Kinetics of proton transfer from phosphonium ions to electrogenerated bases: Polar, steric and structural influences on kinetic acidity and basicity

Article abstract of DOI:10.1039/a708086g

Derivative cyclic voltammetry (DCV) and linear sweep voltammetry (LSV) have been used to measure rates of proton transfer in DMSO solution between different types of electrogenerated base (EGB) and a series of phosphonium ions of relevance to ylide formation for synthetic reactions. Although the electrochemical methods are convenient for the measurement of rates of proton transfer in these systems a major conclusion of the study is that considerable care must be exercised in the application of the methods and in drawing general conclusions from the results. In particular, comparison of kinetic acidity with thermodynamic pK values in DMSO shows that a single Bronsted relationship does not hold for the series of phosphonium ions. The kinetic acidities are profoundly affected by: whether the EGB is a carbon or nitrogen base; the propensity of some of the phosphonium ions to enolise; and steric factors. Other measures of electron-demand at the acidic methylene groups (¹³C and ¹H chemical shifts, reduction potentials) are consistent with the pK(DMSO) values. The kinetic results confirm that the Ph₃P⁺ group is, for non-enolisable phosphonium salts, more activating than Bu₃P⁺.

Full text of DOI:10.1039/a708086g

View Article Online / Journal Homepage / Table of Contents for this issue
Kinetics of proton transfer from phosphonium ions to
electrogenerated bases: polar, steric and structural influences on
kinetic acidity and basicity
Ana-Paula Bettencourt,a Ana Maria Freitas,a M. Irene Montenegro,*,a
Merete Folmer Nielsen*,b and James H. P. Utley*,c
a Department of Chemistry, University of Minho, Largo do Paço, P-4719 Braga Codex,
Portugal
b Department of Chemistry, University of Copenhagen, Symbion Science Park, Fruebjergvej 3,
DK-2100, Copenhagen Ø, Denmark
c Department of Chemistry, Queen Mary and Westfield College, Mile End Road, London, UK
E1 4NS
Derivative cyclic voltammetry (DCV) and linear sweep voltammetry (LSV) have been used to measure
rates of proton transfer in DMSO solution between different types of electrogenerated base (EGB) and
a series of phosphonium ions of relevance to ylide formation for synthetic reactions. Although the
electrochemical methods are convenient for the measurement of rates of proton transfer in these systems a
major conclusion of the study is that considerable care must be exercised in the application of the methods
and in drawing general conclusions from the results. In particular, comparison of kinetic acidity with
thermodynamic pK values in DMSO shows that a single Brønsted relationship does not hold for the series
of phosphonium ions. The kinetic acidities are profoundly affected by: whether the EGB is a carbon or
nitrogen base; the propensity of some of the phosphonium ions to enolise; and steric factors. Other
measures of electron-demand at the acidic methylene groups (13C and 1H chemical shifts, reduction
potentials) are consistent with the pK(DMSO) values. The kinetic results confirm that the Ph3P؉ group is,
for non-enolisable phosphonium salts, more activating than Bu3P؉.
Phosphonium salts of the general formula R3PϩCH2Y XϪ are
R3P+CH2YX–
important in synthesis as precursors of reactive ylides, e.g. in
the Wittig reaction. They are true carbon acids, except where Y
is such that enolisation is possible; it is only recently that the
1a R = Ph, Y = H, X = Br
1b R = Ph, Y = CH3(CH2)2, X = Br
1c R = Ph, Y = Ph, X = Br
thermodynamic acidity (pK) of a number of these salts has
1d R = Bu, Y = Ph, X = Cl
been determined in DMSO.1 In practice it is often kinetic acid-
1e R = Ph, Y = -CH=CHPh, X = Cl
ity which is of significance in preparative reactions; the rate at
1f R = Ph, Y = -CO2Et, X = Br
which a reactive carbanion or ylide may be formed by reaction
1g R = Bu, Y = -CO2Et, X = Br
between a base and carbon acid may determine its usefulness.
1h R = Ph, Y = -COMe, X = Br
Electrogenerated bases (EGBs), formed by cathodic reduc-
tion of organic precursors (probases), have found use2 as
5
4
6
3
reagents which are easy to generate under controlled conditions
and which, depending on the probase structure, cover a range
of effective basicities. As part of an ongoing study of the for-
mation and properties of EGBs the kinetics of proton transfer
between carbon acids and EGBs has been studied over several
years. Initially3 it was hoped that a relationship between the
rates of proton transfer and the thermodynamic pK values of
similar acids (with known thermodynamic acidity) could be
used to predict pK values for acids of unknown pK. However,
this approach is fraught with difficulty because different types
of acids (e.g. oxygen vs. carbon acids) have different intrinsic
rate constants (defined as the statistically corrected second
order rate constant for a process with zero driving force) and
different Brønsted α-values.4 For acids belonging to a single
family good correlations between thermodynamic and kinetic
acidities have previously been found using EGBs.5,6
An examination of the kinetics of proton transfer from
phosphonium ions to EGBs is presented here. This allows a
useful comparison, in one series and depending on the nature
of Y (compounds 1a–h), between acids which are not enolisable
(true carbon acids) and those in which proton transfer can
take place from either carbon or oxygen. Four EGBs have been
used which have differing steric demands and either nitrogen
7
2
9
9
8
1
NC
CN
NC
CO2Et
2
3
N
N
N
N
4
5
or carbon basic centres; the corresponding probases are
compounds 2–5.
Results and discussion
All measurements were made using DMSO as solvent since
DMSO is the dipolar aprotic solvent in which the largest num-
ber of pK values have previously been reliably determined.7
Voltammetric experiments on the probases and measurement
of rates of proton transfer were performed at an Hg-coated
Pt electrode (d = 0.6 mm) with Bu4NPF6 (0.1 ) as supporting
J. Chem. Soc., Perkin Trans. 2, 1998
515

View Article Online
Table 1 Reduction potentials and pK values of phosphonium saltsa
Table 2 Reduction potentialsa for probases
Compound
pK(DMSO)b
22.5
ϪEp,c(1)/V
Compound
ϪEo(1)/V
ϪEo(2)/V
1a
1b
1c
1d
1e
1f
1.989
1.915
1.780
2.690
1.741
1.685
2.611
1.685
2
3
4
5
0.587
0.710
1.298
1.125
1.303
1.235
17.4
8.5
7.1
a Formal potentials with reference to SCE, measured by cyclic voltam-
metry at 1 V sϪ1, as 1 m solutions in DMSO–Bu4NPF6 (0.1 ), at an
Hg/Pt electrode.
