Rhodium complexes possessing S-phosphinite ligands with or without an amino group: Application to hydroformylation of styrene

Article abstract of DOI:10.1016/j.ica.2004.02.005

A chelate cationic rhodium(I) complex with a hemilabile amino- and sulfur-containing phosphinite ligand has been synthesized and, according to the NMR data, the ligand is bound to the metal in a P,S-bidentate coordination mode without any Rh-N interaction. This complex efficiently catalyzes the hydroformylation of styrene. The chelate rhodium complex with the analogous ligand without the amino group has also been synthesized and examined as a catalyst for the same hydroformylation reaction. The reaction rate is higher using the former complex compared to the latter one without the amino group, with, however, a slightly lower regioselectivity towards the formation of the branched aldehyde.

Full text of DOI:10.1016/j.ica.2004.02.005

Inorganica Chimica Acta 357 (2004) 2850–2854
www.elsevier.com/locate/ica
Rhodium complexes possessing S-phosphinite ligands with or
without an amino group: application to hydroformylation of styrene
a,
a
a
b
Ioannis D. Kostas *, Barry R. Steele , Fotini J. Andreadaki , Vladimir A. Potapov
a
Institute of Organic and Pharmaceutical Chemistry, National Hellenic Research Foundation, 48 Vas. Constantinou Ave., 11635 Athens, Greece
b
Favorsky Irkutsk Institute of Chemistry, Siberian Branch of the Russian Academy of Sciences, 1 Favorsky Street, 664033 Irkutsk, Russia
Received 4 December 2003; accepted 7 February 2004
Available online 27 February 2004
Abstract
A chelate cationic rhodium(I) complex with a hemilabile amino- and sulfur-containing phosphinite ligand has been synthesized
and, according to the NMR data, the ligand is bound to the metal in a P; S-bidentate coordination mode without any Rh–N in-
teraction. This complex efficiently catalyzes the hydroformylation of styrene. The chelate rhodium complex with the analogous
ligand without the amino group has also been synthesized and examined as a catalyst for the same hydroformylation reaction. The
reaction rate is higher using the former complex compared to the latter one without the amino group, with, however, a slightly lower
regioselectivity towards the formation of the branched aldehyde.
Ó 2004 Elsevier B.V. All rights reserved.
Keywords: Hemilabile ligands; Phosphinites; P; S-ligands; Rhodium complexes; Hydroformylations; Homogeneous catalysis
1. Introduction
been of great interest due to the improved catalytic ac-
tivity of their transition-metal complexes in certain ho-
mogeneously catalyzed reactions [10–13]. Particular
attention has focused on phosphine ligands containing
nitrogen [14–23] or oxygen [10,24–27]. The use of P; S-
ligands in catalysis, on the other hand, is relatively un-
explored compared to P; N- and P; O-ligands due to the
fear of metal poisoning by sulfur [28–32].
We have previously reported several functionalized
hemilabile P; N-, P; N; P- and N; P; N-ligands and their
application in rhodium-catalyzed hydroformylation,
hydroaminomethylation and hydrogenation [33–37].
Recently, we reported a hemilabile amino- and sulfur-
containing phosphinite ligand (1) and the analogous
sulfur-containing phosphinite (2) without the amino
group, as catalyst precursors for the palladium-cata-
lyzed Heck reaction of aryl bromides with styrene
(Fig. 1) [38]. Treatment of PdCl2(NCPh)2 with one
equivalent of ligand 1 yielded the corresponding chelate
palladium complex, in which the ligand is bound to the
metal in a P; S-bidentate coordination mode without
any Pd–N interaction. Attempts to synthesize a mono-
meric chelate palladium complex with ligand 2 were
unsuccessful and we found that the role of the amino
Rhodium-catalyzed hydroformylation as an attrac-
tive synthetic transformation is a very well-established
reaction with enormous academic and industrial inter-
est, and represents one of the most important homoge-
neously catalyzed reactions world-wide [1–4]. The design
and synthesis of new ligands with improved catalytic
activity and selectivity towards hydroformylation is still
a research subject of great significance. The stereo-
electronic properties of the ligands are crucial for their
catalytic activity and selectivity, and since the intro-
duction of the concept of the ‘‘natural bite angle’’ for
metal complexes with bidentate phosphines by Casey
and Whiteker [5], the influence of the bite angle as well
as steric and electronic effects on the hydroformylation
has received much attention [6–9].
