Electronic effects in the nickel-catalysed hydrocyanation of styrene applying chelating phosphorus ligands with large bite angles

Article abstract of DOI:10.1039/a802269k

Chelating phosphorus ligands with a rigid backbone and a large natural bite angle were applied in the nickel-catalysed hydrocyanation of styrene. The para substituents in the diphenylphosphanyl moiety of the 4,6-bis-(diphenylphosphanyl)-2,8-dimethylphenoxathiine (Thixantphos) ligands were varied and their electronic effects on the activity and selectivity of the catalytic experiments were investigated. The activity of the nickel complexes decreased when electron-donating substituents lead to a more basic phosphorus while electron-withdrawing substituents led to a higher activity. The results of variable temperature ³¹P-{¹H} NMR experiments on the in situ catalysts are discussed in relationship to the catalytic performance. 4,6-Bis(diphenylphosphanyl)-2,8-dimethylphenoxathiine (Thixantphos) L^1d and the complexes [NiCl₂L^1d] 1 and [Ni(CN)₂L^1a] 2 (p-Me₂N on phenyl) have been characterised by single-crystal X-ray diffraction. Complex 2 represents the first crystal structure of a monomeric dicyanonickel(II) complex with a P-P chelating ligand. The geometries of ligand L^1d and complex 1 were predicted by molecular mechanics calculations.

Full text of DOI:10.1039/a802269k

View Article Online / Journal Homepage / Table of Contents for this issue
Electronic effects in the nickel-catalysed hydrocyanation of styrene
applying chelating phosphorus ligands with large bite angles‡
Wolfgang Goertz,a Wilhelm Keim,a Dieter Vogt,*,†,a Ulli Englert,b Maarten D. K. Boele,c
Lars A. van der Veen,c Paul C. J. Kamerc and Piet W. N. M. van Leeuwenc
a Institut für Technische Chemie und Petrolchemie, RWTH Aachen, Templergraben 55,
52056 Aachen, Germany
b Institut für Anorganische Chemie, RWTH Aachen, Templergraben 55, 52056 Aachen, Germany
c J. H. van’t Hoff Research Institute, Department of Inorganic Chemistry, University of
Amsterdam, Nieuwe Achtergracht 166, 1018 WV Amsterdam, The Netherlands
Chelating phosphorus ligands with a rigid backbone and a large natural bite angle were applied in the nickel-
catalysed hydrocyanation of styrene. The para substituents in the diphenylphosphanyl moiety of the 4,6-bis-
(diphenylphosphanyl)-2,8-dimethylphenoxathiine (Thixantphos) ligands were varied and their electronic effects
on the activity and selectivity of the catalytic experiments were investigated. The activity of the nickel complexes
decreased when electron-donating substituents lead to a more basic phosphorus while electron-withdrawing
substituents led to a higher activity. The results of variable temperature 31P-{1H} NMR experiments on the
in situ catalysts are discussed in relationship to the catalytic performance. 4,6-Bis(diphenylphosphanyl)-2,8-
dimethylphenoxathiine (Thixantphos) L1d and the complexes [NiCl2L1d] 1 and [Ni(CN)2L1a] 2 (p-Me2N on
phenyl) have been characterised by single-crystal X-ray diffraction. Complex 2 represents the first crystal structure
of a monomeric dicyanonickel() complex with a P᎐P chelating ligand. The geometries of ligand L1d and complex
1 were predicted by molecular mechanics calculations.
The hydrocyanation of butadiene in the presence of nickel
complexes, known as the DuPont ADN Process, is one of the
most prominent examples of homogeneous catalysis on an
industrial scale.1 Many investigations have been undertaken to
understand its mechanism and to improve the activity and
selectivity.2 For some decades the process has been carried
out with monodentate phosphites which are bound only very
weakly to the zerovalent metal centre.3 For this reason, one
major drawback in this process arises from the formation of
inactive nickel cyanides when the monodentate ligands dissoci-
ate and the catalyst metal is exposed to an excess of HCN. Only
recently bidentate ligands have drawn some attention.4 They are
able to stabilise the active species to a much higher extent due to
the chelate effect. It was reported by Pringle and co-workers5
that diphosphites form complexes with nickel and palladium,
that are active in hydrocyanation catalysis and highly resistant
towards oxidation as well. Striking results have been obtained
in the asymmetric hydrocyanation of vinylarenes by using
chiral diphosphinites derived from sugar backbones.6 Enantio-
meric excesses up to 95% combined with high activities were
achieved.7 It was suggested that the enantioselectivity of the
reaction is determined in the reductive elimination step of the
catalytic cycle which is strongly influenced by the electronic
properties of the ligand.
a strong influence on catalyst selectivity in manifold reac-
tions.10,11 By modifying the ligand backbone the bite angle can
be precisely adjusted to various geometries. For the catalytic
hydrocyanation we proposed that ligands with fixed large bite
angles would (a) disfavour inactive square planar nickel()
dicyano species, and (b) stabilise active tetrahedral nickel(0)
complexes.8
Next to the geometry of the ligands, their electronic proper-
ties play a crucial role for the complex stability and have a great
influence on elementary steps of the catalytic cycle.12 Ligands
with electron-withdrawing substituents at the phosphorus
atoms bear a low basicity and can easily compensate the high
electron density of the d10 configurated zerovalent nickel by
back donation into non-occupied orbitals. A nickel which is
electron-poor due to ligand properties will readily co-ordinate
an olefinic substrate. Moreover, kinetic studies by RajanBabu6
and co-workers indicate that the rate of reductive elimination is
increased by electron-withdrawing substituents at the ligand.