1g
1h
a Peak potentials with reference to SCE, measured by cyclic voltam-
metry at 0.1 V sϪ1, as 1 m solutions in DMSO–Bu4NPF6 (0.1 ), at a
Hg/Pt electrode. b From ref. 1.
anion. The carbanion deprotonates the starting phosphonium
ion to produce a hydrocarbon (YCH3 or RH) and an ylide, the
chemically irreversible reduction of which is responsible for a
second reduction peak. Further reduction peaks may be associ-
ated with reduction of triphenylphosphine (R = Ph) or of the
hydrocarbon (YCH3 or RH).
In the presence of electrogenerated bases, which are gener-
ated at less cathodic potentials than those required for cathodic
cleavage of the phosphonium ions, the slow and usually meas-
urable step is the proton transfer from the phosphonium ion to
the EGB. In previous studies9,10 the linking of this step to ylide
formation has been established.
electrolyte. The supporting electrolyte was chosen to ensure
that neither the cation nor the anion gave rise to ion-pairing
with radical anions/dianions or cations in solution. Reduction
potentials are given vs. the saturated calomel electrode (SCE).
Phosphonium ions
The phosphonium ions R3PϩCH2Y (1) for which the rates of
deprotonation have been measured fall into two families: those
activated by Ph3Pϩ (1a–c, 1e–f, 1h) and those activated by
Bu3Pϩ (1d, 1g). The last family presents a special problem with
respect to which proton is transferred since in addition to the
CH2-protons another six equivalent protons are available
neighbouring the phosphorous atom. The phosphonium ions
may be further divided into three classes: (a) the phosphonium
group is the only activating group [Ph3PϩCH3 (1a), Ph3PϩCH2-
CH2CH2CH3 (1b)]; (b) the carbon skeleton provides an add-
itional weakly acidifying effect [Ph3PϩCH2Ph (1c), Ph3PϩCH2-
EGBs and the measurement techniques
From the thermodynamic acidities in DMSO determined
recently1 for four of the phosphonium ions it appears (Table 1)
that the range of acidities covered by the series is at least 15.4
pK-units. Since no Brønsted α-value has been determined for
the deprotonation of phosphonium ions with any base, we will
initially assume the Brønsted α-value to be ‘normal’ and in the
range 0.3–0.7. As a consequence the kinetic range covered by
the series of phosphonium ions with a single base is estimated
to be 6–10 orders of magnitude.
A kinetic range as large as that predicted above is almost
impossible to cover experimentally using a single technique and
we have therefore chosen to apply two different approaches.
They are: (1) the application of different EGBs with varying
basicity; (2) the application of different electroanalytical tech-
niques for a single (strong) base. The EGBs chosen are the dian-
ions of probases 2 and 3 and the radical anions of probases 4
and 5. They fulfil the criteria that the potential of their form-
ation (see Table 2) is higher (less cathodic) than the potential at
which the phosphonium ions are reduced. Furthermore the
dianions of probases 2 and 3 and the radical anions of probases
4 and 5 are stable in the absence of proton donors on timescales
corresponding to scan rates down to 0.1 V sϪ1. Upon addition
of an excess of appropriately acidic phosphonium salts the
chemical reversibility of the EGB generating reduction is
partially or completely lost and the reduction peak height
increases for 4 and 5.
CH᎐CHPh (1e), Bu PϩCH Ph (1d)]; (c) the acidic protons are
᎐
3
2
doubly activated due to the presence of a strongly electron-
withdrawing group in addition to the phosphonium group
[Ph3PϩCH2CO2Et (1f), Bu3PϩCH2CO2Et (1g), Ph3PϩCH2-
COMe (1h)]. For the last class of phosphonium ions another
problem arises; the possibility for enolisation of the acid may
change the site from which the proton transfer takes place from
carbon to oxygen, and since the intrinsic rates of proton trans-
fer from carbon and from oxygen very often differ by orders of
magnitude4 the kinetic measurements relating to this class of
compounds should be treated very carefully when it comes to
evaluation of relationships between thermodynamic and kinetic
acidities.8
It is important for the proposed kinetic experiments that the
EGBs are formed at higher (less cathodic) potentials than those
at which the phosphonium ions are reduced. The electro-
chemistry of the phosphonium ions was therefore briefly exam-
ined. The data are collected in Table 1. The reduction of phos-
phonium ions in dipolar aprotic solvents in the absence of
added acids or bases has previously been well-investigated.9,10
The main conclusions of that work pertaining to phosphonium
ions with α-protons of the general formula R3PϩCH2Y are
summarized in Scheme 1. The salient features are: (i) the first
The two types of EGBs differ in a number of ways: (a) in the
dianions derived from 2 and 3 the site of protonation is
expected to be a carbon atom whereas in the two radical anions
derived from 4 and 5 the site of protonation is a nitrogen atom;
(b) the proton transfer to the dianions [reaction (8) in Scheme 2]
Ep,c(1)
R3PϩCH2Y ϩ eϪ
(R3P ϩ CH2Y)/(R2PCH2Y ϩ R)
(1)
ؒ
ؒ
Eo(1)
Ϫ
PB ϩ eϪ
PB
(5)
(6)
(7)
(8)
>Ep,c(1)
Ϫ
ؒ
ؒ
᝽
CH2Y/ R ϩ e
ϪCH2Y/ϪR
(2)
(3)
(4)
Eo(2)
Ϫ
R3PϩCH2Y ϩ ϪCH2Y/ϪR → R3P᎐CHY ϩ CH3Y/RH
PB ϩ eϪ
PB2Ϫ
᎐
᝽
Ep,c(2) < Ep,c(1)
R3P᎐CHY ϩ eϪ
?
