Pure & Appl.Chern., Vol.54, No.10, pp.1867—1884, 1982. 0033—4545/82/101867—18$03.OO/OPrinted in Great Britain. Pergarnon Press Ltd.

©1982 IUPAC

SOLVENT EFFECTS ON CHEMICAL REACTIVITY

Christian Reichardt

Department of Chemistry, Philipps-University,

Hans-Meerwein-StraBe, D - 3550 Marburg, Fed. Rep. Germany

Abstract - Typical examples for different types of solventeffects on chemical reactivity are given, e. g. solvent effectson reaction rates, on the position of chemical equilibria,on competitive reaction mechanisms, on dichotomic reactionpaths, on chemoselectivity, and on stereoselectivity.According to their solvent sensitivity, most organic reactionscan be classified into dipolar transition-state reactions,isopolar transition-state reactions, and free-radicaltransition-state reactions, typical examples of which aregiven.Attempts to describe the solvation capability by virtue ofempirical parameters of solvent polarity are mentioned.Particular attention is merited by the ET(30)-parameter, anempirical parameter derived from a negatively solvatochromicpyridinium-N-phenolate betaine dye as reference compound. Someexamples of the application of this ET(30)-scale to solvent-sensitive chemical reactions are given.

INTRODUCTION

In 1862, Marcelin Berthelot, Professor at the College de France in Paris(Fig. 1) and his coworker Leon Péan de Saint-Gilles carried out a remarkableexperiment (Ref. 1): They mixed together equivalent amounts of acetic acidand ethanol and measured the rate of esterification by determining at distincttime intervals the amount of the resulting ethyl acetate.

In order to study the influence of temperature on the reaction rate, onepart of the reaction mixture was put in the cool cellar, whereas the otherpart was kept in the warmer laboratory rooms. The influence of the reactionvolume (i.e. the concentration of the reaction partners) was examined bydiluting the reaction mixture with different inert organic solvents such asdiethyl ether and benzene.

It happend that — at constant temperature, time and concentration — thereaction rate was very low in the ethereal solution, only traces of estercould be detected. In benzene solution, however, a substantial amount of esterwas formed. In their classical paper "Recherches sur les affinités", publishedin 1862, Berthelot and St. Gilles stated that "... the esterification isdisturbed and decelerated on the addition of neutral solvents not belongingto the reaction" (Ref. 1). That was the first experimental observation showingthat solvents may have an influence on chemical reactions.

But these interesting results seem to have been forgotten for many years.It was only in 1887 that Nicolai Menshutkin (Fig. 2), Professor at theUniversity of Sankt Petersburg, again took up these investigations.

After thorough studies on the esterification reaction and on the alkylationof tertiary amines with alkyl halides, he concluded, in 1890, that a reactioncannot be separated from the medium in which it is performed (Ref. 2).

1867

qTacoAeLeq tOL 4J- tTLac cTwG 111 J8 P? rfqMTd GT2GLJ TU vcpiJ (J�Gt 3)LP6 TIJtTnGUCG O 2OJA6IJ2 OIJ 4JJ6 boaTcTou o cpGuGj ednrTpLTg Mga

ao]AeUç GGC2 OJJ L6cçoJJ gcaa 8 ço 2OUJG excejjc cTTT AflqLgcea LPT2 acgcemejc GAGLJ LYOM' J50 kGgLa gce cJJG JL2c qacoAeL7 oce bp7TC BLObGLcTG2 O OLaJJ-c aOJAaIJca gijq cpGTL TJJtTnGIJCG OLT LegcçTouacceq LeaaJJeq]-? Tu .JOO çpgç cpe Ta 0IJTA g AeL? TOO2G couJJeccTou paçeeuGtteCc gijq qT2çIJCf bp7aaJ OL cpGuJTcgJ fLobGLcTGa a cpa 2OJAGU2 naeqt jJ

fceL eçLou eoLçe ço TIJq g CoLLjgçou p64MGGJJ cPT2 ]IJq o 2OAGJJç

ETa 3 O]A6IJç 6GG2 QJJ g GJJ2pflc)qIJ LegccToLT (je

bo1aL!O1A6Ut

516I

Joo 0c(C3H2)3k11 + 0H5-I

OH3

I (C5H2)3V CH5I (C5H2)3V-CH3 + I0

L 20] CH3

2'5 D!2b(OC6W6U b6OCPOU Ot 1L!6pIaw!ue MIJ4J EW'1 oqiqe

20\t\3 ECS Ox.) QCO) 3QSr6.

TJJcLegaG (ETa' 3)•tar cpa LecçaIJ peçaeu cLTecpA]gwTue gug ecpEj aqqe caciaea g 5-aq gce

}Je arrnq tar. exwba' çpgç g aajAeJJç cpude tram u-pexgiJe ça pGij capa

ELLT]usLcJ[CDCp-BTTqLCpTA'

— JO)ETd j WLce]TLr ecperac

ELgLr]tr1Lci'

(J95 —

ETd 5' 11TJWJT weJJapcc}cpJ

C' BEICHYBDI

2QIA6U U - HGXQUG B6U6U6 6pO UO B6U\1 GICOP0I

Solvent effects on chemical reactivity 1869

Arthur Hantzsch in WUrzburg (Ref. 4), Ludwig Knorr in Jena (Ref. 5), andJohannes Wislicenus in Wtirzburg (Ref. 6), simultaneously with the discoveryof keto-enol tautomerism in 3-dicarbonyl compounds, an example of which isshown in Fig. 4 (Ref. 7, 8).

S olven t Effect on Chemical Equilibrium:

Keto-Enol Tautomerism of Ethyl Acetoacetate

KTH3CyOC2H5

33°C

Dketo form cis-EnoL form

Solvent Gas phase n-Hexane Benzene Methanol Aceticacid

0,74 Q64 Q19 Q062 Q019

% Eno 1 43 39 16 5,8 1,9

solvent ___________________________________polarity

Fig. 4. Solvent effects on a tautomeric equilibrium(Ref. 7, 8).

The tautomeric equilibrium constants of ethyl acetoacetate for this cis-enolizing 3-dicarbonyl compound ir1icate an increasing enol content withdecreasing solvent polarity.