For these reasons, phosphites with three electron-withdrawing
oxygen substituents are much more favourable for hydro-
cyanation reactions than phosphines. However, we could
demonstrate that Xantphos type diphosphines are quite
comparably active to monophosphites.8,9
In this paper we describe the application of electronically
tuned Thixantphos type ligands L1a–L1g and L2b (Scheme 1) in
the hydrocyanation of styrene. Substituents were introduced in
the para positions of the phenyl rings only in order to keep
steric effects as small as possible. In this way it should be pos-
sible to study the electronic effects independently from the bite
angle. Natural bite angles for nickel complexes and other ligand
geometries were calculated by molecular modeling, using an
augmented TRIPOS force field with SYBYL software.13 The
natural bite angle is defined as the preferred chelation angle
determined only by ligand-backbone constraints and not by
metal valence angles.14 We have studied the crystal structures of
one free Thixantphos L1d (R = H) and two nickel complexes
[NiCl2L1d] 1 and [Ni(CN)2L1a] 2 to support the suggested
relationship between the geometry and the reactivity of
In preliminary communications, we have reported on the suc-
cessful application of diphosphines in the hydrocyanation of
styrene8 and long-chain, non-activated olefins.9 These newly
developed Xantphos ligands10 have large natural bite angles
and rigid backbones. The compounds are based on hetero-
cyclic xanthene-like aromatics. They were designed to stabilise
geometries with P᎐Ni᎐P angles larger than 100Њ, shown to have
† E-Mail: dieter.vogt@post.rwth-aachen.de
‡ Supplementary data available: computational and other details. For
direct electronic access see http://www.rsc.org/suppdata/dt/1998/2981/,
otherwise available from BLDSC (No. SUP 57410, 5 pp.) or the RSC
Library. See Instructions for Authors, 1998, Issue 1 (http://www.rsc.org/
dalton).
J. Chem. Soc., Dalton Trans., 1998, Pages 2981–2988
2981

View Article Online
S
S
βn /°
L1a Me2N 118.6
R
P
P
O
O
OH
OH
( i )
+ 2
L2b
L1b
L1c
L1d
L1e
L1f
113.8
112.9
112.8
112.8
113.4
114.5
MeO
Me
H
F
Cl
P
P
P
P
- 4 HNEt2
NEt2
Et2N
R
R
NEt2 Et2N
L2a
L1g
Scheme 3 (i) toluene, 90 ЊC, 15 h
CF3
R
R
ised by various spectroscopic methods.16 Ligand dissociation
from a tetrahedrally co-ordinated zerovalent nickel I generates
a trigonal, highly reactive NiL3 species II, which undergoes
rapid oxidative addition of HCN. The resulting complex III
with a hydride and a cyano ligand co-ordinating in trans
positions can easily lose a second monodentate ligand P to yield
the tetrahedral species IV. Depending on the steric bulk of the
ligands P the hydridocyano intermediate IV can also adopt a
square planar structure. The cycle is continued by olefin co-
ordination V and insertion of the substrate into the nickel–
hydrogen bond giving another tetrahedral intermediate VI.
Similar to IV it should be possible that the cyanoalkyl species
VI can have a square planar geometry as well. In the last
step, reductive elimination takes place to form the product
nitrile and the cycle is completed by generating the active
complex II again.
S
O
O
O
P
P
O
O
L2b, βn = 120.4°
Scheme 1 Thixantphos ligands and their calculated natural bite angles
P
I
Ni
P
P
P
Deactivation of the catalyst is often noticed by the formation
of dicyanonickel() species VII, preferably with the cyano
ligands occupying trans positions at the metal. All other inter-
mediates of the catalytic cycle favour P᎐Ni᎐P angles of either
109Њ together with tetrahedral geometry or 120Њ with trigonal
structures. When monodentate ligands are used the complexes
are very flexible to structural changes, but also highly suscep-
tible to dissociation and catalyst deactivation. Consequently,
our conception was focussed on ligands with large natural bite
angles and rigid backbones. These Xantphos bidentates should
suppress the formation of dicyano complexes VII, stabilise the
substantially tetrahedral intermediates IV and VI and thus
enhance the reductive elimination and the overall catalysis.17
Since our aim was the detailed investigation of electronic
properties, we selected the phenoxathiine backbone and the
Thixantphos ligand L1d respectively for further modification.
P
NC
P
R
HCN
Ni
P
P
P
VI
II
CN
Ni
CN
Ni
P
P
P
P
R
P
H
III
P
CN
Ni
CN
Ni
R
P
HCN
P
H
P
H2
P
H
V
IV
Synthesis and molecular modeling
CN
P
P
Ni
Next to a whole series of Thixantphos diphosphines L1a–L1g
that were modified only at the para positions of the diphenyl-
phosphanyl moiety, a new diphosphonite ligand L2b was syn-
thesized (Scheme 1). This latter ligand is much less basic at the
phosphorus atoms because of the biphenyl oxygen substituents.
The diphosphonite ligand L2b is readily accessible by the
reaction of the amido phosphonito phenoxathiine L2a with 2
equivalents of 2,2Ј-dihydroxy-1,1Ј-biphenyl (Scheme 3).
The geometries of the ligands were simulated by molecular
modeling based on force field calculations.13 Since the diphos-
phines L1a–L1g are only modified at the para positions of the
phenyl substituents, the calculated natural bite angles for nickel
complexes are within a very narrow range of 112.8 to 114.5Њ,
except for L1a. A steric influence on the hydrocyanation results
can be excluded. Only ligand L1a with dimethylamino groups
requires a somewhat larger angle of 118.6Њ. For the diphos-
phonite ligand L2b we calculated a bite angle of 120.4Њ. We
suppose that this value is enforced by the limited flexible ring
structures of the biphenoxy groups and their steric demand.
R
NC
VII
Angle
Species
I, IV,VI
II
Geometry
tetrahedral
trigonal
P - Ni - P
109°
120°
120°
III, V
VII
trigonal-bipyramidal
square-planar
90/180°
Scheme 2 Proposed catalytic cycle for hydrocyanation
catalytically active intermediates. We have also carried out 31P-
{1H} NMR investigations with the in situ catalysts which pro-
vide useful information on the performance and deactivation
pathways for the various nickel complexes.
