The kinetics and mechanism of the phosphorus-catalysed dimerisation of acrylonitrile

Article abstract of DOI:10.1039/a707559f

Isopropyl diarylphosphinites (Ar₂POPrⁱ) catalyse the dimerisation of acrylonitrile (AN) to a mixture of cis- and trans-1,4-dicyanobut-1-ene (cis,trans-DCB-1), trans-1,4-dicyanobut-2-ene (DCB-2) and 2,4-dicyanobut-1-ene (MGN). The kinetics and mechanism of the reaction, which is a potential source of hexamethylenediamine, are reported in detail and the factors which govern rate and selectivity to DCB-1 and DCB-2 rather than MGN are elaborated.

Full text of DOI:10.1039/a707559f

View Article Online / Journal Homepage / Table of Contents for this issue
The kinetics and mechanism of the phosphorus-catalysed
dimerisation of acrylonitrile
C. Dennis Hall,*,a Nicholas Lowther,a Bruce R. Tweedy,a Adam C. Halla and
Gordon Shawb
a Dept. of Chemistry, King’s College, Strand, London, UK WC2R 2LS
b ICI, Chemicals and Polymers, PO Box 90, Wilton, Middlesbrough, Cleveland, UK TS6 8JE
Isopropyl diarylphosphinites (Ar2POPri) catalyse the dimerisation of acrylonitrile (AN) to a mixture
of cis- and trans-1,4-dicyanobut-1-ene (cis,trans-DCB-1), trans-1,4-dicyanobut-2-ene (DCB-2) and 2,4-
dicyanobut-1-ene (MGN). The kinetics and mechanism of the reaction, which is a potential source of
hexamethylenediamine, are reported in detail and the factors which govern rate and selectivity to DCB-1
and DCB-2 rather than MGN are elaborated.
London. Some distillations were carried out using a Buchi
GKR-50 Kugelrohr apparatus in which case the boiling points
reflect the oven temperature rather than the true values.
Introduction
The dimerisation of acrylonitrile (AN, 1) gives either 2,4-
dicyanobut-1-ene (MGN, 2) or cis- and trans-1,4-dicyanobut-
1-ene (DCB, 3) as the principle product and hydrogenation of
Solvents
All solvents were dried by standard methods,5 distilled and
stored under nitrogen over molecular sieves prior to use.
CH2CH2CN
CH2CH2CN
CH CHCN
ArnP(OR)3–n
n = 0–3
+
2CH2 CHCN
CH2 CCN
4
Tricoordinated phosphorus compounds
1
2
3
All the tricoordinated compounds used throughout this study
were obtained or prepared and characterised by methods
described in earlier publications.6–9
the latter, via a two stage process, leads to hexamethylene-
diamine, a vital intermediate en route to Nylon 66. Thus the
selective, catalysed dimerisation of AN to DCB is potentially a
very important process which has been achieved using a variety
of tricoordinate organophosphorus compounds (4) as homo-
geneous catalysts.1–4
The successful exploitation of the dimerisation reaction as a
commercial process, however, depends upon a knowledge of the
optimum combination of rate, selectivity and turnover for the
catalyst system. This paper describes structural and kinetic
studies which elucidate the mechanism of the reaction and
hence define the operational window for maximum catalytic
efficiency.
Preparation of 2-cyanoethyldiphenylphosphine
Two methods were used successfully.
(a) Phenylsilane reduction of 2-cyanoethyldiphenylphosphine
oxide. Phenylsilane (0.38 g, 3.5 mmol) was added to 2-cyano-
ethyldiphenylphosphine oxide (2.03 g, 8.0 mmol). The reaction
mixture was maintained under reflux at 100 ЊC for 2 h. After
cooling, the residue was distilled (Kugelrohr) to give 1.10 g
(57%) of a colourless liquid (bp 170–175 ЊC, 0.3 mmHg) which
crystallised on standing to yield a white solid of mp 63–64 ЊC
(lit.,10 64–64.5 ЊC). δP(C6D6) Ϫ16.1, δH(CDCl3) 2.38 (m, 4H) and
7.40 (m, 10H).
(b) The condensation of diphenylphosphine with acrylonitrile.
Diphenylphosphine (9.9 g, 53 mmol) and acrylonitrile (3.6 g,
71 mmol) were introduced to a tube cooled to Ϫ78 ЊC under
an atmosphere of nitrogen. The tube was sealed and heated
to 130 ЊC for 7.5 h. The excess volatiles were removed from
the reaction mixture under reduced pressure to leave a dark
oil which semi-crystallised on standing. This was distilled
(Kugelrohr) to give a colourless liquid (bp 170–180 ЊC, 0.19
mmHg) which crystallised on standing to yield 8.90 g (70%) of
a white solid of mp 63–64 ЊC (lit.,11 64–64.5 ЊC). δP(CDCl3)
Ϫ16.2, δH(CDCl3) 2.39 (m, 4H) and 7.40 (m, 10H).
Experimental
Instrumentation
2
1
31P, 13C, 19F, H and high resolution H NMR spectra were
obtained using a number of Fourier-transform multinuclear
spectrometers (Bruker WM 250, Bruker AMX 400 and JEOL
JNM-FX 200) operating at 250, 400 and 200 MHz (1H)
respectively.
Analytical GC was effected using several chromatographs all
equipped with flame ionisation detectors. At ICI (Chemical and
Polymers), three instruments were used: a Pye-Unicam series
104 GC, equipped with FID, a PD10 GC computer and tele-
type printer or two internally constructed GCs with integrators
linked to a PDP-11 GC, minicomputer and a 32k Commodore
PET computer and printer. One instrument was fitted with
carrier gas backflush system to remove high boiling fractions.