Ligands with mixed donors of different coordination
abilities, in particular those which contain ‘‘soft’’ and
‘‘hard’’ donor atoms, so-called hemilabile ligands, have
* Corresponding author. Tel.: +30-210-7273878; fax: +30-210-
7273877.
E-mail address: ikostas@eie.gr (I.D. Kostas).
0020-1693/$ - see front matter Ó 2004 Elsevier B.V. All rights reserved.
doi:10.1016/j.ica.2004.02.005

I.D. Kostas et al. / Inorganica Chimica Acta 357 (2004) 2850–2854
2851
which the ligand is bound to rhodium via phosphorus
and sulfur atoms, while nitrogen is situated away from
the coordination sphere around the metal. Complex 4
displays similar behavior. Its 31P NMR spectrum (d
135.06, d, JRhP ¼ 164:03 Hz) is essentially the same as
that of complex 3, indicating an analogous chelation. It
is clear, therefore, that the coordination ability of ligand
2 towards rhodium is in contrast to that observed to-
wards palladium, in which the analogous chelation
could not be achieved [38].
Fig. 1.
group in the formation of a P; S-chelate complex as well
as in the Heck reaction was decisive. Within our ongoing
research, we wish to report here the coordination be-
havior of the two above-mentioned ligands towards
rhodium. Rhodium complexes with both of these li-
gands were synthesized and applied to the hydro-
formylation of styrene, in order to investigate the
influence of the amino group on the catalyst behavior of
the complex in this reaction. Although phosphinites
have been widely used in homogeneous catalysis
[16,35,37–43], to our knowledge, sulfur-containing
phosphinites have not been applied to rhodium-cata-
lyzed hydroformylation until now.
In the 1H- and 13C NMR spectra of 3 and 4, the
olefinic COD resonances are non-equivalent. In complex
3, the olefinic COD resonances are split into four sig-
nals, two signals for the olefinic protons or carbons trans
to the phosphorus atom (1H NMR: 5.68, 5.60–5.47; 13
C
NMR: 113.40, 108.45–108.25), and two signals for the
olefinic protons or carbons trans to the sulfur atom (1H
NMR: 5.15, 5.02–4.89; 13C NMR: 86.99, 84.56). The
COD behavior in complex 4 is quite similar to that
observed in complex 3 for the olefinic protons and car-
bons (1H NMR: 5.76, 5.61–5.49, 5.29, 5.05–4.92; 13C
NMR: 112.68–112.45, 106.89–106.69, 85.26, 82.65).
2. Results and discussion
2.2. Hydroformylation
2.1. Synthesis and characterization of the rhodium
complexes
The rhodium complexes 3 and 4 were applied to the
hydroformylation of styrene using a high substrate:cat-
alyst ratio of 1455:1 (Scheme 2, Table 1). The reaction
was first catalyzed by complex 3 under variable condi-
tions of pressure and temperature. The catalyst displays a
high chemoselectivity for aldehydes (over 98%), and the
variation observed in the regioselectivities towards the
branched aldehyde 6 is as expected for hydroformylation
of styrene using rhodium systems. At 100 bar total
pressure of syngas, the conversion of styrene was almost
quantitative after 22 h at 30 °C, with a 96.0% regiose-
lectivity towards 6 (entry 1). Increasing the temperature
dramatically increases the reaction rate but with a lower
regioselectivity. On lowering the syngas pressure to 50 or
30 bar, the reaction rate as well as the regioselectivity is
decreased. However, even under these mild reaction
conditions, the activity of complex 3 is still acceptable.
The catalytic activity of complex 4 was tested for a
representative range of reaction conditions (entries 2, 4,
7). Under identical conditions, it displays lower TONs
compared to those obtained for complex 3. The higher
activity observed with the amino-substituted complex 3
compared to that observed with complex 4 is not sur-
prising, since it is widely accepted that ligands contain-
ing electron-withdrawing substituents lead to higher
Treatment of [Rh(COD)2]BF4 in dichloromethane
solution with one equivalent of the ligand 1 or 2 yielded
the cationic rhodium complexes 3 or 4, respectively, in
high yields (Scheme 1).