Results and Discussion
Mechanistic aspects
Crystal structures
The catalytic cycle for the hydrocyanation of olefins applying
monodentate phosphorus ligands P as shown in Scheme 2 has
been proposed by Tolman et al.15 Intermediates and catalytic-
ally active species have been isolated or observed and character-
The molecular structure of ligand L1d is drawn in Fig. 1,
together with the atom numbering scheme, while selected bond
lengths and angles are given in Table 1. The structure clearly
shows that only very little adjustment of the ligand is necessary
2982
J. Chem. Soc., Dalton Trans., 1998, Pages 2981–2988

View Article Online
Table 1 Selected bond lengths (Å) and angles (Њ) for ligand L1d with
estimated standard deviations in parentheses
Table 2 Selected bond lengths (Å) and angles (Њ) for complex 1 with
estimated standard deviations in parentheses
S᎐C(6)
1.763(3)
1.846(3)
1.823(3)
1.824(3)
1.371(3)
1.510(4)
S᎐C(7)
1.760(3)
1.841(3)
1.836(3)
1.824(3)
1.375(3)
1.518(5)
Ni᎐Cl(1)
Ni᎐P(1)
S᎐C(6)
P(1)᎐C(2)
P(1)᎐C(30)
O᎐C(12)
P(1) ؒ ؒ ؒ P(2)
2.191(2)
2.307(2)
1.767(6)
1.829(6)
1.823(6)
1.397(6)
3.77
Ni᎐Cl(2)
Ni᎐P(2)
S᎐C(7)
P(1)᎐C(20)
O᎐C(1)
C(4)᎐C(13)
Ni ؒ ؒ ؒ O
2.206(2)
2.304(2)
1.758(6)
1.826(6)
1.386(6)
1.497(8)
3.37
P(1)᎐C(2)
P(1)᎐C(30)
P(2)᎐C(40)
O᎐C(1)
P(1)᎐C(20)
P(2)᎐C(11)
P(2)᎐C(50)
O᎐C(12)
C(4)᎐C(13)
C(9)᎐C(14)
C(6)᎐S᎐C(7)
C(2)᎐P(1)᎐C(20)
C(11)᎐P(2)᎐C(40) 101.6(1)
C(1)᎐O᎐C(12)
O᎐C(1)᎐C(6)
O᎐C(12)᎐C(11)
S᎐C(6)᎐C(5)
101.4(1)
100.5(1)
P(2)᎐C(40)᎐C(45) 117.3(3)
C(20)᎐P(1)᎐C(30) 103.2(1)
P(2)᎐C(11)᎐C(10) 125.1(2)
Cl(1)᎐Ni᎐Cl(2)
Cl(1)᎐Ni᎐P(2)
Cl(2)᎐Ni᎐P(2)
C(6)᎐S᎐C(7)
Ni᎐P(1)᎐C(20)
C(2)᎐P(1)᎐C(20)
132.37(7)
102.88(7)
102.08(7)
98.2(3)
108.3(2)
103.0(3)
Cl(1)᎐Ni᎐P(1)
Cl(2)᎐Ni᎐P(1)
P(1)᎐Ni᎐P(2)
Ni᎐P(1)᎐C(2)
Ni᎐P(1)᎐C(30)
C(2)᎐P(1)᎐C(30) 105.7(3)
C(1)᎐O᎐C(12) 116.2(5)
105.82(7)
103.03(7)
109.52(6)
116.8(2)
116.8(2)
124.3(2)
123.4(2)
114.4(2)
117.3(2)
O᎐C(1)᎐C(2)
O᎐C(12)᎐C(7)
S᎐C(6)᎐C(1)
115.9(2)
124.4(2)
123.4(2)
P(2)᎐C(50)᎐C(51) 115.6(2)
P(2)᎐C(40)᎐C(41) 124.9(2)
P(2)᎐C(50)᎐C(55) 126.5(2)
C(20)᎐P(1)᎐C(30) 104.7(3)
Fig. 1 Molecular structure of ligand 1d
to form a chelate complex since the orientation of the diphenyl-
phosphanyl moieties is nearly ideal for metal co-ordination.
The non-bonding P ؒ ؒ ؒ P distance is 4.02 Å. While molecular
modeling calculations suggest a bending in the phenoxathiine
moiety, the two aromatic rings in the backbone of Thixantphos
L1d are coplanar to each other. Probably, the planarity in the
solid state arises from crystal structure packing effects. Simu-
lated structures are just idealised gas phase molecules with no
interaction to other compounds.
In contrast, the dihedral angle between the two phenyl planes
in the backbone of the related Xantphos ligand (CMe2 instead
of S, H instead of Me) is 158Њ, determined by X-ray analy-
sis.10,18 π-Stacking interactions between the phenyl substituents
at the phosphorus atoms are thought to be responsible for this
bending.19,20
Despite the numerous bidentate phosphorus ligands known
in the literature, only a few dichloro nickel complexes have been
characterised by X-ray analysis. Most common geometries to
be found for [NiCl2(P᎐P)] are tetrahedral21 and square planar.22
The molecular structure and numbering scheme for complex
[NiCl2L1d] 1 are given in Fig. 2. Selected bond lengths, non-
bonding distances and angles are presented in Table 2. The
compound is paramagnetic which made recording of NMR
spectra impossible. Complex 1 was simulated by molecular
mechanics and predicted to have a distorted tetrahedral geom-
etry around the metal. The calculated P᎐Ni᎐P angle is 110.6Њ
which is in very good agreement with the X-ray determined
angle of 109.52(6)Њ. The backbone bending also could be pre-
dicted with a C᎐S᎐C angle of 97.4Њ and a C᎐O᎐C angle of
114.6Њ [X-ray: 98.2(3) and 116.2(5)Њ respectively]. The dihedral
angle between the two aromatic rings in the phenoxathiine
moiety is 147.7(3)Њ.
Fig. 2 Molecular structure of complex 1
[NiCl2(PPh3)2]23 and 2.333 Å (average) for [NiBr2(PPh3)2].24
Although the bite angle P᎐Ni᎐P of 109.52(6)Њ is ideal for a
tetrahedral geometry around the nickel, the structure is signifi-
cantly distorted with a Cl᎐Ni᎐Cl angle of 132.37(7)Њ due to
the lone-pair repulsion between the two chlorine atoms. A simi-
lar geometry was found in the complexes [NiCl2(pop)]25 [pop =
2,2Ј-bis(diphenylphosphino)diethyl ether] and [NiCl2(diop)]26
[diop = 4,5-bis(diphenylphosphinomethyl)-2,2-dimethyl-1,3-di-
oxolane] with Cl᎐Ni᎐Cl angles of 127.1(2) and 130Њ respect-
ively. A discussion of distortion of nickel() tetrahedra in terms
of ligand field theory has been given by Venanzi.27
The complex [Ni(CN)2L1a] 2 represents the first known crys-
tal structure of a monomeric dicyanonickel() complex with a
P᎐P chelating ligand (Fig. 3 and Table 3). Even complexes of
the type [Ni(CN)2L2] (L = monodentate phosphorus ligand)
have been determined by X-ray analysis very rarely.28 Related
dimeric species with bidentate ligands were described by Holah
et al.,29 [Ni2(CN)2(dppm)2], and by Manojlovic-Muir et al.,30
[Ni2(CN)4(Me2PCH2PMe2)2], only the latter being character-
ised by X-ray diffraction. While the calculated natural bite
angle for the ligand L1a is 118.6Њ, the P᎐Ni᎐P angle found in the
complex 2 is widened to 151.51(5)Њ. This value is out of the
flexibility range31 for Xantphos type ligands, which is typically
about 35Њ for an excess strain energy of 15 kJ mol .