At King’s College a Perkin-Elmer model 8310 GC equipped
with a data handling station was used in conjunction with a
Perkin-Elmer 8300 autosampler and printer. An in-line carrier
gas purification system comprising a 50 cm by 3 cm copper
column packed with reduced BASF R3-11 oxygen-removing
catalyst and an Alltech 4004 indicating oxytrap was used. Pre-
parative GC was effected using a Pye-Unicam series 105 Mk. 2
automatic preparative chromatograph. Mass spectra were
obtained by the ULIRS service at the School of Pharmacy,
Preparation of 2-cyanoethyldiphenylphosphine oxide
Acrylonitrile (6.64 g, 0.125 mol) was added, dropwise with stir-
ring under nitrogen, to a cooled (0 ЊC) solution of diphenyl-
phosphine oxide (21.80 g, 0.108 mol) and potassium hydroxide
(10  aqueous solution, 8 ml) in acetonitrile (30 ml). Stirring
was continued for a further 2.5 h, during which time the reac-
tion vessel was allowed to warm to ambient temperature and
the mixture was then allowed to stand at ambient temperature
for a further 48 h. The reaction mixture was washed with potas-
sium hydroxide solution (10 , 3 × 10 ml) water (3 × 20 ml)
and the organic layer was separated and dried (MgSO4). The
volatiles were removed, under reduced pressure, to leave a
brown oil which was triturated with diethyl ether (250 ml). The
J. Chem. Soc., Perkin Trans. 2, 1998
2047

View Article Online
resultant white, crystalline solid was collected and recrystallised
from benzene–hexane to yield 25.2 g (90%) of the product as
fine white crystals, mp 101–102 ЊC (lit.,12 102–103 ЊC). δP(C6D6)
ϩ25.8, δH(CDCl3) 2.74 (m, 4H) and 7.70 (m, 10H).
Acrylonitrile
Commercial acrylonitrile (Aldrich) was dried (×2) over crushed
calcium hydride, degassed and distilled onto freshly activated
3 Å molecular sieves under an atmosphere of nitrogen.
Kinetic procedures
1
2
GC and high resolution H and H NMR were used as the
analytical tools in the evaluation of the kinetic parameters k1,
K2 and k8/k5.
The monitoring of [AN], [MGN], [DCB] and [trimer] at King’s
College London
Toluene (20 ml), AN (6 ml, 9 × 10Ϫ2 g mol) and isopropanol
(IPA, 2 ml, 2.66 × 10Ϫ2 g mol) were introduced to a glass vessel
which was sealed (Subaseal) under an atmosphere of nitrogen.
The flask was thermostatted at 60 ЊC in an oil bath for 15 min.
An aliquot of phosphinite catalyst (2 mmol, ca. 500 µl) was
introduced to the thermostatted reaction mixture using a gas-
tight microsyringe with shaking to effect complete solution.
Aliquots of the reaction mixture (500 µl) were removed period-
ically and were added to solutions of 0.01  HCl (6 µl) in acet-
one (300 µl) containing dimethyl phthalate (10 µl) in GC micro-
vials. The vials were sealed with crimped caps and were shaken
throughly. This process deactivated the catalyst by solvolysis
and introduced a ‘low boiler’ internal standard (acetone) and a
‘high boiler’ internal standard (dimethyl phthalate) to each
sample. These samples were analysed on a Perkin-Elmer 8310
Fig. 1 A typical gas chromatogram of the AN dimerisation mixture
1
–
using a 2.5 cm × Љ od stainless steel (ss) column of 7.5% LAC
8
2R-446 on Chromosorb WH-P treated with 0.5% H3PO4. The
primary carrier gas pressure was 30 psi. The injector block
temperature was 180 ЊC and the oven temperature was 190 ЊC.
Peaks were eluted in the order: solvent, MGN, cis-DCB-1,
dimethyl phthalate, and trans-DCB-1 (plus DCB-2). Each
analysis was performed in triplicate to check for consistency.
With the aid of the internal standard (dimethyl phthalate) the
total product distribution was quantified enabling percentage
selectivities of each product to be calculated. Oligomer quan-
tities were assessed on the basis of ‘non-GC accountables’, i.e.
by a weight difference method.
GC with an integral data handling station linked to a Perkin-
1
–
Elmer 8300 autosampler. A 1 m × 4Љ od glass column packed
with Ultrabond 20 M 100/120 mesh was used. The carrier gas
flow rate was 9 ml minϪ1. The temperature program was as
follows: 40 ЊC isothermal (8 min) followed by a ramp (10 ЊC
minϪ1) to 130 ЊC with a further isothermal period (15 min) fol-
lowed by a second ramp (5 ЊC minϪ1) to 240 ЊC with a final
isothermal period of 55 min. The injector block temperature
was 270 ЊC and the detector temperature was 300 ЊC. Peaks
were eluted in the order: acetone, IPA, AN, toluene, MGN, cis-
DCB-1, dimethyl phthalate, trans-DCB-1, trimer, solvolysed
catalyst and a typical chromatogram is displayed in Fig. 1. The
amount of DCB-2 formed in the reaction was small (generally
<2%) and with this GC method was only detectable as a
shoulder (at ca. 19.5 min) on the peak due to trans-DCB-1.
With the aid of the two internal standards the raw GC data
were converted into [AN], [MGN], [DCB] and [trimer] vs. time
data.
The monitoring of [AN] at ICI Chemicals and Polymers
Toluene (10 ml), AN (3 ml) and IPA (1 ml) together with
n-heptane (1 ml) were loaded into a glass vessel which was
sealed under nitrogen and the flask was thermostatted at 60 ЊC
for 15 min. An aliquot of phosphinite catalyst (1 mmol, ca. 260
µl) was introduced to the reaction mixture using a gas-tight
microsyringe with shaking to effect complete solution. Small
aliquots (ca. 1–2 µl) were extracted periodically from the reac-
The analysis of the total product distribution at ICI Chemicals
and Polymers
1
–
tion mixture and were analysed by GC using a 2.5 m × Љ od ss
8
column of silicon OV-225 with a pre-column of 7.5% LAC
2R-446 on Chromosorb WH-P treated with 0.5% H3PO4. A
carrier gas backflush mechanism was fitted to the system to
remove higher boilers. The primary carrier gas pressure was 7
psi, the injector block temperature was 70 ЊC and the oven tem-
perature was 80 ЊC. Peaks were eluted in the order: n-heptane,
IPA, AN, toluene. The n-heptane was employed as an internal
standard and the raw GC data were converted into [AN] vs.
time data.
