Sandmeyer reactions. Part 7.1 An investigation into the reduction steps of Sandmeyer hydroxylation and chlorination reactions

View Online / Journal Homepage / Table of Contents for this issue

Sandmeyer reactions. Part 7.¹An investigation into the reduction

steps of Sandmeyer hydroxylation and chlorination reactions

Peter Hanson,*^aJason R. Jones,^aAlec B. Taylor,^aPaul H. Walton^aand Allan W. Timms^b

2

^aDepartment of Chemistry, University of York, Heslington, York, UK YO10 5DD

^bGreat Lakes Fine Chemicals Ltd, Halebank, Widnes, UK WA8 8NS

Received (in Cambridge, UK) 21st January 2002, Accepted 21st March 2002

First published as an Advance Article on the web 19th April 2002

For Sandmeyer hydroxylation and chlorination in aqueous solution, the reduction steps have been investigated by

means of correlation analyses of the eﬀects of diazonium ion substitution on the rates of reduction. For simple

hydroxylation, a change of behaviour between diazonium ions substituted by electron donor groups and those

substituted by electron acceptor groups is interpreted as a change within an inner-sphere process from rate-

determining electron transfer to rate-determining association of the reactants. By contrast, for citrate-promoted

hydroxylation, a similar change in behaviour may be interpreted as a change between inner- and outer-sphere

electron transfers. For chlorination, there is no mechanistic variation within the range of substituents examined

but the pattern of behaviour is consistent with an inner-sphere mechanism. The various patterns of behaviour are

rationalised in terms of the eﬀects of diazonium ion substitution and catalyst ligation on the reduction potentials and

self-exchange rates of the various reacting redox couples. Comparative correlation analyses of reductions and other

electrophilic reactions of diazonium ions are used to support the arguments advanced in respect of Sandmeyer

reduction steps. It is suggested that the Cu^reductants react via a nucleophilic bridging ligand at the diazonium N_βto

give transient Z-adducts which are the precursor complexes and that activation for electron transfer involves rotation

about the N–N bond.

The general mechanism for a Sandmeyer reaction sequence is

given in Scheme 1.^2,3Reaction (1) is the one-electron reduction

formed in alternative systems such as the homolysis of diazo-

nium compounds in alkaline conditions,¹⁰of azo com-

pounds^11,12or of diazosulﬁdes,¹³the reactions of aryldiazenyl

radicals can be investigated. An upper limit for the rate con-

stant for the fragmentation of phenyldiazenyl radical [reaction

(2)] at 298 K has been measured¹²as 2 × 10⁸s^Ϫ1and most other

estimates exceed 10⁷s^Ϫ1.^10,11

However formed, the aryl radicals occurring in a Sandmeyer

sequence may undergo a variety of fates. Useful reaction results

from the transfer of a ligand X from a copper() complex giv-

ing ArX and returning the copper to the univalent state [reac-

tion (3)] but, since aryl radicals are highly reactive species, other

reactions may compete [reaction (4)]. The Sandmeyer sequence

proper [reactions (1)–(3)] thus depends upon a catalytic cycle in

copper. Copper() is labile in respect of the exchange of

monodentate ligands (e.g. the lifetime for exchange of water on

hexa-aquacopper() is ca. 10^Ϫ9s)¹⁴so the uptake of X by the

catalyst [reaction (5)] is expected to be very rapid. Although

there is a thermodynamically determined upper limit on the

standard reduction potential of a reductant of diazonium ions

(estimated by Galli⁵to be ca. 1 V), Sandmeyer reaction

sequences are most eﬃcient when the relevant Cu^/Cu^couple

has a relatively high reduction potential for, in this circum-

stance, aryl radicals are produced relatively slowly and con-

sumed relatively rapidly by the required sequence. Cyano-

dediazoniation exempliﬁes this well. Ligation by cyanide so

stabilises copper() that, on mixing, Cu^2ϩand CN^Ϫ, the initially

formed cyanocuprate() complexes have only a transient exist-

ence, decomposing to cyanogen and cyanocuprate().^15,16Never-

theless, cyanocuprate() complexes are able to reduce diazo-

nium ions and the cyanocuprate() complexes formed in the

process then transfer cyanide ligands to aryl radicals produced

concomitantly within their solvent cage.¹Cyanogen formation,

being a bimolecular process,^15,16does not compete. Thus in

Sandmeyer cyanation, the resting state of the catalyst is Cu^and

the Cu^state has a transient caged existence only.

Scheme 1

of the diazonium ion by a copper() complex. It is the rate

determining step⁴and the rate has been shown to depend on

substituents in the aryl ring and on the nature and number of

ligands in the copper complex,^1,5,6but its precise nature is

uncertain. The precipitates often formed during synthetic pro-

cedure when solutions of diazonium salts and cuprous halides

are mixed were widely believed to be copper()–diazonium

complexes⁷but an X-ray crystallographic investigation demon-

strated that, in the case of benzenediazonium bromide and

CuBr at least, the precipitate is a simple salt, PhN₂Cu₂Br₃,

which shows no evidence of bonding contact between the metal

and the diazonium group.⁸Although complexes involving

ligation of many transition metals by diazonium ions are

known, none has been reported involving copper.⁹Reaction (1)

might therefore proceed by either an outer-sphere mechanism

or an inner-sphere mechanism in which the precursor complex

has a ﬂeeting existence.

ϩ

Ϫ

If the former, the product of the reduction step would be an

aryldiazenyl radical, ArN₂; if the latter, the successor complex

formed upon electron transfer might decompose stepwise

giving the diazenyl radical as an intermediate or, directly, giving

dinitrogen and an aryl radical. The involvement of aryldiazenyl

radicals in Sandmeyer reactions is thus presumed. When

ؒ

DOI: 10.1039/b200748g

J. Chem. Soc., Perkin Trans. 2, 2002, 1135–1150

This journal is © The Royal Society of Chemistry 2002

1135

View Online

For halodediazoniations, the aﬃnities of chloride and brom-

ide ions for copper() are not such that these ions reduce Cu^2ϩ

and, in Sandmeyer reactions, their ligand transfers [reaction (3)]

are slower than those of their cyano analogues. In this circum-

stance, aryl radicals can escape the solvent cage in which they

are produced either to be lost to side-reactions or, in the

presence of added alkenes, to be trapped in Meerwein reaction-

products.^17–19Galli^20,21showed that the yields of Sandmeyer

halogenation reactions are improved by having initial Cu()

present in addition to Cu(); the improvement occurs because

the aryl radicals are not then restricted to ligand transfer only

from co-produced halocuprate() complexes. Thus, in Sand-

meyer halogenation the resting state of the catalyst is usually

Cu^but yields are improved by having signiﬁcant initial

amounts of Cu^present too.

must be an insuﬃciency of Cu^. This presents something of a

diﬃculty in Sandmeyer reactions where the catalytic cycle

regenerates Cu^in the ligand transfer step (Scheme 1) and, in an

ideal system, the Cu^concentration would remain constant after

initiation with both reactions running to completion. In real

systems, however, Cu^may be lost through aerial oxidation³⁰

and side reactions which do not regenerate Cu^and thus, pro-

vided a suﬃciently small amount of ascorbic acid is added and

the reactivities of the two diazonium ions are not too dissimilar,

both reactions will be arrested before completion. What consti-

tutes a suﬃciently small amount of ascorbic acid varies with

the pair of diazonium ions being compared. The Cu^concen-

tration is set high so as to optimise the ligand transfer step and

ensure that the products of the reduction step (aryl radicals) are

eﬃciently converted to Sandmeyer products, the amounts of

which can be quantiﬁed by GC analysis.

Hydroxydediazoniation is not one of the ‘traditional’ family

of Sandmeyer reactions. It was Cohen and co-workers²²who

showed homolytic hydroxydediazoniation catalysed by copper

ions to be feasible. Here, the catalytic copper couple is, in prin-

ciple, Cu^2ϩaq/Cu^ϩaq but, since Cu^ϩaq is liable to aerial oxid-

ation and disproportionation, experimentally, Cohen supplied

Cu() as suspended Cu₂O. High concentrations of Cu^2ϩaq are

required if most radicals are not to be lost to side reactions

because the transfer of water (or hydroxide) ligands to aryl

radicals is comparatively slow. We have shown previously that

water is transferred from Cu(OH₂)₆^2ϩ400-fold less rapidly than

chloride is transferred from the chlorocuprate() complexes

occurring in Sandmeyer reaction conditions.^23,24Thus for

Sandmeyer hydroxylation, the resting state of the catalyst is Cu^

and Cu^is transient. We have also discovered that coordination

of the copper by citrate and certain other multi-dentate ligands

promotes the catalytic cycle,²⁵i.e. maximal yields of phenol can

be obtained using a smaller concentration of copper when it is

ligated by the non-transferrable ligand than when it is not so

ligated. The eﬀect of coordination of the catalyst by a multi-

dentate ligand on the ligand transfer step will be addressed

separately; in this paper some light is shed on the eﬀect for the

reduction step.

Hydroxylation. The reduction step of Sandmeyer hydroxyl-

ation was studied by measuring the amounts of the two phenols

produced from the competitive reaction of two diazonium ions

in aqueous solutions of Cu(NO₃)₂as a function of the concen-

tration of added ascorbic acid. Diazonium ions with substitu-

ents having similar net electronic eﬀects were paired in order to

avoid large disparity between the two phenol yields and hence

the error in GC analysis. As an illustration, the competition

between the unsubstituted benzenediazonium ion and its

4-chloro-derivative is displayed in Fig. 1. It can be seen that the

Results

(i) Relative rates of reduction in Sandmeyer hydroxylation and

chlorination reactions

Since the reduction steps of Sandmeyer hydroxylation and

chlorination reactions occur much faster than those of cyan-

ation, direct measurement of absolute rate constants from the

nitrogen evolution in the manner described in the foregoing

paper¹is not feasible. However, relative rates for either process

can be obtained by carrying out competitive reductions of pairs

of diazonium ions. An experimental procedure was devised in

which a pair of substituted diazonium ions was dissolved in a

suitable solution of Cu^to which was added a very small

amount of ascorbic acid (< 2 mol% relative to Cu^) to initiate

the reaction (Scheme 2). Ascorbate is known to reduce Cu^2ϩ

Fig. 1 Variation of percentage phenol yields from competitive

Sandmeyer hydroxylations with the concentration of initiator used;

plot 1, 4-chlorophenol; plot 2, phenol.

diazonium ion bearing the net electron withdrawing substitu-

ent, Cl, is reduced the more rapidly and reacts to completion for

lower initial concentrations of the initiator than the unsubsti-

tuted ion and that only at low ascorbic acid concentrations

(< 3 × 10^Ϫ4mol dm^Ϫ3) is reaction truly competitive so that the

diﬀerence in phenol yields reﬂects the diﬀerence in rates. All the

competed pairs of diazonium ions behaved in similar ways

(though were competitive over diﬀerent ranges of ascorbic acid

concentration) and all gave maximum yields of phenols around

80%. For kinetic analysis, only the phenol yields demonstrably

falling within the competitive region were used.

Scheme 2

rapidly^26,27whereas its reaction with diazonium ions is com-

paratively slow²⁸and intermediate diazo-ether adducts may be

isolated.²⁹The initiation process is therefore more probably the

reduction by ascorbate of Cu^than reduction of diazonium

ions; whatever the case, the small amounts of initiator used and

the cycling of the catalyst ensure that the majority of diazo-

nium ion reduction is eﬀected by Cu^.

Kinetic analysis. For second order reduction steps of two

competing diazonium ions XC₆H₄N₂and YC₆H₄N₂we have

eqn. (6) and (7):

ϩ

Ϫd[XC₆H₄N₂^ϩ]/dt = k_X[XC₆H₄N₂^ϩ][Cu^ϩ]

Ϫd[YC₆H₄N₂^ϩ]/dt = k_Y[YC₆H₄N₂^ϩ][Cu^ϩ]

(6)

(7)

In order to ensure truly competitive reductions, reaction must

cease before complete reduction of both diazonium ions—there

1136

J. Chem. Soc., Perkin Trans. 2, 2002, 1135–1150

View Online

Table 1 Percentage yields of phenols from the competitive reduction of 4-chloro- and unsubstituted benzenediazonium ions and derived relative

reduction rate

a

10⁴[Ascorbic acid]/mol dm^Ϫ3

Percentage yield 4-ClC₆H₄OH

Percentage yield C₆H₅OH

k_Cl/k_H

0.58

1.16

1.74

2.32

14.26

32.86

49.92

80.05

3.20

7.43

9.94

25.51

4.73

5.16

6.61

5.47

Mean 5.49 0.80^b

^aCalculated as ln [100/(100Ϫ%_{4-chlorophenol})]/ln [100/(100Ϫ%_phenol)], cf. eqn. (10). ^bThe uncertainty is twice the standard error on the mean.

i.e.

which retain water or hydroxide ligands^31,32and, on the

assumption that such monodentate ligands may be transferable,

the formation of phenols in this reaction is explicable. However,

the nature of the reducing species formed by copper in the

presence of trisodium citrate is uncertain. Although soft Cu^

is not expected to form strong complexes with ligands binding

through hard oxygen atoms, the multi-dentate, anionic citrate

ligand should bind in preference to water for coulombic and

entropic reasons (the chelate eﬀect) and hence inﬂuence its

redox characteristics by comparison with the simple aquated

ion. The polarographic half-wave potential of Cu^2ϩhas been

reported to be lowered in the presence of citrate³³as have

voltammetric peak potentials for sulfato-complexes of two

diﬀerent stoichiometries.³⁴Competitive hydroxylations were

run in the presence of trisodium citrate and at reduced copper

concentration (see Experimental) but which, otherwise, were

analogous to the simple hydroxylations described above and

relative rate constants were found in the same way; these are

given in Table 3. The values span two fewer orders of magni-

tude than those from simple hydroxylation but again indicate

greater reactivity for those ions with electron withdrawing

substituents.