Ϫ1 10 The two
cyano groups occupy trans positions at the nickel with a
C᎐Ni᎐C angle of 161.0(2)Њ. Considering the small Ni᎐O dis-
tance of 2.72 Å and the four angles P᎐Ni᎐C in the range of
89.8(1) to 95.9(1)Њ, the geometry around the metal can be
described as distorted square pyramidal with the oxygen on
The bond distances Ni᎐P [2.307(2) and 2.304(2) Å] appear
quite normal compared with the values of 2.28 Å found for
J. Chem. Soc., Dalton Trans., 1998, Pages 2981–2988
2983

View Article Online
Table 3 Selected bond lengths (Å) and angles (Њ) for complex 2 with
estimated standard deviations in parentheses
Table 4 Catalytic hydrocyanation of styrenea
Regio-
selectivityb
(%)
Ni᎐P(1)
Ni᎐C(1)
S᎐C(15)
2.208(1)
1.863(5)
1.764(5)
1.840(4)
1.808(5)
1.806(4)
1.410(5)
1.154(6)
1.379(6)
1.363(6)
4.27
Ni᎐P(2)
Ni᎐C(2)
S᎐C(16)
2.197(1)
1.861(5)
1.760(5)
1.813(4)
1.829(4)
1.795(4)
1.385(5)
1.145(6)
1.475(7)
1.442(6)
2.72
Conversionb Yieldb Selectivityb
Entry Ligand (%)
(%)
(%)
1
2
3
4
5
6
7
8
L1a
L1b
L1c
L1d
L1e
L1f
L1g
L2b
16
33
55
77
46
75
98
49
12
22
48
70
38
52
90
47
76
66
87
92
83
70
92
97
>99
99.5
99.8
99.4
99.4
99.3
99.0
99.4
P(1)᎐C(11)
P(1)᎐C(40)
P(2)᎐C(50)
O᎐C(10)
N(1)᎐C(1)
N(3)᎐C(33)
N(5)᎐C(53)
P(1) ؒ ؒ ؒ P(2)
P(1)᎐C(30)
P(2)᎐C(20)
P(2)᎐C(60)
O᎐C(21)
N(2)᎐C(2)
N(3)᎐C(36)
N(5)᎐C(56)
Ni᎐O
a Reaction conditions: toluene (2 cm3), nickel:ligand:styrene:HCN =
1:1.05:20:25; 60 ЊC, 16 h, preformation time = 30 min. b Conversion
(based on the substrate); yield, selectivity and regioselectivity deter-
mined by temperature-controlled GC analysis. Regioselectivity is
defined as the percentage of branched nitrile.
P(1)᎐Ni᎐P(2)
P(1)᎐Ni᎐C(1)
P(2)᎐Ni᎐C(1)
C(15)᎐S᎐C(16)
Ni᎐C(1)᎐N(1)
151.51(5)
90.5(1)
89.8(1)
98.5(2)
173.9(4)
C(1)᎐Ni᎐C(2)
P(1)᎐Ni᎐C(2)
P(2)᎐Ni᎐C(2)
C(10)᎐O᎐C(21) 117.3(3)
Ni᎐C(2)᎐N(2) 172.1(4)
161.0(2)
95.9(1)
93.0(1)
Ni᎐P(1)᎐C(11) 105.2(1)
Ni᎐P(1)᎐C(40) 115.9(1)
Ni᎐P(2)᎐C(50) 117.1(1)
Ni᎐P(1)᎐C(30) 122.4(2)
Ni᎐P(2)᎐C(20) 105.9(2)
Ni᎐P(2)᎐C(60) 118.9(2)
L1a–L1g
[Ni(cod)2]
P
P
P
P
Ni
3a–3g
Scheme 4 Formation of bis-chelate nickel complexes
other two halogen derivatives L1e and L1f perform less success-
fully than expected (entries 5 and 6). A plausible explanation
may be that ligand decomposition has occurred by reaction of
the aryl halogenides with styrene mediated by nickel species.
Similar Heck type reactivity is best known with palladium.33
We have observed the same deactivation mechanism with
nickel complexes in the presence of chlorinated or brominated
aromatic substrates. The fluoride in ligand L1g is not bound
directly to an aromatic ring and therefore inert towards reaction
with the metal of the catalyst. Moderate yield combined with a
very high selectivity is obtained in the presence of the diphos-
phonite ligand L2b (entry 8). This unprecedented behaviour will
be explained later.
With styrene as a substrate, the branched nitrile is the
strongly favoured product over the linear one. This regio-
selectivity is attributed to stabilisation of the branched alkyl
intermediate by a η3-benzyl interaction of the nickel.2 Still, the
selectivity induced by the Thixantphos ligands L1a–L1g and L2b
(>99%) is significantly higher than those obtained with com-
mon diphosphines (up to 95%)8 and the commercial o-tolyl
phosphite system (91%),34 and comparable to that of diphos-
phonites based on sugar backbones.35
Fig. 3 Molecular structure of complex 2
top. However, an interaction between the nickel and the
oxygen cannot be predicted from the crystal structure only. For
nickel complexes with ether ligands found in the Cambridge
Crystallographic Database the average Ni᎐O distance is 2.15 Å.
The metal–phosphorus and –carbon bond lengths are in good
agreement with those observed in other dicyano d8-nickel com-
plexes.32 Both of the Ni᎐C᎐N linkages are slightly distorted
from ideal linearity with angles of 173.9(4) and 172.1(4)Њ. The
backbone of the ligand is bent by the dihedral angle of 33.6(3)Њ.
Ligand L1a is the most basic out of the series of diphosphines
and the only one that formed a dicyano complex with nickel.
We assign that to the electron-donating capability towards the
Ni(CN)2 system.