Ϫ
K7
PB ϩ PB2Ϫ
2 PB
᎐
᝽
k
Scheme 1 Direct cathodic reduction of phosphonium ions
PB2Ϫ ϩ R3PϩCH2Y
PBHϪ ϩ R3P᎐CHY
᎐
Scheme 2 Deprotonation of phosphonium salts by the EGBs 22Ϫ or
32Ϫ by an (E)EC mechanism
chemically irreversible reduction peak corresponds to a two
electron process but appears as a one electron peak since half of
the phosphonium ion is used as proton donor [reaction (3)]; (ii)
the first reduction peak corresponds to cleavage of a C᎐P bond
with rapid further reduction of the carbon radical to a carb-
may in principle be reversible on the timescale of the measure-
ments. This is because neither of the products (the more stable
516
J. Chem. Soc., Perkin Trans. 2, 1998

View Article Online
ylides and the monoanions PBHϪ equate to 2HϪ or 3HϪ) are
consumed in fast follow-up reactions under the conditions of
the measurements, whereas the neutral radicals formed by pro-
tonation of the radical anions of 4 and 5 are further reduced
and subsequently protonated in a normal DISP1 reaction
(Scheme 3), rendering the initial proton transfer essentially
irreversible.
Ϫ
PB ϩ eϪ
PB
(9)
᝽
k
Ϫ
PB ϩ R3PϩCH2Y
PBH ϩ R3P᎐CHY
(10)
ؒ
᎐
᝽
fast
Ϫ
PBHϪ ϩ PB
(11)
ؒ
PBH ϩ PB
᝽
fast
PBHϪ ϩ R3PϩCH2Y
PBH2 ϩ R3P᎐CHY
(12)
᎐
Ϫ
Scheme 3 Deprotonation of phosphonium salts by the EGBs 4 or
᝽
5
Ϫ by a DISP1 mechanism
᝽
The possible reversibility of protonation of the dianions of 2
and 3 limits their applicability for simple, uncomplicated meas-
urements of pure proton transfer rate constants. A driving force
for the proton transfer reaction large enough to make the back
reaction insignificant on the timescale of the measurements is
required for a simple (E)EC scheme to be adequate, i.e. only the
more acidic phosphonium ions can be used with these EGBs.
On the other hand very acidic phosphonium ions cannot be
studied since eventually they will be able to protonate the rad-
ical anions of 2 and 3 and thereby invalidate the measurements.
Rates of proton transfer from carbon or oxygen acids to rad-
ical anions or dianions have previously been made using linear
sweep voltammetry (LSV),11 cyclic voltammetry,5,12 derivative
cyclic voltammetry6,13 (DCV) and double potential step
chronoamperometry (DPSC).3,5,13–15 The method used must be
carefully chosen. For instance, it is now clear that DPSC is
unsuitable for use with the phosphonium ions using the
dianions of 2 and 3 because the method involves stepping
cathodically to potentials at which direct reduction of the
phosphonium ion may take place. We now believe that earlier
measurements for phosphonium ions using this method15 are
unreliable. The applicability of the LSV method is also limited
since determination of rate constants is only possible when the
electrochemical process is under purely kinetic control, i.e. the
rates of electron and proton transfers must be relatively high.
In this study we have chosen to use DCV16 as the main tech-
nique, and in order to check whether the acid–base systems are
well-behaved and indeed follow the mechanisms in Schemes 2
and 3 we have chosen to record the values of RIЈ [RIЈ = ϪipЈ(ox)/
ipЈ(red), where ipЈ(ox) and ipЈ(red) are the heights of the peaks of
the derivative of the current] over a broad range of scan rates
(ν) and to examine the fit of the experimental data [RIЈ vs.
Ϫlog(ν)] to the theoretical working curve for the assumed
mechanism. In most cases two or more concentrations of the
phosphonium salt were used for this type of measurement, and
the rate constants calculated from the fit of experimental to
theoretical data (see Experimental section).
Fig. 1 Methods adopted for kinetic measurements using probases 2
and 3. (a) Choice of rest potential indicated on a normal CV. (b)
Derivative cyclic voltammogram (iЈ vs. t) for method 1 (broken line) and
method 2 (full line). For both methods the quantity determined is
RЈI(2) = ϪiЈp(ox2)/iЈp(red2).
potentials for consecutive reduction of 2 or 3. Due to the stabil-
ity of the radical anions, 2 or 3 are completely converted into
the corresponding radical anions in the diffusion layer next to
the electrode which will be completely depleted of 2 or 3. The
mechanistic scheme is therefore considerably simplified (to an
EC mechanism) since only reactions (6) and (8) in Scheme 2
need to be considered when calculating the theoretical data
necessary to obtain the rate constant for the proton transfer.
This way of making the measurements will be referred to as
method 2. The quantities measured for each of the two
methods are outlined in Fig. 1, and both methods were applied
for most of the phosphonium salts studied using 2 or 3 as
Ϫ
Ϫ
probases. The stability of the radical anions 2 or 3 in the
᝽
᝽
presence of the phosphonium salts was always checked by
confirming that RIЈ(1) ≈ 1 even at low scan rates, when the scan
was reversed between the two reduction peaks.
For probases 4 and 5 only a single electron transfer is
involved and the data are analyzed according to theoretical
data for the DISP1 mechanism in Scheme 3. For some of the
proton transfer reactions (see below) involving 4 Ϫ and 5 Ϫ the
᝽
᝽
rate was so high that DCV was inconvenient, and the rate con-
stants were therefore determined by LSV, i.e. by measuring the
values of Ep Ϫ Eo at several scan rates followed by application
of eqn. (13) pertaining to the mechanism in Scheme 3.17
nFν
2nF
When the rest potential from which the cathodic scan is initi-
ated is anodic relative to the first reduction of probases 2 or 3,
the mechanism includes all the reactions in Scheme 2 in an EEC
process. Accordingly, the theoretical data necessary to calculate
the rate constant, k, for the proton transfer must include the
electron transfer reaction between the dianion and the substrate
[reproportionation, reaction (7)], the equilibrium constant for
which depends on whether 2 or 3 is used as probase due to the
difference in their values of ∆Eo = Eo(1) Ϫ Eo(2), i.e.