Ϫd[XC₆H₄N₂^ϩ]/[XC₆H₄N₂^ϩ] = k_X[Cu^ϩ]dt

Ϫd[YC₆H₄N₂^ϩ]/[YC₆H₄N₂^ϩ] = k_Y[Cu^ϩ]dt

Integrating both between time limits of 0 and t gives eqn. (8)

and (9):

ln ([XC₆H₄N₂^ϩ]₀/[XC₆H₄N₂^ϩ]_t) = k_X[Cu^ϩ]t

ln ([YC₆H₄N₂^ϩ]₀/[YC₆H₄N₂^ϩ]_t) = k_Y[Cu^ϩ]t

(8)

(9)

For competitive reactions, [Cu^ϩ]t, though varying, is com-

mon to both at any instant hence elimination of [Cu^ϩ]t between

eqn. (8) and (9) gives:

If it is assumed that the phenol produced measures the dia-

zonium ion which has reacted, eqn. (10) follows:

Chlorination. Chloride ions bind to Cu^much more strongly

than do water molecules^30,35and chloride ligands are trans-

ferred to aryl radicals by Cu^much more rapidly than water

ligands.²⁴Consequently, addition of a suitable concentration of

KCl to a solution of Cu(NO₃)₂will ensure any copper() pro-

duced in it will occur as Cu^Cl₂^Ϫor Cu^Cl₃^{2Ϫ 35,36}and the product

of Sandmeyer reaction will be the aryl halide. Reductions of

pairs of diazonium ions were carried out as above but in the

presence of KCl and the yields of aryl chlorides produced

under demonstrably competitive conditions were used to evalu-

ate relative rate constants. The results, given in Table 3, span ﬁve

orders of magnitude and, as expected, show electron deﬁcient

ions to be reduced the more readily.

(10)

[In competitive rate studies the simple molar ratio of prod-

ucts is often taken to represent the relative rate constant. This is

acceptable for competitive reactions of a single species such as

occurs in radical clock experiments but, in experiments involv-

ing competition between diﬀerent species which react at sub-

stantially diﬀerent rates an integration over the competitive

interval which gives a logarithmic ratio such as eqn. (10) is

advisable].

Eqn.(10) indicates that the relative reduction rate of com-

peting diazonium ions can be obtained from the percentage

yields of the derived phenols: k_X/k_Y= ln [100/(100 Ϫ X%)]/ln

[100/(100 Ϫ Y%)]. Table 1 gives the evaluation of k_Cl/k_Hfrom

the phenol yields shown in the competitive domain of Fig. 1.

The relative rates of reduction for other competed pairs of

diazonium ions were obtained in like manner then, by serial

multiplication where necessary, all were re-expressed relative to

the rate for benzenediazonium ion itself. As a percentage of

the mean values from competed pairs of diazonium ions, the

uncertainties (twice the standard error on the mean) averaged

9%. In propagating error in the serial multiplications, we have

therefore accorded each value an error of 10%. The rates of the

reduction steps in Sandmeyer hydroxylation of substituted

diazonium ions relative to that of the unsubstituted ion are

given in Table 2. It is seen that the values span four orders of

magnitude and those ions with electron withdrawing groups are

reduced the more readily.

(ii) Correlation analysis of the relative rates of reduction of

substituted diazonium ions

In this section is presented an analysis of the response to vari-

ation in substitution of the relative rates of reduction given in

Tables 2 and 3. The logarithms of the relative rate constants

were correlated using simple or multiple linear regression with

single or dual substituent parameters as explanatory variables

and with an intercept being allowed in each case. The simple

ϩ

substituent parameters σ_m, σ_pand σ_pwere taken from the

compendium of Hansch, Leo and Taft³⁷and σ⁰values were

those of Wells, Ehrenson and Taft.³⁸The dual substituent

ϩ

parameters, σ_Iwith one of σ_R^BA, σ_R⁰, σ_Ror σ_R^Ϫ, were taken

from the review of Ehrenson, Brownlee and Taft.³⁹The results

are summarised in Table 4. Correlations involving explanatory

variables other than those indicated in the table were all poorer

as judged in terms of the appropriateness of a linear correlation

model or the precision of ﬁt to such a model. The goodness of

ﬁt of data to a linear model is judged on the basis of the

F-statistic and its signiﬁcance, the magnitude of the stand-

ard deviation of the estimate, s, and by the use of Exner’s

Citrate-promoted hydroxylation. Several mononuclear and

binuclear citratocopper() complexes have been described

J. Chem. Soc., Perkin Trans. 2, 2002, 1135–1150

1137

View Online

Table 2 Relative rate constants for the reduction of substituted diazo-

requires the separation of data for diﬀerently positioned sub-

nium ions by Cu^ϩaq

stituents because inductive (I) and mesomeric (M) eﬀects com-

bine in proportions which are position-dependent. Preliminary

to the application of multiple linear regression, entries 3 and 4

report simple regressions for the subset of eleven 4-substituted

diazonium ions. Comparison of entries 3 and 4 shows that the

use of σ_p^ϩin place of σ_pfor the 4-substituents having ϩI or ϩM

eﬀects in the reduced data set improves the correlation but it is

still not precise, 0.10 < ψ < 0.20 representing only ‘fair’ good-

ness of ﬁt.⁴¹Nor does multiple linear regression using σ_Iwith

σ_R^ϩas explanatory variables improve the ﬁt (compare entries 5

and 4). Evidently, the variation in the 11-point subset is not

explained satisfactorily in terms of any single combination of

inductive and resonance eﬀects which applies to the whole sub-

set. However, if the data are further subdivided, the six points

for those 4-substituents having ϩI or ϩM eﬀects (and including

H) are precisely accommodated by multiple linear regression

using σ_Iwith σ_R^ϩas explanatory variables (entry 6, Fig. 3) and

Substituent,

X

Substituent,

X

k_4X/k_H

k_3X/k_H

4-OMe

4-OPh

4-Me

4-H

4-F

(5.01 0.71) × 10^Ϫ33-Me

(4.95 0.70) × 10^Ϫ23-OMe

(5.81 0.58) × 10^Ϫ23-Cl

(7.47 0.75) × 10^Ϫ1

2.52 0.25

16.4 2.3

1.00 0.10

30.9 5.7

49.5 9.9

50.4 10.1

116 23

1.00 0.10

2.02 0.20

5.49 0.55

20.2 3.5

27.8 3.9

56.2 7.9

83.6 14.5

110 19

3-H

3-CF₃

3-CN

3-NO₂

3,5-bis-CF₃

4-Cl

4-COMe

4-CO₂Et

4-CF₃

4-CN

4-NO₂

Table 3 Relative rate constants for the reduction by citrato- and

chloro-copper() complexes of various 4-substituted benzenediazonium

ions

Substituent

Citratocopper() k_4X/k_H

Chlorocopper() k_4X/k_H

4-OMe

4-OPh

4-Me

4-H

4-F

0.0785 0.0111

0.249 0.035

0.342 0.034

1.00 0.10

1.37 0.14

1.99 0.20

2.74 0.47

3.31 0.47

4.31 0.75

5.57 0.96

8.06 1.40

0.00529 0.00075

0.0525 0.0074

0.0655 0.0066

1.00 0.10

2.69 0.27

10.7 1.1

42.2 7.3

78.6 11.1

74.5 10.5

579 100

4-Cl

4-COMe

4-CO₂Et

4-CF₃

4-CN

4-NO₂

1700 290

Fig. 3 Plots of log (k_4X/k_H) versus the eﬀective substituent constant,

σ_eff, for the reduction step in Sandmeyer hydroxylation of 4-substituted

benzenediazonium ions; plot 1, σ_eff= (σ_Iϩ 0.74σ_R^ϩ); plot 2, σ_eff= σ_I.

there is no signiﬁcant intercept. The ﬁve points for the 4-

substituents having ϪI and ϪM substituents are not correlated

on the same line but, taken as a group, they are correlated by σ_I

alone (entry 7, Fig. 3); no admixture of resonance character

(i.e. via any scale of σ_R) is statistically signiﬁcant. The quality of

the correlation is only ‘fair’ but no worse than that of entry 5.

The variation of the 11-point subset is thus better described by

further subdivision of the subset than otherwise. We note at this

juncture that the susceptibility to inductive eﬀects, ρ_I, diﬀers

between the two subsets of 4-substituents and also that the

correlation line of the ϪI and ϪM subset shows a signiﬁcant

intercept.

Fig. 2 Curved Hammett plot for the rate determining reduction step

in Sandmeyer hydroxylation of 3- and 4-substituted benzenediazonium

ions.

parameter ψ which facilitates comparisons between corre-

lations diﬀering in size of data set or degrees of freedom.^40,41

Entry 8 in Table 4 gives the results of multiple linear regres-

sion applied to the data for the subset of 3-substituents which

spans the full range of electronic characteristics. The more

precise ﬁt of the majority of points is obtained if those for the

3-NO₂and 3,5-bis-CF₃substituents are excluded from the

correlation (entry 9, Fig. 4); the intercept is not signiﬁcantly

diﬀerent from zero. Entry 10 correlates the data for the 3-CF₃,

3-CN, 3-NO₂and 3,5-bis-CF₃substituents with σ_Iin a manner

comparable with entry 7 and it also includes the intercept of the

latter entry as a data-point (see Discussion).

Hydroxylation. Fig. 2 shows the Hammett plot of the relative

rates of reduction during Sandmeyer hydroxylation (cf. Table

2). The plot exhibits both curvature and scatter and clearly

a linear model is not appropriate for the correlation of

the complete data set which includes points for both 3- and

ϩ

4-substituents. The use of σ_pin place of σ_pfor 4-substituents

having ϩI or ϩM eﬀects reduces the curvature somewhat but

a linear model is still not appropriate. (In Table 4 no standard

deviations of the estimate, s, intercept, or regression coef-

ﬁcients, ρ, are quoted where the trend-line of the data points is

curved and a linear model is inappropriate, e.g. entries 1 and 2).

The application of multiple linear regression to a data set

Citrate-promoted hydroxylation. Entries 11 to 17 in Table 4

report correlations for the data set of eleven 4-substituted

1138

J. Chem. Soc., Perkin Trans. 2, 2002, 1135–1150

View Online

Table 4 Correlation analysis of the logarithms of the relative rate constants for the reduction steps in Sandmeyer reactions

Regression coeﬃcients

Explanatory

Entry Substituents variables

d

n^aR^{2 b}

F^c

F_signif

ψ^e

s^f

Intercept c

ρ

ρ_I

ρ_R

Hydroxylation

1

2

3

3 and 4

4, all

σ_mor σ_p

σ_m, σ_por σ_p

σ_p

17 0.909 150.07 3.26 × 10^Ϫ90.321

17 0.938 226.30 1.86 × 10^Ϫ100.265

11 0.923 107.21 2.68 × 10^Ϫ60.307

ϩ

4

5

4, all

σ_por σ_p

σ_Iand σ_R

11 0.971 304.88 3.00 × 10^Ϫ80.188 0.264

0.423 0.187 2.816 0.322

11 0.970 130.89 7.73 × 10^Ϫ70.203 0.285 Ϫ0.133 0.410

ϩ

3.362 0.965 2.774 0.481

4.626 0.673 3.408 0.405

1.915 0.543

ϩ

6

7

4, ϩI & ϩM σ_Iand σ_R

4, ϪI & ϪM σ_I

6

5

8

6

5

0.996 396.32 2.32 × 10^Ϫ40.098 0.092 Ϫ0.100 0.221

0.974 114.30 1.75 × 10^Ϫ30.208 0.058

0.925 74.11 1.35 × 10^Ϫ40.316

0.834 0.269

8

9

3, all

σ_m

BA

3, Me to CN σ_Iand σ_R

0.996 337.40 2.94 × 10^Ϫ40.089 0.068

0.990 286.17 4.50 × 10^Ϫ40.135 0.054

0.040 0.152

0.855 0.153

2.906 0.385 0.687 0.363

1.370 0.258

10

3, CF₃to

(CF₃)₂

σ_I

Citrate-promoted hydroxylation

11

12

13

14

15

16

4, all

σ_p

σ_por σ_p

σ_Iand σ_R

11 0.899 79.91 9.02 × 10^Ϫ60.351

11 0.924 109.79 2.42 × 10^Ϫ60.305

ϩ

11 0.966 112.65 1.38 × 10^Ϫ60.216 0.130 Ϫ0.084 0.187

1.389 0.440 1.186 0.220

1.992 0.306 1.585 0.184

1.135 0.357

4, ϩI & ϩM σ_Iand σ_R

4, ϪI & ϪM σ_I

with H

6

5

6

0.996 399.12 2.29 × 10^Ϫ40.089 0.042

0.971 102.09 2.06 × 10^Ϫ30.220 0.036

0.979 187.23 1.65 × 10^Ϫ40.177 0.050

0.000 0.100

0.141 0.168

0.045 0.116

1.328 0.270

17

4, ϪI & ϪM σ_I

with H,

8

0.489 5.75

5.34 × 10^Ϫ20.825

F & Cl

Chlorination

18

19

20

21

22

4, all

σ_p

σ_por σ_p

σ_Iand σ_R

11 0.947 160.12 4.89 × 10^Ϫ70.254

ϩ

11 0.967 260.13 6.00 × 10^Ϫ80.201 0.341

0.282 0.241 3.359 0.471

11 0.992 508.78 3.70 × 10^Ϫ90.105 0.175 Ϫ0.096 0.251

4.443 0.591 3.223 0.295

5.069 0.415 3.592 0.250

3.983 2.784

ϩ

4, ϩI & ϩM σ_Iand σ_R

4, ϪI & ϪM σ_I

6

5

0.999 1186.4 4.49 × 10^Ϫ40.045 0.057 Ϫ0.035 0.135

0.874 20.72 1.99 × 10^Ϫ20.561 0.281

0.493 1.310

^aNo. of data points. ^bR is the correlation coeﬃcient. ^cF = R²(n Ϫ m)/[(1 Ϫ R²)(m Ϫ 1)] where (m Ϫ 1) is the number of explanatory variables.