NMR characterisation of nickel complexes
Various in situ catalyst solutions were investigated by means
of 31P-{1H} NMR spectroscopy. Surprisingly, the addition of
either 1 or 2 equivalents of Thixantphos diphosphine ligands
L1a–L1g to toluene solutions of [Ni(cod)2] resulted in similar
spectra. The complexity of the peak pattern observed can only
be explained by multiple couplings between various phos-
phorus atoms. The species which are responsible for this are
the bis(chelate) complexes [Ni(P᎐P)2] 3a–3g (Scheme 4). The
31P-{1H}NMRspectrumofcomplex3d(P᎐P = ThixantphosL1d)
was simulated with g-NMR software36 (Fig. 4). The peak
pattern could most accurately be simulated by an AAЈXXЈ
system with four large coupling constants of 50 Hz and two
smaller ones of 12.5 Hz (Table 5). Therefore the structure of
complex 3d is supposed to be distorted tetrahedral.
The complex peak pattern at 273 K is caused by a highly
asymmetric geometry of the bis(chelate) complex 3d. We con-
clude that due to the non-planarity of the ligand backbones
there are at least two different co-ordination modes (Scheme 5),
which make the phosphorus nuclei Pa and Pb magnetically
inequivalent. With a fixed roof-like co-ordination of the first
Catalytic hydrocyanation of styrene
Reactions of the Thixantphos ligands L1a–L1g and L2b with 1
equivalent [Ni(cod)2] yield active catalyst precursors for the
hydrocyanation of styrene (Table 4). The compounds L1a–L1c
are more basic at the phosphorus atoms than the unsubstituted
ligand L1d, caused by a positive mesomeric effect of the amino
and methoxy groups and a positive inductive effect of the
methyl group. Less basic than Thixantphos L1d are the halogen
derivatives L1e–L1g (with R = F, Cl or CF3). Nevertheless, the
negative inductive effect of Cl is reduced by a positive
mesomeric effect.
The electron-rich diphosphines L1a–L1c (entries 1–3, Table 4)
lead to lower activities in catalytic experiments than the unsub-
stituted Thixantphos L1d. Best results are obtained with the
most electron-poor diphosphine L1g (entry 7, R = CF3). The
2984
J. Chem. Soc., Dalton Trans., 1998, Pages 2981–2988

View Article Online
Table 5 Chemical shifts and J(PP) values of the NMR simulation for
[NiL1d2] 3d
J/Hz
Nucleus
δ
1
2
3
P1
P2
P3
P4
15.00
11.70
15.00
11.70
50
50
12.5
12.5
50
50
Fig. 4 Recorded (left, 273 K, toluene-d8) and simulated (right) 31P-
{1H} NMR spectra of [NiL1d2] 3d
Pa
Pb
Pa′
Pa
Pb′
Pa′
Pb
Ni
Ni
Pb′
Pa′
Pb′
backbone flipping
ligand exchange
Scheme 5 Possible coordination modes and exchange processes of a
bis-chelate complex (phenyl groups are omitted for clarity)
equivalent of ligand the backbone of the second diphosphine
can be folded either to the left or to the right side. These two
species are enantiomeric complexes. However, since there are
two diastereotopic phosphorus atoms, the 31P-{1H} NMR spec-
trum shows two different chemical shifts for Pa and Pb at δ 11.7
and 15.0 respectively.
Subsequently, the bis(chelate) complex [NiL1d] 3d was
investigated in detail by variable temperature 31P-{1H} NMR as
shown in Fig. 5. On heating the sample from room temperature
to 371 K, line broadening occurs caused by a fluxional process
and the phosphorus atoms become magnetically equivalent,
with a coalescence temperature of 353 K. For this exchange
process a butterfly flipping of the ligand backbones as well as
ligand dissociation can be proposed. At higher temperatures
this process is fast in comparison to the ν(Pa) Ϫ ν(Pb) frequency
separation. Especially since the bis(chelate) complex is as
catalytically active as the monochelate complex [Ni(cod)L1d],
ligand dissociation cannot be excluded as a reason for the
fluxionality.
Fig. 5 Variable temperature 31P-{1H} NMR spectrum (toluene-d8) of
[NiL1d2] 3d, recorded (left) and simulated (right)
∆S‡ is Ϫ16.7 KϪ1 molϪ1. The latter value is relatively small,
indicative of an intramolecular rearrangement process, thus
a butterfly flipping of the ligand backbones. For a temperature
of 293 K, the free energy value ∆G‡ = 66.3 kJ molϪ1 has been
determined from the Gibbs equation ∆G‡ = ∆H‡ Ϫ T∆S‡.
When our investigations were extended to the Thixantphos
diphosphonite L2b a similar behaviour was observed. The 31P-
{1H} NMR spectrum ([2H8]toluene, 298 K) of an in situ catalyst
solution containing equimolar amounts of [Ni(cod)2] and lig-
and L2b shows a singlet at δ 196.6 5 min after mixing, resulting
from the monochelate [Ni(cod)L2b]. After 30 min, about 80% of
From the recorded and simulated 31P-{1H} NMR spectra, the
rate constants (k) for the fluxional process have been deter-
mined for complex 3d. In the Eyring plot in Fig. 6 the rate
constants (k) for phosphorus exchange are given as a func-
tion of the NMR sample temperature for 3d. From the Eyring
equation and the Eyring plot, the thermodynamic values for
the activation of the dynamic process were calculated. The
assumption used is that ∆H‡ and ∆S‡ are constant over the
temperature range (293–371 K) employed. The enthalpy of
activation ∆H‡ is 61.4 kJ molϪ1 and the entropy of activation
the ligand has already formed the bis(chelate) complex [NiL2b
]
2
4 showing two pseudo-triplet signals at δ 187.5 and 177.6 (Fig.
7). Surprisingly, NMR simulation for complex 4 requires equal
J. Chem. Soc., Dalton Trans., 1998, Pages 2981–2988
2985

View Article Online
present in large amounts (routes A and B). Regarding a pre-
formation time of 30 min and a deactivation of about 80% of
the nickel, the moderate values for conversion and yield in the
presence of ligand L2b given in Table 4 must be reconsidered.
Taking all this into account, L2b shows high catalytic activity
due to electronic properties.
Conclusion
A series of electronically tuned Thixantphos ligands was
applied to the nickel catalysed hydrocyanation of styrene.
Electron-withdrawing substituents at the phosphorus atoms
make the ligands more acidic and lead to best activities. The
concept of large bite angles combined with rigid backbones is
demonstrated by their importance for intermediates of the
catalytic cycle and is supported by both crystal structures and
31P-{1H} NMR spectroscopy. A diphosphonite Thixantphos
ligand with comparable geometry forms highly active nickel
complexes. Bis(chelate) complexes of this ligand are very stable
and catalytically inactive.