K7 = exp[(F∆Eo)/(RT)]. This way of making the measurements
is called method 1 in the following discussion. Since the radical
anions derived from probases 2 or 3 are stable in the absence of
strong proton donors, the measurements may be made in a sim-
pler way if the rest potential is chosen to be between the two
(13)
k =
exp
ͫ
(Ep Ϫ Eo) ϩ (2 × 0.783) ϩ ln 2
ͬ
RT
RT
Kinetic results and the influence of experimental difficulties
The results of the kinetic measurements for the series of phos-
phonium salts and the four EGBs are summarised in Table 3. In
general the rate constants determined by methods 1 and 2 using
22Ϫ and 32Ϫ as EGBs were identical within the experimental
error. Before discussing the data it is proper to consider in detail
the measurements which lead to the values given in Table 3.
Measurements using 22؊ as EGB. There are only three num-
bers in Table 3 for this EGB. The least acidic salts do not react
on the timescale of the measurements, and an additional
problem was encountered when using 22Ϫ as the base for
J. Chem. Soc., Perkin Trans. 2, 1998
517

View Article Online
Table 3 Rate constants, kEGB in Ϫ1 sϪ1, for proton transfer in DMSOa
Ϫ
Ϫ
R3PϩCH2Y (1)
22Ϫ
32Ϫ
4
5
᝽
᝽
1a, R = Ph, Y = H
[12.31]b
(1.9 0.1) × 102 d
(2.4 0.4) × 103 d
{8.25}c
{1.27 × 102}c
{1.61 × 103}c
1b, R = Ph, Y = CH3(CH2)2
1c, R = Ph, Y = Ph
1d, R = Bu, Y = Ph
1e, R = Ph, Y = PhCH᎐CH
1f, R = Ph, Y = EtO2C
1g, R = Bu, Y = EtO2C
1h, R = Ph, Y = CH3CO
[13.15]b
(2.1 0.2) × 102 e
(1.6 0.1) × 103 d
(2.8 0.5) × 103 f
(2.7 0.2) × 103 h
(1.8 0.2) × 101 i
(4 2) × 101 g
(3.8 0.1) × 102 j
(1.6 0.2) × 105 l
(3.2 0.2) × 104 n
(4.6 × 105)q
(2.1 0.1) × 104 e
(2.8 0.2) × 108 m
(4 1) × 107 o
᎐
(6.6 0.3) × 104 k
(5.0 0.5) × 104 l
4.3 × 105 p
a Co(probase) = 1 m, t = 22 2 ЊC, Hg/Pt working electrode, Eo(2) Ϫ Esw = 0.3 V for 2 and 3, Eo(1) Ϫ Esw = 0.3 V for 4 and 5. DMSO–Bu4NPF6
(0.1 ). b By extrapolation—see Fig. 3. c Corrected for number of acidic protons, i.e. ×0.67. d Co(HB) = 10 m, average of rate constants determined
by fit to working curve (DISP1) and from the ν0.5-value. e Co(HB) = 10 and 20 m, average of rate constants determined by fits to working curve
(DISP1) and from ν0.5-values. f Co(HB) = 10 and 20 m, average of rate constants determined from ν0.5-values (DISP1). g Co(HB) = 80 m, average of
rate constants determined from ν0.5-values (methods 1 and 2). h Co(HB) = 10, 20 and 40 m, average of rate constants determined by fits to working
curve (DISP1) and from ν0.5-values. i Co(HB) = 40 m, average of rate constants determined by fit to working curve (DISP1) and from ν0.5-value.
j Co(HB) = 10 and 40 m, average of rate constants determined by fits to working curve and from ν0.5-values (method 1). k Co(HB) = 10 m, average
of rate constants determined by fit to working curve and from ν0.5-value (method 1). l Co(HB) = 10 m, average of rate constants determined by fits
to working curves and from ν0.5-values (methods 1 and 2). m Co(HB) = 20 m, rate constant determined from values of Ep Ϫ Eo measured at
0.1 Ϫ 1.0 V sϪ1 and eqn. (13). n Co(HB) = 10 and 20 m, average of rate constants determined by fits to working curves and from ν0.5-values (methods
1 and 2). o Co(HB) = 10 and 20 m, average of rate constants determined by LSV at ν = 0.5, 1 and 2 V sϪ1 (DISP1). p Co(HB) = 1 m, rate constant
determined by fit to working curve (method 1). q Co(HB) = 10 m, rate constant determined by fit to working curve (method 1) for ν = 100–800 V sϪ1
.
phosphonium ions of intermediate acidity: the working curves
become too steep at low scan rates (i.e. the problem is serious
for weak acids) apparently due to some kind of acid-catalysed
tautomerisation of the dianion leading to a splitting of the oxi-
dation peak into two peaks. The problem is not related to the
phosphonium ions since the same phenomenon was observed
for other types of weak acids as well. The new peak is at a
slightly more anodic potential than the original one. The con-
sequence is an artificially fast decrease of the height of the
original oxidation peak with decreasing scan rate thereby
invalidating measurements on this timescale. The group of
phosphonium ions of intermediate acidity caused the above
mentioned peak splitting at low scan rates and results from
these measurements are therefore excluded. The rate constants
obtained using 1g and 1f, R3PϩCH2CO2Et (R = Bu, Ph), are
based on measurements with only a single concentration of the
acid (10-fold excess) due to the high rate of the reactions. The
fits to the working curves are good for both compounds and the
rate constants obtained are surprisingly similar. The data
obtained for 1h, Ph3PϩCH2COCH3, are questionable. Since the
rate constant for proton transfer is high, data for working curve
fit by method 1 (rest potential anodic to the first reduction)
could only be obtained using equimolar concentrations of
phosphonium salt and probase.
Measurements using 32؊ as EGB. This EGB does not cause
the problem encountered for 22Ϫ with splitting of the oxidation
peak and 32Ϫ appears to be slightly more basic than 22Ϫ. A
larger number of rate constants could therefore be determined.