Toluene (10 ml), AN (3 ml) and IPA (1 ml) were loaded into a
glass vessel and were sealed under nitrogen. The flask was
thermostatted at 60 ЊC for 15 min. An aliquot of phosphinite
catalyst (1 mmol, ca. 250 µl) was introduced to the thermo-
statted reaction mixture using a gas-tight microsyringe with
shaking to effect complete solution. At the required level of AN
conversion the reaction was ‘killed’ by the addition of water
(1 ml) and methanol (1 ml) which deactivated the catalyst by
solvolysis. The resulting mixture was passed through a pre-
weighed sintered filter in order to collect any crystalline hex-
amer. The sinter was dried to constant weight. The low boiling
materials were removed from the filtrate under reduced pressure
(ca. 20 mm) at temperatures of 30–50 ЊC. There remained an
oily residue containing dimeric species, low molecular weight
oligomers and ‘dead’ catalyst; the weight of this residue was
recorded. A weighed aliquot (ca. 0.1 g) of the ‘oils’ was added
to a weighed quantity of dimethyl phthalate (ca. 0.1 g) in
acetone (ca. 10 ml). The resulting sample was analysed by GC
The monitoring of [MGN] and [DCB] at ICI Chemicals and
Polymers
Toluene (10 ml), AN (3 ml) and IPA (1 ml) together with
adiponitrile (0.2 ml) were loaded into a glass vessel which was
sealed under nitrogen. The flask was thermostatted at 60 ЊC for
15 min and an aliquot of phosphinite catalyst (1 mmol, ca. 250
µl) was introduced using a gas-tight microsyringe with shaking
to ensure complete solution. Small aliquots (ca. 1–2 µl) were
2048
J. Chem. Soc., Perkin Trans. 2, 1998

View Article Online
periodically extracted from the reaction mixture and were
1
–
analysed on a Pye-Unicam series 104 GC using a 1.5 m × 4Љ od
glass column containing 7.5% LAC 2R-446 Chromosorb
WH-P treated with 0.5% H3PO4. The primary carrier gas pres-
sure was 16 psi, the injector block temperature was 170 ЊC
and the oven temperature was 185 ЊC. Peaks were eluted in the
order: solvent, MGN, cis-DCB-1, adiponitrile, trans-DCB-1
(ϩ DCB-2). The adiponitrile was employed as an internal
standard for quantitation and the raw GC data were converted
into [MGN] and [DCB] vs. time data.
The incorporation of deuterium into unreacted AN
Toluene (10 ml), AN (3 ml) and IPA-OD (1 ml or in preliminary
experiments, 4 ml) were loaded into a glass vessel and sealed
under nitrogen. The flask was thermostatted at 60 ЊC for 15 mn
and an aliquot of phosphinite catalyst (1 mmol, ca. 250 µl)
was introduced using a gas-tight microsyringe with shaking to
ensure complete solution. A number of reactions were set up
under these conditions and at different percentage conversions
of AN the reactions were ‘killed’ by the addition of water (1 ml)
and methanol (1 ml). Each reaction mixture was filtered to
remove the crystalline hexamer by-product. The low boiling
materials were removed from the filtrates under reduced pres-
sure (ca. 20 mm) and were collected in a cold (Ϫ78 ЊC) trap.
Both the low boiling and the high boiling fractions from each
reaction were separated on a Pye-Unicam series 105 Mark 2
Fig. 2 A plot of [trimer] vs. % conversion; 10:3:1 (v/v) ratio of
toluene:AN:IPA, at 333 K; [catalyst] = 0.037 
_
ROH k2
k1
_
+
+
AN
+
P
PCHCH2CN
PCH2CHCN
k
ROH
k–1
–2
B1
Y
P = ArnP(OPri)3–n
automatic preparative chromatograph prepared with a 1.5 m ×
4Љ od glass column containing 7.5% LAC 2R-446 on Chromo-
1
–
AN, k8
AN, k5
n = 1, 2
sorb WH-P treated with 0.5% H3PO4. For the low boiling
fractions the oven temperature was 70 ЊC and the primary
carrier gas pressure was 8 psi. For the high boiling fractions the
oven temperature was 190 ЊC and the primary carrier gas pres-
sure was 20 psi. Using these conditions the partially deuteriated
unreacted AN was collected from each reaction and assessed by
high resolution 1H and 2H NMR spectroscopy.
+
+
PCH2CHCN
PCHCH2CN
_
_
CH2CHCN
B2
CH2CHCN
B3
The incorporation of hydrogen into unreacted deuterio-
acrylonitrile13
ROH, k6
ROH, k9
Deuteriotoluene (0.5 ml), deuterioacrylonitrile (0.15 ml) and
IPA (0.05 ml) were loaded into a 5 mm NMR tube and the
mixture was thermostatted at 60 ЊC for 15 min. An aliquot of
phosphinite catalyst (0.05 mmol, ca. 13 µl) was introduced using
a gas-tight microsyringe with shaking to ensure complete solu-
_
_
+
+
PCHCHCN
PCH2CCN
CH2CH2CN
B4
–P
MGN
1
CH2CH2CN
B5
tion. The reaction was monitored by high field (250 MHz) H
NMR in the thermostatted (60 ЊC) probe of a Bruker WM250
instrument. The levels of protonation in the unreacted AN were
assessed using the methyl resonance signals in the IPA as an
internal standard.
-P
DCB-1
DCB-2
Results and discussion
Scheme 1
The mechanistic pathway proposed for the formation of dimers
2 and 3 is shown in Scheme 1. In addition the reaction affords a
number of by-products including 1,4-dicyanobut-2-ene, a var-
iety of trimeric species and a crystalline hexamer (CNCH2-
Scheme 1). Thus a plot of [trimer] vs. % conversion of AN gave
an S-shaped curve (Fig. 2) consistent with consecutive form-
ation of the trimer from dimers and AN.
A similar plot was observed for the formation of hexamer
again indicating a consecutive process. Deliberate loading of
mixtures of AN and the individual dimers gave relative rate
data for the formation of hexamer and trimer under dimeris-
ation conditions (Table 1). Thus one may envisage formation of
hexamer, trimer and higher oligomers from DCB-1, DCB-2 and
MGN via the routes shown in Scheme 2, where the base (B) is
probably the initial betaine B1 or the ylide, Y.
CH ) C(CN)CH᎐CHC(CN)(CH CH CN) in quantities depen-
᎐
2
2
2
2
2
dent upon the reaction conditions and the catalyst used.