Fig. 6 Eyring plot: data obtained from variable-temperature 31P-{1H}
NMR of [NiL1d2] 3d
Experimental
Computational details
Molecular mechanics calculations were performed on a Silicon
Graphics Indigo2 workstation using SYBYL software and a
modified TRIPOS force field.13 Natural bite angles were calcu-
lated using a method similar to that described by Casey and
Whiteker,14 based on a Ni᎐P bond length of 2.177 Å and a
P᎐Ni᎐P bending force constant of 0 kcal molϪ1 degreeϪ2
(cal = 4.184 J).
Syntheses
All preparations were carried out under an atmosphere of puri-
fied argon using standard Schlenk techniques. Solvents were
dried and freshly distilled prior to use. 2,8-Dimethylphenox-
athiine and the diphosphines L1a–L1g were prepared as des-
cribed in recent publications.10 The compound [Ni(cod)2] was
synthesized according to literature methods.37 The NMR
spectra were recorded on a Bruker DPX300 spectrometer (1H,
300; 13C, 75; 31P, 121 MHz).
Fig. 7 31P-{1H} NMR spectrum (toluene-d8, 298 K) of in situ catalyst
preformation with [Ni(cod)2] and 1 equivalent of ligand L2b, recorded
30 minutes after mixing
P
Hydrogen cyanide. CAUTION: HCN is a highly toxic, volatile
liquid (b.p. 27 ЊC) that is susceptible to exothermic and
uncontrolled polymerisation in the presence of basic catalysts.
It should be handled only in a well ventilated fume hood and by
teams of at least two technically qualified persons who have
received appropriate medical training for treating HCN poison-
ing. Sensible precautions include also the use of HCN monitor-
ing equipment. It can be generated by the addition of H2SO4
to sodium cyanide. Uninhibited HCN should be stored at a
temperature lower than its melting point (Ϫ13 ЊC).
P
P
Ni(cod)
P
A
B
D
P
[Ni(cod)2]
catalysis
P
2 P
P
C
Ni
P
P
Scheme 6 Pathways of in situ catalyst preformation
2J(PaPaЈ), J(PaPb), J(PaPbЈ), J(PbPbЈ) couplings of 27 Hz sug-
gesting a structure of high symmetry. The bis(chelate) complex
4 is very stable. Heating to 373 K does not change the signals in
the 31P-{1H} NMR spectrum. Furthermore, 4 is catalytically
inactive up to 423 K.
Together with the results obtained in the hydrocyanation of
styrene, the NMR spectra of the catalyst solutions allow a
deeper insight into the pathways of catalyst activity and
deactivation for the various ligands (Scheme 6). The chelating
Thixantphos ligands L1a–L1g and L2b preferably form bis(chel-
ates) via route C, even if they are added substoichiometrically.
In the case of the diphosphines L1a–L1g the bis(chelate) nickel
complexes 3a–3g are still catalytically active since dissociation
of the ligand occurs even at room temperature (route D). In
contrast, the bis(chelate) complex [NiL2b2] 4 is uncommonly
stable and therefore not active in catalysis. High conversion
of the substrate is only possible when the preformation time
is short and the active catalyst precursor [Ni(cod)L2b] is still
2
2
2
4,6-Bis(diethylaminophosphino)-2,8-dimethylphenoxathiine
L2a. 2,8-Dimethylphenoxathiine (3.26 g, 14.3 mmol) and
N,N,NЈ,NЈ-tetramethylethane-1,2-diamine (4.14 g) were dis-
solved in diethyl ether (50 cm3) and cooled to 230 K. n-Butyl-
lithium (14.3 cm3 of a 2.5  solution in hexanes, 35.7 mmol)
was added slowly, giving a bright yellow solution. The mixture
was stirred at room temperature for 16 h and then added to a
solution of chlorobis(diethylamino)phosphine (6.32 g, 35.7
mmol) in pentane (30 cm3) at 230 K. After stirring for 16 h at
room temperature and evaporating the solvents a crude yellow
product was obtained. The compound L2a was recrystallised
from pentane, yield 3.90 g (48%) (Found: C, 61.6; H, 8.60; N,
9.3. C30H50N4OP2S requires C, 62.5; H, 8.75; N, 9.7%). NMR
1
(C6D6): H, δ 7.28 (s, 2 H), 6.74 (s, 2 H), 3.21 (m, 16 H, CH2),
3
2.09 (s, 6 H, CH3) and 1.13 (t, 12 H, J 7.0 Hz); 13C-{1H},
2986
J. Chem. Soc., Dalton Trans., 1998, Pages 2981–2988

View Article Online
δ 151.0, 132.8, 131.0, 127.7, 127.1, 118.9, 43.5 (CH2), 20.7 (CH3)
and 14.8 (CH2CH3); 31P-{1H}, δ 90.4.
Complex 1. C38H30Cl2NiOP2Sؒ3C6H6, Mr 960.63 (including
solvent of crystallisation), monoclinic, space group P21/c,
a = 19.755(8), b = 19.314(8), c = 13.072(3) Å, β = 103.14(3)Њ,
U = 4857(3) Å3, Dc = 1.314 g cmϪ3, Z = 4, F(000) 2000, λ(Mo-
Kα) = 0.710 73 Å. Crystal size 0.40 × 0.15 × 0.20 mm.
A total of 12 207 reflections were collected at 203 K in the
range 2.0 < θ < 25.0Њ, corresponding to 5601 unique data with
I > 1.0σ(I), which were used for further computations. An
empirical absorption correction based on ψ scans was applied.
Refinement converged with 568 parameters using a statistical
weighting scheme at values of R = 0.085 and RЈ = 0.065 with a
goodness of fit of 1.173.
4,6-Bis(dibenzo[d,f ][1,3,2]dioxaphosphepino-1-yl)-2,8-
dimethylphenoxathiine L2b. Compound L2a (577 mg, 1.0 mmol)
and dihydroxybiphenyl (372 mg, 2.0 mmol) were dissolved in
toluene (5 cm3) and the solution was stirred at 383 K for 15 h.
The mixture was allowed to evaporate to dryness to give a
bright yellow crystalline product, yield 650 mg (99%), m.p.