Compound 1c, Ph3PϩCH2Ph, reacts very slowly with 32Ϫ, the
RIЈ vs. Ϫlog(ν) plots at low concentration of 1c are too flat
compared with the working curves, and the apparent reaction
order in phosphonium salt is larger than unity. This behaviour
Fig. 2 Fit of experimental DCV data obtained for 1g (10 m) using 3
(1 m) as probase in DMSO (0.1  Bu4NPF6) to working curves for (a)
method 1, based on ∆Eo = Eo(1) Ϫ Eo(2) = 0.52 V and (b) method 2. In
both cases Eo(2) Ϫ Esw = 0.3 V.
indicates reversibility in the proton transfer reaction on the
timescale of the measurements.18 The better fits of the experi-
mental data to the working curves are therefore obtained at a
high concentration of 1c (80 m) where the proton transfer
equilibrium is displaced towards the products. The analogous
compound 1d, Bu3PϩCH2Ph, is unreactive towards 32Ϫ. The
presents a problem using 3 as probase since protonation of the
radical anion, 3 Ϫ, takes place at low scan rates. Using a 10-fold
᝽
data obtained for 1e, Ph PϩCH CH᎐CHPh, give acceptable fits
excess of 1h and scan rates in the range 100–800 V sϪ1 only
᎐
3
2
to the working curves for both methods. In contrast to the
results obtained using 22Ϫ as EGB, the rate constants deter-
mined for reaction between 32Ϫ and the compounds 1g and 1f,
R3PϩCH2CO2Et (R = Ph, Bu), differ by a factor of five. In both
cases the quality of the data is good as seen in Fig. 2, which
demonstrates the data treatment by fit to working curves for the
two methods. The high acidity of 1h, Ph3PϩCH2COCH3,
allows an uncertain estimate of the rate constant by method 1
since a very limited part of the working curve is accessible.
Ϫ
Measurements using 4 as EGB. The least acidic phospho-
᝽
nium cations do not protonate either of the dianionic EGBs
derived from 2 and 3 on the timescale of the measurements, and
therefore the faster reacting radical anions derived from phen-
azine (4) and azobenzene (5) were used for reaction with weakly
518
J. Chem. Soc., Perkin Trans. 2, 1998

View Article Online
Ϫ
acidic cations. Using 4 as an EGB allows determination of
᝽
the kinetic acidity of all the thermodynamically weak phos-
phonium cations—except for 1d, Bu3PϩCH2Ph, which does not
react on the timescale of the measurements. However, the qual-
ity of the data varies considerably; good fits to the working
curve for the DISP1 mechanism (Scheme 3) are obtained for
1a, Ph PϩCH , and 1e, Ph PϩCH CH᎐CHPh, in contrast to the
᎐
3
3
3
2
poor fits obtained for the salts 1b and 1c, Ph3PϩCH2CH2-
CH2CH3 and Ph3PϩCH2Ph. For the most acidic phosphonium
cations the rates of protonation of 4 Ϫ are so high that the LSV
᝽
technique and eqn. (13) had to be used for the determination of
the rate constants. However, the rate constants determined are
somewhat uncertain due to non-ideal behaviour; the peaks were
too broad and the values of ϪdEp/dlog(ν) too large compared
to the theoretical values for a simple DISP1 mechanism. The
non-ideal behaviour may be due to contribution from the rate
of the heterogeneous electron transfer process but was not
Fig. 3 Comparisons of rate data for different EGBs.—Data from
further investigated.
Measurements using 5 as EGB. The kinetic data obtained
Table 3.
Ϫ
᝽
using the thermodynamically very weak acid 1a, Ph3PϩCH3,
give an excellent fit to the working curve for the DISP1 mechan-
ism, whereas the data obtained using 1b, Ph3PϩCH2CH2-
CH2CH3, for unknown reasons do not. The rate constant cal-
culated for proton transfer from 1b is therefore only an
approximate value. The rate constant for proton transfer from
1c, Ph3PϩCH2Ph, was obtained by an excellent fit to the theo-
retical data but surprisingly the value is close to that obtained
for the thermodynamically much weaker (cf. Table 1) phospho-
nium ion 1a, Ph3PϩCH3, (corrected for the different number of
equivalent protons in the two compounds). This contrasts with
the relative rate constants found for deprotonation of the same
two compounds using 4 Ϫ as base where the rate constant found
᝽
for 1c, Ph3PϩCH2Ph, is approximately an order of magnitude
larger than the rate constant determined for 1a, Ph3PϩCH3. The
azobenzene EGB 5 Ϫ is the only base used for which any reac-
᝽
Fig. 4 Attempted Brønsted plot using EGB derived from 3
tion of 1d, Bu3PϩCH2Ph, was detected, and the reaction is so
slow that the rate constant is more uncertain than the others
reported using this EGB. The most interesting feature is the fact
that whereas the change from R = Ph to Bu in the pair,
R3PϩCH2CO2Et, 1g and 1f, gives rise to a change in the rate
constant less than a factor of five, the change from R = Ph to
Bu in the pair R3PϩCH2Ph, 1c to 1d, gives rise to a change in the
rate constant of more than two orders of magnitude.
factor for a change in the driving force for the proton transfer
(normally the Brønsted α-value) is different for the two bases.
As a consequence it is not possible with any confidence to use
overlapping rate data from pairs of different probases to con-
struct a meaningful table of relative rates of deprotonation. The
possible origin of this difficulty is discussed in the following
sections.
Using only the rate constants obtained using the dianion of
3, and estimating from the plot in Fig. 3 an approximate rate
constant for the reaction of 1a with 32Ϫ, a Brønsted plot results
in a non-linear relationship (Fig. 4). Since there was an appar-
Rates of proton transfer and thermodynamic acidity
Relative kinetic acidities of the series of phosphonium cations.
In order to be able to compare the kinetic acidities of the entire
series of phosphonium cations on a single scale, it is necessary
to ‘overlap’ the four series of kinetic data obtained using the
four different EGBs. This, however, proved to be difficult. In an
ideal situation the change from one EGB to another EGB with
a different basicity would amount to a change by a constant
factor of the measured rate constants for the same series of
acids (HAi), i.e. kEGB2(HAi) = constant × kEGB1(HAi)—or using
the logarithms of the rate constants: log[kEGB2(HAi)] =
log[kEGB1(HAi)] ϩ constant. The kinetic data in Table 3 do not
follow this relationship: using as the basis of the comparison the
values (k4 Ϫ) obtained with phenazine 4 as probase, since this
ent linear correlation between the data obtained using 32Ϫ and
Ϫ
those obtained using 4 a similar non-linear relationship will
᝽
be obtained using the rate constants related to 4 in the Brønsted
plot. This non-linearity is in contrast to what we have observed
for the same type of EGBs when the proton donors were bis-
sulfones of the type RSO2CH2SO2R.19 For a given EGB,
acceptable Brønsted plots (logk vs. pK) are obtained provided
that sulfones likely to offer steric hindrance to proton transfer
were excluded.19 Linear relationships have also previously been
obtained using the radical anions of azobenzene derivatives as
EGBs for series of phenylacetonitriles and for series of benzyl-
sulfones,5 or the radical anion of anthracene for deprotonation
of a series of phenols.6 The possible origin of the non-linearity
in the present case will also be discussed in the following
sections.