The coordination mode in the complexes was deter-
mined by spectroscopic techniques. In the 31P NMR
spectrum, complex 3 shows a doublet at d 135.33
(JRhP ¼ 167:3 Hz) as a result of Rh–P interaction
(dPðligand 1Þ112:21Þ. The observation of the [1Rh(COD)]þ
ion (m=z 620) and the absence of higher aggregates in the
ESI MS spectrum suggest the monomeric nature of
complex 3. In the 13C NMR spectrum of 3, the SCH3
resonance is shifted to low field compared to that in the
free ligand, in contrast to the N(CH3)2 resonance which
is the same in both the complex and the free ligand
(SCH3: dCðligand 1Þ 14.38, dCðcomplex 3Þ 18.80; N(CH3)2:
dCðligand 1Þ 45.56, dCðcomplex 3Þ 45:60Þ. These data indicate
the formation of a seven-membered chelate complex in
Scheme 1.
Scheme 2.

2852
I.D. Kostas et al. / Inorganica Chimica Acta 357 (2004) 2850–2854
Table 1
Hydroformylation of styrene catalyzed by rhodium complex 3 or 4
Catalysta
T (°C)
Pb(bar)
Time (h)
Conversion (%)
Rc (%)
Rbr (%)
TONe
c
d
Entry
1
2
3
4
3
4
3
3
4
3
3
3
30
30
40
40
40
60
60
60
40
40
100
100
100
100
100
100
100
100
50
22
22
4
98.8
89.0
85.7
62.8
99.2
98.4
81.0
99.9
95.9
75.5
98.8
99.1
99.2
99.7
99.3
98.0
99.3
99.3
98.8
99.2
96.0
96.3
94.7
96.0
95.3
88.5
90.6
88.9
92.4
87.2
1420
1283
1237
911
3
4
4
5
8
1433
1403
1170
1443
1379
1090
6
1
7
1
8
4
9
10
22
22
30
a A 4 mM solution in CH2Cl2. Styrene:catalyst ¼ 1455:1.
b P ¼ initial total pressure of CO/H2 (1/1).
c Rc ¼ chemoselectivity towards aldehydes.
d Rbr ¼ regioselectivity towards branched aldehyde.
e Turnover number (TON) ¼ aldehydes fraction ꢀ substrate:catalyst ratio.
reaction rates [2]. For Rh-catalyzed hydroformylation
with ligands possessing additional nitrogen donors, it
has been noted also that, despite the absence of a Rh–N
interaction, the existence of an additional coordination
site as a stabilizing group during the course of a metal-
mediated reaction can improve the catalytic efficiency of
the complexes [35,44]. Although complex 3 exhibits a
higher activity compared to 4, the regioselectivity to-
wards the branched aldehyde 6 observed with complex 3
was slightly lower than that observed with 4. Although
mechanistic studies have not been performed yet and the
reasons are not understood, this may perhaps be due to
steric effects, since it is known that the regioselectivity of
hydroformylation is governed by steric factors [8].
chloride hydrate according to a literature procedure [45–
47]. Syntheses of the complexes and catalysis were car-
ried out under argon by using dry and degassed reagents
and solvents. Hydroformylation was performed in a
stainless steel autoclave (300 ml) with magnetic stirring.
Syngas: CO/H2 (1/1). NMR: Bruker AC 300 (300.13
MHz, 75.47 MHz, and 121.50 MHz for 1H, 13C and 31P,
respectively); 1H and 13C NMR shifts were referenced to
the solvent and the 31P NMR shifts were referenced to
external 85% H3PO4. The distinction of the CH, CH2
and CH3 carbons in the 13C NMR spectra was per-
formed by DEPT NMR experiments. ESI MS: Finnigan
MAT TSQ 7000. GC–MS (EI): Varian Saturn 2000 with
a 30 m ꢀ 0.25 mm DB5-MS column. GC: Varian Star
3400 CX with a 30 m ꢀ 0.53 mm DB5 column.
2.3. Conclusion
3.2. Rhodium(1+)-[(1,2,5,6)-1,5-cyclooctadiene]-[2-
[(1-methylthio-S)-3[(diphenylphosphino-P)oxy]propyl]-
In summary, we have synthesized two seven-mem-
bered chelate rhodium complexes with S-phosphinite
ligands, one of which contains an amino group with the
nitrogen situated away from the coordination sphere
around the metal. The amino-substituted complex effi-
ciently catalyzes the hydroformylation of styrene and
exhibits a higher activity compared to the complex
without the amino group, presumably due to the exis-
tence of nitrogen as a stabilizing group at an additional
coordination site during the course of the metal-medi-
ated reaction. However, the regioselectivity towards the
branched aldehyde was slightly lower when the amino
group was present on the aryl ring of the ligand, and this
may perhaps be due to steric effects.