>260 ЊC (Found: C, 69.1; H, 4.25. C38H26O5P2S requires C, 69.5;
1
4
H. 4.00%). NMR (CDCl3): H, δ 7.33 (d, J 2.1, 2 H), 7.30 (d,
4J 2.4, 2 H), 7.18–7.09 (m, 8 H), 6.96 (d, 4J 1.8 Hz, 2 H), 6.93–
6.90 (m, 6 H) and 2.01 (s, 6 H, CH3); 13C-{1H}, δ 152.2, 151.4,
134.0, 132.0, 130.0, 129.8, 129.4, 129.0, 128.8, 122.1, 119.4 and
20.5 (CH3); 31P-{1H}, δ 178.3.
Complex 2. C48H50N6NiOP2Sؒ2C7H8, Mr 1063.98 (including
solvent of crystallisation), triclinic, space group P1, a =
11.436(5), b = 12.496(3), c = 21.699(8) Å, α = 77.20(2), β =
[NiCl2L1d] 1. Ligand L1d (853 mg, 1.43 mmol) and nickel()
dichloride hexahydrate (340 mg, 1.43 mmol) were suspended in
benzene (10 cm3) and stirred at 323 K for 3 h. Evaporating the
solvent gave a reddish brown solid in quantitative yield. Red
crystals suitable for X-ray analysis were obtained from boiling
benzene, m.p. >260 ЊC (Found: C, 64.0; H, 4.37. C38H30Cl2Ni-
OP2S requires C, 62.8; H, 4.16%).
86.18(3), γ = 65.86(2)Њ, U = 2758(2) Å3, Dc = 1.282 g cmϪ3
,
Z = 2, F(000) 1124, λ(Mo-Kα) = 0.710 73 Å. Crystal size
0.52 × 0.48 × 0.28 mm.
A total of 10 824 reflections were collected at 203 K in the
range 2.0 < θ < 25.0Њ, corresponding to 7164 unique data
with I > 1.0σ(I), which were used for further computations.
Refinement converged with 658 parameters using a statistical
weighting scheme at values of R = 0.081 and RЈ = 0.072 with a
goodness of fit of 1.414.
CCDC reference number 186/1065.
See http://www.rsc.org/suppdata/dt/1998/2981/ for crystallo-
graphic files in .cif format.
[Ni(CN)2L1a] 2. One large blue crystal of complex 2 was iso-
lated from a toluene solution (2 cm3) of a catalytic experiment
after standing for several days at room temperature. Before
catalysis, the solution contained [Ni(cod)2] (17.9 mg, 0.065
mmol), ligand L1a (53.6 mg, 0.070 mmol), styrene (135.4 mg, 1.3
mmol) and hydrogen cyanide (42 mg, 1.625 mmol).
Acknowledgements
Catalytic hydrocyanation of styrene
Financial support from the E.U. Human Capital and Mobility
Program (MMCOS Network) and the Dutch Foundation for
Chemical Research and Foundation for Technological Sciences
(SON/STW) is gratefully acknowledged.
In a typical experiment, a bright yellow 0.065 m solution of
[Ni(cod)2] in toluene (2 cm3) was added to a Schlenk tube con-
taining a stirring bar and 1.05 equivalents of ligand. The mix-
ture was stirred for 30 min to ensure complete formation of the
catalyst precursor. Then styrene (1.3 mmol) was added. The
solution was cooled to 220 K, liquid HCN (63 µl, 1.625 mmol)
was added at once and the tube was placed in a heating bath.
After 16 h at 333 K the excess of HCN was removed by a gentle
stream of argon, solid particles were removed by centrifugation
and the remaining solution was analysed by temperature-
controlled gas chromatography.
References
1 K. Huthmacher and S. Krill, in Applied Homogeneous Catalysis with
Organometallic Compounds, eds. B. Cornils and W. A. Hermann,
VCH, Weinheim, 1996, p. 465.
2 R. J. McKinney, in Homogeneous Catalysis, ed. G. W. Parshall,
Wiley, New York, 1992, p. 42.
3 W. C. Drinkard (DuPont), US Pat., 3 766 237, 1973; M. Rapoport
(DuPont), US Pat., 4 371 474, 1983; W. Tam (DuPont), US Pat.,
5 543 536, 1996.
4 K. A. Kreutzer and W. Tam (DuPont), US Pat., 5 512 696, 1996;
A. I. Breikss (DuPont), US Pat., 5 523 453, 1996; K. A. Kreutzer
and W. Tam (DuPont), US Pat., 5 663 369, 1997; W. Tam, K. A.
Kreutzer and R. J. McKinney (DuPont), US Pat., 5 688 986, 1997.
5 M. J. Baker, K. N. Harrison, A. G. Orpen, P. G. Pringle and
G. Shaw, J. Chem. Soc., Chem. Commun., 1991, 803; M. J. Baker and
P. G. Pringle, J. Chem. Soc., Chem. Commun., 1991, 1292.
6 A. L. Casalnuovo, T. V. RajanBabu, T. A. Ayers and T. H. Warren,
J. Am. Chem. Soc., 1994, 116, 9869.
7 T. V. RajanBabu and A. L. Casalnuovo, J. Am. Chem. Soc., 1996,
118, 6325.
8 M. Kranenburg, P. C. J. Kamer, P. W. N. M. van Leeuwen, D. Vogt,
and W. Keim, J. Chem. Soc., Chem. Commun., 1995, 2177.
9 W. Goertz, P. C. J. Kamer, P. W. N. M. van Leeuwen and D. Vogt.
Chem. Commun., 1997, 1521.
Crystallography
Colourless crystals of ligand L1d were obtained from boiling
acetone, red crystals of complex 1 from boiling benzene. Blue
crystals of complex 2 were isolated from a catalysis experiment
applying ligand L1a as described earlier.
Data were collected on a CAD-4 diffractometer (graphite
monochromator) and ω–2θ scans. Structure solution was by
direct methods. Full-matrix least-squares refinement on F was
carried out with anisotropic displacement parameters for all
non-hydrogen atoms. Hydrogen atoms were placed in idealised
positions with isotropic displacement parameters of U(H) =
1.3B(C) and allowed to ride on their C atoms. Calculations were
performed using the SDP system of programs.38
10 M. Kranenburg, Y. E. M. van der Burgt, P. C. J. Kamer, P. W. N. M.
van Leeuwen, K. Goubitz and J. Fraanje, Organometallics, 1995, 14,
3081; M. Kranenburg, P. C. J. Kamer, P. W. N. M. van Leeuwen and
B. Chaudret, Chem Commun., 1997, 373; M. Kranenburg, J. G. P.
Delis, P. C. J. Kamer, P. W. N. M. van Leeuwen, K. Vrieze, A. L.