TYPES OF SOLVENT EFFECTS ON CHEMICAL REACTIVITY

This short historical introduction has already given two examples ofsolvent effects on reaction rates and chemical equilibria. Dealing withsolvent effects on chemical reactivity includes, however, a greater varietyof solvent effects. Four typical examples, taken mainly from recent publi-cations, are shown in Fig. 5 ... 8.

First, a solvent can drastically change the mechanism of a reaction. Anexample is the thermolysis of arene diazonium salts in solution (Fig. 5),

Solvent Effect on Competitive Reaction Mechanisms:

Dediazoniation of Arenediazonium Salts

Solvents of low r e®LAr.NNJ Ar ÷NEN

nuci eop hi lici ty

(HFIP, TFE, AcOH, Heterolytic bond cleavageH20 (pH< 1)]

Ar—NEN k1 products

/5@ ô@:

[Ar."NNS] Ar + NEN +Solvents of high

nucleophiticityHomolyfic bond cleavage

IPyridineHMPT. DMSOI

Fig. 5. Solvent effects on competitive reaction mechanisms (Ref. 9).

1870 C. REICHARDT

where two competitive reaction mechanisms have been established. Theseinvestigations were mainly carried out by Zollinger and his coworkers(Ref. 9).

In solvents of low nucleophilicity such as hexafluoroisopropanol (HFIP)or trifluoroethanol (TFE) the first and rate-limiting step is a heterolyticcleavage of the diazonium ion to give an aryl cation and molecular nitrogen,followed by rapid reaction of the cation with any available nucleophile.

In solvents of high nucleophilicity and low redox potential such as pyri-dine or dimethylsulfoxide (DMSO), however, a homolytic decomposition of thediazonium ion is favoured to give an aryl radical and consequent reactionproducts.

Thus, depending on the solvent used, a substrate can react by two com-pletely different reaction mechanisms, as shown by this dediazoniation re-action.

Different reaction mechanisms and sometimes different reaction productscan be obtained from the same reaction partners simply by changing thesolvent. A nice example of this is the cycloaddition reaction of an imidazo-[2,1-bjthiazole with dimethyl acetylene dicarboxylate (Fig. 6), which hasbeen studied by Abe and coworkers (Ref. 10).

Solvent Effect on Dichotomic Reaction Paths:

Cyctoaddition Reaction of Imidazo[2,1-blthiazole withDimethyt Acetylenedicarboxylate

E

in Xytene S.frN ______ S

NJC6H5 NCC6H5

EESN EXE __CH3 [çj1ç, çTJ.-C6H5 -1l.-Dipolor )— N •(H--C-CH3)

Cycloodition in — S

dipolar oprotic CH3solventImidazotl,2-o1pyridinn

Fig. 6. Solvent effects on dichotomic reaction paths (Ref. 10).

In an apolar aprotic solvent such as xylene, a normal Diels—Alder reactiontakes place to give as the final product a pyrrolo[2,1-b}thiazole. However,when the same reaction is carried out in a dipolar aprotic solvent such asacetonitrile, an imidazo[1,2-a}pyridine is obtained as the result of an1 ,4-dipolar cycloaddition reaction.

This cycloaddition reaction clearly demonstrates that the choice ofsolvent can determine the principal product of a reaction.

A further example shows the solvent effect on chemoselectivity, i.e. onfunctional group differentiation (Fig. 7).

The reduction of the bifunctional compound il-bromoundecyl tosylate withlithium aluminium hydride in different solvents has been examined byKrishnamurthy (Ref. 11). In diethyl ether, the reaction proceeds with theselective reduction of the tosylate group to give n-undecyl bromide, whereasin diglyme, the bromine substituent is selectively reduced to yield n-undecyltosylate.

This reaction is an impressive example of how solvents can be used tomodify both the chemoselectivity and reactivity of a complex metal hydride.Thus, by a careful choice of solvent, it should be possible, using the samereagent, to selectivity react the various functional groups in an organicmolecule.

23%

Solvent effects on chemical reactivity 1871

So/vent Effect on Chemoselectivity:

Reduction of 11-Bromoundecyl Tosytcite with LiA(H4

÷LIALH4 25°C

in DiethyL ether in DigLyme

H H

83% 78%

Fig. 7. Solvent effects on chemoselectivity (Ref. 11).

One unusual example of solvent effects on the stereoselectivity of areaction is the Grignard—reaction between racemic 3-phenylbutanone andphenylmagnesium bromide (Fig. 8). This reaction has been studied in solventsof different polarity by Pêrez-Ossorio and coworkers (Ref. 12).

So/vent Effect on Stereoselectivity:

Grignard-Reaction between (±)-3 -Phenylbuta none and C6H5Mg Br

Me

PhH MetPh +PhMe

1)+Pht..MgBr/3o0C (R,R) (S,R)

2)+H20. -Mg(OH)Br + +Me

Me Me MePh ,L. Me Me J.. Ph

H Ph H)fPh+

H

(S)(S,S) (R,S)

SoLvent Et3N MeOCH2CH2OMe

[(R,S)+(S,R)]In— -10 +10[(R,R)+(S,Sfl

Fig. 8. Solvent effects on stereoselectivity (Ref. 12).

The reaction of this ketone with the Grignard reagent should lead to twopairs of enantiomers, the (R,R)- and (S,S)- as well as the (S,R)- and (R,S)-forms. Interestingly enough, the ratio in which these two pairs of enantio-mers are formed, is strongly solvent—dependent as shown by the figures inthe table (Fig. 8). In going from the apolar triethylamine to the more polar

glyme, an increasing amount of the (S,R)- and (R,S)-form is formed (from26 to 73 %).

Therefore, the activated complex leading to the (S,R)- and (R,S)-productmust be more dipolar and hence more strongly solvated in polar solvents.

In conclusion, the extent of asymetric induction in nucleophilic reactionsof chiral carbonyl compounds is not only sterically controlled (cf. Cram'sRule), but may also be influenced to some extent by the solvent.

A similar example of solvent effects on stereoselectivity has been foundrecently by Benoiton et al. in the solvent—dependent synthesis of epimericdipeptides (Ref. 13).

1872 C. REICHARDT

These four examples have already given us a strong impression of thedifferent ways in which solvents can affect chemical reactivity. However,the second part of this review will deal with only one aspect, namely theinfluence of the solvent on reaction rates together with the various solventproperties responsible for it. Rate constants are fairly sensitive indicatorsof even small, solvent—induced energy differences: a change of 10 % inthe rate at 25 °C corresponds to a change of only 250 Joule/Mol (ca. 60 callmol) in the free energy of activation.