^dThe signiﬁcance of F is greater the lower the value of F_signif^eψ = [n(1 Ϫ R²)/(n Ϫ m)]^1/2; goodness of ﬁt criteria: ψ < 0.02, very good; 0.02 < ψ < 0.1,

good; 0.1 < ψ < 0.2, fair; 0.2 < ψ < 0.5, poor (cf. refs. 40, 41). ^fStandard deviation of the estimate.

.

ϩ

is inappropriate. Multiple linear regression using σ_Iand σ_R

gives a ﬁt of the full data set which is ‘fair’ (entry 13) but sub-

division of the set according to the electronic characteristics of

the substituents improves the precision for the ϩI and ϩM

substituents, including H (entry 14 and Fig. 5) and the ϪI and

Fig. 4 Plots of log (k_3X/k_H) versus the eﬀective substituent constant,

σ_eff, for the reduction step in Sandmeyer hydroxylation of 3-substituted

benzenediazonium ions; plot 1, σ_eff= (σ_Iϩ 0.24σ_R^BA); plot 2, σ_eff= σ_I; ×

marks the intercept of plot 2 in Fig. 3.

diazonium ions. The ﬁndings parallel those for the same ions

Fig. 5 Plots of log (k_4X/k_H) versus the eﬀective substituent constant,

when reduced under simple hydroxylation conditions: corre-

σ_eff

,

for the reduction step in citrate-promoted Sandmeyer

ϩ

lations in terms of σ_por σ_pwith σ_pfor ϩI and ϩM substitu-

hydroxylation of 4-substituted benzenediazonium ions; plot 1, σ_eff

=

(σ_Iϩ 0.79σ_R^ϩ); plot 2, σ_eff= σ_I.

ents (entries 11 and 12) produce curves for which a linear model

J. Chem. Soc., Perkin Trans. 2, 2002, 1135–1150

1139

View Online

ϪM substituents are again best correlated by σ_Ialone (entry 15,

Fig. 5). As for simple hydroxylation, the variation in the

dependent variable for citrate-promoted hydroxylation is best

accounted for by subdivision of the 11-point set according to

the nature of the substituents. The range of log (k_4X/k_H) values

for citrate-promoted hydroxylation is narrower than that for

simple hydroxylation and the intercept for entry 15 is small and

appears not to be signiﬁcantly diﬀerent from zero; indeed, the

precision of correlation of the data for ϪI and ϪM substituents

is improved if the datum for H is included with them (entry 16).

At ﬁrst sight, it would appear that if it were valid to include H

with the ϪI and ϪM substituents, it would be necessary also to

include the data for F and Cl as these groups exert net electron

withdrawing eﬀects intermediate in magnitude between H and

the ϪI and ϪM substituents; however, these halogens do not

correlate with H and the ϪI and ϪM substituents (entry 17)

(see Discussion).

determining step but which arise from a subsequent fast step

(ligand transfer) where aryl radicals are trapped. Curvature

might conceivably occur if the trapping eﬃciency were to vary

signiﬁcantly across the substituent sequence so that the

apportioning of aryl radicals between phenolic products and

by-products varied. Water (or hydroxide) transfer is the slowest

of the Sandmeyer ligand transfers and electron deﬁcient

radicals react the least rapidly;²⁴furthermore, the success of

the competition experiments used to obtain the results requires

suﬃcient side-reaction to arrest the catalytic cycle. There is

therefore the potential for curvature in Fig. 2 to occur trivially

as an analytical artefact. In order to minimise this possibility,

we used a concentration of Cu^2ϩthat was suﬃcient to ensure

that, given an eﬃcient catalytic cycle, each conversion of diazo-

nium ion to phenol would be maximised: outside the competi-

tive domain, reactant accountability as phenol exceeded 80%

for the 4-substituted series and 70% for the 3-substituted series.

In view of this, we are conﬁdent that the curvature in Fig. 2 is

due to underlying mechanistic cause and not artefact. The ﬁnd-

ing that the data used in constructing the ﬁgure are amenable to

correlation analysis supports this conclusion.

Chlorination. Entries 18–22 of Table 4 summarise corre-

lations for the set of eleven 4-substituted diazonium ions. Here

the pattern of behaviour is rather diﬀerent from that found for

the two hydroxylations. As for them, the simple Hammett plot

is curved (entry 18) and correlation is improved by using σ_p^ϩfor

the ϩI and ϩM substituents (entry 19) but multiple linear

regression using σ_Iand σ_R^ϩnow gives ‘good’ correlation of the

whole data set (entry 20, Fig. 6). If the ϩI and ϩM groups are

(ii) Summary of substituent eﬀects

In Results (ii) the best ﬁttings of experimental relative rate con-

stants (k_X/k_H) to eqn. (11) were identiﬁed.

log (k_X/k_H) = ρ_Iσ_Iϩ ρ_Rσ_Rϩ c

(11)

In every case requiring two explanatory variables, c proved to

be zero within experimental error (cf. Table 4, entries 6, 9, 14

and 20). Hence, if λ = ρ_R/ρ_I,

log (k_X/k_H) = ρ_I[σ_Iϩ λσ_R] = ρ_Iσ_eff

(12)

where σ_effis the eﬀective substituent parameter. In three cases

ρ_R(and hence λ) proved to be zero (cf. Table 4, entries 7, 10 and

15).

For Sandmeyer hydroxylation eqn. (13)–(17) give the best

account of the experimental data:

4-substituents (ϩI or ϩM and including H),

ϩ

log (k_4X/k_H) = (4.63 0.67)σ_Iϩ (3.41 0.40)σ_R

ϩ

= (4.63 0.67)[σ_Iϩ (0.74 0.14)σ_R

4-substituents (ϪI or ϪM),

log (k_4X/k_H) = (1.91 0.54)σ_Iϩ (0.83 0.27) (14)

3-substituents (Me to CN),

]

(13)

Fig. 6 Plot of log (k_4X/k_H) versus the eﬀective substituent constant,

σ_eff, for the reduction step in Sandmeyer chlorination of 4-substituted

benzenediazonium ions, σ_eff= (σ_Iϩ 0.72σ_R^ϩ).

taken separately, their correlation using σ_Iand σ_R^ϩis improved

(compare entries 20 and 21) but the ϪI and ϪM substituents

are not then adequately correlated as a separate subset by σ_I

(entry 22) nor is any scale of σ_Rstatistically signiﬁcant in

correlating the subset. We conclude that the variation within

the 11-point data set for chlorination is best accounted for by

retaining the integrity of the set over the full range of substitu-

ent character.

BA

log (k_3X/k_H) = (2.91 0.38)σ_Iϩ (0.69 0.36)σ_R

= (2.91 0.38)[σ_Iϩ(0.24 0.13)σ_R

BA

]

(15)

3-substituents [CF₃to 3,5-(CF₃)₂],

log (k_3X/k_H) = (1.37 0.26)σ_Iϩ (0.85 0.15) (16)

The deﬁnition of eqn. (16) is discussed later.

For citrate-promoted hydroxylation the best equations are

(17) and (18):

Discussion

(i) The curved Hammett plot for simple Sandmeyer hydroxylation

4-substituents (ϩI or ϩM and including H),

Diﬀerent origins for ‘downward concavity’ in a Hammett plot

can be envisaged according to the reaction concerned. For the

particular case of Fig. 2, the plot is constructed from analyses

of products which are not directly formed in the actual rate

ϩ

log (k_4X/k_H) = (1.99 0.31)σ_Iϩ (1.58 0.18)σ_R

= (1.99 0.31)[σ_Iϩ (0.79 0.15)σ_R^ϩ] (17)

1140

J. Chem. Soc., Perkin Trans. 2, 2002, 1135–1150

View Online

4-substituents (ϪI or ϪM),

log (k_4X/k_H) = (1.13 0.36)σ_Iϩ (0.14 0.17) (18)

equilibrium constant for the same reaction, k₁₁and k₂₂are the

respective self-exchange rate constants of the reactant species

and Z is the collision frequency (i.e. 10¹¹^Ϫ1).

s

The standard reduction potentials of the couples

Fe^(EDTA)^Ϫ/Fe^(EDTA)^2Ϫand Cu^2ϩaq/Cu^ϩaq are similar

(0.12 and 0.159 V vs. NHE, respectively)^43,44and consequently

the reductions of any given species by Fe^(EDTA)^2Ϫand Cu^ϩ-

aq are expected to have similar free energy changes of reaction

and hence similar values of K₁₂. However, the respective self-

exchange rates of the couples Fe^(EDTA)^Ϫ/Fe^(EDTA)^2Ϫand

Cu^2ϩaq/Cu^ϩaq are ca. 3 × 10⁴dm³mol^Ϫ1s^Ϫ1and 1 × 10^Ϫ5dm³

For Sandmeyer chlorination the experimental data are best

described solely by eqn. (19):

all 4-substituents

ϩ

log (k_4X/k_H) = (4.44 0.59)σ_Iϩ (3.22 0.29)σ_R

= (4.44 0.59)[σ_Iϩ (0.72 0.12)σ_R^ϩ] (19)

mol^Ϫ1s^{Ϫ1 45,46}

eqn. (20) and (21) thus lead to the expectation

,

that, in an outer-sphere process, Fe^(EDTA)^2Ϫshould reduce a

given substrate more rapidly than aquacopper().

Previously, we have measured absolute rate constants for the

outer-sphere reduction of a series of 4-substituted diazonium

ions by Fe^(EDTA)^2Ϫat pH 9 and have correlated these by

eqn. (22).⁴⁷

Correlation analysis has thus allowed two distinct types

of relative rate behaviour to be discerned; one is dependent

solely on the inductive eﬀects of substituents whereas the

other depends additionally on their mesomeric eﬀects. For

4-substituted diazonium ions in both simple and promoted

hydroxylation and for 3-substituted diazonium ions in simple

hydroxylation, the eﬀects of the most strongly electron with-

drawing substituents on the relative reaction rate are due to

their inductive eﬀects alone and are independent of their

mesomeric eﬀects (σ_eff= σ_I). By contrast, in both hydroxylations

and chlorination the eﬀects of 4-substituents with ϩM eﬀects

are measured by σ_eff= [σ_Iϩ (0.75 0.08)σ_R^ϩ]; the dependence

ϩ

)

log (k_4X/k_H) = 1.74(σ_Iϩ 0.78σ_R

(22)

Although an absolute rate constant for the reduction for

4-nitrobenzenediazonium ion was not directly measurable

owing to the limitations of the EPR technique employed,

eqn. (22) and the value of 1.8 × 10⁵dm³mol^Ϫ1s^Ϫ1measured for

k_Hmay be used to derive for it a value of 3.9 × 10⁶dm³mol^Ϫ1

s^Ϫ1which is 3 to 4 orders of magnitude below the diﬀusion-

controlled rate. It is very improbable therefore that the curv-

ature in Fig. 2 arises because diazonium ions with electron

withdrawing substituents, such as 4-nitrobenzenediazonium

ion, are reduced by aquacopper() in an outer-sphere process at

rates approaching the diﬀusion-controlled limit.

The curvature of Fig. 2 thus suggests the possibility of a

change in rate-determining step within a multi-step mechan-

ism and the resolution of the relative rate data for both

3- and 4-substituted diazonium ions into two distinct corre-

lations [eqn. (13) with (14) and eqn (15) with (16)] provides

conﬁrmation. In both series, the diazonium ions with sub-

stituents capable of electron donation and the unsubstituted

ion undergo reactions with rate determining steps which are

diﬀerent from those of ions having electron withdrawing sub-

stituents. The intercepts in eqn. (14) and (16) occur because

the correlated substituted ions react with rate determining

steps diﬀerent from that of the parent ion with which they

are compared.

ϩ

on σ_Rimplies that the reactivity perturbed by the 4-substi-

tuents is strongly inﬂuenced by their capacity for electron

donation to the π-system to which they are attached. For

3-substituents in simple hydroxylation, the eﬀects of ϩM sub-

stituents and the weakly electron withdrawing substituents are

measured by σ_eff= [σ_Iϩ(0.24 0.13)σ_R^BA]. The lower value of λ

and the change of scale of σ_Rfound for 3-substituents by com-

parison with the similar 4-substituents no doubt reﬂect the

reduced conjugation in the 3-substituted series between the

substituent and the diazonium function.

The ﬁnding that the value of ρ_Iin eqn. (13) is larger than

its value in eqn. (15) indicates that the relative rates of the

4-substituted diazonium ions show a greater susceptibility to

inductive eﬀects than those of the 3-substituted ions despite the

substituent being at a greater distance from the diazonium

function. This is well precedented and the I_πeﬀect, i.e. the

inductive polarisation of the π-electrons of the substituted ring

(cf. 1 and 2) is invoked in explanation.³⁸The ﬁnding that ρ_I(para)

> ρ_I(meta)may be construed as evidence that the π-system of the

diazonium ions is particularly important for the reactivity being

perturbed by substituent eﬀects.

Although an outer-sphere reduction is envisaged as a one-

step electron transfer process within an encounter complex of

the reactants, such a process is not necessarily excluded for the

reduction of diazonium ions by the evidence of two diﬀerent

rate determining steps within the reaction series. For example,

ؒ

it remains possible that if an aryldiazenyl radical, ArN₂, were

to be formed reversibly in the reduction, its fragmentation to

nitrogen and an aryl radical might conceivably become rate

determining in cases where electron transfer is fast, i.e. for ions

with electron withdrawing substituents (Scheme 3).