Spek, K. Goubitz and J. Fraanje, J. Chem. Soc., Dalton Trans., 1997,
1839.
11 M. Kranenburg, P. C. J. Kamer and P. W. N. M. van Leeuwen, Eur. J.
Inorg. Chem., 1998, 1, 25; 2, 155.
12 C. A. Tolman, J. Chem. Educ., 1986, 63, 199.
13 SYBYL, version 6.3, TRIPOS Associates, St. Louis, MO, 1996.
14 C. P. Casey and G. T. Whiteker, Isr. J. Chem., 1990, 30, 299.
15 C. A. Tolman, R. J. McKinney, W. C. Seidel, J. D. Druliner and
W. R. Stevens, Adv. Catal., 1985, 33, 1.
L1d. C38H30OP2S, Mr 596.67, monoclinic, space group P21/c,
a = 11.904(2), b = 21.377(2), c = 13.285(2) Å, β = 112.18(1)Њ,
U = 3130.4(7) Å3, Dc = 1.266 g cmϪ3, Z = 4, F(000) 1248,
λ(Cu᎐Kα) = 1.541 84 Å. Crystal size 0.40 × 0.25 × 0.20 mm.
A total of 9557 reflections were collected at 293 K in the
range 5.0 < θ < 75.0Њ, corresponding to 4989 unique data with
I > 1.0σ(I), which were used for further computations. An
empirical absorption correction based on ψ scans was applied.
Refinement converged with 499 parameters using a statistical
weighting scheme at values of R = 0.062 and RЈ = 0.063 with a
goodness of fit of 1.280.
J. Chem. Soc., Dalton Trans., 1998, Pages 2981–2988
2987

View Article Online
16 C. A. Tolman, J. Am. Chem. Soc., 1970, 92, 2956; Inorg. Chem.,
1971, 10, 1540; J. Am. Chem. Soc., 1972, 94, 2994; J. D. Druliner,
A. D. English, J. P. Jesson, P. Meakin and C. A. Tolman, J. Am.
Chem. Soc., 1976, 98, 2156; W. R. Jackson and C. G. Lovel, Aust. J.
Chem., 1982, 35, 2053; C. A. Tolman, W. C. Seidel and L. W. Gosser,
Organometallics, 1983, 2, 1391; J. E. Bäckvall and O. S. Andell,
Organometallics, 1986, 5, 2350; R. J. McKinney and D. C. Roe,
J. Am. Chem. Soc., 1986, 108, 5167.
17 J. M. Brown and P. J. Guiry, Inorg. Chim. Acta, 1994, 220, 249.
18 S. Hillebrand, J. Bruckmann, C. Krüger and M. W. Haenel,
Tetrahedron Lett., 1995, 36, 75.
19 M. Kranenburg, Ph.D. Thesis, University of Amsterdam, 1996.
20 C. A. Hunter and J. K. M. Sanders, J. Am. Chem. Soc., 1990, 112,
5525.
26 V. Gramlich and G. Consiglio, Helv. Chim. Acta, 1979, 62, 1016.
27 L. M. Venanzi, J. Chem. Soc., 1958, 719.
28 B. Corain, M. Basato, G. Favero, P. Rosano and G. Valle, Inorg.
Chim. Acta, 1984, 85, L27; H. Hope, M. M. Olmstead, P. P. Power
and M. Viggiano, Inorg. Chem., 1984, 23, 326.
29 D. G. Holah, A. N. Hughes and N. I. Khan, Can. J. Chem., 1984, 62,
1016.
30 L. Manojlovic-Muir, K. W. Muir and M.-A. Rennie, Acta
Crystallogr., Sect. C, 1995, 51, 1533.
31 C. P. Casey, G. T. Whiteker, M. G. Melville, L. M. Petrovich, J. A.
Gavney and D. R. Powell, J. Am. Chem. Soc., 1992, 114, 5535.
32 H. M. Powell, D. J. Watkin and J. B. Wilford, J. Chem. Soc. A, 1971,
1803; J. K. Stalick and J. A. Ibers, Inorg. Chem., 1969, 8, 1090.
33 A. de Meijere and F. E. Meyer, Angew. Chem., Int. Ed. Engl., 1994,
33, 2379.
34 C. A. Tolman, W. C. Seidel, J. D. Druliner and P. J. Domaille,
Organometallics, 1984, 3, 33.
35 T. V. RajanBabu and A. L Casalnuovo, J. Am. Chem. Soc., 1992,
114, 6265.
21 M. D. Fryzuk, P. A. McNeil, S. J. Rettig, A. S. Secco and J. Trotter,
Organometallics, 1982, 1, 918.
22 S.-T. Liu, G.-J. Liu, C.-H. Yieh, M.-C. Cheng and S.-M. Peng,
J. Organomet. Chem., 1990, 387, 83; R. Busby, M. B. Hursthouse,
P. S. Jarrett, C. W. Lehmann, K. M. A. Malik and C. Philips,
J. Chem. Soc., Dalton Trans., 1993, 3767; A. L. Spek, B. P. van
Eijck, R. J. F. Jans and G. van Koten, Acta Crystallogr., Sect. C,
1987, 43, 1878; F. Bachechi and L. Zambonelli, Acta Crystallogr.,
Sect. C, 1992, 48, 788.
36 g-NMR, version 3.6, Ivory Soft, Cherwell Scientific, Oxford, 1996.
37 B. Bogdanovic, M. Kröner and G. Wilke, Liebigs Ann. Chem., 1966,
699, 1; R. A. Schunn, Inorg. Synth., 1974, 15, 5.
38 SDP program (Structure Determination Package), V5.0, B. A. Frenz
& Associates, Inc., College Station, TX, 1989.
23 G. Garton, D. E. Henn, N. M. Powell and L. M. Venanzi, J. Chem.
Soc., 1963, 3625.
24 J. A. J. Jarvis, R. H. B. Mais and P. G. Owston, J. Chem. Soc. A,
1968, 1473.
25 P. T. Greene and L. Sacconi, J. Chem. Soc. A, 1970, 866.
Received 23rd March 1998; Paper 8/02269K
2988
J. Chem. Soc., Dalton Trans., 1998, Pages 2981–2988