Although no betaine (B) or ylide (Y) intermediates were
observed by NMR, the intermediacy of the ylide species was
established by trapping the latter with benzaldehyde.1b The
kinetics of the reaction were studied by monitoring the rate of
disappearance of AN and the rate of appearance of DCB or
MGN by gas-liquid chromatography (GLC) on a capillary
column which was also capable of analysing the reaction mix-
tures quantitatively for trimers. Any crystalline hexamer formed
was filtered off and estimated gravimetrically. Independent
experiments showed that the hexamer was derived from
either DCB-1 or DCB-2 and that the trimers (and oligomers)
were derived in a consecutive fashion from dimers and AN
rather than in a parallel manner from B2 or B3 plus AN (see
4AN
–
DCB-1 ϩ Base (B)
NCCH2CHCH᎐CHCN 
Hexamer
or
᎐


AN
–

DCB-2 ϩ B
NCCH2CH᎐CHCHCN
Trimer

᎐
AN
–
MGN ϩ B
CH2᎐᎐C(CN)CH2CHCN
Trimer etc.
Scheme 2
J. Chem. Soc., Perkin Trans. 2, 1998
2049

View Article Online
Table 1 Relative rates of formation of the hexamer and trimers from
Table 3 Average k1, K2k8/k5, k8/k5 and K2 values for a series of
phosphinite catalysts (XC6H4) (XЈC6H4)POPri a
DCB-2, DCB-1 and MGN under standard dimerisation conditionsa
Relative rates under dimerisation
conditions
k1/10Ϫ4
Catalyst
K2k8/k5 K2
k8/k5
1 molϪ1 sϪ1
Σσ
Hexamer
formation
X = XЈ = p-F
23.1
19.0
15.3
16.7
2.29 10.1
0.64
2.57
4.85
5.34
8.16
7.20
77.1
3.60
0.120
0
Dimer
Trimer formation
X = XЈ = H
2.02
1.55
9.4
9.9
X = XЈ = m-Me
X = H, XЈ = p-Me
Ϫ0.138
Ϫ0.170
Ϫ0.268
Ϫ0.290
Ϫ0.700
—
trans-DCB-2
cis-DCB-1
trans-DCB-1
MGN
12
5
1
7
4.5
1
1.65 10.1
1.35 10.5
X = H, XЈ = p-MeO 14.2
X = H, XЈ = p-PriO
X = H, XЈ = p-Et2N
X = XЈ = o-MeO
13.8
7.58
1.32
1.47
1.01
1.37
9.4
7.5
0.96
0
4
a T 60 ЊC. In 10:3:1 (v/v) toluene–AN–IPA; catalyst, Ph2POPri.
a T 60 ЊC. In 10:3:1 (v/v) toluene–AN–IPA.
Table 2 kobs vs. [PIII] obtained using [AN], f [MGN] and f [DCB] dataa
kobs/10Ϫ5 sϪ1 obtained using
[Catalyst]
K2k8/k5
[AN]
f [MGN]
f [DCB]
0.0426
0.0515
0.0618
0.0741
0.0754
0.124
18.9
20.6
18.9
18.8
20.8
19.9
18.9
1.06
—
1.52
1.71
2.18
3.23
3.72
0.90
1.28
1.67
1.74
1.91
3.19
3.57
0.90
1.28
1.67
1.75
1.92
3.19
3.57
0.146
a T 60 ЊC. In 10:3:1 (v/v) toluene–AN–IPA; catalyst, Ph2POPri.
Fig. 4 Plot of kobs vs. [phosphinite]; 10:3:1 (v/v) ratio of toluene:
AN:IPA, at 333 K; catalyst Ph2POPri
ively, should also give slopes = 2k1[P]. In fact it was found that,
provided [AN], [MGN] and [MCB] were all corrected for hex-
amer and trimer formation, the logarithmic plots vs. time all
gave excellent linear correlations (e.g. Fig. 3 for AN data) and
the resultant values for kobs using a range of concentrations of
Ph2POPri as catalyst are shown in Table 2. A plot of the average
kobs values from reactant product monitors vs. catalyst concen-
tration [P] gave a linear correlation (Fig. 4) with slope = 2k1.
Thus the kinetics of dimerisation were shown to be first order in
AN and catalyst.
The value of K2k8/k5 was taken as the ratio of DCB–MGN
found in the product and is seen (Table 2) to be independent of
the concentration of catalyst used but dependent upon the
nature of the catalyst (Table 3). Table 3 shows quite clearly that
as the reactivity of the catalyst increases, from X = XЈ = p-F to
X = H, XЈ = p-NEt2, the selectivity to DCB decreases by a
factor of ca. 3. A plot of log K2k8/k5 vs. σ gave a good linear
correlation (r = 0.992) with ρ = ϩ0.58 showing that the selectiv-
ity to DCB is increased by electron withdrawal in the aryl
groups of the phosphinite catalyst. Since the experimental ρ
value is a composite of K2 (ϩρ), k8 (Ϫρ) and k5 (probably a
small Ϫρ value) it is apparent that the selectivity is dominated
by the equilibrium between the betaine B1, and the ylide Y, of
Scheme 1.
Fig. 3 A typical plot of ln[AN] vs. t (s)
Having established the rates of formation of by-products
under any set of dimerisation conditions it was possible to cor-
rect concentrations of AN, DCB or MGN observed in the reac-
tion mixtures at any time, t, for the formation of hexamer or
trimers. Assuming k5[AN] and k8K2[AN] of Scheme 1 are both
ӷkϪ1 (justified by deuterium exchange studies, vide infra) the
SSO may be applied to B1 and Y from which eqn. (1) may be
Ϫd(AN)/dt = 2k1 [P][AN]
(1)
derived. Thus a plot of ln [AN] vs. t should give a slope = 2k1[P]
where [P] is the constant phosphorus catalyst concentration.