᝽
EGB is the one for which there are data for the largest number
of phosphonium ions, the values of log[kEGB(HAi)] were plot-
ted against log[k4 Ϫ(HAi)] for each of the three other EGBs.
᝽
The results are shown in Fig. 3, and obviously we do not have
three horizontal lines. Due to the limited amount of data for
Driving force for proton transfer. Unfortunately the thermo-
dynamic basicity of the EGBs used is not known but the final
products of hydrogenation of the probases are known; hydro-
genation of the C᎐C double bond for 2 and 3,20 9,10-
dihydrophenazine for 421 and 1,2-diphenylhydrazine for 5.22
The most basic site in 22Ϫ and 32Ϫ is expected to be C-9 rather
than the exocyclic carbon atom, C-10. This is based on the ca.
10 pK unit difference in the thermodynamic acidity in DMSO
Ϫ
EGBs 22Ϫ and 5 these points could be on two horizontal
᝽
lines, but for the data relating to 32Ϫ this cannot be the case.
2Ϫ
The apparent linear relationship between log[k3 (HAi)] and
log[k4 Ϫ(HAi)] is intriguing. It means that the multiplier con-
᝽
necting rates of deprotonation of individual phosphonium cat-
ions is varying continuously from salt to salt when the EGBs
4
Ϫ and 32Ϫ are compared or, in other words, that the response
᝽
J. Chem. Soc., Perkin Trans. 2, 1998
519

View Article Online
Fig. 5 1H (᭹) and 13C (᭿) chemical shifts (R3PϩCH2Y), measured
in [2H6]DMSO, plotted vs. pK(DMSO)1
Fig. 6 1H (᭝) and 13C (᭿) chemical shifts (R3PϩCH2), measured
in [2H6]DMSO, plotted vs. the irreversible reduction potentials for
(R3PϩCH2) (see Table 1)
between fluorene (pK = 22.6)7 and CH2(CN)2 (pK = 11.1)7 or
CH2(CN)CO2Et (pK = 13.1)23 which argues strongly in favour
of the C-9 position being the more basic. The ‘extra’ negative
charge in 22Ϫ and 32Ϫ compared to the fluorenide anion is
expected to increase the thermodynamic basicity of the dian-
ions compared to the fluorenide anion, and consequently
pK(2HϪ) and pK(3HϪ) are expected to be higher than 22.6. The
pK of 9,10-dihydrophenazine (4H2) is not known in DMSO but
the pK of 1,2-diphenylhydrazine (5H2) in DMSO is 26.2.7 The
thermodynamic acidity of the conjugate acids of the radical
deprotonation of carbon acids often are orders of magnitude
slower than protonation of oxygen and nitrogen anions.4
Since the rates of proton transfer to the dianions are far from
diffusion controlled, the coulombic attraction between the
dianions and the cationic acids is not expected to have any
influence on the rate of proton transfer. However, the (smaller)
Ϫ
Ϫ
coulombic attraction between 4 or 5 and the most acidic
᝽
᝽
phosphonium cations for which the measured rate constants are
approaching the rate constant for a diffusion controlled process
may contribute to the magnitude of the measured rate constant.
Role of the atom from which the proton is transferred—the role
of enolisation. The difference between those phosphonium ions
which are unambiguously carbon acids (compounds 1a–e) and
those which contain carbonyl groups α to the -CH2- group
(compounds 1f–h) is first of all a considerably higher thermo-
dynamic acidity of the latter type (cf. the pK values for 1f and
1h in Table 1). This is also associated with a significant increase
in proton transfer rate: for a given probase the latter group gives
rates of proton transfer ca. 103 times faster than the first group
which may not only be associated with the higher thermo-
dynamic acidity but also with the enolisation of the phospho-
ؒ
ؒ
anions (4H ) and (5H ) are probably somewhat higher than the
thermodynamic acidity of (4H2) and (5H2) in DMSO in ana-
logy with the relative thermodynamic acidities of the conjugate
acids of e.g. anthracene radical anion and the monoanion of
9,10-dihydroanthracene.24 In conclusion, the thermodynamic
basicity of the four applied EGBs may be very similar with pK
values in the range 23–25 in DMSO.
These estimates of the thermodynamic basicities of the
EGBs should be compared with the thermodynamic acidities of
the studied phosphonium cations. The pK values in DMSO are
known for four of them,1 cf. Table 1, and an independent
assessment of the polar effects at the methylene group in
1
R3PϩCH2Y may be obtained from the H and 13C chemical
shifts which are known25 to be significantly influenced by elec-
tron demand at the carbon atom. The relevant 1H and 13C
chemical shifts have been measured for the phosphonium salts
nium cation. Using H NMR spectroscopy the phosphonium
1
salt 1h was shown to contain 14% enol at equilibrium in
DMSO, whereas the equilibrium constant for enolisation of 1f
and 1g is so small that the enol form is not detected by NMR
spectroscopy, i.e. the enol content is <1%.
However, since the intrinsic rate constant for the oxygen
based acid may be several orders of magnitude larger than the
rate constant of the corresponding carbon acid,4 the enhanced
kinetic acidity of the enol-protons certainly contributes to the
very high measured rate constant in the case of 1h and may to
an unknown extent contribute to the kinetic acidities measured
for 1f and 1g. Some of the apparent upward curvature in the
Brønsted plot in Fig. 4 (which includes 1f and 1h) may be
related to this phenomenon.
Possible intervention of steric factors. The indications of elec-
tron demand at the methylene group of the phosphonium ions,
taken from NMR chemical shifts, relate to initial state properties
and are, to a first approximation, independent of steric factors.