CLASSIFICATION OF ORGANIC REACTIONS BY MEANS OF THEIR SOLVENTSENSITIVITY

In the light of the transition-state theory, solvent effects on rate con-stants depend on the relative stabilization of the reactant molecules and thecorresponding activated complex through solvation. An understanding of thefactors which control the solvent dependence of reaction rates requiresknowledge of both the direction and magnitude of the changes which occur inthe educt and transition state solvation.

The early pioneering work of Hughes and Ingold in the 1930's in nucleo-philic substitution and elimination reactions is well known (Ref. 14, 15, 16).Using a simple qualitative solvation model which allows only pure electro-static interaction between ions or dipolar molecules and solvent moleculesin both the initial and transition states, they divided all nucleophilicsubstitution and elimination reactions into different charge types. Based onreasonable assumptions as to the expected extent of solvation in the presenceof electric charges, they stated that, a change to a more polar solvent willeither increase or decrease the reaction rate, depending on whether theactivated complex is more or less dipolar than the initial state. A changein solvent polarity will have a negligible effect on the rates of reactionswhich involve little or no change in the charge density on going fromreactants to the activated complex.

A more general classification of organic reactions with respect to solventeffects, not restricted to substitution and elimination reactions, is shownin Fig. 9 (Ref. 17).

Classification of Organic Reactions

(a)Dipolar ?ransition State Reactions: Large solvent effectse.g.

R-X products

Y:+ R-X X] —.products

(b)gpolar Transition State Reactions: Small solvent effects

e.g.,, __+ . products

R r *I

- products

(c) Free-Radical Transition State Reactions: Small solvent effects

e.g. __ r& &1* ___R-R — IR j R@ + ®R —b products

+ A-X —' R-A + X® —. productsFig. 9. Classification of organic reactions

According to this scheme, organic reactions can be roughly divided intothree classes depending on the character of the activated complex involved:dipolar, isopolar and free-radical transition state reactions.

Here are now some typical examples, chosen from the recent literature toillustrate the theoretical and practical usefulness of this classification.These examples show for which reactions a solvent change may be useful inobtaining a desired rate acceleration.

Solvent effects on chemical reactivity 1873

Dipolar transition state reactions with large solvent effects can befound amongst ionization, displacement, elimination, and fragmentationreactions. The Figs. 10 and 11 show two examples.

A Dipolar Transition State Reaction: Large solvent effect

5N1 Solvolysis of tert-Butylchloride

solventpolarityFig. 10. Solvent effects on the 5N1 solvolysis of tert-butyl-

chloride (Ref. 18, 19).

A classical example is the 5N1 solvolysis of tert-butylchloride (Fig. 10).The transition state of this ionization reaction involves a partial separa-tion of unlike charges. The increased solvation of the resulting dipolaractivated complex relative to the less dipolar starting molecule, leads toa huge rate acceleration with increasing solvent polarity. The rates differby 10 powers of ten in going from benzene to water (Ref. 18, 19). Proticsolvents such as methanol and water play an additional role in this case:they act as an hydrogen-bond donating electrophile by solvating the incipientchloride ion.

In strong contrast to the ion-forming solvolysis of tert-butylchloride,the 5N2 displacement reaction between chloride ions and uncharged methyl-bromine is much slower in water than in dipolar aprotic solvents (Fig. 11).

ADipolar Transition State Reaction: Large solvent effect

5N2 Displacement Reaction of Methyl bromide with Chloride Ion

H

CEe + HVF:Br k2

25°CH

H

[CE5e...L... Brâ®J

C1H + BrL H' H

Water Methanol DMF Acetone No solvent

1,3 8i0 6io 210 15I t—

: ca. : cci. 1015

Fig. 11. Solvent effects on a SN2 displacement reaction(Ref. 20, 21, 22).

Meki

kA ",25°C

Me LMeV' CI

Me

Me

I 'MeMe

p: 2,9'10 Cm 27 i00 Cm

Solvent

+ Cle products

1rel"1

Benzene Acetone DMF Methanol Water

1 1,9' 102 ,8' io3 1,1 106 ,2,1010

Solvent

Protic solvents Dipolar aprotic solvents Gas phase

k rel2

Ea[kJ/molJ

solventpolarity

1

103

Ca. 1

11

1874 C. REICHARDT

This difference can be as large as 5 powers of ten (Ref. 20) . Due to thedispersal of negative charge during the activation process, the transitionstate is less solvated in solvents of increasing polarity. In going fromdipolar aprotic to protic solvents, the chloride ion is further stabilisedthrough hydrogen-bonding. The result is that in water the initial reactantsare much more stabilised than the large transition state anion, so the freeenergy of activation is higher in water and the reaction is thereforeslower.

This ionic S 2 reaction is one of the rare examples which have been studiedboth in solutioT and in the gas phase (Ref. 21, 22). A new technique, calledPulsed-Ion Cyclotron -Resonance Spectrometry (ICR mass spectrometry) , makesit possible to examine reactions of isolated ions and molecules in theabsence of any solvation (Ref. 23, 24).

Until recently, mainly solution-phase data have been used in attemptsto describe nucleophilicities and leaving-group abilities in terms of thestructural and thermodynamic properties of the reactants. Solvent inter-ference, however, makes it difficult to distinguish the intrinsic propertiesof the reactants from the solvation effects. In the absence of solventmolecules the intrinsic reactivity of the naked reactants can be measured anddistinguished from the effects attributable to solvation.

The result in this case is truly remarkable. In the gas phase the reactionof chloride ion with methylbromide is faster by a factor of 15 powers of tenthan it is in water. Whereas in solution the activation energy amounts upto 103 kJ/mol, in the gas phase the reaction possesses an intrinsic acti-vation energy of only 11 kJ/mol (Ref. 20, 21). These and other results leadto the conclusion that reaction rates in solution are primarily determinedby the amount of energy needed to destroy the solvation shells during theactivation process, and only to a minor extent by the intrinsic propertiesof the reactants. In other words, in solution reactions the solvent isalmost solely responsible for the observed rate constants.