(iii) Reduction mechanism

If Fig. 2 were to relate to an outer-sphere electron transfer to

diazonium ions by aquacopper(), the curvature could imply

that the rates of reduction of ions with electron withdrawing

substituents approach the diﬀusion-controlled limit. We con-

sider this is unlikely to be the case from the following appli-

cation of the Marcus cross-relation:⁴²

Scheme 3

k₁₂= (k₁₁k₂₂K₁₂f₁₂)^1/2

(20)

(21)

With the notation given in Scheme 3, application of the

ln f₁₂= (ln K₁₂)²/4ln (k₁₁k₂₂/Z²)

ؒ

steady state hypothesis to ArN₂gives eqn. (23).

where k₁₂is the second order rate constant for an outer-

sphere electron transfer between species 1 and 2, K₁₂is the

Rate = k ^Fk ^ET[ArN₂^ϩ][Cu^]/(k ^ϪET[Cu^] ϩ k ^F) (23)

J. Chem. Soc., Perkin Trans. 2, 2002, 1135–1150

1141

View Online

It is evident that for rate determining fragmentation of the

diazenyl radical (k ^FӶ k ^ϪET[Cu^]) the rate is inversely pro-

portional to [Cu^] whereas for fast diazenyl radical fragmen-

tation (k ^Fӷ k ^ϪET[Cu^]) the rate is independent of [Cu^]. By

competitively reducing a pair of diazonium ions which do

not have the same rate determining steps, the eﬀect of chan-

ging [Cu^] should be to change their relative rate. No signiﬁ-

cant increase in the relative rate for the 4-CO₂Et/4-Cl pair

(from 5.5 when [Cu^] = 0.3 mol dm^Ϫ3) was observed when

[Cu^] was halved nor was any signiﬁcant decrease observed in

the relative rate of the 4-CF₃/4-Cl pair (from 10.3 when [Cu^]

= 0.3 mol dm^Ϫ3) when [Cu^] was doubled. There is thus no

evidence for any involvement of rate determining diazenyl

radical fragmentation. The conclusion from the results for

simple Sandmeyer hydroxylation is that, whatever the nature

of the two rate determining steps discovered, they are not

consistent with the reduction by Cu^ϩaq of diazonium ions

with electron donating substituents occurring via an outer-

sphere process which becomes non-rate determining for ions

with electron withdrawing substituents. The reduction must

consequently occur by an inner-sphere process, i.e. one

involving the formation of an intermediate which facilitates

electron transfer.

In principle, any of the three steps—formation of the pre-

cursor complex, activation to enable electron transfer proper,

or dissociation of the successor complex—might be the over-

all rate determining step. Application of the steady state

hypothesis to the precursor and successor complexes of

Scheme 4, using the notation shown in the scheme, results in

eqn. (24).

Rate = K ^Ek ^Ak ^ETk ^D[ArN₂^ϩ][Cu^]/(k ^Dk ^ϪAϩ

k ^Dk ^ETϩ k ^ϪETk ^ϪA

)

(24)

As Cu^undergoes rapid ligand exchange,¹⁴the rate of

dissociation of the successor complex comprising Cu^and a

diazenyl radical, itself liable to rapid fragmentation,¹²is

expected to be fast and irreversible with the consequence that

electron transfer is irreversible. Eqn. (24) is thus simpliﬁed to

eqn. (25).

Rate = {K ^Ek ^Ak ^ET/(k ^ϪAϩ k ^ET)}[ArN₂^ϩ][Cu^] (25)

When k ^ETӶ k ^ϪA, electron transfer is rate determining and

eqn. (25) is simpliﬁed to eqn. (26):

Although little is known of the species responsible for

reduction in the citrate-promoted hydroxylation, the lower

reduction potentials observed^33,34for the Cu^/Cu^couple in

the presence of citrate than in its absence are consistent

with the presence of citrate resulting in a more powerful

reductant. If reduction rates increase with a falling reduction

potential of the reductant, the citratocopper() should reduce

diazonium ions faster than aquacopper() and the diminished

selectivity observed for reactions in the citrate-containing sys-

tem is thus consistent with the reactivity–selectivity principle.

The substituent-dependent pattern of behaviour reﬂected by

eqn. (17) and (18) parallels that reﬂected by eqn. (13) and

(14); evidently, increased absolute rates of reduction in the

citrate-containing system do not aﬀect the occurrence of a

behavioural change at the same position in the substituent

sequence as occurs for simple hydroxylation and the citrato-

copper() reductant, like Cu^ϩaq, is inferred to reduce diazo-

nium ions which are substituted by electron donating groups

by an inner-sphere mechanism (see later).

Rate = (K ^Ek ^Ak ^ET/k ^ϪA)[ArN₂^ϩ][Cu^]

(26)

and when k ^ETӷ k ^ϪA, association to form the precursor com-

plex is rate determining and eqn. (25) is simpliﬁed to eqn. (27):

Rate = K ^Ek ^A[ArN₂^ϩ][Cu^]

(27)

The change in rate determining step uncovered by the corre-

lation analysis for diazonium ion reduction during simple

Sandmeyer hydroxylation is thus identiﬁed as being between

electron transfer and association of the reactants to form the

precursor complex.

In order to assign the correlations found for simple hydroxy-

lation [eqn. (13) and (14)] to the appropriate rate determining

step, we note again that they have disparate gradients and that

both are positive. If it is assumed that eqn. (13) relates to rate

determining electron transfer, it follows from eqn. (26) that the

observed relative rate constants, k_4X/k_H, correlated by eqn. (13)

Scheme 4 illustrates an inner-sphere reduction mechanism

correspond to (k ^Ak ^ET/k ^ϪA)_X/(k ^Ak ^ET/k ^ϪA

)

(K ^Ecancels being

H

independent of substitution) and the gradient of eqn. (13) is

hence (ρ_I^ETϩ ρ_I^AϪ ρ_I^ϪA). The relative rate constants correlated

by eqn. (14) correspond to k ^A_X/(k ^Ak ^ET/k ^ϪA

) since H falls in

H

the subset of substituents for which rate determining electron

transfer has been assumed. However, k ^A_X/(k ^Ak ^ET/k ^ϪA)_H= (k ^A

/

X

k ^A_H) × (k ^ϪA_H/k ^ET_H) = (k ^A_X/k ^A_H) × P, where P > 1 since the

condition k ^ET< k ^ϪAholds generally when electron transfer is

rate determining; log₁₀P is identiﬁed as the intercept of

eqn. (14). The gradient of eqn. (14) is thus ρ_I^Awhich is positive,

i.e. association to form the precursor complex from the

encounter complex is assisted by inductively electron withdraw-

ing substitution in the diazonium ion. It follows that the reverse

process will be assisted by the opposite electronic character and

that ρ_I^ϪAis hence expected to be negative. Electron transfer

from Cu^to the diazonium moiety within the precursor complex

will be assisted by electron withdrawing substituents making

ρ_I^ETpositive. The gradient of eqn. (13), (ρ_I^ETϩ ρ_I^AϪ ρ_I^ϪA), is

therefore the sum of three positive terms whereas that of

eqn. (14) equals only one of these terms. The assumption that

eqn. (13) relates to rate determining electron transfer thus

requires eqn. (13) to have a gradient larger than eqn. (14); this

accords with observation and so validates the initial assump-

tion. [The contrary assumption that eqn. (13) relates to rate

determining association fails because it requires electron trans-

fer to become rate determining only for the most electron

withdrawing substituents which is nonsensical.]

Scheme 4

for diazonium ions. The reactant species form an encounter

complex, the concentration of which will depend on the dif-

fusive characteristics of the particular species and their ionic

charges; substitution in the diazonium ions is not expected to

be important for the encounter process. The components of

the encounter complex then associate to produce the precur-

sor complex.⁴⁸This process, involving bond formation, is

expected to be susceptible to substituent eﬀects. Electron

transfer, which is also expected to be susceptible to substitu-

ent eﬀects in the diazonium moiety, occurs when vibrational

excitation takes the precursor complex to a state which is also

accessible to the successor complex, i.e. the complex formed

by the reacting species after electron transfer. The latter com-

plex then dissociates in a process expected to be substituent

dependent.

1142

J. Chem. Soc., Perkin Trans. 2, 2002, 1135–1150

View Online

(iv) Electron transfer as the rate determining step

eqn. (15) and (16)† to deduce a corresponding expression for the

3-substituted series.

A similar analysis may be applied to the case of citrate-

promoted hydroxylation and the results compared to obtain

insight into how the promotion of the reduction step is derived.

With reference to eqn. (14) and (18), the values of ρ_I^Aare

similar for both hydroxylations, (1.91 0.54) and (1.13 0.36)

for the simple and citrate-promoted processes, respectively.

Evidently the promotional eﬀect does not reside in any greatly

changed sensitivity to substituent variation on the part of the

relative rates at which the precursor complexes are formed from

the encounter complexes. The gradients of eqn. (13) and (17)

correspond to the respective values of (ρ_I^ETϩ ρ_I^AϪ ρ_I^ϪA) for

the two hydroxylations. Since the values of ρ_I^Aare similar, the

diﬀerence in gradients must correspond mainly to diﬀerences in

the values of (ρ_I^ETϪ ρ_I^ϪA). As the values of ρ_I^Ado not diﬀer

greatly between the two hydroxylations, the same might reason-

ably be presumed for ρ_I^ϪAand hence the diﬀerence in gradients

of eqn. (13) and (17) is inferred to stem essentially from diﬀer-

ences in ρ_I^ET. We have seen that the intercepts of eqn. (14) and

We obtain the inequality (30):

log (k_3X/k_H)^ET< (1.54 0.46)σ_Iϩ (0.69 0.36)σ_R

BA

< (1.54 0.46)[σ_Iϩ (0.45 0.27)σ_R

]

(30)

We note that ρ_I(meta)< ρ_I(para)for the electron transfer step

consistent with the similar ﬁnding for the overall ρ_Ivalues; as

expected, ρ_R(meta)< ρ_R(para)

.

The relative rates of reduction during Sandmeyer chlorin-

ation do not exhibit evidence of a change in rate determining

step and the single correlation observed [eqn. (19)] is consistent

with electron transfer as the rate determining step over the

entire range of substituents considered. The ﬁnding that

eqn. (19) has the same algebraic form as eqn. (13) supports this

suggestion. Indeed, the two equations appear to be identical

within the indicated uncertainties. This must be fortuitous; they

cannot apply to the same actual reaction as they encompass

diﬀerent ranges of substitution and the former involves mixed

chlorocuprate reductants. At the chloride concentration of our

experiments (0.3 mol dm^Ϫ3) only the less chlorinated couples

(18) are given by their respective values of log₁₀P = log₁₀(k ^ϪA

/

H

k ^ET_H); a small intercept in eqn. (18) is thus consistent with a

much greater value of k ^ET_Hrelative to k ^ϪA_Hin the case of pro-

moted hydroxylation. Generalising across the range of substi-

tution, the evidence therefore supports the inference that the

origin of the eﬀect that citrate ligation of the copper catalyst

has on the reduction step of Sandmeyer hydroxylation is via

increased rates of electron transfer with an associated reduction

in selectivity.

are important^35,36and a self-exchange rate of 20 dm³mol^Ϫ1s^Ϫ1

Ϫ

/

has been estimated for the predominant of these (Cu^Cl₂

Cu^Cl₂).⁴⁹The ratio of self-exchange rates of Cu^Cl₂^Ϫ/Cu^Cl₂

and Cu^ϩaq/Cu^2ϩaq⁴⁶is hence 20/(1 × 10^Ϫ5) = 2 × 10⁶. Thus, if

aquacopper() were to reduce a given substrate in an outer-

sphere process, the Marcus cross-relation implies the eﬀect of

introducing two chloride ligands on the intrinsic barrier to elec-

tron transfer should be to increase the reduction rate by a factor

of up to 2 × 10⁶. However, the chloride ligation also raises the

standard reduction potential of the reductant. This attenuates

the rate at which the dichloro-complex will reduce a particular

substrate, relative to the rate of the aqua-complex, by a factor

For simple hydroxylation, ρ_I^A= (1.91 0.54) [eqn. (14)] and

since Ϫρ_I^ϪAis a positive number it follows that (ρ_I^AϪ ρ_I^ϪA) >

(1.91 0.54). When electron transfer is rate determining [eqn.

(13)] the observed gradient, ρ_I, is (4.63 0.67) i.e. ρ_I= (4.63

0.67) = (ρ_I^ETϩ ρ_I^AϪ ρ_I^ϪA), hence, ρ_I^ET< [(4.63 0.67) Ϫ (1.91

0.54)] = (2.72 0.86).

–– –

᭺

–– –

᭺

of exp(F∆E /RT) where ∆E is the diﬀerence in their stand-

ard reduction potentials, F and R are, respectively, the Faraday

and gas constants and T is the absolute temperature. Inserting

values, the factor is exp[9.469 × 10⁴× (0.159 Ϫ 0.50)/8.314 ×

298] = 2.2 × 10^Ϫ6. The eﬀect of the chloride ligation on reduc-

tion potential in diminishing the relative rate of reduction

would be to negate its eﬀect in reducing the intrinsic barrier to

outer-sphere electron transfer. It follows from this reasoning

that if, as we observe, the aquacopper() complex does not

reduce diazonium ions in an outer-sphere process, the dichloro-

cuprate is also unlikely to do so. Yandell⁴⁹has suggested that

the self-exchange rate constant for Cu^ϩaq/Cu^2ϩaq is 5.2 dm³

mol^Ϫ1, much larger than the value of 1 × 10^Ϫ5dm³mol^Ϫ1

obtained by Sutin and co-workers.⁴⁶Such a large value would

imply that the rate-enhancing eﬀect of the chloride ligation on

the intrinsic barrier to outer-sphere electron transfer would be

small and greatly oﬀ-set by its detrimental eﬀect on reduction

potential. The available evidence thus favours an inner-sphere

reduction step in the Sandmeyer chlorination reaction which

remains rate determining across the whole range of substituents

examined.