The same is true for the reduction potentials and for the thermo-
dynamic acidity which is independent of the base involved. The
kinetic results, on the other hand, are crucially influenced by
both polar and steric factors which may emanate from either the
EGB or the phosphonium salt as well as by the precise nature of
the proton transfer reaction as discussed above.
1
in [2H6]DMSO and both the 13C and H chemical shifts are
consistent with the order of acidities known1 for four of the
phosphonium salts studied, although the relationship is quali-
tative (Fig. 5). Also the potentials of the first irreversible reduc-
tion of the triphenylphosphonium ions correlate quite well with
1
both H and 13C chemical shifts and the results are displayed
in Fig. 6. Superimposed on those plots are values for the two
tributylphosphonium ions in the study which clearly form a
very distinct set.
The estimated thermodynamic basicities of the EGBs and
of the thermodynamic acidities of the phosphonium cations
indicate that (with perhaps the exception of 1d) all the proton
transfer reactions studied should be thermodynamically
favourable.
Role of the atom receiving the proton. The kinetic basicities of
the two radical anions derived from 4 and 5 in which the site of
protonation is definitely nitrogen, are much higher than those
of the dianionic EGBs despite the similarities in thermo-
dynamic basicity as estimated above. For the phosphonium
salts, e.g. 1c, 1e, 1f and 1g, the rate constants for proton transfer
to the radical anions of phenazine (4) or azobenzene (5) are 2–3
orders of magnitude greater than for proton transfer to the
dianions of 2 or 3 which is almost certainly to carbon. This is
not unexpected since protonation of carbanions analogously to
Steric effects are probably responsible for the curious ‘level-
ling’ of rates of proton transfer to the radical anion of azoben-
Ϫ
zene 5 in the series 1a–c. For both phenazine radical anion
᝽
520
J. Chem. Soc., Perkin Trans. 2, 1998

View Article Online
4
Ϫ and the dianion of 3 (accepting the values given by extrapo-
enone and, respectively, malononitrile and ethyl cyanoacetate
᝽
lation) the rate of proton transfer from the benzylphosphonium
ion 1c is significantly greater than that from the alkylphospho-
nium salts 1a and 1b. This is entirely consistent with the
using modifications of literature methods.26
Acids. The phosphonium salts, Ph3PϩCH3 BrϪ (Aldrich) and
Ph3PϩCH2COMe ClϪ (Aldrich) were dried under vacuum
before use. The phosphonium salts, Ph3Pϩ(CH2)3CH3 ClϪ,
known1 greater acidity of 1c. But this increase in rate is not
Ϫ
᝽
found for the azobenzene derived EGB 5 in which the nega-
tive charge, on the nitrogens, is hindered relative to the nitro-
gens in phenazine. Thus it is likely that the more hindered base
5
more hindered benzylic cation 1c vis à vis the less acidic and less
hindered ions 1b and 1a.
Ph PϩCH CH᎐CHPh ClϪ, Ph PϩCH Ph BrϪ and Ph3PϩCH2-
᎐
3
2
3
2
CO2Et BrϪ were prepared by reaction between triphenylphos-
phine and the corresponding halides.27 The phosphonium salts
Bu3PϩCH2Ph, ClϪ and Bu3PϩCH2CO2Et BrϪ were prepared by
reaction between tributylphosphine and the corresponding
halides.27
Ϫ
reacts more slowly than expected with the more acidic but
᝽
Bu3P؉ cf. Ph3P؉ substituent effect. The Bu3PϩCH2Y and
Ph3PϩCH2Y ions clearly behave very differently in the proton
transfer reactions but analysis of this difference is problem-
atical. The relative reactivities depend on the type of cation
(enolisable or not) and on the probase. The phosphonium ions
1f and 1g react at comparable rates with the EGB derived from
2 but the triphenylphosphonium cation reacts 5–7 times faster
with EGBs from 3 and 4. In contrast the triphenylphosphonium
Solvents and electrolytes. The solutions of tetrabutyl-
ammonium hexafluorophosphate (0.1 ) in dimethyl sulfoxide
(Fluka, puriss.) were passed through a column filled with neu-
tral alumina (Woelm, W200) before the measurements were
made. [2H6]DMSO (Fluka, puriss.) solutions of phosphonium
salts were used to record 1H and 13C NMR spectra.
Electrodes, cells and instrumentation. For the kinetic meas-
urements, the electrodes, cells and the electrochemical instru-
mentation, as well as the measurement and data handling
procedures for derivative cyclic voltammetry and linear sweep
voltammetry were identical to those previously described.28
NMR spectroscopy. The NMR spectra were recorded on a
Varian Unity Plus spectrometer. Both 1H NMR (300 MHz) and
13C NMR (75 MHz) spectra used the solvent peak as an
internal reference.
Digital simulations. The theoretical DCV response for the
DISP1 mechanism, Scheme 3, the simple EC mechanism
(method 2) and the full EEC mechanism in Scheme 2 (method
1) was obtained by digital simulation using the fully implicit
method described by Rudolph29 using locally developed soft-
ware as described elsewhere.30 For the EEC mechanism two
different sets of theoretical data were generated, one using
Eo(1) Ϫ Eo(2) = 0.72 V corresponding to probase 2, and one
using Eo(1) Ϫ Eo(2) = 0.52 V corresponding to probase 3.
cation 1c is about two orders of magnitude more reactive
Ϫ
towards EGB 5 than its tributyl counterpart 1d. A possible
᝽
explanation is that where enolisation is precluded, as in 1c and
1d, the triphenylphosphonium group is the more activating for
proton transfer from the adjacent methylene group. Zhang and
Bordwell1 attribute the greater anion-stabilising effect of Ph3Pϩ
cf. Me3Nϩ to a combination of inductive and polarisability
effects; similarly Bu3Pϩ is less polarisable than Ph3Pϩ. But for
the enolisable salts 1f, and 1g the picture is complicated by
possibilities for fast proton transfer from the enols, the contents
of which may differ between the two series.