N,N-dimethyl-benzenamine]-tetrafluoroborate(1)) (3)
A solution of the ligand 1 (0.1942 g, 0.475 mmol) in
dichloromethane (15 ml) was added dropwise to the
dark red solution of [Rh(COD)2]BF4 (0.1933 g, 0.476
mmol) in dichloromethane (5 ml) with a dry ice/acetone
cooling bath. The reaction mixture was warmed slowly
to room temperature within 1 h and stirred at this
temperature for an additional 2 h. The resulting orange
solution was evaporated under reduced pressure to a
volume of 1 ml and addition of ether (20 ml) caused the
precipitation of a solid. The solvents were decanted and
the remaining solid was washed with ether (2 ꢀ 10 ml)
and dried, yielding 3 (0.3009 g, 90%) as a mustard-yel-
low solid, m.p. (dec.) 132–140 °C. 1H NMR (CD2Cl2): d
8.06–7.99 (m, 2H, Ar), 7.76–7.67 (m, 3H, Ar), 7.44–7.39
(m, 4H, Ar), 7.28–7.06 (m, 5H, Ar), 5.68 (br m), 5.60–
5.47 (m), 5.15 (br m) and 5.02–4.89 (m) (4 ꢀ 1H, COD–
CH), 4.79–4.73 (m, 1H, CHS), 4.05 and 3.78 (2 ꢀ br m,
2 ꢀ 1H, CH2O), 2.68–2.23 (m, 10H, CH2 and COD–
3. Experimental
3.1. General
Ligands 1 and 2 were prepared as reported recently
[38]. [Rh(COD)2]BF4 was prepared from rhodium tri-

I.D. Kostas et al. / Inorganica Chimica Acta 357 (2004) 2850–2854
2853
CH2), 2.18 (s, 6H, N(CH3)2), 1.78 (s, 3H, SCH3).
References
13C{1H} NMR (CD2Cl2): d 154.00–121.01 (Ar), 113.40
(dd, JRhC ¼ 11:6 Hz, JPC ¼ 5:2 Hz, COD–CH), 108.45–
108.25 (m, COD–CH), 86.99 (d, JRhC ¼ 10:7 Hz, COD–
CH), 84.56 (d, JRhC ¼ 10:7 Hz, COD–CH), 71.61 (s,
CH2OP), 45.60 (s, N(CH3)2), 44.47 (s, CHS), 34.78,
32.92, 31.92, 29.95 and 28.66 (s, CH2 and COD–CH2),
18.80 (s, SCH3). 31P{1H} NMR (CD2Cl2): d 135.33 (d,
JRhP ¼ 167:3 Hz). ESI MS: m=z 620 ([M–BF4]þ). Anal.
Calc. for C32H40BF4NOPRhS (707.42): C, 54.33; H,
5.70; N, 1.98. Found: C, 53.68; H, 5.74; N, 1.56%.
[1] C.D. Frohning, C.W. Kohlpaintner, Hydroformylation, Chapter
2.1.1, in: B. Cornils, W.A. Herrmann (Eds.), Applied Homoge-
neous Catalysis with Organometallic Compounds, vol. 1, VCH,
Weinheim, 1996.
[2] P.W.N.M. van Leeuwen, C. Claver (Eds.), Rhodium Catalyzed
Hydroformylation, Kluwer Academic Publishers, Dordrecht,
2000.
[3] B. Breit, Acc. Chem. Res. 36 (2003) 264.
[4] Survey of hydroformylation for 2002: F. Ungvary, Coord. Chem.
Rev. 241 (2003) 295.
ꢀ
[5] C.P. Casey, G.T. Whiteker, Isr. J. Chem. 30 (1990) 299.
[6] L.A. van der Veen, M.D.K. Boele, F.R. Bregman, P.C.J. Kamer,
P.W.N.M. van Leeuwen, K. Goubitz, J. Fraanje, H. Schenk, C.