ET

It follows from eqn. (26) that ρ_Rin eqn. (13) equals (ρ_R

ϩ ρ_RϪ ρ_R^ϪA). The ﬁnding that σ_Iis the only required

A

explanatory variable in eqn. (14) indicates the association

process to be independent of the mesomeric eﬀects of sub-

stituents and the principle of microscopic reversibility

requires that its reverse should also be independent of such

A

ϪA

eﬀects, i.e. ρ_R= ρ_R

= 0. The value of ρ_Rin eqn. (13)

therefore corresponds to ρ_R^ET. For the electron transfer pro-

cess in the simple hydroxylation of 4-substituted diazonium

ions we hence deduce the inequality (28), where, since the

value of ρ_I^ETis an upper limit, the value of λ^ET, i.e. (1.25

0.42), is a lower limit.

ϩ

log (k_4X/k_H)^ET< (2.72 0.86)σ_Iϩ (3.41 0.40)σ_R

ϩ

< (2.72 0.86)[σ_Iϩ (1.25 0.42)σ_R

]

(28)

Analogous reasoning with respect to citrate-promoted

hydroxylation results in the inequality (29)

ϩ

log (k_4X/k_H)^ET< (0.86 0.47)σ_Iϩ (1.58 0.18)σ_R

ϩ

< (0.86 0.47)[σ_Iϩ (1.84 1.03)σ_R

]

(29)

† Eqn. (16) was derived somewhat diﬀerently from the other correl-

ations on account of the few data and their narrow range. Because they

occur near the intersection of the lines given by eqn. (15) and (16) (cf.

Fig. 4) the points for 3-CF₃and 3-CN are accommodated equally well

by both equations and we note also that eqn. (16) and eqn. (14) are

expected to have an intercept in common, i.e. log₁₀(k ^ϪA_H/k ^ET_H). We

therefore used data for 3-CF₃, 3-CN, 3-NO₂, 3,5-bis-(CF₃) and the

intercept found for eqn. (14) in evaluating regression coeﬃcients for

Although the uncertainties on the values of λ^ETestimated for

the two hydroxylation processes are large, the values themselves

are larger than 1, indicating the electron transfer process to be

strongly inﬂuenced by the mesomeric eﬀects of substituents.

The explanatory variable needed to represent the ϩM eﬀects of

substituents (σ_R^ϩ) suggests that, in the transition state for elec-

tron transfer, the lone pairs of the ϩM substituents interact

strongly with the π-system to which they are attached and, the

stronger this interaction, the more diﬃcult it is for the diazo-

nium moiety to accept the electron.

eqn. (16); error in log (k_{3,5-(CF )}/k_H) was minimised by competing

3

decomposition of the 3,5-bis(tri²ﬂuoromethyl)benzenediazonium ion

against diazonium ions in both the 3- and 4-substituted series (substitu-

ent parameter additivity was assumed for the disubstituted diazonium

ion). An attempt to extend the range of the data set by including

3,5-dinitrobenzenediazonium ion failed; no phenolic products were

produced from its decomposition in the same conditions as the other

diazonium ions, only tarry products being obtained.

Just as the expressions (28) and (29) were found for electron

transfers to 4-substituted diazonium ions, we can argue from

J. Chem. Soc., Perkin Trans. 2, 2002, 1135–1150

1143

View Online

Table 5 Comparative correlation analysis of selected electrophilic reactions of diazonium ions

λ = ρ_R/ρ_In^aR^{2 b}

d

Entry Reaction

Ref. Best correlation

F^c

F_signif

ψ^e

s^f

Anion addition to β-N^g

1

2

3

4

5

OH^Ϫ

50

51

2.86σ_p

2.35σ_p

3.65σ_p

1

1.27

0.94

6

0.993 580.4

0.996 912.3

0.997 1476.3

0.992 193.3

1.000 5734.8

1.76 × 10^Ϫ50.102 0.098

7.16 × 10^Ϫ60.077 0.064

2.74 × 10^Ϫ60.067 0.081

6.72 × 10^Ϫ40.126 0.098

4.23 × 10^Ϫ60.000 0.028

CN^Ϫ

Ϫ

BA

PhSO₂

2.13σ_Iϩ 2.70σ_RBA

3.64σ_Iϩ 3.42σ_R

Azo-coupling

1-Naphtholate

ϩ

6

7

8

9

10

11

(i)^g

52

53

3.99σ_Iϩ 2.75σ_R

4.29σ_Iϩ 0.68σ_R

4.00σ_p

4.04σ_Iϩ 1.39σ_R

0.69

0.16

1

0.34

0.62

0.30

6

5

6

5

0.999 1341.7

0.998 705.9

0.976 120.3

3.73 × 10^Ϫ50.045 0.081

9.76 × 10^Ϫ50.063 0.081

1.62 × 10^Ϫ30.200 0.300

ϩ

(ii)^h

1-Naphtholate-4-sulfonate (i)^g

(ii)^h

o

1.000 15668.8 9.37 × 10^Ϫ70.000 0.016

2-Naphthylamine-6-sulfonate (i)^g

(ii)^h

4.03σ_Iϩ 2.50σ_RBA

0.999 1763.5

0.999 1074.3

5.67 × 10^Ϫ40.050 0.058

9.30 × 10^Ϫ40.050 0.059

ϩ

4.30σ_Iϩ 1.32σ_R

Electron transfer^g

Polarographic E_1/2: sulfolane

nitromethane

ϩ

12

13

54

55

0.225σ_p^ϩ/V, 3.80σ_p

1

9

5

0.973 254.1

0.999 1558.0

9.28 × 10^Ϫ70.186 0.030

6.41 × 10^Ϫ40.050 0.008

(0.317σ_Iϩ 0.198σ_R^ϩ)/V, 0.62

ϩ

5.36σ_Iϩ 3.35σ_R

14

15

16

17

18

19

20

21

22

23

24

25

Decamethylferrocene, MeCN

Fe(CN)₆^4Ϫ, pH 7, H₂O

56

47

57

58

59

60

61

3.25σ_p

4.56σ_p

5.97σ_p

1.74σ_Iϩ 1.37σ_R

1.99σ_Iϩ 1.05σ_R

1

5

7

5

7

5

6

7

0.982 164.4

0.991 541.2

0.991 575.4

0.997 349.8

0.975 77.2

0.974 110.2

0.992 193.6

0.996 468.9

1.02 × 10^Ϫ30.173 0.099

2.73 × 10^Ϫ60.112 0.192

2.35 × 10^Ϫ60.128 0.244

2.85 × 10^Ϫ30.087 0.037

6.37 × 10^Ϫ40.209 0.131

1.84 × 10^Ϫ30.208 0.075

6.70 × 10^Ϫ40.126 0.080

1.80 × 10^Ϫ50.083 0.110

3.95 × 10^Ϫ20.089 0.045

ϩ

ؒ

ArN₂/ArN₂exchange

Fe(EDTA)^2Ϫ, pH 9, H₂O

0.78

0.53

1

1.05

1.33

1.53

1.25

0.45

1.84

ϩ

TMPD, pH 4.5, H₂O

PhCHOH radical, H₂O

ϩ

0.59σ_p

BA

Semiquinone radical anion, pH 7, H₂O

Hydroquinone anion, pH 7, H₂O

Phenothiazine, acidiﬁed MeCN

2.38σ_Iϩ 2.50σ_Ro

3.09σ_Iϩ 4.10σ_Ro

2.39σ_Iϩ 5.03σ_R

< 2.72σ_Iϩ 3.41σ_R

< 1.54σ_Iϩ 0.69σ_R

< 0.86σ_Iϩ 1.58σ_R

4ⁱ0.998 320.5

Cu^ϩ_aq, H₂O

(i)^g

6

ϩ

(ii)^h

6

CitratoCu^, H₂O

6ⁱ

^a–fSee corresponding footnotes to Table 4. ^g4-Substituted diazonium ions. ^h3-Substituted diazonium ions. ⁱMinimal data set but signiﬁcance

acceptable.

diazonium ions are characterised by λ ≥ 1 and ρ_Iof magnitude

2.5 0.4.

Entries 6 to 11 relate to azo-coupling reactions of substi-

(v) Comparative correlation analyses of electrophilic reactions of

diazonium ions

Preparatory to addressing the nature of the association process

it occurred to us to enquire whether correlation analysis which

enables patterns characteristic of the type of process to be

recognised, would allow distinction between inner- and outer-

sphere reduction processes for diazonium ions. Data sets from

the literature of selected electrophilic reactions of diazonium

ions were subjected to correlation analysis and the ﬁndings,

given in Table 5, judged by the same criteria as those in Table 4.

Of necessity, some of the original data sets have been reduced

because of the unavailability of Taft dual substituent param-

eters for a few of the substituents used and because con-

sideration is restricted to mono-substituted systems. Data for

3-substituents were also removed which ensured that compar-

isons between single and dual parameter correlations for sub-

sets relating to 4-substituents are more reliably made but, where

there were suﬃcient 3-substituted diazonium ions to constitute

an adequate subset for treatment, these were analysed separ-

ately. The correlations are compared in terms of ρ_Ifor when a

dual parameter expression gives the best account of the data, ρ_I

tuted diazonium ions with naphthalenic substrates. Entry 8 is

the only correlation in the group best described in terms of a

single substituent parameter but this correlation is consider-

52

ˇ

ably less precise than the others and Steˇrba and co-workers

noted aberrant behaviour of the 4-cyano- and 4-nitro-

benzenediazonium ions; unfortunately, exclusion of these

points results in a data set too small for the dual parameter

treatment. However, the precise correlations of the other

entries in the group show that the reaction of N_βof the

diazonium ions with these aromatic π-systems is characterised

by reaction constants of about 4, that the coupling rates are

more susceptible to the inductive eﬀects of 3-substituents

than those of 4-substituents [i.e. ρ_I(meta)> ρ_I(para)], and that

substituents’ mesomeric eﬀects are less important than their

inductive eﬀects (λ < 1) especially, as expected, for 3-substi-

tuted series (entries 7, 9 and 11). Most of the correlations

ϩ

involve σ_Ras the descriptor of resonance eﬀects which is

unsurprising as σ^ϩ, from which it is derived, was deﬁned for

electrophilic aromatic substitution;⁶²those which do not involve

σ_R^ϩrefer to 3-substituted series where the extent of conjugation

is reduced relative to the 4-substituted series. The diﬀerent

ranges of ρ_Iand λ values evident between two classes of reac-

tion of diazonium ions, both involving electron-pair donation

to N_β, led to the hope that inner- and outer-sphere electron

transfer processes might be distinguishable in terms of

correlation analysis.

is the coeﬃcient of the eﬀective substituent parameter [σ_eff

=

(σ_Iϩ λσ_R)]; when a single explanatory variable gives the best

account of a data set, λ (= ρ_R/ρ_I) is given as 1 since Hammett’s

σ_p= (σ_Iϩ σ_R^BA), σ_p^ϩ= (σ_Iϩ σ_R^ϩ) etc.

Entries 1 and 2 relate to the rates of addition of OH^Ϫand

CN^Ϫto the β-nitrogen atom of 4-substituted diazonium ions to

give Z-azo-adducts.⁵⁰The latter nucleophile adds reversibly and

entry 3 applies to the equilibrium constants. Similarly, entries 4

and 5 relate to the addition rates and equilibrium constants for

the reversible association of 4-substituted diazonium ions with

benzenesulﬁnate anions.⁵¹The reaction constants for the

equilibria are, of course, larger than those for the correspond-

ing addition rates owing to the contribution from the reaction

constants of the reverse steps. It is evident from entries 1 to 5

that rates of addition of simple anionic nucleophiles at N_βof

Entries 12 to 22 are all concerned with single electron trans-

fers to diazonium ions from a variety of donors and entries 23

to 25 inclusive give, for comparison, the limiting correlations of

the inner-sphere electron transfers from Cu^deduced in this

paper [inequalities (28)–(30)]. Entry 12 relates to the correlation

of polarographic half-wave potentials, E_1/2, found for diazo-

nium ions in sulfolane solution by Elofson and Gadallah⁵⁴who

plotted values for fourteen 3- and 4-substituted diazonium ions

1144

J. Chem. Soc., Perkin Trans. 2, 2002, 1135–1150

View Online

ϩ

versus σ, or σ_pfor 4-substituents having ϩM eﬀects. We ﬁnd

of entry 13 may be better estimates of standard reduction

that for the reduced set of nine 4-substituted diazonium ions

that we have considered, the best correlation is also obtained

using σ or σ_p^ϩ, as appropriate, and with negligible change in the

reaction constant. In order to obtain a dimensionless gradient

for comparability with other reaction constants given in Table 5

we divide the gradient of 0.225 V by (RTln 10)/F, i.e. 0.0592 V

and obtain 3.80. There is a considerable scatter in this data set

by comparison with entry 13 which relates to a more recent

polarographic investigation of diazonium ions in nitromethane

solution.⁵⁵

potentials than those of entry 12.