Furthermore, for the formation of MGN and DCB the appro-
priate integrated rate expressions (see Appendix) are given by
eqns. (2) and (3).
ln {[AN]0 Ϫ 2[MGN](1 ϩ k8K2/k5)} = 2k1[P]t
ln {[AN]0 Ϫ 2[DCB](1 ϩ k5/k8K2)} = 2k1[P]t
(2)
(3)
An attempt was made to evaluate K2 independently from the
Thus plots of ln f [MGN] vs. t and ln f [DCB] vs. t where
f [MGN] and f [DCB] are given by eqns. (2) and (3), respect-
ratio of HA (or HB) to HX (Scheme 3) incorporated into un-
2050
J. Chem. Soc., Perkin Trans. 2, 1998

View Article Online
Table 4 The incorporation of hydrogen into unreacted [2H3]AN vs. %
Table 6 Deuterium uptake in cis-DCB-1 by 2H NMRa
conversiona
% D incorporated
% AN
conversion
% H incorporated
% AN
conversion
HA
HX
HC
HD
HA/HX
HA
HB
HX
¹(HA ϩ HB)/HX
¯
²
5.0
11.6
16.8
17.3
18.1
28.2
40.2
44.2
45.9
58.3
51.4
47.3
42.4
35.3
27.0
25.2
29.2
37.1
32.2
32.6
27.0
23.7
17.0
14.7
3.5
3.9
4.0
3.2
3.1
2.9
4.0
2.6
20.2
26.4
24.5
20.0
17.8
15.1
12.5
11.0
1.57
1.57
1.60
1.45
1.57
1.49
1.59
1.71
2.2
3.8
8.1
11.9
14.6
19.4
23.7
28.6
32.4
38.9
0.21
0.43
0.51
1.13
1.33
1.67
2.47
2.27
3.17
3.26
0.20
0.39
0.67
1.16
1.38
1.65
2.42
2.27
3.03
3.41
0.13
0.27
0.30
0.45
0.60
0.50
1.27
1.37
1.78
2.17
1.62
1.56
1.97
2.54
2.25
3.32
1.93
1.66
1.75
1.62
a T 60 ЊC. In 10:3:1 (v/v) toluene–AN–IPA-OD; catalyst Ph2POPri
(0.073 ).
a T 60 ЊC. In 10:3:1 (v/v) toluene–[2H3]AN–IPA; catalyst Ph2POPri
(0.073 ).
incorporated vs. % conversion of AN was linear with a slope
of ca. 2.5 indicating that all the available D (or H) would be
exchanged at ca. 40% AN conversion. Thereafter the species in
solution could be expected to equilibrate slowly with H/D avail-
able in solution. The analysis generated the data in Table 5 and
Table 5 Incorporation of deuterium into unreacted AN vs. % conver-
sion of ANa
2
it was checked by H NMR (at 30.6 MHz) of the partially
% D incorporated
%
Conversion
deuterated AN which gave three distinct signals for DA, DB and
DX integrating for the same ratio of DA/DX or DB/DX as
HA
HB
HX
¹(HA ϩ HB)/HX
¯
²
1
observed from the H data. The experiment was repeated in
5.0
11.6
16.8
17.3
28.2
40.2
0.63
0.74
1.45
1.60
1.71
2.36
0.79
0.92
1.53
1.50
1.60
2.66
0.23
0.43
0.68
0.63
0.61
0.89
3.09
1.93
2.19
2.58
1.64
2.82
reverse by observing the incorporation of 1H into [2H3]AN
which gave the data recorded in Table 4. Although there is some
scatter, it is clear that the two sets of data agree in terms of the
ratio of exchanged HA/HX or HB/HX with an average value of
ca. 2 using Ph2POPri as catalyst. Incorporation of D (or H)
into sites A and B must occur through an ylide structure and
incorporation of D (or H) into site X together with the reverse
reaction to regenerate AN, must occur through a betaine struc-
ture. If the betaine–ylide equilibrium was fast and the H/D
exchange at all three sites was also fast relative to the reverse
reaction or the onward reaction to MGN or DCB then the
exchange observed within sites A, B and X of unreacted AN
would be the same and should merely reflect the amount of D
(or H) available from the alcohol solvent. In fact, using Ph2PO-
Pri as catalyst, the amount of H incorporated in sites A and B
of [2H3]AN is approximately twice that in site X. This observ-
ation might be ascribed to a deuterium isotope effect on the
reverse (elimination) step but such an argument is negated by
the fact that incorporation of D into unreacted AN also gives
more D in sites A and B than in site X whereas a deuterium
isotope effect on the elimination would have predicted the
opposite. Incidentally, this suggests that the elimination to
regenerate AN occurs via an E1cB mechanism with little or no
isotope effect (vide infra). An examination of the incorporation
of H (or D) in isolated cis- or trans-DCB-1 (13a or 13b) from
a T 60 ЊC. In 10:3:1 (v/v) toluene–AN–IPA-OD; catalyst Ph2POPri
(0.073 ).
DA
DB
DX
i (P)
DOPr
HA
HB
HX
CN
CN
exchanged via ylide
exchanged via betaine-1
Scheme 3
reacted [2H3]AN (Table 4) or the incorporation of D into un-
reacted AN (Table 5) over a range of conversions from 2 to
40%. K2 was found to be about 2 for Ph2POPri but this varied
with the nature of the catalyst (Table 3). The evaluation of K2 is
rationalised as follows. It is obvious that there are potentially
eight separate species (5–12) in the partially deuterated AN. All
HA
HB
HA
HB
HA
DB
DA
HB
CN
HX
CN
DX
CN
HX
CN
HX
C
D
HX
NC
HX
NC
HA
CH2CH2CN
C
D
HA
CH2CH2CN
5
8
6
7
13b
13a
[2H3]AN (or AN) gave a similar result. Using H NMR† the
average ratio of DA/DX up to 44% conversion was 1.5 in cis-
DCB-1 (Table 6) and 1.7 in trans-DCB-1 (Table 7). The average
ratio (1.60) is slightly lower than that observed for unreacted
AN (ca. 2) but this may be explained by a small deuterium
isotope effect on the final intra- or inter-molecular elimination
step to form DCB-1 (Scheme 4). The kH/kD value for this pro-
2
HA
DB
DA
DB
DA
DB
DA
HB
CN
DX
CN
DX
CN
DX
CN
HX
12
9
10
11
except 12 could be observed separately in the resolution
1
enhanced 400 MHz H NMR spectrum of unreacted AN. The
† The results from the incorporation of D into DCB derived from
[1H3]AN were particularly useful since with cis-DCB-1, four distinct 2H
signals were observed for DA, DX, DC and DD although with trans-
DCB-1 only three distinct signals were observed since DC and DD
overlapped.