Another class of organic reactions showing only small solvent effectsare isopolar transition state reactions. By definition, isopolar activatedcomplexes differ very little in charge distribution from the initialreactants and can be found in pericyclic reactions such as cycloaddition,sigmatropic, electrocyclic, and cheletropic reactions. Two examples shallillustrate the lack of solvent sensitivity of pericyclic reactions(Figs. 12 and 13).

Only a very small solvent effect, a factor of two, is found in theDiels-Alder-cycloaddition reaction of 2,3-dimethylbutadiene with 4-chloro-1-nitrobenzene (Fig. 12). This supports the formulated one-step mechanismwith synchronous bond formation without any creation, destruction ordispersal of charge during the activation process (Ref. 25, 26).

The [3,3]sigmatropic rearrangement of allyl S-methyl xanthate alsoexhibits a low sensitivity to solvent polarity (Fig. 13). The reaction inpolar glycol is only 21 times faster than it is in apolar n-hexadecane(Ref. 27). This confirms the concerted reaction mechanism which involvesvery little change in charge separation between the initial and transitionstates.

.

An Isopolar Transition State Reaction: Small solvent effect

[4+21 Cycloaddition of 2,3-Dimethylbutadiene with 4-Chloro-l-nitrosobenzene

*Me 0 Me ... Me+ ____ " ?Me-'k N 25°C MeAr

Ar = -C6H4-p-CI

Solvent Totuene Dichloromethane Ethanol Nitrobenzene

ket 1,0 1,L 2,0 22LGtkJ/mot1 87,0 86,2 85,L

solvent _____________________________________polarity

Fig. 12. Solvent effects on a Diels-Alder cycloaddition reaction (Ref. 25)

Solvent effects on chemical reactivity 1875

An Isopolar Transition State Reaction.• SmaU solvent effect

13,3lSigmatropic Rearrangement of Allyl S-Methyl Xonthate

OS 67°C

OSSMe SMe SMe

Solvent h-Hexadecane Benzene Acetonitrile Glycol

kel 1 2,5 8 21

solvent ______________________________________polarity

Fig. 13. Solvent effects on the rearrangement of allylS-methyl xanthate (Ref. 27).

A third category of organic reactions involves free-radical transitionstates, formed by the creation of unpaired electrons during radical pairformation or atom transfer reactions. In reactions of this type onlynegligible solvent effects have been normally observed. When the solventitself, however, participates as a reactant and is incorporated into theproducts of the reaction, the situation is different. Two typical examplesare shown in Figs. 14 an 15.

A Free-Radical Transition State Reaction: Small solvent effect

Thermolysis of Azobisisobutyronitrile

CN Me

LMe

k1 - MeMe- NN4Me + N:N + ®I 67°CMe

CN CN

trans-ozo compound

Solvent Decal in Chlorobenzene o-DichLorobenzene N,N-Dimethylaniline

kcet 1,0 1,5 1,6

E0[kJ/moll 11.8 133 131 128

solventpolarity

Fig. 14. Solvent effects on the thermolysis ofazobisisobutyronitrile (Ref. 28).

The thermolysis of azobisisobutyronitrile (Fig. 14), an apolar trans-azocompound, is not at all sensitive to medium effects as shown by the nearlyequal relative first—order rate constants measured in different solvents(Ref. 28). This is in agreement with a homolytic , concerted two-bondscission mechanism, producing two neutral radicals as intermediates withoutany charge separation.

Contrary to the negligible solvent effect obtained in the decompositionof trans—azo alkanes, the thermolysis of cis-azo alkanes such as 3,3,5,5-tetramethyl-1—pyrazoline reveals a small but significant solvent effect(Fig. 15). Again, there is no charge separation during the activation process.But a cis—azo compound exhibits a small permanent dipole moment, which islost during the reaction. Therefore, with increasing solvent polarity thereaction rate decreases due to the better solvation of the dipolar startingmolecule (Ref. 29, 30).

In conclusion, it can be stated that a great variety of organic reactionscan be reasonably classified according to their sensitivity to changes inthe surrounding medium. Vice versa, the application of the solvent influence

1876 C. REICHARDT

on a reaction as a criterion for the elucidation of its mechanism is alsopossible. Unfortunately, investigations of solvent effects on rates haveoften been studied only in a very small number of solvents (sometimes onlytwo) . In such cases, no far—reaching conclusions about the interrelationbetween solvent effect and reaction mechanism can be drawn. As a rule,the reaction under consideration should be investigated in as many solventsof different polarity as possible. It is recommended to use a minimum ofat least five solvents of different polarity in order to get a reliablepicture of the solvent effect under study.

A Free -Radical Transition State Reaction: Small solvent effect

Thermolysis of 3,3.5,5-Tetramethyt-1-pyrazoline

MeN L-Me

+ UI

N -N2 MeMe

Solvent Gas phase Ethylbenzene Cyclohexanone Glycol

kçel 11 8 6 1

solvent ____________________________polarity

Fig. 15. Solvent effects on the thermolysis of 3,3,5,5-tetramethyl-1-pyrazoline (Ref. 29, 30).

EMPIRICAL PARAMETERS OF SOLVENT POLARITY

Having discussed the solvent susceptibility of reactions, the questionremains, which solvent property is responsible for medium effects, andwhether it is possible to describe these solvent effects in a morequantitative manner.

Organic chemists usually attempt to understand solvent effects in termsof the polarity of the solvent. It is, however, not easy to define thisproperty, and it is even more difficult to assess it quantitatively.

Seduced by the simplicity of electrostatic solvation models, attemptsat expressing solvent polarity quantitatively, usually involve physicalsolvent properties such as dielectric constant, dipole moment, orrefractive index. But this procedure is often inadequate, especially sinceit does not take account specific solute-solvent interactions like hydrogen-bonding, charge-transfer forces etc..

Hence, from a more practical point of view, it seems reasonable to under-stand solvent polarity in terms of the overall solvation capability of asolvent, this in turn being determined by the sum of all the molecularproperties responsible for the solute-solvent interaction (Ref. 15, 31).It is obvious that solvent polarity thus defined cannot be describedquantitatively by a single physical parameter.

In such situations other indices of solvent polarity are sought.

The lack of comprehensive theoretical expressions for calculating solventeffects and the inadequacy of defining solvent polarity in terms of simplephysical constants has led to the introduction of pure empirical scalesof solvent polarity (Ref. 31, 32, 33).