The large span of exchange rate constants indicates a

requirement for a substantial structural and orbital reorganis-

ation on receipt of an electron by a diazonium ion which is very

sensitive to the nature of substituents. X-Ray crystallographic

studies have shown diazonium ions to be linear in their CNN

moiety^64,65whereas EPR coupling constants demonstrate aryl-

diazenyl radicals to be bent.^66,67Theoretical work by Glaser’s

group^68–70has indicated that a diazonium ion is more accurately

represented by 3 than by the traditional Lewis structure 4. (In 3

an aryl cation is coordinated datively at C(1) by a dinitrogen

molecule which, though strongly polarised, carries little nett

charge, the overall cationic charge being dispersed inductively

to hydrogen atoms). The electron-deﬁcient orbital of the

phenyldiazenyl radical is a σ-orbital largely conﬁned to the

diazo moiety, 5.^66,67Clearly, when a diazonium ion receives an

electron, there are substantial changes in both geometry and

electronic structure. Recent work which reports calculation

Entries 14 and 15 relate to reductions of diazonium ions

where the Fe^reductants were chosen by Doyle and co-

workers,⁵⁶on account of their being substitution inert, to react

by outer-sphere mechanisms. The absolute rates of reduction

found for these two reductants and their known self-exchange

rates were used to extract, via the Marcus cross-relation

[eqn. (20) and (21)], self-exchange rates for diazonium/diazenyl

ϩ

ؒ

couples, ArN₂/ArN₂(entry 16). In doing so, Doyle’s group

took the E_1/2values of entry 12 to represent standard reduc-

tion potentials; furthermore, because their rate constant for

4-methoxybenzenediazonium ion did not correlate well with

E_1/2fortheion(0.14V), theyinferredthelattertobeincorrectand

adopted a value of 0.24 V. In fact, the E_1/2values found for solu-

tion in sulfolane⁵⁴are correlated with those in nitromethane⁵⁵by

theexpressionE_1/2^sulf=(0.994E_1/2^nitroϩ0.112)fromwhichitcanbe

inferred that Elofson and Gadallah’s value of E_1/2in sulfolane

was, probably, essentially correct. For these reasons, the value

of the reaction constant of entry 16 must be treated with

circumspection.

ϩ

ؒ

of the barriers to self exchange in ArN₂/ArN₂couples by a

variety of theoretical methods supports a strong substituent

dependence.⁷¹

ϩ

ؒ

Nevertheless, the reaction constant for ArN₂/ArN₂self-

exchange can be inferred to be large as follows. The Leﬄer

index,⁶³α, relates the free energy change of activation for a

reaction, ∆G^‡, to the overall free energy change of reaction,

∆G^o, via α = d∆G^‡/d∆G^o, where α may take values between 0

and 1. The reaction constant for a set of rate data is given by

ρ^rate= d(log₁₀k)/dσ = Ϫ(RTln 10)^Ϫ1d∆G^‡/dσ and that for the

corresponding overall reactions considered as formal equilibria

is ρ^equil= d(log₁₀K)/dσ = Ϫ(RTln 10)^Ϫ1d∆G^o/dσ whence it fol-

lows that ρ^rate= αρ^equil, i.e. for a reaction series, the upper limit

on the reaction constant for forward rates is determined by the

reaction constant for the equilibria. The upper limit on reaction

constants relating to the rates of outer-sphere reduction of

diazonium ions to aryldiazenyl radicals is therefore determined

by the reaction constant for the overall reduction of diazonium

ions to diazenyl radicals. For a self-exchange, the rates of

the forward ( f ) and backward (b) steps are equal and both

participants bear the same substituent. Hence, for a set of

ϩ

ؒ

The unsubstituted couple PhN₂/PhN₂is therefore expected

to exhibit a self-exchange rate much reduced in comparison

with that ca. 10⁸dm³mol^Ϫ1s^Ϫ1shown by the couple PhCN/

PhCN^Ϫwhich, though isoelectronic, does not undergo substan-

tial structural or orbital reorganisation⁷²and, given a reaction

ϩ

ؒ

constant > 5 for self-exchange in ArN₂/ArN₂couples, it

follows that the self-exchange rates of those with electron

donating substituents must be comparatively slow.

Entries 17 to 21 represent other outer-sphere electron trans-

fers to diazonium. Entry 22 refers to our previous work

examining electron transfer between various phenothiazines

and diazonium ions.⁶¹The main kinetic study used 4-methoxy-

benzenediazonium ion and we suggested a bonded mechanism

because, in the cases of slower reacting phenothiazines, initial

absorbance due to intermediates was observed. This interpret-

ation has been questioned on the grounds that the rates of

reaction of phenothiazine and its 10-methyl derivative could be

calculated fairly closely by using the Marcus cross relation with

assumed⁷³or suspect⁵⁶self-exchange rates; the initial absorb-

ance was suggested to be due to a charge transfer complex

formation.⁵⁶Such cannot be the case: organic charge transfer

complexes are formed rapidly at essentially the diﬀusion-

controlled rate⁷⁴whereas the observed absorbance developed

over intervals of up to 100 s, dependent on the substrate. Fur-

thermore, we also reported that diazonium ions which are more

ϩ

ؒ

self-exchanges such as those in ArN₂/ArN₂,

ρ

^exch= (ρ_f^rateϪ ρ_b^rate) = [ρ_f^rateϪ

(Ϫρ_f^rate)] = 2ρ_f^rate= 2αρ^equil= ρ^equil

,

(34)

since the symmetry requires α to be 0.5. The reaction constant

for self-exchange rates should therefore equal that for overall

reduction. From this reasoning it follows that either the E_1/2

values of entry 12 are not valid measures of standard reduction

potentials as their adjusted reaction constant (3.80) is not larger

than all the reaction constants for outer-sphere reduction rates:

that for reduction by ferrocyanide (4.56) exceeds it; or, ferro-

cyanide does not react by an outer-sphere mechanism. Assum-

ing that reduction by ferrocyanide does occur by an outer-

sphere process, the reaction constant for self-exchange must

exceed 4.56. It is likely, therefore, that over the range of sub-

stituent character from 4-MeO to 4-NO₂the self-exchange rates

of diazonium ions probably span at least ﬁve orders of magni-

tude although their absolute values remain uncertain. The

adjusted reaction constant for E_1/2values found in nitro-

methane (5.36) suggests that values interpolated from the data

ϩ

electrophilic than 3-MeOC₆H₄N₂behaved diﬀerently, pro-

ducing a high concentration of phenothiazine radicals immedi-

ately on mixing. Referring to eqn. (20) and (21), it seems to us

that if diazonium ions have self-exchange rates which span ﬁve

orders of magnitude and standard reduction potentials which

show an equivalent strong substituent dependence, the possi-

bility arises for a given reductant, that the mechanism might

change from outer-sphere for electron-deﬁcient diazonium ions

having relatively rapid self-exchange rates and high reduction

potentials to inner-sphere for electron-rich diazonium ions

having slow self-exchange rates and low reduction potentials.

Kochi and co-workers⁷⁵have described a gradation of mechan-

ism in electron transfers from tetraalkyltins to tetracyanoethene

or to hexachloroiridate() which is dependent on the steric

J. Chem. Soc., Perkin Trans. 2, 2002, 1135–1150

1145

View Online

hindrance oﬀered by the alkyl groups. Although simple 3- and

4-substituted diazonium ions will not present signiﬁcantly dif-

ferent steric aspects to a given reductant, it seems reasonable to

suggest for them a continuum of activated complexes for elec-

tron transfer arising from the interplay of driving force and

intrinsic barrier.

It thus seems possible that phenothiazine (entry 22), a

relatively poor reductant (E = 0.696 V vs. NHE)⁷⁶should

undergo inner-sphere reaction with electron-rich diazonium

ions, its fast self-exchange rate⁷²not compensating suﬃciently

for the combination of slow self-exchange rate and weak driv-

ing force in the cases of diazonium ions with electron donating

substituents. The electron transfer mechanism may then change

to outer-sphere for diazonium ions more electrophilic than

ϩ

3-MeOC₆H₄N₂

.

Aquacopper() (entry 23), despite being thermodynamically a

relatively good reducing agent (E = 0.159 V vs. NHE),⁴⁴also

undergoes inner-sphere electron transfer with diazonium ions:

when the latter have electron donating substituents, both react-

ants have low self-exchange rates and hence there is a high

intrinsic barrier to electron transfer. When the diazonium ions

have electron withdrawing substituents, their self-exchange

rates and reduction potentials are higher but still insuﬃcient to

compensate the very low self-exchange rate of Cu^ϩaq⁴³and

electron transfer remains of inner-sphere character though no

longer the rate determining step (association becomes rate

determining).

Given a continuum of transition states between inner- and

outer-sphere extrema, sharply distinguished correlation anal-

yses are perhaps not to be expected for entries 13 to 24. How-

ever, a gradation in the values of λ is evident. Entries 12 to

19 inclusive have λ ≤ 1 and entries 22 to 25 which we believe

to relate to inner sphere processes have λ > 1. It is interesting

that entry 21 has a value λ which exceeds 1 although it is

believed to refer to an outer-sphere transfer for the full range of

diazonium ions. It is perhaps signiﬁcant that rates calculated by

the Marcus cross relation for reaction of hydroquinone with

diazonium ions (albeit using the questionable self-exchange

rates of entry 16) agree better with experiment for electron-

deﬁcient diazonium ions than for ions bearing electron-

donating substituents which are wrong by more than two orders

of magnitude.⁶⁰Conceivably, for reductions of diazonium ions

by hydroquinone the transition states for both mechanisms

have conﬁgurations which are similar.

Fig. 7 Schematic representation of reaction proﬁles for electron

transfer reactions.

occurs later in the reaction coordinate than in Fig. 7a and the

crossing points (z) and (y) are such that the former occurs at

the higher value of G, the conversion of E into P (i.e. the

process we have called association) is then rate determining

and the situation corresponds to that observed for the reduc-

tion of diazonium ions with ϪM substituents by aqua-

copper(). Here, as the substituent sequence is traversed towards

stronger electron acceptance, the values of G at intersection

of the curves for E and P (z) fall but remain higher than

those (y) of the corresponding intersections of the curves for

P and S.

In (iii) above we identiﬁed the intercepts of eqn. (14) and (18)

as log₁₀(k ^ϪA_H/k ^ET_H) for the simple and citrate-promoted

hydroxylations, respectively, and it was also noted that the

intercept for eqn. (18) is small and, within the experimental

uncertainty, could be zero (cf. Table 4 entry 15). A zero inter-

cept could occur in two circumstances. The ﬁrst is that, fortuit-

ously, k ^ET_H= k ^ϪA_H, i.e. the precursor complex in an inner-sphere

reduction of the unsubstituted diazonium ion could react in the

forward and reverse directions with equal probability; in terms

of Fig. 7a or b the intersections (z) and (y) now occur at the

same value of G. The second circumstance is that the inter-

sections (x), (y) and (z) lie close together (Fig. 7c) so that the

energy minimum corresponding to P is shallow and close to the

maximum produced by the intersection of the curves of E and

S. In this situation, the diﬀerence in magnitudes of k ^ϪAand k ^ET

for all substituents decreases, log₁₀(k ^ϪA/k ^ET) tends to 0 and, as

x, y and z converge, inner-sphere association resembles ever

more closely the activation for outer-sphere electron transfer.

Such convergence can thus explain the gradation from inner- to

outer-sphere mechanisms suggested above. For suﬃciently shal-

low energy minima for P, the distinction between inner- and

outer-sphere processes would be diﬃcult to discern in terms of

reaction rate or correlation analysis. The reductions of diazo-

nium ions by some phenothiazines and by hydroquinone may

represent this condition.

(vi) Outer-sphere reduction of electron-deﬁcient diazonium ions

by citratocopper(I)

Fig. 7 represents schematically the relative positioning of the

free energy curves corresponding to the encounter (E), pre-

cursor (P) and successor (S) complexes in electron transfer

reactions. In Fig. 7a and b, E and S are well separated in the

reaction coordinate with the consequence that the right- and

left-hand limbs, respectively, of their curves intersect (x) at a

high value of G. There is, accordingly, a high activation free

energy barrier to the outer-sphere electron transfer which

would convert E directly into S. If a precursor complex P is

interposed between E and S, relatively early in the reaction

coordinate, such that the crossing point (z) of the curves of E

and P occurs for a lower value of G than that (y) of the

curves of P and S (Fig. 7a), the conversion of P into S (i.e.

electron transfer) is rate determining and the situation corre-

sponds to that observed for the reduction of 4-substituted

diazonium ions by chlorocuprate() complexes or the reduction

of diazonium ions having 3- or 4-substituents of ϩM

character by aquacopper(). Here, as the substituent sequence

is traversed from electron donating towards electron accept-

ing, the values of G at intersection (y) of the curves of P and

S fall but each remains higher than that of the corresponding

intersection (z) of E and P. In Fig. 7b the position of P

1146

J. Chem. Soc., Perkin Trans. 2, 2002, 1135–1150

View Online

We suggest that the reduction of diazonium ions bearing

4-substituents of ϪM character by the citratocopper() reduc-

tant may also be represented by Fig. 7c; thus eqn. (17) and (18)

do not denote a change of rate determining step within an

inner-sphere electron transfer mechanism [as in the case of

reduction by aquacopper()] but rather, a change of mechanism

from inner-sphere reduction of diazonium ions with ϩM sub-

stituents to evident outer-sphere reduction of diazonium ions

with ϪI/ϪM substituents. The multidentate citrate ligand pro-

vides the metal centre with a more rigid inner coordination

sphere than occurs in the aqua-complex (the separation of E

and S in the reaction coordinate is relatively reduced) with the

consequence that the cost in inner-sphere reorganisation energy

on electron transfer is reduced. As a result, the rate of self-

exchange in the citratocopper redox couple will be increased

relative to that in the aquacopper couple. This, together with

higher self-exchange rates for diazonium ions with electron

withdrawing substituents, will result in a signiﬁcant lessening of

the intrinsic barrier to outer-sphere electron transfer [the value

of G at intersection (x) of the E and S curves is signiﬁcantly

reduced]. It seems therefore that the promotional eﬀect of

citrate on the reduction step of Sandmeyer hydroxylation arises

not only from an enhanced driving force of the reaction (lower

syn-approach are consequently small in comparison with the

N_βlone-pair–aryl ring interactions of the anti-approach. By

contrast, if the transition state for addition is late, the forming

bond is well developed and hence shorter and nucleophile–aryl

ring interactions are corresponding larger in the conﬁguration

leading to the Z-adduct and the E-stereoisomer then becomes

the preferred primary addition product (e.g. azo-coupling). The

diﬀerence in magnitude of the reaction constants between

entries 1 and 2 of Table 5, on the one hand, and entries 6–11 on

the other, is also a consequence of the relative earliness of the

transition states for addition of OH^Ϫand CN^Ϫ.^77,78

The reaction constants found for the association steps in the

reaction of 4- and 3-substituted diazonium ions with aqua-

copper(), respectively 1.91 and 1.37 [cf. eqn. (14) and (16)], are

thus consistent with a very early transition state for nucleophilic

attack by a bridging ligand at N_β. The requirement of σ_Ias the

only explanatory variable in eqn. (14) and (16) suggests that the

direction of approach of the reductant lies in the molecular

plane of the diazonium ion (and hence is orthogonal to the

π-system through which the substituent’s mesomeric eﬀects are

exerted); this, too, is consistent with a very early transition state

for the addition of the reductant to N_βas any signiﬁcant steric

interaction of the reductant with the syn-aryl ring would be

expected to deﬂect its line of approach out of the molecular

plane. We therefore postulate that Sandmeyer precursor com-

plexes are Z-adducts of the reductant at the β-nitrogen atom of

the diazonium ion, the linkage between the two moieties being a

bridging ligand. In the case of aquacopper() as the reductant,

we suggest the bridge is likely to be a hydroxide ligand rather

than a water ligand as the loss of a proton would both increase

the nucleophilicity of the bridging oxygen atom and provide a

neutral, rather than cationic, reaction partner for the diazo-

nium ion.