In the case of 1g participation of the six equivalent protons
in the butyl groups to the proton transfer can be ruled out on
the basis of their expected acidity compared to the doubly acti-
vated Bu3PϩCH2Y protons. In the case of 1d participation of
the butyl protons cannot be ruled out since for 1c, Ph3Pϩ-
CH2Ph, and 1b, Ph3PϩCH2CH2CH2CH3, the difference in
proton transfer rate is less than an order of magnitude for all
Acknowledgements
The collaboration was supported under the EU Human Capital
and Mobility Programme No. ERBCHRXCT 920073.
applied EGBs but only a factor of 1.1 for 5 Ϫ which is the only
᝽
EGB deprotonating 1d on the timescale of the measurements.
Conclusions
References
Kinetic acidities of phosphonium cations defined as the second
order rate constant for proton transfer to electrogenerated
bases (EGBs) can conveniently be measured by electroanalyti-
cal techniques (derivative cyclic voltammetry or linear sweep
voltammetry) in a very broad range of rate constants (6–7
orders of magnitude). Two structurally different types of elec-
trogenerated bases with comparable thermodynamic basicities
but with different kinetic basicities have been used to character-
ise the kinetic acidities of the series of phosphonium salts. Since
some of the phosphonium cations are subject to enolisation a
single Brønsted relationship (logk[proton transfer] vs. pK) does
not hold for the series of phosphonium cations even for a single
common EGB. In addition, for certain combinations of phos-
phonium cations and EGBs the proton transfer reaction
appears to be subject to steric hindrance. Chemical shifts (13C
and 1H) and reduction potentials of the phosphonium salts are,
as measures of electron-demand, consistent with the few
reported pK(DMSO) values. The kinetic results also confirm
that the Ph3Pϩ group is, for non-enolisable phosphonium ions,
more activating than Bu3Pϩ.
1 X.-M. Zhang and F. G. Bordwell, J. Am. Chem. Soc., 1994, 116, 968.
2 J. H. P. Utley, Topics in Current Chemistry, Springer-Verlag,
Heidelberg, 1987, vol. 142, p. 133.
3 S. Ling-Chung, K. D. Sales and J. H. P. Utley, J. Chem. Soc., Chem.
Commun., 1990, 662.
4 J. E. Crooks, in Proton Transfer Reactions, ed. E. Caldin and
V. Gold, Chapman and Hall, London, 1975, ch. 6.
5 P. J. Kinlen, R. D. Goodin and F. E. Woodard, J. Electroanal.
Chem., 1985, 190, 185.
6 M. F. Nielsen and O. Hammerich, Acta Chem. Scand., 1992, 46, 883.
7 F. G. Bordwell, Acc. Chem. Res., 1988, 21, 456.
8 M. F. Nielsen, H. Eggert and O. Hammerich, Acta Chem. Scand.,
45, 1991, 292.
9 J. M. Savéant and S. K. Binh, Bull. Soc. Chim. Fr., 1972, 3549;
J. M. Savéant and S. K. Binh, J. Org. Chem., 1977, 42, 1242.
10 J. H. P. Utley and A. Webber, J. Chem. Soc., Perkin Trans. 1, 1980,
1154.
11 M. F. Nielsen and O. Hammerich, Acta Chem. Scand., 1987, B41,
668.
12 F. Maran, A. Goldblum, F. D’Angeli and E. Vianello, Electrochim.
Acta, 1988, 33, 227.
13 M. F. Nielsen, O. Hammerich and V. D. Parker, Acta Chem. Scand.,
1987, B41, 50.
14 C. Amatore, M. Gareil and J. M. Savéant, J. Electroanal. Chem.,
147, 1983, 1 and refs. therein.
15 S. K. Ling-Chung, P. J. I. Runciman, K. D. Sales and J. H. P. Utley,
J. Electroanal. Chem., 1988, 250, 373.
Experimental
Chemicals
16 V. D. Parker, in Topics in Organic Electrochemistry, ed. A. J. Fry and
W. E. Britton, Plenum Press, New York, 1986, ch. 2.
17 L. Nadjo and J. M. Savéant, J. Electroanal. Chem., 1973, 48, 113.
18 M. F. Nielsen and O. Hammerich, unpublished results.
19 M. F. Nielsen, J. H. P. Utley and H. Viertler, unpublished results.
Probases. Phenazine (Fluka, purum) and azobenzene (Merck,
para sintesis) were used as received. The dicyano(fluoren-9-
ylidene)methane and 2-cyano-2-(fluoren-9-ylidene)ethyl acetate
were synthesised by Knoevenagel condensation from fluor-
J. Chem. Soc., Perkin Trans. 2, 1998
521

View Article Online
20 J. H. P. Utley and S. K. Ling-Chung, Electrochim. Acta., 1997, 42,
2109.
21 D. T. Sawyer and R. Y. Koami, Anal. Chem., 1972, 44, 715.
22 S. Cheng and M. D. Hawley, J. Org. Chem., 1985, 50, 3388.
23 F. G. Bordwell and H. E. Fried, J. Org. Chem., 1981, 46, 4237.
24 V. D. Parker, M. Tilset and O. Hammerich, J. Am. Chem. Soc., 1987,
109, 7905.
28 M. F. Nielsen, S.Aa. Laursen and O. Hammerich, Acta Chem.
Scand., 1990, 44, 392; M. F. Nielsen, O. Hammerich and
V. D. Parker, Acta Chem. Scand., Ser. B, 1986, 40, 101.
29 M. Rudolph, J. Electroanal. Chem., 1991, 314, 13.
30 M. L. Andersen, M. F. Nielsen and O. Hammerich, Acta Chem.
Scand., 1995, 49, 503.
25 D. F. Ewing, in Correlation Analysis in Chemistry, ed. N. B.
Chapman and J. Shorter, Plenum Press, New York, 1978, ch. 8.
26 S. M. McElvain and D. H. Clemens, in Organic Synthesis, ed.
N. Rabjohn, Wiley, New York, 1963, vol. 4, p. 463; E. J. Cragoe,
C. M. Robb and J. M. Sprague, J. Org. Chem., 1950, 15, 381.
27 A. W. Johnson, in Organic Chemistry—a series of monographs, ed.
A. T. Blomquist, Academic Press, New York, 2nd edn., 1968, vol. 7,
ch. 3.
Paper 7/08086G
Received 10th November 1997
Accepted 17th December 1997
522
J. Chem. Soc., Perkin Trans. 2, 1998