Based on the assumption, that particular solvent—dependent chemical reac-tions or spectral absorptions may srve as suitable model processes,reflecting all the possible solute-solvent interactions in the solvation game,several empirical parameters of solvent polarity have been introduced bydifferent authors.

cis-azo compound

Solvent effects on chemical reactivity 1877

This approach is very characteristic of the experimental chemist, but itis regarded with mistrust by theoretical chemists. However, organic chemistryhas always made use of the qualitative, empirical rule that similar compoundsreact in similar ways, and that similar changes in the structure or in thereaction medium produce similar changes in the chemical reactivity.

One of the first verifications of this rule was the introduction of theso—called Hammett equation for the calculation of substituent effects onchemical reactivity, using the ionization of substituted benzoic acids inwater at 25 °C as a reference process (Ref. 34, 35).

The model processes used to establish such empirical scales of solventpolarity have been reviewed (Ref. 31, 32, 33). Only one of these parameters,introduced some years ago as the so-called ET(30)-scale (Ref. 36), shall bementioned and its application for the description and prediction of solventeffects on reaction rates will e shortly demonstrated.

Fig. 16 Fig. 17

Large permanent Large polarizable hvD'- electron system iji(1.Lrr-electrons)

101

150 Pew 6D Lpw9D

dipole mament

I

Weak electron pairacceptar?

LU LU—e

hgtm 150Strong electron-pair donorHydrogen-bond acceptor

LSolvent

C6H5OCH3 CH3COCH3 i-C5H110H C2H5OH CH3OH

A (nmj 769 677 608 550 515

LE1[kcal/mol)

olution colour37.2

yellow42.2

green'47.0blue

51.9violet

555red

ETlkcal/mot] h•cvN 2,85910 i (cm I

Fig. 16. Ground state properties of the standard 2,4,6-tn-phenyl-N- (2, 6-diphenyl-4-phenoxide) -pyridiniumbetaine (no. 30 in Ref. 36).

Fig. 17. Solvent-dependent intramolecular charge-transferabsorption of the 2,4, 6-triphenyl-N- (2, 6-diphenyl-4-phenoxide)-pyridinium betaine, proposed as solventpolarity indicator (Ref. 37).hg und are the permanent dipole moments in theelectronic ground and excited state of a betaine dyewith two tert-butyl groups instead of the two2,6-diphenyl groups next to the oxygen atom.

By virtue of its exceptionally large negative solvatochromism, thepyridinium-N-phenOxide betaine dye shown in Fig. 16 is particularly suitableas a standard dye for the determination of an empirical solvent parameter(Ref. 37). This betaine dye exhibits (i) a large permanent dipole moment ofabout 15 Debye, suitable for the registration of dipole-dipole and dipole-induced dipole interactions between solute and solvent, (ii) a largepolanizable it—electron system, consisting of 44 it—electrons, suitablefor the registration of dispersion interactions, and (iii) the phenolicoxygen atom represents a highly basic electron-pair donor centre, suitablefor specific hydrogen—bond interactions with protic solvents.

With increasing solvent polarity, the long-wavelength absorption band ofthis betaine dye is shifted hypsochromically due to the increasingstabilization of the dipolar ground-state relative to the less dipolar firstexcited state (Fig. 17).

The long-wavelength solvatochromic absorption band is situated at 810 nm indiphenyl ether and at 453 nm in water as solvents. With the correspondingsolvent-induced hypsochromic shift of 357 nm ( IET = 28 kcal/mol = 117 kJ/mol),this betaine dye still holds the world record in solvatochromism.

The molar transition energies, ET, of this betaine dye measured in kcal/mol,have been proposed as empirical parameters of solvent polarity (Ref. 36). Ahigh ET-value corresponds to high solvent polarity, i.e. the solvent has ahigh solvating ability.

PAAC 54:10 - H

1878 C. REICHARDT

Since the greater part of the solvatochromic range lies within the visibleregion, it is even possible to make a visual estimate of the solvent polarity.The solution color of this betaine dye is red in methanol, violet in ethanol,green in acetone, blue in isoamyl alcohol, and greenish-yellow in anisole.With suitable binary solvent mixtures, almost every colour of the visiblespectrum can be obtained.

The limitations of the ET-scale are: (i) no ET-values can be obtained foracidic solvents since protonation of the phenolic oxygen atom of the betainedye leads to disappearance of the solvatochromic absorption band; (ii) dueto the low volatility of the betaine dye, no gas-phase ET-values can bedetermined; and (iii) ET-values for nonpolar solvents such as hydrocarbonscannot be directly measured due to the low solubility of the standard betainedye in those solvents.

The last problem can be overcome by the use of alkyl—substituted betainedyes (Ref. 36, 38). One recent example is shown in Fig. 18 (Ref. 38).

Fig. 18. A new secondary penta-tert-butyl-substituted betainedye for the determination of ET(30)-values for non-nolar solvents (Ref. 38).

55. .COH •

C:kOM

50 -r-C3H7H

• ;-c3H,OH2-o •'CHCN CP-4NO2

c3so: /cH2—çH—c4z c3ca ••OAc Ac OAc

HCO4(CN3)2

C42CL2 ••a •,/ :coci3

1.0- /tç

.6 c:o2c2 ))35 -

,,,,,,

ET(30) [kca(/mo(j

Fig. 19. Correlation between the ET-values of 2,4,6-triphenyl-N-(2, 6-diphenyl-4-phenoxide) -pyridinium betaine[ET (30) —values] and its penta-tert-butyl—substitutedderivative (E-values) for twenty solvents (Ref. 38).E = 0.914 ET(3O) + 3.43 (n = 20; r = 0.997).

lipophi/ic tert-butyl groups

'Thbetter solubility inapolar so/yen ts

TMS Formamide

no changein the basicchromophore

Amax

[nm]911 519

= 392 nm

1E1cx26 kcal/mol

C(CHJ1

HCOt2

0E

45'I

.1-

I I I I35 40 45 50 55

Solvent effects on chemical reactivity 1879

The introduction of five tert-butyl groups into the peripheric phenylgroups does not change the basic chromophore, but does yield a more lipophilicbetaine dye. Although not more soluble in water it is now soluble in hydro-carbons such as n—hexane and even in tetramethylsilane (TMS).