E ^{–– –}for the reducing couple) but also from a reduction of the

᭺

intrinsic barrier to electron transfer enabling a change from

inner- to outer-sphere mechanism. The ﬁnding of such a

change according to the type of substitution in the diazonium

ion lends further credence to the idea of a continuum of

electron transfer transition states between formal inner- and

outer-sphere extremes.

(vii) Association as the rate determining step

The term ‘association’ requires clariﬁcation. The implication of

Scheme 4 is that it constitutes that reorganisation of the

encounter complex to the precursor complex which enables

inner-sphere electron transfer to occur and, if ‘inner-sphere’ is

to have its usual inorganic connotation⁴⁸in respect of the

copper() complex, either the diazonium ion should enter into

coordination of the copper or a ligand coordinated to copper

should act as a nucleophile towards the diazonium ion and

serve as the bridge for relay of the electron. Whatever is the

nature of the association process, its dependence on only the

inductive eﬀects of substituents [cf. eqn. (14), (16) and (18)]

requires explanation and the product of the association process

should be such as to facilitate electron transfer in a process

which is strongly sensitive to the mesomeric eﬀects of the

substituents in the diazonium moiety.

(viii) The activation of electron transfer

In the vibrational ground state of the Z-adduct proposed for

the precursor complex, it is expected that there will be a signiﬁ-

cant dihedral angle, LN_βN_αC(1), (where L is the bridging

ligand) on account of the steric interactions of the bulky

attachments in Z-conﬁguration on the azo function. The prin-

ciple of microscopic reversibility requires that the heterolytic

dissociation of the adduct directly into the components from

which it was formed must be associated with a vibration which

stretches the L–N_βbond and closes the dihedral angle (Scheme

5a). An alternative heterolysis of the precursor complex may

also be envisaged which involves the breaking of the L–Cu^

bond. Such a dissociative process would eliminate Z-ArN᎐NL

᎐

Many transition metal complexes are known in which the

metal centre is ligated formally by a diazonium ion via one or

both of its nitrogen atoms but none has been reported to

as a composite ‘ligand’ from the inner coordination sphere of

the metal ion. Z-ArN᎐NL would then itself be expected readily

᎐

to lose L^Ϫgiving the diazonium ion. Thus the formation of

the precursor complex, formulated as suggested, would be

expected, one way or another, to be rapidly reversible to give the

species which combined to produce it, as required by the kinetic

analysis (cf. Scheme 4). Another vibration of the precursor

complex during its limited lifetime must cause activation of the

electron transfer; we suggest this to be rotation about the N–N

bond (Scheme 5b).

ϩ

involve copper.⁹Comparing PhN₂with known ligands of Cu^

such as CO, MeCN and PhNC, the diazonium ion is expected

to be a poorer σ-donor. Also, on account of the d¹⁰conﬁgur-

ation of Cu^, the d-orbitals are contracted and back-donation

from metal to ligand is also expected to be weak. These expec-

tations are particularly true for the Glaser structure,^68–703, in

which N_βconstitutes the positive end of the N–N dipole and

the diazo moiety experiences back-donation of π-electron dens-

ity from the aromatic ring. Until evidence is found to the con-

trary, we presume these factors preclude the ligation of Cu^by

diazonium ions in any of the ways found for other transition

metals and discount ‘association’ involving such ligation.

Simple nucleophiles such as OH^Ϫor CN^Ϫadd to diazonium

ions at N_βto give Z-adducts as kinetically controlled products

which may undergo subsequent isomerisation to the thermo-

dynamically stabler E-isomers. Zollinger^77,78has explained the

stereoselectivity of addition in terms of early transition states in

which the separation of the incoming nucleophile and N_βis

relatively large and the nucleophile–aryl ring interactions of the

As a result of the reduced N–N π-overlap in the precursor

complex arising from the twist due to steric interactions, N–N

bond rotation should be easier than that about a sterically

unencumbered N–N π-bond. Opening of the LN_βN_αC(1)

dihedral angle towards 90Њ would have the consequence of fur-

ther uncoupling the π-bond and bringing of the N_βL σ-bond

into eclipse with the singly occupied p_zorbital on N_α. This

would allow a β-scission producing an aryldiazenyl radical and

a change of oxidation state of the copper (Scheme 5b). There

are grounds for expecting N–N bond rotation, and hence homo-

lysis, to be particularly sensitive to the nature of the ring substi-

tution. The Z to E isomerisation of diazoate anions, ArN᎐NO^Ϫ

᎐

J. Chem. Soc., Perkin Trans. 2, 2002, 1135–1150

1147

View Online

Scheme 5

is strongly substituent dependent, exhibiting a dichotomy

increased proportion. In 6 the N–N bond rotation necessary for

between ϩM and ϪM substituents comparable with that found

radical formation is inhibited by the fact that the π-bond is

stronger than those in Z-adducts, due to lack of steric eﬀects,

and also because the E-conﬁguration is additionally stabilised

against isomerisation by the H-bond. Evidently, the activation

energy for access to the albeit stabilised ascorbyl radical from 6

is too high for the reaction to be feasible except perhaps for the

4-nitro derivative which has been reported to decompose on

heating²⁹and where the eﬀect of the substituent would be to

reduce the N–N bond order and facilitate rotation. The form-

ation of an increased proportion of stable E-adduct also

explains the low homolytic yields (< 50%) in instances where

some reduction of diazonium ions by ascorbate does occur.²⁸

for the hydroxylation reactions reported here.⁷⁹Aryl substitu-

ents with ϪM eﬀects which conjugate with the N᎐N double

᎐

bond reduce its bond order and so facilitate the rotation about

the N–N axis which leads to isomerisation. Aryl substituents

with ϩM eﬀects such as 4-Cl do not so reduce the barrier to

rotation and a diﬀerent mechanism obtains, i.e. inversion of N_β

in the diazohydroxide, ArN᎐NOH, which has a considerably

᎐

higher activation energy.⁸⁰Diazoate anions and diazohydrox-

ides have no propensity for direct N–O bond homolysis but

other O-linked compounds such as azo-ethers, ArN᎐NOArЈ

᎐

and diazoanhydrides, ArN᎐NON᎐NAr do homolyse.⁸¹Such

᎐

homolyses can be understood in terms analogous to those

proposed for a Sandmeyer precursor complex, i.e. N–N bond

rotation in a sterically congested Z-adduct, the only diﬀerence

being that the β-scission now produces a second, stabilised

ؒ

radical (respectively, Ar–O and ArN᎐NO ) in addition to the

᎐

diazenyl radical. In the Sandmeyer precursor complex, the

capacity of the copper to change its oxidation state provides

a particularly eﬀective way of stabilising the ‘radical’ co-

produced with the aryldiazenyl radical.

If the extent of N–N bond rotation required to give suﬃcient

orbital overlap for β-scission to occur should be less when there

is a relatively strong driving force than when the driving force is

weak, the occurrence of precursor complexes manifesting low

barriers to electron transfer can be understood (cf. Fig. 7b), e.g.

the reductions of diazonium ions bearing ϪM substituents by

aquacopper() where association rather than electron transfer is

rate determining. Conversely, if β-scission of the Cu^–L bond

should be relatively diﬃcult despite eﬀective eclipsing of the p_z

orbital on N_α, the occurrence of diﬀerent electron transfer rates

for reductants having similar reduction potentials can be under-

stood. In the foregoing paper we reported that cyanocuprate()

complexes reduce diazonium ions much more slowly than

chlorocuprate() complexes of similar reduction potential. In

the former reductants the bridging ligand is diatomic and the

electron transfer, formally from the metal to nitrogen in order

to maintain the octet of the latter on β-scission, must involve

the intervening carbon atom. We suggest the additional atom in

the path of the transferred electron inhibits β-scission and

reduces the apparent ‘conductivity’ of the cyanide ligand

relative to that of a chloride ligand.¹

The bond-rotation mechanism of electron transfer also

allows a rationalisation of the initially surprising observation

that ascorbate does not readily reduce diazonium ions but

instead may give stable diazo-ether adducts, e.g. 6.²⁹If the

interactions which normally ensure the kinetically preferred

formation of the Z-adduct on addition of a nucleophile to a

diazonium ion should be attenuated in the case of ascorbate by

the possibility of hydrogen bonding between N_αand the ascor-

bate 2-OH group, the E-adduct 6 could be formed directly in

Conclusions

Correlation analysis of relative reaction rates reveals two dis-

tinct types of behaviour in the reduction of diazonium ions by

aquacopper(). Kinetic analysis indicates these to correspond,

within an inner-sphere process, to rate determining electron

transfer for the unsubstituted diazonium ion and those bearing

ϩM substituents and to rate-determining association of the

reactants for those bearing ϪM substituents. In the presence of

citrate two types of behaviour are again observed: rate deter-

mining electron transfer within an inner-sphere process for

diazonium ions bearing ϩM substituents but for those bearing

ϪM substituents the behaviour is now consistent with a change

to an outer-sphere reduction mechanism. When mixed chloro-

cuprate() complexes reduce diazonium ions, inner-sphere rate

determining electron transfer is observed across the whole

substituent range.

A critical survey of reductions of diazonium ions reported in

the literature leads to the conclusion that there exists a grad-

ation of mechanism between formal inner- and outer-sphere

extremes which arises because the reduction potentials and

self-exchange rates of diazonium ion/diazenyl radical couples

manifest a strong substituent dependence.

The precursor complex in inner-sphere reductions of diazo-

nium ions is suggested to be the Z-adduct of the reductant at

N_βof the ion; it is short lived on account of its ready heterolysis

back to initial components. The activation for electron transfer

is suggested to be a rotation and uncoupling of the N–N bond

which allows a homolytic β-scission producing an aryldiazenyl

radical and either a second stabilised radical or, in the case of a

1148

J. Chem. Soc., Perkin Trans. 2, 2002, 1135–1150

View Online

Table 6 Yields and melting/decomposition points of diazonium

tetraﬂuoroborates

ten minutes of stirring, the reaction mixture was extracted with

ethyl acetate (25 cm³) containing internal standard (4-bromo-

benzonitrile, 0.02 mol dm^Ϫ3). The extract was analysed by GC,

performed on an FFAP column (temperature program 60 ЊC for

5 min then 16 ЊC min^Ϫ1until 230 ЊC). The only detectable prod-

ucts in all cases were the expected phenols or aryl chlorides.

Their yields were determined from the peak areas using

response factors previously determined for each product by use

of authentic materials.

Substituent

Yield (%)

Mp/ЊC

Lit. mp/ЊC

Ref.

4-OMe

4-OPh

4-Me

H

4-F

95

32

97

99

90

71

69

68

76

97

85

53

51

53

77

50

62

83

141–142

177–178

101–102

96–98

161–162

133–134

105–106

85–87

105–106

140–141

152–153

82–83

142

83

84

83

85

83

86

87

88

89

83

90

91

83

159–160

109–110

108–110

161–161.5

136–137

116–117

94–95

4-Cl

4-CF₃

4-COMe

4-CO₂Et

4-CN

4-NO₂

3-OMe

3-Me

Acknowledgements

93–95

150

We thank Great Lakes Fine Chemicals and the University of

York for research studentships (J. R. J. and A. B. T.), A. A. J.

Hendrickx for assistance with the experiments on disubstituted

diazonium ions and A. C. Whitwood, B. C. Gilbert and J. E.

McGrady for helpful discussions.

157–158

87–88

97–101

146–148

146–147

168–170

—

96–97

3-Cl

142–143

149–150

179–180

202–204

203–205

3-CF₃

3-NO₂

3,5-(CF₃)₂

3,5-(NO₂)₂

References

203

1 Part 6. P. Hanson, A. B. Taylor, S. C. Rowell, P. H. Walton and A. W.

Timms, J. Chem. Soc., Perkin Trans. 2, 2002, preceding paper

(DOI: 10.1039/b200747a).

Sandmeyer reduction step, a concomitant change of oxid-

ation state of the copper. The mechanism of bond rotation and

β-scission in an initial adduct also allows the low reactivity of

ascorbate and cyanocuprate in the reduction of diazonium ions

to be rationalised.

