstuk 3 wordt een kwantitatieve analyse beschreven van oplosmiddeleffecten van 24 monohydrische en polyhydrische alcoholen op de pseudo-eerste-orde in verdunde snelheidsconstante voor de hydrolyse van 1-benzoyl-3-fenyl-l,2,4-triazool (<2 mol k g ' ) watemjke oplosmiddelmengsels. Het model, zoals dat is afgeleid in hoofdstuk 2, wordt kritisch getest met betrekking tot de toepasbaarheid van het additiviteitsprincipe. Lineaire alcoholen waarbij zowel het aantal als de relatieve positie van de hydroxygroepen identiek zijn, vertonen een verrassend goede additiviteit van de bijdrage van de hydrofobe groepen tot het totale oplosmiddeleffect. De bijdrage van de hydroxygroepen tot het totale oplosmiddeleffect van polyhydrische alcoholen kan echter niet op grond van het additiviteitsprincipe worden vastgesteld. De reactie, die gekenmerkt wordt door een apolaire reactant en een polair geactiveerd complex, wordt vertraagd door alle organische cosolvents. De methyleengroepen vertragen de reactie, terwijl de hydroxygroepen de reactie versnellen. De belangrijkste conclusie is dat hydrofobe interacties tussen het cosolvent enerzijds en de reactanten en het geactiveerd complex anderzijds primair verantwoordelijk zijn voor het waargenomen oplosmiddeleffect. Deze interacties leiden tot de deelname van het cosolvent in de solvatatiemantel van de reactant en het geactiveerd complex. Dit proces gaat gepaard met overlap van hydratatieschillen. De hydroxygroepen verminderen de hydrofobiciteit van de organische cosolvents. Bovendien verkleinen de hydroxygroepen het belang van hydrofobe solvatatie van apolaire groepen voor de totale solvatatie van het organische cosolvent. Dit leidt ertoe dat apolaire groepen minder worden afgeschermd voor interacties met een apolair reactant enlof geactiveerd complex. Dit verklaart het ogenschijnlijk hydrofobe karakter van overigens zeer goed

in water oplosbare polyhydrische alcoholen. Hierbij wordt een verband gelegd met moleculaire herkenning. Tenslotte wordt aangetoond dat ureum de mediumeffecten van de monohydrische alcoholen reduceert door het belang van hydrofobe interacties te venninderen. Hoofdstuk 4 laat zien dat paarsgewijze en hogere orde interactieparameters toereikend zijn voor een nauwkeurige en kwantitatieve beschrijving van oplosmiddeleffecten op de snelheidsconstante voor de neutrale hydrolyse van p-methoxyfenyl dichlooracetaat in waterige oplossingen van ureum en ureumderivaten, zelfs als deze in zeer hoge concentratie aanwezig zijn (0-8 mol kg-'). Hoewel het belang van hogere orde interacties nadrukkelijk toeneemt bij toenemende concentratie aan cosolvent, domineren niettemin paarsgewijze interacties de waargenomen mediumeffecten op de snelheidsconstanten. Anders is dit voor de enthalpie en entropie van activering. De resultaten tonen aan dat oplosmiddeleffecten op activeringsparameters vooral structurele veranderingen van het oplosmiddel tijdens het activeringsproces weerspiegelen. Een kwantitatieve analyse van oplosmiddeleffecten op A+HOen TA*SOdoor gebruik te maken van onze theorie is slechts zinvol in zeer verdunde waterige oplosmiddelen (<I-1.5 mol kg-'). Het tweede deel van dit hoofdstuk beschrijft een succesvolle kwantitatieve analyse van oplosmiddeleffecten op, achtereenvolgens, (i) de (ii) de intramoleculaire Diels-Alderreactie van N-furfuryl-N-methylmaleaminezuur, bimoleculaire Diels-Alderreactie van cyclopentadieen met methylvinylketon en (iii) het ketoenol evenwicht van 2,Cpentaandion in verdunde waterige oplossingen van monohydrische alcoholen. Hoofdstuk 5 geeft de resultaten van een studie naar het substituenteffect van alkylgroepen op de pseudo-eerste-orde snelheidsconstante voor de neutrale hydrolyse van 18 1-acyl-3-alkyl(of feny1)-1,2,4-triazolen in waterrijke oplossingen van ethanol en 1-propanol. De bestaande opvattingen over substituenteffecten van alkylgroepen worden kort uiteengezet. Hierin wordt de invloed van het oplosmiddel meestal geheel verwaarloosd. Onze resultaten tonen aan dat dit met name voor substituenteffecten van alkylgroepen op reacties in waterige oplosmiddelen volledig ten onrechte is. Het vergroten van de hydrofobiciteit van reactanten door het invoeren of vergroten van alkylsubstituenten blijkt te leiden tot een destabilisatie van de reactant ten opzichte van het geactiveerd complex. De vergroting van de snelheidsconstante die hiervan het gevolg is, wordt "hydrofobe versnelling" genoemd. In dit hoofdstuk worden bovendien opmerkelijke spectroscopische gegevens van de reactanten, oplosmiddel deuterium isotoopeffecten en snelheidsconstanten voor de zuurgekatalyseerde hydrolyse beschreven. Hoofdstuk 6 gaat in op mediumeffecten op bimoleculaire Diels-Alderreacties van cyclopentadieen met alkylvinylketonen en gesubstitueerde naftochinonen in mengsels van water met respsctievelijk ethanol en 1-propanol. Hoewel de snelheidsconstanten