level of deuterium incorporation into 6–11 could therefore be
assessed by integration and the level of deuterium incorporated
at all three sites (represented by 12) was estimated as one third
of the average of 9–11. A plot of % of theoretical D (or H)
J. Chem. Soc., Perkin Trans. 2, 1998
2051

View Article Online
Table 7 Deuterium intake in trans-DCB-1 by 2H NMRa
Table 8 Activation parameters for the dimerisation of ANa at [P] =
0.073 
% D incorporated
% AN
conversion
k1 (60 ЊC)/10Ϫ4
EA/kcal
∆S‡(298 K)/
HA
HX
(HC ϩ HD)b
HA/HX
Catalyst
1 molϪ1 sϪ1
molϪ1
cal molϪ1 KϪ1
5.0
11.6
17.3
18.1
28.2
40.2
44.2
39.1
42.4
44.5
26.8
30.4
22.6
22.8
23.8
26.4
26.5
15.4
18.0
12.4
11.4
13.0
15.4
11.4
8.0
8.9
7.1
1.64
1.61
1.68
1.74
1.69
1.82
2.00
(p-Me2NC6H4)2POPri
(p-MeOC6H4)2POPri
(p-MeC6H4)2POPri
Ph2POPri
2300
27.1
3.91
2.57
10.3
6.8
6.0
5.0
Ϫ33
Ϫ52
Ϫ56
Ϫ64
a In 10:3:1 toluene–AN–IPA (v/v).
6.3
a T 60 ЊC. In 10:3:1 (v/v) toluene–AN–IPA-OD; catalyst Ph2POPri
(0.073 ). b Not observable as separate signals.
_
+
+
PCHCHCN
PCHCH2CN
Intramolecular
_
CH2CH2CN
CH2CHCN
ROH
P
+
_
DCB-1
_
OR
+
+
PCHCHCN
PCHCH2CN
Intermolecular
CH2CH2CN
CH2CH2CN
Scheme 4
cess may be approximated by comparing the ratio of HA, HB/
HX for the reverse reaction with that of the elimination to DCB,
giving a kH/kD value for the latter process of ca. 1.3 (2/1.6)
consistent with an E1cB mechanism.
Fig. 5 Hammett plot of log k1 vs. Σσ
Thus we are faced with a situation where more H (or D) is
incorporated in sites A and B (via ylide) than in site X (via
betaine) and this suggests kinetic control of a process in which
the rate of exchange at all three sites is determined by the
relative concentrations of ylide and betaine. In other words
there is competition for the unstable betaine and ylide inter-
mediates between a reversal to starting materials (via the
betaine), H/D exchange with the alcohol and reaction with
more AN to form dimers. This is a reasonable hypothesis since
the alcohol concentration remains constant throughout and
kexchange values for ylide or betaine are likely to be very similar if
not identical. As pointed out by a referee however, this analysis
to extract K2 is clearly only valid if the equilibration of betaine
and ylide does not involve proton (or deuteron) transfer via
alcohol as depicted in Scheme 5a. An alternative which does
not involve proton transfer (Scheme 5b) is feasible but if
Scheme 5a (the expected pathway) is correct, the amounts of H
(or D) at sites A/B vs. X would only be an indication of the
relative concentrations of ylide and betaine an hence an indi-
cation of the trend in K2 as the substituent, X, was varied.
The values of K2 and k1 vary with the catalyst (Table 3) and
Hammett plots gave ρ-values (against σ) of Ϫ2.3 for k1 (Fig. 5,
r = 0.99) and ϩ0.4 ( 0.2) for K2 (r = 0.974). The k8/k5 ratio
however, remained fairly constant throughout except for the
o-MeO substituent (vide infra).
The data are entirely consistent with formation of the betaine
as the rate-limiting step of the reaction with K2 having a pro-
found influence on the selectivity to DCB as the substituents in
the aryl groups are changed. Thus increasing electron donation
by X increases the rate of reaction (negative ρ) by stabilising the
developing positive charge on phosphorus. Conversely, electron
withdrawing groups stabilise Y relative to B1 thus increasing K2
(positive ρ) and hence increasing the selectivity to DCB as the
rate of reaction decreases. The ratio k8/k5 remains essentially
constant throughout the range of meta- and para-substituents
but ortho-substituents in the phosphinite decrease the ratio of
k8/k5 by an order of magnitude. This accounts for the low select-
ivity found with ortho-substituents which probably decrease the
reactivity of the ylide by steric hindrance. The activation
parameters (Table 8) indicate an associative rate-limiting step
with the most reactive having the highest EA value (10.3 kcal
R
O
H
+
ROH
H
P C
H
_
_
+
+
δ–
δ–
δ–
PCH2CHCN
PCHCH2CN
CHCN
Scheme 5a
molϪ1) but the least negative ∆S‡ value (Ϫ33 cal molϪ1
K
Ϫ1).