2 W. A. Waters, J. Chem. Soc., 1942, 266.

3 J. K. Kochi, J. Am. Chem. Soc., 1957, 79, 2942.

4 H. Zollinger, Diazo Chemistry I, VCH, Weinheim, 1994, ch. 8.

5 C. Galli, J. Chem. Soc., Perkin Trans. 2, 1984, 897.

6 W. A. Cowdrey and D. S. Davies, J. Chem. Soc., 1949, S48.

7 W. A. Cowdrey and D. S. Davies, Chem. Soc. Rev., 1952, 6, 358.

8 C. Rømming and K. Waerstad, J. Chem. Soc., Chem. Commun.,

1965, 299.

9 D. Sutton, Chem. Rev., 1993, 93, 995.

10 E.-L. Dreher, P. Niederer, A. Rieker, W. Schwarz and H. Zollinger,

Helv. Chim. Acta, 1981, 64, 488.

11 N. A. Porter, L. J. Marnett, C. H. Lochmüller, G. L. Closs and

M. Shobataki, J. Am. Chem. Soc., 1972, 94, 3664.

12 N. A. Porter, G. R. Dubay and J. G. Green, J. Am. Chem. Soc., 1978,

100, 920.

13 T. Suehiro, S. Masuda, R. Nakausa, M. Taguchi, A. Mori, A. Koike

and M. Date, Bull. Chem. Soc. Jpn., 1987, 60, 3321.

14 S. F. Lincoln and A. E. Merbach, Adv. Inorg. Chem., 1995, 42, 1.

15 A. Katagiri, S. Yoshimura and S. Yoshizawa, Inorg. Chem., 1981, 20,

4143.

Experimental

(i) Diazonium tetraﬂuoroborates

Diazonium tetraﬂuoroborates were obtained from the corre-

sponding anilines using the following general method (scaled

down in some instances).⁸²The appropriate aniline (0.1 moles)

was dissolved in a mixture of 50% ﬂuoroboric acid (34 cm³) and

distilled water (40 cm³). After cooling to 0 ЊC, sodium nitrite

(6.9 g) in 15 cm³of distilled water was added in 2 cm³portions.

The mixture was stirred for 30 minutes and the thick precipitate

was collected and re-dissolved in acetone. The diazonium

tetraﬂuoroborate was then precipitated by the addition of

diethyl ether. The yields and melting/decomposition points

observed are given in Table 6.

16 A. Katagiri, S. Yoshimura and S. Yoshizawa, Inorg. Chem., 1986, 25,

2278.

17 J. K. Kochi, J. Am. Chem. Soc., 1956, 78, 1956.

18 C. S. Rondesvedt, Org. React. (N. Y.), 1960, 11, 189.

19 C. S. Rondesvedt, Org. React. (N. Y.), 1976, 24, 225.

20 C. Galli, J. Chem. Soc., Perkin Trans. 2, 1981, 1459.

21 C. Galli, Chem. Rev., 1988, 88, 765.

22 T. Cohen, A. G. Dietz and J. R. Miser, J. Org. Chem., 1977, 42, 2053.

23 P. Hanson, R. C. Hammond, P. R. Goodacre, J. Purcell and

A. W. Timms, J. Chem. Soc., Perkin Trans. 2, 1994, 691.

24 P. Hanson, R. C. Hammond, B. C. Gilbert and A. W. Timms,

J. Chem. Soc., Perkin Trans. 2, 1995, 2195.

(ii) Phenols and aryl halides

The phenols and aryl chlorides required for product authentica-

tion and calibration purposes were all commercial materials as

were the 4-bromobenzonitrile used as internal standard for GC

and other reagents.

25 B. C. Gilbert, P. Hanson, J. R. Jones and A. W. Timms, European

Patent 0 596 684 B1, 29/01/1997.

(iii) Competitive hydroxylation and chlorination

26 P. Martinez, J. Zuluaga and C. Sieiro, J. Phys. Chem. (Leipzig),

1984, 265, 1225.

27 A. P. Moravskii, Y. N. Skurlatov, E. V. Shtamm and V. F. Shuvalov,

Bull. Acad. Sci. USSR, Div. Chem. Sci., 1977, 61, 49.

28 U. Costas-Costas, E. Gonzalez-Romero and C. Bravo-Díaz, Helv.

Chim. Acta, 2001, 84, 632.

29 M. P. Doyle, C. L. Nesloney, M. S. Shanklin, C. A. Marsh and

K. C. Brown, J. Org. Chem., 1989, 54, 3785.

30 A. Günter and A. Zuberbühler, Chimia, 1970, 24, 340.

31 E. R. Stille and P. Wikberg, Inorg. Chim. Acta, 1980, 46, 147.

32 D. Mastropaolo, D. A. Powers, J. A. Potenza and H. J. Shugar, Inorg.

Chem., 1976, 15, 1444.

33 J. Heyrovský and P. Zuman, Practical Polarography, Academic

Press, London, 1968, p.188.

34 A. Madinzi, M. El Amane and H. Atmani, Ann. Chim. Sci. Mat.,

2000, 25, 179.

35 S. Ahrland and J. Rawsthorne, Acta Chim. Scand., 1970, 24, 157.

36 R. W. Ramette, Inorg. Chem., 1986, 25, 2482.

37 C. Hansch, A. Leo and R. W. Taft, Chem. Rev., 1991, 91, 165.

38 P. R. Wells, S. Ehrenson and R. W. Taft, Prog. Phys. Org. Chem.,

1968, 6, 147.

The following procedures are typical of those used to measure

the relative rates of reduction of various pairs of substituted

diazonium ions. For simple hydroxylation, the appropriate two

diazonium tetraﬂuoroborate salts (1 mmol of each) were dis-

solved in a solution of Cu(NO₃)₂(100 cm³, 0.3 mol dm^Ϫ3). For

chlorinations, KCl (0.03 mol) was added to the mixture. For

measurements of citrate-promoted hydroxylations, the method

was altered to take account of the promotional eﬀect. Thus for

the competed pairs of diazonium ions with the following

4-substituents: MeO/Me, PhO/Me and Me/H, the reaction

solution was Cu(NO₃)₂(0.05 mol dm^Ϫ3) and trisodium citrate

(0.1 mol dm^Ϫ3) whereas for all other pairs these concentrations

were doubled. To initiate the reaction, whilst stirring vigorously,

the appropriate amount of an aqueous solution of ascorbic

acid was added to produce a concentration in the reaction mix-

ture in the range 5 × 10^Ϫ6–5 × 10^Ϫ3mol dm^Ϫ3; in order to ensure

rapid mixing, volumes in the range 0.1–2.0 cm³of appropriate

molarity were added. Nitrogen evolution was immediate. After

J. Chem. Soc., Perkin Trans. 2, 2002, 1135–1150

1149

View Online

67 T. Suehiro, S. Masuda, T. Tashiro, R. Nakausa, M. Taguchi,

A. Koike and A. Rieker, Bull. Chem. Soc. Jpn., 1986, 59, 1877.

68 R. Glaser and C. J. Horan, J. Org. Chem., 1995, 60, 7518.

69 R. Glaser and C. J. Horan, Can. J. Chem., 1996, 74, 1200.

70 R. Glaser, C. J. Horan, M. Lewis and H. Zollinger, J. Org. Chem.,

1999, 64, 902.

39 S. Ehrenson, R. T. C. Brownlee and R. W. Taft, Prog. Phys. Org.

Chem., 1973, 10, 1.

40 O. Exner, Collect. Czech. Chem. Commun., 1966, 31, 3222.

41 J. Shorter, Correlation Analysis of Organic Reactivity, Research

Studies Press (Wiley), Chichester, 1982, ch. 7.

42 R. A. Marcus, J. Phys. Chem., 1963, 67, 8853.

43 W. H. Koppenol and J. Butler, Adv. Free Radical Biol. Med., 1985, 1,

91.

71 M. N. Weaver, S. Z. Janicki and P. A. Petillo, J. Org. Chem., 2001, 66,

1138.

44 U. Bertocci amd D. D. Wagman, Standard Potentials in Aqueous

Solution, Eds. A. J. Bard, R. Parsons, J. Jordan, IUPAC, 1985,

p. 292.

45 R. G. Wilkins and R. E. Yelin, Inorg. Chem., 1968, 7, 2667.

46 M. A. Hoselton, C.-T. Lin, H. A. Schwarz and N. Sutin, J. Am.

Chem. Soc., 1978, 100, 2383.

47 B. C. Gilbert, P. Hanson, J. R. Jones, A. C. Whitwood and A. W.

Timms, J. Chem. Soc., Perkin Trans. 2, 1992, 629.

48 A. Haim, Prog. Inorg. Chem., 1983, 30, 273.

49 J. K. Yandell, Aust. J. Chem., 1981, 34, 99.

72 B. A. Kowert, L. Marcoux and A. J. Bard, J. Am. Chem. Soc., 1972,

94, 5538.

73 L. Eberson, Adv. Phys. Org. Chem., 1982, 18, 79.

74 R. Foster, Organic Charge-transfer Complexes, Academic Press,

London, 1969, p. 106.

75 S. Fukuzumi, C. L. Wong and J. K. Kochi, J. Am. Chem Soc., 1980,

102, 2928.

76 P. Wardman, J. Phys. Chem. Ref. Data, 1989, 18, 1637.

77 H. Zollinger, Acc. Chem. Res., 1973, 6, 335.

78 H. Zollinger, Diazo Chemistry I, VCH, Weinheim, 1994, pp. 156–

157.

50 C. D. Ritchie and D. J. Wright, J. Am. Chem. Soc., 1971, 93, 6574.

51 C. D. Ritchie, J. D. Saltiel and E. S. Lewis, J. Am. Chem. Soc., 1961,

83, 4601.

79 E. S. Lewis and M. P. Hanson, J. Am. Chem. Soc., 1967, 89, 6268.

80 W. Schwarz and H. Zollinger, Helv. Chim. Acta, 1981, 64, 573.

81 H. Zollinger, Diazo Chemistry I, VCH, Weinheim, 1994, pp. 114–

116.

ˇ

52 H. Kropácˇova, J. Panchartek, V. Steˇrba and K. Valter, Collect.

Czech. Chem. Commun., 1970, 35, 3287.

53 H. Zollinger, Helv. Chim. Acta, 1953, 36, 1730.

54 R. M. Elofson and F. F. Gadallah, J. Org. Chem., 1969, 34, 854.

55 H. Böttcher, A. V. El’cov and N. I. Rtisˇcˇev, J. Prakt. Chem.

(Leipzig), 1973, 315, 725.

56 M. P. Doyle, J. K. Guy, K. C. Brown, S. N. Mahapatro, C. M.

VanZyl and J. R. Pladziewicz, J. Am. Chem. Soc., 1987, 109, 1536.

57 B. Ya. Medvedev, L. A. Polyakova, K. A. Bilevich, N. N. Bubnov

and O. Yu. Okhlobystin, Teor. Eksp. Khim., 1973, 9, 838.

58 J. E. Packer, C. J. Heighway, H. M. Miller and B. C. Dobson, Aust. J.

Chem., 1980, 33, 965.

59 J. Jirkovský, A. Fojtík and H. G. O. Becker, Collect. Czech. Chem.

Commun., 1981, 46, 1560.

60 K. C. Brown and M. P. Doyle, J. Org. Chem., 1988, 53, 3255.

61 J. M. Bisson, P. Hanson and D. Slocum, J. Chem. Soc., Perkin Trans.

2, 1978, 1331.

82 A. Roe, Org. React. (N. Y.), 1949, 5, 193.

83 T. J. Broxton, J. F. Bunnett and C. H. Paik, J. Org. Chem., 1977, 42,

643.

84 G. Vasiliu and O. Maior, Analele Univ. “C. I. Parhon”, Ser. Stiint.

Nat., 1962, 11, 167; G. Vasiliu and O. Maior, Chem. Abs., 1964, 60,

14498h.

85 S. H. Korzeniowski, A. Leopold, J. R. Beadle, M. F. Ahern, W. A.

Sheppard, R. K. Kanna and G. W. Gokel, J. Org. Chem., 1981, 46,

2153.

86 M. D. Ravenscroft and H. Zollinger, Helv. Chim. Acta, 1988, 71,

507.

87 R. M. Southam and M. C. Whiting, J. Chem. Soc., Perkin Trans. 2,

1982, 597.

88 G. Schiemann and W. Winkelmüller, Org. Synth., 1943, Coll. Vol. II,

299.

62 H. C. Brown and Y. Okamoto, J. Am. Chem. Soc., 1958, 80, 4979.

63 J. E. Leﬄer and E. Grunwald, Rates and Equilibria of Organic

Reactions, Wiley, New York, 1963, p. 156.

64 M. Cygler, M. Przybylska and R. M. Elofson, Can. J. Chem., 1982,

60, 2852.

65 C. Rømming and T. Tjornhorn, Acta Chem. Scand., 1968, 22, 2934.

66 T. Suehiro, T. Tashiro and R. Nakausa, Chem. Lett., 1980, 1339.

89 P. Allongue, M. Delamar, B. Desbat, O. Fagebaume, R. Hitmi,

J. Pinson and J.-M. Savéant, J. Am. Chem. Soc., 1997, 119, 201.

90 P. S. J. Canning, K. McCrudden, H. Maskill and B. Sexton, J. Chem.

Soc., Perkin Trans. 2, 1999, 2735.

91 No precedent found for mp; ¹³C NMR spectrum agrees with that

given for the hexaﬂuorophosphate by T. M. Bockman, D. Kosynkin

and J. K. Kochi, J. Am. Chem. Soc., 1997, 62, 5811.

1150

J. Chem. Soc., Perkin Trans. 2, 2002, 1135–1150

Article Doi

Sandmeyer reactions. Part 7.1 An investigation into the reduction steps of Sandmeyer hydroxylation and chlorination reactions

DOI: 10.1039/b200748g

Source and publish data:

Authors:

Article abstract of DOI:10.1039/b200748g

Full text of DOI:10.1039/b200748g

Products guided by the article

R&D Labs maybe for 1422-76-0

Relevant to this article

Hot Product