voor Diels-Alderreacties slechts weinig gevoelig zijn voor veranderingen van het oplosmiddel, worden gaande naar waterrijke media uitzonderlijk grote versnellingen gevonden. De experirnentele resultaten tonen aan dat aggregatie van de reactanten, vaak verondersteld als de reden voor de versnelling van intermoleculaire DielsAlderreacties in waterige oplossingen, niet optreedt. Op grond van de standaard chemische potentialen van reactanten en geactiveerde complexen en activeringsparameters voor de Diels-Alderreacties in water en in mengsels van 1-propanol en water wordt de "hydrofobe versnelling" toegeschreven aan de "gedwongen hydrofobe interactie" tussen het dieen en het dienofiel. De reductie van het hydrofobe oppervlak van de reactanten tijdens het activeringsproces speelt hierbij een doorslaggevende rol. De interactie wordt gedwongen genoemd omdat (i) de associatie ongunstig is met

betrekking tot de Gibbs energie en (ii) omdat het complex een door de reactie bepaalde geometrie bezit. Gesuggereerd wordt dat water in staat is een verhoogde polariteit van het geactiveerd complex te induceren. Dit hoofstuk bevat tevens een studie naar de oplosmiddeleffecten op de endolexo productverhouding waaruit blijkt dat in waterige oplossingen vrijwel uitsluitend endo product wordt gevormd. Het toevoegen van organische cosolvents aan het waterige milieu leidt tot een dramatische verlaging van deze snelheidsconstanten. Opmerkelijk is dat een daling van de reactiviteit plaats vindt in een betrekkelijk smal concentratiegebied. Deze kwantitatieve analyse van deze oplosmiddeleffecten toont aan dat in dit concentratiegebied het gemengde oplosmiddel wordt gekenmerkt door de aanwezigheid van zeer dynamische cosolventrijke en watemjke fasen. Het blijkt dat de daling van de snelheidsconstante een gevolg is van voorkeursolvatatie van de reactanten door het organische cosolvent. Dat ook intramoleculaire Diels-Alderreacties worden versneld in waterige oplosmiddelen, wordt verder beschreven in hoofdstuk 7. Het hoofdstuk bevat een kwantitatieve analyse van oplosmiddeleffecten op de intramoleculaire Diels-Alderreacties van Nfurfuryl-N-alkylmaleaminezurenin waterige oplossingen van ethanol en 1-propanol. Oplosmiddeleffecten op cis-trans isomerisatie rond de amidebinding blijken niet verantwoordelijk te zijn voor de waargenomen versnelling van de intramoleculaire ringsluiting in waterig milieu. Opmerkelijk is ook dat intramoleculaire associatie van het dieen en het dienofiel in water niet wordt waargenomen. Oplosmiddeleffecten op de intramoleculaire Diels-Aldeneactie in water/alcohol mengsels zijn geheel vergelijkbaar met die op de intermoleculaire Diels-Alder reacties. De "hydrofobe versnelling" kan dan ook worden toegeschreven aan "gedwongen hydrofobe interacties" tussen het covalent gebonden dieen en dienofiel. In hoofdstuk 8 wordt de kwantitatieve analyse van oplosmiddeleffecten in gemengde oplosmiddelen, zoals die in dit proefschrift wordt beschreven, kritisch geevalueerd. Het grootste deel van dit hoofdstuk is echter gewijd aan de introductie van een nieuwe visie op hydrofobe effecten. Een belangrijk aspect van deze visie is dat de vorming van een hydrofobe hydratatiemantel wordt verondersteld te leiden tot een gunstigere interacties tussen een hydrofobe verbinding en water. Dit impliceert dat hydrofobe hydratatie de oplosbaarheid van hydrofobe verbindingen vergroot en de sterkte van hydrofobe interacties verkleint. In het model wordt bovendien veel aandacht besteed aan de temperatuurafhankelijkheid van hydrofobe effecten. Aangezien een hydrofobe hydratatiemantel "smelt" bij toenemende temperatuur, nemen zowel de hydrofobiciteit als de sterkte van hydrofobe interacties toe met toenemende temperatuur. Tenslotte wordt benadrukt dat de enthalpie voor London-dispersie interacties tussen de hydrofobe verbindingen enerzijds, en de zwakke interactie tussen de hydrofobe 9erbindingen en water andexzijds, een dominante rol speelt bij hydrofobe interacties. Het hoofdstuk wordt besloten met een beschouwing waarin in algemene zin wordt aangegeven dat organische reacties waarbij apolaire reagentia en reactanten zijn betrokken in aanmerking kunnen komen voor een "hydrofobe versnelling" en verandering van de stereospecificiteit wanneer de reactanten (een deel van) hun apolaire karakter verliezen tijdens het activeringsproces. Hierbij zijn met name reacties met een niet geladen geactiveerd complex aantrekkelijke kandidaten. De oplosbaarheid van de organische reactanten kan hierbij worden verbeterd door het toevoegen van organische cosolvents waarbij voorkomen dient te worden dat de reactanten grotendeels of geheel worden gesolvateerd door het toegevoegde organische cosolvent. Tenslotte wordt een eerste aanzet gegeven tot een systematische studie naar de toepasbaarheid van water als oplosmiddel voor synthetisch organische toepassingen.