The excellent linear correlation between the ET-values of these two betainedyes in twenty solvents allows the calculation of ET(30)-values for solvents inwhich the standard betaine dye is not soluble.

At present, ET(30)-values for more than 200 organic solvents and for manybinary solvent mixtures are available (Ref. 31, 33).

Definition of ET-ValueS and Normalized E!-VaLues

E1[kcal/molJ h'c.Il.N = 2,85910 7 [cm1] (1)

ET(Solvent) - E1(TMS) ET(Solvent) - 30,7EN (2)IET(Water)- ET(TMS) 32,4

Correlation between the two scales:

TMS Anisole OMAC 1—Butanol Formamide Water

ET I

30,7 37,2 43,7 50,1 56,6 63 1

EN IT 00 0,2 0,4 0,6 08 10

Fig. 20. Definition and correlation of ET-values andnormalized E-values.

Unfortunately, ET-values have by definition the dimension of kcal/mol, aunit which should be abandoned in the framework of SI-units.

Therefore, the use of the so—called normalized E-values has been recentlyrecommended (Ref. 38, 39). E-Values are defined according to equation (2)in Fig. 20 as the quotient of two differences: the difference betweenthe Em—values of the solvent under consideration and of tetramethylsilane,and tlie difference between the ET-values of water and of tetramethylsilane.Hence,the corresponding E-scale ranges from 0.0 for tetramethylsilane, Nthe least polar solvent, to 1.0 for water, the most polar solvent. These ETvalues are thus dimensionless numbers such as Hammett's a substituentconstants (Ref. 34, 35).

APPLICATIONS OF SOLVENT POLARITY PARAMETERS TO CHEMICALREACTIVITY

Since solvent polarity parameters such as the ET-scale are establishedin anempirical way, their usefulness has to be tested in the same manner.Therefore, some applications of the ET—parameter to chemical reactivityshall be given, using some of the solvent-dependent reactions as exampleswhich have been mentioned in the first part of this review (Figs. 21 ... 24).

The 5N1 solvolysis of tert-butylchloride has been already mentioned asan example of solvent effects on chemical reactivity (Ref. 18, 19).Fig. 21 shows the correlation between the ET-values and the ionization rateof tert-butylchloride. The linear correlation between the solvent polarityparameter and the rate of this typical dipolar-transition-state reactionis so good that rate constants for further solvents can easily becalculated.

This astonishingly good correlation between the spectroscopicallydetermined solvent polarity parameter and the kinetically determined rateconstants, shows that the light absorption of the solvatochromic betaine dyeis in fact a good model for the estimation of solute-solvent interactionsin dipolar-transition state reactions.

1880 C. REICHARDT

0./. 0.6

Fig. 21. Linear correlation between ET(30)-values and lg k1of the SN1 solvolysis of tert-butylchloride at 25°C:

ig k1 0.312 . ET(3O — 22.55 (n = 17; r = 0.972).

40 50E1(30)[kcat/mot]

Fig. 22. Linear correlation between E (30)-values and lg k2 ofthe [4+2] cycloaddition reaction of 2,3-dimethyl-butadiene with 4-chloro-1-nitrosobenzene at 25 °C

lg ]c2 = 0.016 .

ET(3O)— 2.88 (n = 6; r = 0.936).

0.2 0.8 1.0

E1(30)[kcal/motl

ENI0,2 04 0,6

C6H5N02

[0]

-2,4 -

C1-CH2CH2-C1 •.

C2H5OH

• CH2CL2

CCLC6H5CI

• C6H5CH3 Me0

Me+

Ar Me..ft+.Ar

I I

Solvent effects on chemical reactivity 1881

In spite of only small solvent effects, even isodipolar-transition-statereactions such as the Diels-Alder cycloaddition of 2,3-dimethylbutadienewith 4-chloro-1-nitrosobenzene (Ref. 25) exhibit a satisfactory linearcorrelation with the solvent polarity parameter ET (Fig. 22). The gentleslope of the correlation line is typical for concerted cycloadditionreactions with a one-step mechanism involving little or no dipolarity changeduring the activation process. Similar gentle slopes have been observedfor other pericyclic reactions such as the 1,3-dipolar cycloadditionreactions studied by Huisgen and coworkers (Ref. 40).

Not only dipolar and isopolar transition-state reactions but free-radicaltransition—state reactions with small solvent sensitivity can also becorrelated with the solvent polarity parameter ET as shown in Figs. 23 and24.

EN0,2 0.4

T o, 0,8

40 50E1(30) [kcal/mo(]

Fig. 23. Linear correlation between ET(30)-values and lg k1 ofthe thermolysis of 3,3,5,5-tetramethyl-1-pyrazolineat 189 °C:

lg k1 = —0.040 E(3O) — 2.10 (n = 4; r = —0.975).

The rate of the aforementioned thermolysis of 3,3,5,5-tetramethyl-1-pyrazoline (Ref. 29, 30) correlates well with the ET-parameter (Fig. 23).The observed rate decrease with increasing solvent polarity is bestexplained by assuming a decrease in the dipole moment of the dipolar cis—azocompound during activation. Increased solvation of the dipolar reactantmolecule relative to the apolar activated complex is compatible with thesmall solvent effect observed.

The aforementioned Grignard reaction between racemic 3—phenylbutanone andphenylmagnesium bromide is a striking example of solvent influence onstereoselectivity (Ref. 12). As shown in Fig. 24, a surprisingly goodlinear relationship between the ET—parameter and the diastereomeric productratio is found in a large selection of solvents of varying polarity. Thefact that the observed stereoselectivity may be additionally influencedby solvent polarity in a predictab4 manner is of invaluable importancein other asymmetric syntheses.

Four examples of the many Em-correlations described in the literature(Ref. 31, 33, 41) demonstrate that empirical parameters of solvent polarityare useful tools for predicting chemical reactivity in a semiquantitativemanner in different solvents.

-3,5

I I

C6H5C2H5 .

• i-C5H110H

, N+

H02C-CH20H •

1882C. REICF1ARDT

I

+ +

C

+1,0

-1,0

0,1 0,2

Fig. 24. Linear correlation between ET(30)-values and thestereoselectivity observed in the Grignard reactionof (±) -3-phenylbutanone with phenylmagnesiumbromide at 30 °C:

ln[(R,S)+(S,R)]/[(R,R)+(S,S)] = 0.365 ET(3O) — 13.2(n = 7 : r = 0.984).