Bo, J. Am. Chem. Soc. 120 (1998) 11616.
3.3. Rhodium(1+)-[(1,2,5,6)-1,5-cyclooctadiene]-[[(1-
methylthio-S)-3-[(diphenylphosphino-P)oxy]propyl]-
benzene]-tetrafluoroborate(1)) (4)
[7] L. Gonsalvi, H. Adams, G.J. Sunley, E. Ditzel, A. Haynes, J. Am.
Chem. Soc. 124 (2002) 13597.
[8] Z. Freixa, P.W.N.M. van Leeuwen, Dalton Trans. (2003) 1890.
[9] R.P.J. Bronger, P.C.J. Kamer, P.W.N.M. van Leeuwen, Organo-
metallics 22 (2003) 5358.
Treatment of [Rh(COD)2]BF4 (0.0499 g, 0.123 mmol)
in dichloromethane (5 ml) with ligand 2 (0.0451 g, 0.123
mmol) in dichloromethane (10 ml), as described above
for the synthesis of 3, yielded complex 4 (0.0725 g, 89%)
as a yellow solid, m.p. (dec.) 147–156 °C. 1H NMR
(CDCl3): d 8.06–8.01 (m, 2H, Ar), 7.69 (br m, 3H, Ar),
7.43 (br m, 3H, Ar), 7.25 (br m, 5H, Ar, obscured with the
signal of CHCl3; clearly observed in CD2Cl2 solution),
6.84 (br m, 2H, Ar), 5.76 (br m), 5.61–5.49 (m), 5.29 (br
m) and 5.05–4.92 (m) (4 ꢀ 1H, COD–CH), 3.98 (br m,
1H, CHS), 3.74–3.71 (m, 2H, CH2O), 2.77–2.13 (m, 10H,
CH2 and COD–CH2), 1.92 (s, 3H, SCH3). 13C{1H}
NMR (CDCl3): d 137.16–128.20 (Ar), 112.68–112.45 (m,
COD–CH), 106.89–106.69 (m, COD–CH), 85.26 (d,
JRhC ¼ 12:2 Hz, COD–CH), 82.65 (d, JRhC ¼ 12:2 Hz,
COD–CH), 71.89 (s, CH2OP), 50.57 (s, CHS), 32.58,
31.68, 31.16, 29.28 and 28.54 (s, CH2 and COD–CH2),
14.86 (s, SCH3). 31P{1H} NMR (CD2Cl2): d 135.06 (d,
JRhP ¼ 164:03 Hz). Anal. Calc. for C30H35BF4OPRhS
(664.35): C, 54.24; H, 5.31. Found: C, 53.74; H, 5.19%.
[10] A. Bader, E. Lindner, Coord. Chem. Rev. 108 (1991) 27.
[11] C.S. Slone, D.A. Weinberger, C.A. Mirkin, Prog. Inorg. Chem. 48
(1999) 233.
€
[12] A. Borner, Eur. J. Inorg. Chem. (2001) 327.
[13] P. Braunstein, F. Naud, Angew. Chem. Int. Ed. 40 (2001) 680.
[14] K.N. Gavrilov, A.I. Polosukhin, Russ. Chem. Rev. 69 (2000)
661.
[15] G. Helmchen, A. Pfaltz, Acc. Chem. Res. 33 (2000) 336.
[16] F. Agbossou-Niedercorn, I. Suisse, Coord. Chem. Rev. 242 (2003)
145.
[17] M.L. Clarke, A.M.Z. Slawin, M.V. Wheatley, J.D. Woollins,
Dalton Trans. (2001) 3421.
[18] J. Andrieu, P. Richard, J.-M. Camus, R. Poli, Inorg. Chem. 41
(2002) 3876.
ꢁ
[19] G. Francio, D. Drommi, C. Graiff, F. Faraone, A. Tiripicchio,
Inorg. Chim. Acta 338 (2002) 59.
[20] T. Bunlaksananusorn, K. Polborn, P. Knochel, Angew. Chem.
Int. Ed. 42 (2003) 3941.
ꢀ
[21] L. Barloy, G. Malaise, S. Ramdeehul, C. Newton, J.A. Osborn, N.
Kyritsakas, Inorg. Chem. 42 (2003) 2902.
[22] P. Braunstein, J. Zhang, R. Welter, Dalton Trans. (2003) 507.