The isokinetic temperature for the dimerisation reaction is
found to be ca. Ϫ100 ЊC. The effect of temperature on the ratio
K2k8/k5 was also investigated for a range of catalytic activity
(Table 9). In general, lower temperatures are associated with
higher selectivity so if we assume that the temperature coef-
ficients for k5 and k8 are equal, the equilibrium between betaine
and ylide (represented by K2) is an exothermic process with ∆HЊ
R
H
δ+
O
H
_
_
+
PCHCH2CN
+
δ–
C
δ–
CHCN
+
P
ROH
PCH2CHCN
H
Scheme 5b
2052
J. Chem. Soc., Perkin Trans. 2, 1998

View Article Online
Table 9 Effect of temperature on the ratio K2k8/k5 for a series of
catalysts in 10:3:1 C7H8–AN–IPA (v/v)
Table 11 kr for a series of catalyst XC6H4(XЈC6H4)POPri loaded at
0.073  in 10:3:1 (v/v) toluene–AN–IPA at 60 ЊC
K2k8/k5 at
kr/10Ϫ5 sϪ1 obtained using
Average kr/
Catalyst
333 K
314 K
283 K
Catalyst
HA
HB
HX
10Ϫ5 sϪ1
(p-Et2NC6H4)2POPri
(p-Et2NC6H4)PhPOPri
(p-MeC6H4)2POPri
Ph2POPri
2.9
9.7
14.0
18.9
3.4
11.8
17.3
22.9
4.1
14.2
21.3
28.5
X = XЈ = H
1.52
1.84
3.11
4.49
2.80
6.82
1.51
1.85
3.20
4.52
2.81
7.11
1.99
1.98
3.33
4.50
2.88
7.18
1.7
1.9
3.2
4.5
2.8
7.0
X = XЈ = m-Me
X = H, XЈ = p-Me
X = H, XЈ = p-MeO
X = H, XЈ = p-PriO
X = H, XЈ = p-Et2N
Table 10 kobs values from [AN], f [MGN] and f [DCB] data in different
solvent systemsa
the implication is that electron donation accelerates betaine
decomposition as well as betaine formation. The apparent
paradox is resolved by noting that the rate of the back reaction
is given by kϪ1[B] = kϪ1[Y]/K2. Thus the reverse reaction must be
dominated by the ylide/betaine interconversion since the posi-
tive ρ value for K2 (0.4 0.2) would then give a negative ρ value
for the reverse reaction.
In conclusion the data may be used to create a reactivity
profile embracing the rate of reaction (controlled by k1, tem-
perature and the medium) and the selectivity to DCB (con-
trolled by K2, the ratio k8/k5, temperature and the medium).
From this it may be seen that the catalyst which gives the most
convenient rate at 60 ЊC with a selectivity to DCB of ca. 92% is
isopropyl di-p-tolylphosphinite. Moreover, under the reported
experimental conditions, this catalyst gives a turnover in excess
of 5000.
kobs/10Ϫ5 sϪ1 obtained using
Solvent system (v/v)
toluene–AN–IPA–PNb
K2k8/k5
[AN]
f [MGN]
f [DCB]
10:3:1:0
18.9
19.4
20.0
20.8
21.3
1.71
1.07
0.86
0.61
0.52
1.75
1.15
0.92
0.83
0.52
1.75
1.15
0.92
0.85
0.52
10:2.5:1:0.64
10:2:1:1.28
10:1.5:1:1.92
10:1:1:2.26
a T 60 ЊC; catalyst Ph2POPri (0.073 ). b PN = propionitrile.
in the region of Ϫ2.0 kcal molϪ1. The composition of the reac-
tion mixture may also be expected to change both reaction rate
and selectivity.
This is illustrated in Table 10 where an attempt was made to
keep the relative permittivity of the medium constant by add-
ition of propionitrile. As the ratio of AN–IPA decreased, the
selectivity of DCB increased but the rate of reaction (as
denoted by k1) decreased.
Acknowledgements
We are grateful to the EPSRC and to ICI Chemicals and Poly-
mers, Wilton, Cleveland for CASE Awards to N. L. and B. R. T.
and a summer studentship to A. C. H. We also thank Jane
Hawkes and John Cobb for NMR spectra and the University of
London Intercollegiate Research Service for NMR (KCL) and
mass spectrometry (KCL and School of Pharmacy) data.
Thus any practical catalytic dimerisation system requires a
compromise between reactivity and selectivity as controlled by
the nature of the catalyst, the temperature window and the
composition of the reaction mixture in terms of the relative
amounts of AN and the proton transfer agent, IPA. It is
reasonable to assume that the pentacoordinate structure of
Scheme 6 might constitute an intermediate in the proton trans-
Appendix
Arn(RO)4–nPCH2CH2CN
Derivation of kinetic expressions
First consider the variation of [AN] vs. time. From Scheme 1,
ROH, k3
ROH, k4
we obtain eqn. (a). Hence we require expressions for [B1] and
Ϫd[AN]/dt =
ROH k2
ROH k-2
+
–
+–
Arn(RO)3–nPCHCH2CN
k1[P][AN] Ϫ kϪ1[B1] ϩ k5[B1][AN] ϩ k8[Y][AN] (a)
Arn(RO)3–nPCH2CHCN
Scheme 6
[Y]. In order to simplify the problem the equilibrium to P(v)
was ignored since it is a non-productive process. Thus eqns. (b)
fer process. In fact, this hypothesis is extremely unlikely since
the dialkoxyphosphorane of Scheme 6 (Ar = Ph, R = Pri and
n = 2) when prepared independently by the sulfenate ester
route6 failed to catalyse the dimerisation reaction under condi-
tions identical to those used with Ph2POPri. Thus the proton
transfer is mediated by alcohol, preferably secondary and prob-
ably via an intermediate (or TS) depicted in Scheme 5a or 5b.
It is clear from the data in Tables 4 and 5 that the first step of
the reaction (with rate coefficient k1) is reversible to a small
extent. At ca. 40% conversion of AN to dimer, about 7.5% of
the total available hydrogen (or deuterium) in sites HA, HB or
HX has been exchanged. The rate of exchange may then be used
to estimate an observed rate coefficient (kr) for the reverse reac-
tion and values of the latter parameter as derived from HA, HB
and HX are shown for a range of catalysts in Table 11.
d[B1]/dt = k1[P][AN] Ϫ kϪ1[B1] Ϫ k2[ROH][B1] ϩ
k
Ϫ2[ROH][Y] Ϫ k5[B1][AN] (b)
and (c) can be derived, while at the steady state eqn. (d) applies.
d[Y]/dt = k2[ROH][B1] Ϫ kϪ2[ROH][Y] Ϫ k8[Y][AN] (c)
d[B1]/dt = d[Y]/dt = 0
(d)
This is justified by a 31P NMR analysis of a dimerisation reac-
tion where no build up of phosphonium intermediates is
observed. Hence from eqns. (b) and (c) eqns. (e) and ( f ) can be
[Y] = K2*[B1], where
The magnitude of kr increases with electron donation by X
and a Hammett plot against σ gives ρ = Ϫ0.9 0.5. The corre-
lation coefficient (0.90) is poor and thus the data should be
regarded as a trend rather than a correlation. Nevertheless
K2* = k2[ROH]/(kϪ2[ROH] ϩ k8[AN]) (e)
[B1] = k1[P][AN]/(kϪ1 ϩ k5[AN] ϩ K2*k8[AN])
( f )
J. Chem. Soc., Perkin Trans. 2, 1998
2053

View Article Online
derived. Substitution for [B1] and [Y] in eqn. (a) gives eqn.