For an estimate of solvent effects on chemical reactions, empirical para-meters of solvent polarity are as useful as the empirically derived Harnmettconstants used in the prediction of substituent effects in chemical reactions.Both approaches are verifications of the principle of "Linear Free-EnergyRelationships" (Ref. 31, 42, 43).

One important reservation should be made. Solvent effects are basicallymore complicated and more specific than substituent effects. The applicationof solvent polarity parameters is based on the assumption that, the natureof the solute-solvent interactions in the reference process used toestablish a solvent scale is similar to those influencing the reaction understudy. This is obviously true only for closely related solvent-sensitivereactions. Therefore, the use of solvent parameters to predict solventeffects should be limited to Largely analogous processes.

A large variety of empirical parameters of solvent polarity are presentlyat the chemist's disposal. This topic has been recently reviewed (Ref. 31,32, 33). Therefore, it should not be too difficult to find a suitablesolvent parameter for a particular reaction under study.

Despite considerable progress in calculating solvent effects on a purelytheoretical basis, actually the use of carefully selected empirical solventpolarity parameters is still the most practical and successful method forreliable solvent effect predictions.

During the last ten years there have been various attempts at consider-ing two ore more specific solute-solvent interactions simultaneously usingmultiparameter equations instead of single—parameter correlations (Ref. 15).Different empirical parameters for solvent polarizability, dipolarity,electrophilicity, and nucleophilicity have been separately determined andcombined in multiparametric equations. The most ambitious recent approachis that of Taft and coworkers (Ref. 44).

1

6590 2)H2O N C5H5

HO C6H5

H3C-0-O,0,CH3 .

00

U-C3H720

34 36 38

ET(3O) [kcol/mol]

Solvent effects on chemical reactivity 1883

The use of multiparametric equations instead of single—parameter equationshas in many cases produced a dramatic improvement in the correlations betweensolvent—dependent reactions and inherent solvent properties. However, a finalsolution to the problem of separating solvent polarity into the varioussolute—solvent components is not yet in sight.

A more detailed explanation of multiparametric equations can be found inRef. 15, 33 and 44.

FINAL REMARKS

120 years after the discovery by Berthelot and St. Gilles of the solventinfluence on chemical reactivity the problem of how to understand and des-cribe solvent influence on chemical reactivity in a more quantitative manneris far from being finally solved.

Most organic reactions can be classified into dipolar, isopolar, and free-radical—transition—state reactions according to their solvent sensitivity.On this basis, the extent and direction of solvent influence on reactionrates can be used as a valuable criterion in the elucidation of reactionmechanisms. Thereby, extreme rather than intermediate solvent effects onchemical reactivity are more conclusive.

The use of empirical parameters of solvent polarity is still the simplestand most practical method of predicting solvent effects in a more quantita-tive way. Various empirical solvent scales are now available for thisprocedure. Multiparameter equations have led to further improvements.

As the selection of suitable reference processes for the determinationof empirical parameters of solvent polarity leaves room for subjectivedecisions, the well-known remark that organic chemistry is not a sciencebut an art (Ref. 35) seems to be not wholly unfounded.

Acknowledgement - I thank Dr. Edeline Wentrup-Byrne, Marburg,for improvements of the English manuscript. The author's ownpapers cited in this review were sponsored by the Fonds derChemischen Industrie, Frankfurt (Main), for which the authorwould like to express his gratitude.

REFERENCES

1. N. Berthelot and L. Pan de Saint—Gilles, Ann. Chim. et Phys., 3. Sir., 65, 385 — 422(1862); 66, 5 — 110 (1862); 68, 225 — 359 (1863); translated into German and edited by

N. and A. Ladenburg in: Ostwald's Kiassiker der exakten Naturwissenschaften, Nr. 173,

Engelmann, Leipzig 1910.2. N. Menshutkin, Z. Phys. Chem. 1, 611 — 630 (1887); 5, 589 — 600 (1890); 6, 41 — 57

(1890); 34, 157 — 167 (1900).3. L. Claisen, Liebigs Ann. Chem. 291, 25 — 111 (1896).4. A. Hantzsch and 0.W. Schultze, Ber. Dtsch. Chem. Ges. 29, 2251 — 2267 (1896).5. L. Knorr, Liebigs Ann. Chem. 293, 70 — 120 (1896).6. W. Wislicenus, Liebigs Ann. Chem. 291, 147 — 216 (1896).7. K.H. Meyer, Liebigs Ann. Chem. 380, 212 — 242 (1911); Ber. Dtsch. Chem. Ges. 47, 826 —

832 (1914).8. M.T. Rogers and J.L. Burdett, J. Am. Chem. Soc. 86, 2105 — 2109 (1964); Canad. J. Chem.

43, 1516 — 1526 (1965).9. I. Szele and H. Zollinger, Helv. Chim. Acta 61, 1721 — 1729 (1978); H. Zollinger, Angew.

Chem. 90, 151 — 160 (1978); Angew. Chem., mt. Ed. Engi. 17, 141 (1978).10. N. Abe, T. Nishiwaki, and N. Komoto, Chem. Lett. (Tokyp) 223 — 224 (1980).

11. S. Krishnamurthy, J. Org. Chem. 45, 2550 — 2551 (1980).12. R. Prez—Ossorio, A. Pérez—Rubalcaba, and M.L. Quiroga, Tetrahedron Lett. 21, 1565 —

1566 (1980); 0. Arjona, R. Pêrez—Ossorio, A. Prez—Rubalcaba, and M.L. Quiroga, J. Chem.Soc., Perkin Trans. II 597 — 603 (1981).

13. N.L. Benoiton, K. Kuroda, and F.M.F. Chen, Tetrahedron Lett. 22, 3361 — 3364 (1981).

14. C.K. Ingold, Structure and Mechanism in Organic Chemisjy, 2nd edition, CornellUniversity Press, Ithaca, London (1969).

15. C. Reichardt, Solvent Effects in Organic Chemistry, Verlag Chemie, Weinheim, New York

(1979).16. C. Reichardt, Chemie in unserer Zeit 15, 139 — 148 (1981).17. E.M. Kosower, An Introduction to Physical Organic Chemistry, Wiley, New York (1968).18. S. Winstein and A.H. Fainberg, J.Am. Chem. Soc. 78, 2770 — 2777 (1956); 79, 5937 —

5950 (1957).