[23] C. Borriello, M.E. Cucciolito, A. Panunzi, F. Ruffo, Inorg. Chim.
Acta 353 (2003) 238.
€
[24] N.J. Hovestad, E.B. Eggeling, H.J. Heidbuchel, J.T.B.H. Jastr-
3.4. Hydroformylation of styrene
zebski, U. Kragl, W. Keim, D. Vogt, G. van Koten, Angew.
Chem. Int. Ed. 38 (1999) 1655.
In a typical experiment, styrene (2 ml, 17.456 mmol)
and a 4 mM solution of rhodium complex 3 or 4 in di-
chloromethane (3 ml, 0.012 mmol) were placed under
argon in an oven-dried autoclave, which was then
closed, pressurized with syngas (CO/H2 ¼ 1:1) and
brought to the appropriate temperature. After the re-
quired reaction time, the autoclave was cooled to room
temperature, the pressure was carefully released and the
solution was passed through Celite and analyzed by GC
and GC–MS. Conversions were determined by GC.
[25] X. Bei, H.W. Turner, W.H. Weinberg, A.S. Guram, J. Org. Chem.
64 (1999) 6797.
€
[26] R. Weber, U. Englert, B. Ganter, W. Keim, M. Mothrath, Chem.
Commun. (2000) 1419.
[27] W.-M. Dai, K.K.Y. Yeung, J.-T. Liu, Y. Zhang, I.D. Williams,
Org. Lett. 4 (2002) 1615.
ꢀ
ꢀ
[28] J.C. Bayon, C. Claver, A.M. Masdeu-Bulto, Coord. Chem. Rev.
193–195 (1999) 73.
[29] E. Hauptman, P.J. Fagan, W. Marshall, Organometallics 18
(1999) 2061.
ꢁ
ꢀ
[30] O. Pamies, M. Dieguez, G. Net, A. Ruiz, C. Claver, Organomet-
allics 19 (2000) 1488.
[31] J.R. Dilworth, D. Morales, Y. Zheng, Dalton Trans. (2000)
3007.
Acknowledgements
ꢀ
[32] D. Morales-Morales, R. Redon, Y. Zheng, J.R. Dilworth, Inorg.
Chim. Acta 328 (2002) 39.
The investigation was supported by the Greek Gen-
eral Secretariat of Research and Technology and the
Russian Foundation for Basic Research.
[33] I.D. Kostas, C.G. Screttas, J. Organomet. Chem. 585 (1999) 1.
[34] I.D. Kostas, J. Chem. Res. (S) (1999) 630.
[35] I.D. Kostas, J. Organomet. Chem. 626 (2001) 221.

2854
I.D. Kostas et al. / Inorganica Chimica Acta 357 (2004) 2850–2854
[36] I.D. Kostas, J. Organomet. Chem. 634 (2001) 90.
[37] I.D. Kostas, Inorg. Chim. Acta 355 (2003) 424.
[38] I.D. Kostas, B.R. Steele, A. Terzis, S.V. Amosova, Tetrahedron
59 (2003) 3467.
[42] R.B. Bedford, S.L. Hazelwood, M.E. Limmert, J.M. Brown, S.
Ramdeehul, A.R. Cowley, S.J. Coles, M.B. Hursthouse, Organo-
metallics 22 (2003) 1364.
[43] W. Tang, X. Zhang, Chem. Rev. 103 (2003) 3029.
[44] M.T. Reetz, S.R. Waldvogel, R. Goddard, Tetrahedron Lett. 38
(1997) 5967.
€
[39] V.I. Tararov, R. Kadyrov, T.H. Riermeier, J. Holz, A. Borner,
Tetrahedron Lett. 41 (2000) 2351.
[40] Y.-Y. Yan, T.V. RajanBabu, J. Org. Chem. 66 (2001) 3277.
[41] K. Ohe, K. Morioka, K. Yonehara, S. Uemura, Tetrahedron:
Asymmetry 13 (2002) 2155.
[45] G. Giordano, R.H. Crabtree, Inorg. Synth. 28 (1990) 88.
[46] M.A. Bennett, J.D. Saxby, Inorg. Chem. 7 (1968) 321.
[47] M. Green, T.A. Kuc, S.H. Taylor, J. Chem. Soc. A (1971) 2334.