(g). From deuteriation studies we see that: kϪ1 Ӷ k5[AN],
f[MGN] = [AN]0 Ϫ 2[MGN](1 ϩ K2k8/k5)
ln f [MGN] = Ϫ2k1[P]t
(o)
(p)
Ϫd[AN]/dt = 2k1[P][AN](k5[AN] ϩ k8K2*[AN])/
(kϪ1 ϩ k5[AN] ϩ k8K2*[AN]) (g)
argument we obtain eqn. (q), where f [DCB] = [AN]0 Ϫ
ln f [DCB] = Ϫ2k1[P]t (q)
k8K2*[AN]. Hence Ϫd[AN]/dt = 2k1[P][AN]. Thus plots of
ln[AN] vs. time should reveal straight lines of slopes = kobs
2k1[P].
Now we consider the formation of dimers, DCB and MGN
vs. time. For small kϪ1 and assuming that subsequent reactions
through B2 and B3 are fast, then eqn. (h) applies for MGN, and
eqn. (i) for DCB.
=
2[DCB](1 ϩ k5/K2k8). Hence the plots of f [DCB] and f [MGN]
vs. time should reveal straight lines of slopes = kobs = 2k1[P].
Therefore we have three routes to the value of k1 through
monitoring the consumption of AN and the formation of both
the linear and branched dimers. Also, from the ratio [DCB]t/
[MGN]t we can assess a value of the ratio K2k8/k5. However, the
above kinetic analysis essentially ignores any pathway giving
rise to the small quantities of oligomer (normally trimer and
crystalline hexamers) that are observed during a dimerisation
reaction. Thus the ‘raw’ [AN]t, [MGN]t and [DBC]t data were
refined by the incorporation of small corrections which were
dependent on the knowledge of the origin and quantities of
both trimer and hexamer.
d[MGN]/dt = k5[AN][B1]
= (k5[AN] × k1[P][AN])/(k5[AN] ϩ K2*k8[AN])
= k1[P][AN]/(1 ϩ K2*k8/k5)
(h)
d[DCB]/dt = k8[AN][Y]
= k8K2*[AN][B1]
= (k8K2*[AN] × k1[P][AN])/(k5[AN] ϩ K2*k8[AN])
= K2*k8k1[P][AN]/k5(1 ϩ K2*k8/k5) (i)
References
Now, [DIMER]t = [DCB]t ϩ [MGN]t and 2[DIMER]t =
[AN]0Ϫ [AN]t. Since the ratio K2k8/k5 is taken to reflect the
observed ratio [DCB]t/[MGN]t we obtain eqn. (j), and thus
eqn. (k), which gives eqn. (l).
1 (a) J. D. McClure, J. Org. Chem., 1970, 35, 3045; (b) R. Oda,
T. Kawabata and S. Tanimoto, Tetrahedron Lett., 1964, 1653;
(c) P. D. Beer, R. C. Edwards, C. D. Hall, J. R. Jennings and
R. J. Cozens, Phosphorus Sulfur, 1983, 17, 283; (d) C. D. Hall,
N. Lowther, B. R. Tweedy, R. Kayhanian, M. Piffl and G. Shaw,
Phosphorus, Sulfur Silicon, 1996, 109–110, 521.
2 J. R. Jennings and R. J. Cozens, Appl. Catalysis A, 1995, 124, 297.
3 J. R. Jennings, R. J. Cozens and K. Wade, Appl. Catalysis A, 1995,
130, 175.
4 J. R. Jennings and R. J. Cozens, Appl. Catalysis A, 1996, 135, 163.
5 A. I. Vogel, A Textbook of Practical Organic Chemistry, 1956,
Longmans, London.
6 C. D. Hall, B. R. Tweedy and N. Lowther, Phosphorus, Sulfur Silicon,
1997, 123, 341.
2[MGN](1 ϩ K2k8/k5) = [AN]0 Ϫ [AN]t
(j)
d[MGN]dt = k1[P]{[AN]0 Ϫ
2[MGN](1 ϩ K2k8/k5)}/(1 ϩ K2*k8/k5) (k)
d[MGN]/[AN]0 Ϫ 2[MGN](1 ϩ K2k8/k5) =
k1[P]dt/(1 ϩ K2*k8/k5) (l)
7 J. R. Lloyd, N. Lowther and C. D. Hall, J. Chem. Soc., Perkin Trans.
2, 1985, 245.
8 J. R. Lloyd, N. Lowther, G. Szabo and C. D. Hall, J. Chem. Soc.,
Perkin Trans. 2, 1985, 1813.
9 C. D. Hall, P. D. Beer, R. L. Powell, M. P. Naan, Phosphorus, Sulfur
Silicon, 1995, 105, 145.
10 G. P. Schiemenz, Chem. Ber., 1966, 98, 514.
11 F. G. Mann and I. G. Millar, J. Chem. Soc., 1952, 4453.
12 W. H. Dietsche, Tetrahedron Lett., 1966, 6347.
13 M. G. Futica and H. J. Harwood, J. Macromol. Sci., Chem., 1978,
A12, 1099.
Now, if kϪ2 [ROH] ӷ k8[AN] which is reasonable for a proton
transfer process and confirmed by the rapid deuteriation in the
olefinic sites of the isolated linear dimer, then from eqn. (e):
K2*→K2. Using standard integrals [eqn. (m)] we derive the
dx
= Ϫln (b Ϫ ax)/a
(m)
(b Ϫ ax)
integrated rate equation, eqn. (n). From this, and assuming
ln {[AN]0 Ϫ 2[MGN](1 ϩ K2k8/k5)} =
Ϫ2k1[P]t(1 ϩ K2k8/k5)/(1 ϩ K2k8/k5) = Ϫ2k1[P]t (n)
Paper 7/07559F
Received 20th October 1997
Accepted 3rd June 1998
eqn. (o) applies, then eqn. (p) can be derived. By a similar
2054
J. Chem. Soc., Perkin Trans. 2, 1998