(1861) O9 — c87 'T •maiia 2o sAj °d ':ijj •kta 'UI)J rw 'pnoqqy w'ir •177

(L86t) 7cc — LE 'E[ •PUS •maTt3 °V 'PIOM 'S UB WO1SC w . (8L6 I ) uopuoj ' SS3i urnuata ' sazuApy uej ' Li spuaq

iiTsTs!Z-LuV UOTI1O3 '(spa) itoqg r pu uu1dq3 (zL6L) uopuoj 'ssei

WflU3Id 'sdiqsuoij (2;taua—aa2d iau rt saup '(spa) ieiot j pu uuid3 N (8961) EL — I 'U TjOSlOd '4D '2tpS2Off 'TO2u1Q )I p129 P'°T3 3 •I7 (0861) ZOEZ — C8Z 'zc "'-tD IddV 9Ifld 'U92STflH i

___________ (1861) czi — 611 'LE

UO234B13] 'piqoiwj D PU 'UO2! ''d ')jWELIOZS8Zj y 'plSwOpJ 'f' 'pjst1O2Ac2)1 'J

snsai peqsqndun '12au2qfJ—qosnq2H 'a pu jp T91 ' 'c '(1861) E6Z — SLZ 'd 'uopuo' 'J2OA 'SS3Id UIflU3Id 'C11StU13T43 UOtfl[O UT S33UApy '('Spa) au ' pu 'Tzzrnirv 'Upi3 'I wç 's2UfpU.j ta auio s2ouuI iCip9Eod

U3AIOS s saA aua apxouaqd _mnuppiC '1aLnI4 'j pu 'qosnq2U} 'a 'P2TaI '3 'LE '(1L61) L9 — i9 'ZSL

'pqi 'piqzraa ' (696i) coi — £6 'LZL 'pTqT '1p2qa)J '3 pu TO2tUTU ')I (E96I) LE — 1 '199 '111343 'UU sqai 'UUUUIp[O '[ puU 'uuuida 'J 'p2qora '3 '41O2111]j ')I '9E

____________ '(EL61) 'J 9t'E 'd 'PLv1 '31143 2UI13A '31111343 aqosTU2O

aqoSTTSqa (oL61) I20X N 'IITH_/2O0M 'LtlSTulaq3 ZYfUU2JO [°TS4d '13U1UJ5H 'd'I 'SE '(LE6I) £01 — 96 '6S '°S "43 "W r '3WLUBH 'd"I 'E '5932d u (z861) Z8Z — ItlZ 'd 'E 'I°A ')j1OA 3N

'23S34OT43 'cnit 'SUO ZB23UI 151fl0310J4 '('Spa) SsWOt!,L_3IITA1O 'f'J pUS szftsj H UT '(FApOS3 ISOUI343 PUS 1SIOd U3ATOS 30 S233WS1Sd TsoT1dma '3 EE

'(8L61) 16Z — 69Z d ')J10X 'a1Sq°43 'ETM 'SaT22adO1dISzTtuaq3 PUS 'SOpUSU(Q '31flZ1fl15 'S.pflbrI OUSLt0 '('spa) solsla 'g pus

'3ddTi 'a 'u 4uTZrn 'Q'y U 'AZS3j jSOU1343 pus S3[SDS 1U3ATOS '1P2540T3I '3 'ZE

____________ ___________ ___________ '(6L61) 011

— 86 '81 '12ua 'pa 'ui ''111343 a2uV (6L61) lEt — 611 '16 '111343 '13T1 '3 'IE '(8L61) 509 — 965 '111 '13[ '111343 '1p2St43nj '3 pUs uUsu1Sna 'M 'OE

'(9961) 696E — E96E '88 '°S '111343 'U1 'f 'sfljSfl4 'y pus p203L513 'f'H '6Z

'(0861) 0E8—EZ8 'SE 'PS 'U2ua '111343 'AcU15tST520Q ')' U5 '15)113455W 'y' 'TU15)flfl 'O'14 '8Z

___________ _______________ (9L61) 3 0L591

'8 'isqy '111343 (SL61) ISLI — 1711 '? n)1s2thjsA 'ppn2sj 'j pus 'ousisa 'i '5i55)J ' (cL61) ZLi — L9i ' (oi()1oJ) 'TINS 'W254 '111343 'T40fl253 'J pUS OUSISH ')I

____________ __________ _________ '(0861) 6LL '61

'iSua 'pa 'luI "111943 auy (0861) 108 — ELL ' '111343 t32Uy 'U511115T15 ' pUS 13fl55 '1 '9Z '(6L61) iZS — 61S II SUSIJ UT)llaa ''OOs 'm343 'r ''11 'H pUS UOI3T&S '0 'SZ

__________________________ ___________________ '(1861) LZ

1354OSU9SST 18p wfllPjacf5 (0861) LS1 — 871 'S 'ON 'EI7Z US3TI3U1y 3T3TU3TO5 'J3AIOW 'L'I ' '(0861) 8 — 9L 'El '31 '111343 'O3 '3pO 'H'f 'EZ

'(LL6I) 8W7 — 61Z7 '66 '305 '111343 'my 'r 'usu1ns1 ']'f pus psaswio 'M'M (L61) 1E0t — OEO1i '96 '30 '111343 'uiy 'f 'iapay 'y'3 pus 'PS3ISmIO 'N'M 'U5111fl515 'I'f 'ZZ

_______________ '(9L61)

6S91 — E791 '7S '11'13 'f 'psUs3 '31114O ')J'Q pus '1USZACSd 'cr'r 'Ls)1osW '') 's)jsUsJ ')j (t?L61) 8Z0? — LZ07 '96 'S '1343 '11V 'r 'U5Z1C5d 'Q'f )U5 '51°5W 'I'9 '3W40 ')I'Q

_______________ '(6961) ZE —

1 '69 'A1 '111943 '13)1lsd 'r'V (8961) 690S

— 61705 '() 'O0 '111843 ""V '1' 'Uox0ig 'f'J PUS '13)j15 'f'V '0)1 ''3' '13)U5X3 'j 'OZ '(zL61) LSEI — E17EI 'II 'SU513 UT)119d ''°S '111343 'f 'wsqsiqy 'H'W '61

1'dVH3IHH '3 17881