Formation of anionic PdX3(PPh3)- complexes by reaction of halide ions with PdX2(PPh3)2

Article abstract of DOI:10.1002/(sici)1099-0682(199907)1999:7<1081::aid-ejic1081>3.0.co;2-2

Halide ions react with PdX₂(PPh₃)₂ (X = Cl, Br, I) complexes in an equilibrated reaction in which the phosphane ligand is substituted by the halide X^- to form anionic complexes PdX₃(PPh₃)

Full text of DOI:10.1002/(sici)1099-0682(199907)1999:7<1081::aid-ejic1081>3.0.co;2-2

FULL PAPER
Formation of Anionic PdX3(PPh3)– Complexes by Reaction of Halide Ions
with PdX2(PPh3)2
[a]
[a]
[a]
¨
Christian Amatore,* Anny Jutand,* and Loıc Mottier
Keywords: Palladium / Anionic complexes / Phosphane substitution
Halide ions react with PdX2(PPh3)2 (X = Cl, Br, I) complexes
PdX3(PPh3)– in THF and DMF. The equilibrium constants
in an equilibrated reaction in which the phosphane ligand is have been determined by cyclic voltammetry.
substituted by the halide X– to form anionic complexes
signals appeared in the presence of 120 equivalents of
nBu4NCl (Figure 1): one signal at δ ϭ Ϫ5.27 characteristic
Introduction
It has been established that halide ions are good ligands
of the phosphane PPh3 and a second one at δ ϭ 27.42 as-
for palladium(0) complexes.[1Ϫ3] The best way to investigate
signed to the anionic species PdCl3(PPh3)Ϫ by comparison
the effect of halide ions on the structure and reactivity of
palladium(0) complexes in oxidative addition with aryl hal-
with an authentic sample of [PdCl3(PPh3)Ϫ, nBu4Nϩ].[7]
ides[2Ϫ5] was to generate them by electrochemical reduction
of PdX2(PPh3)2 in the presence of various amounts of hal-
ides XϪ (X ϭ Cl,[2][3] Br,[3] I[3]) in THF. However, during
the course of these reactions, it was observed that the re-
duction peak of PdX2(PPh3)2 was affected by the presence
of added XϪ so that its usual single bielectronic reduction
peak became split into two peaks whose relative currents
depended on the halide concentration: the higher the con-
centration of XϪ, the higher the second reduction peak cur-
rent relative to the first one.[2][3] This phenomenon was first
Figure 1. 31P-NMR spectrum (162 MHz) of
a solution of
discarded and tentatively interpreted as the formation of
an intermediate palladium(I) complex by reduction of the
palladium(II) complex. Its stabilization by halide ions as an
PdCl2(PPh3)2 (3 mmol·dmϪ3) and nBu4NCl (360 mmol dmϪ3) in 3
mL of THF and 0.2 mL of [D6]acetone with H3PO4 as an exter-
nal reference
2Ϫ
anionic 18-electron dimer e.g. [PdI(PPh3)2X2]2 would in-
deed make it less easily reduced than PdIIX2(PPh3)2, thus
The signal of PdCl2(PPh3)2 was still present even in the
presence of a large excess of chloride ions suggesting that
ClϪ substituted one phosphane from PdCl2(PPh3)2 through
an equilibrated reaction (Equation 1).
giving rise to the second reduction peak.[2][3]
However, careful investigation of this reaction led us now
to reconsider our first tentative hypothesis and to establish
that reactions of halides XϪ with PdX2(PPh3)2 result in sub-
stitution of one phosphane ligand with formation of an
anionic PdX3(PPh3)Ϫ species; this palladium(II) complex is
more difficult to reduce. In other words, the two peaks re-
present the reduction of two different palladium(II) com-
plexes.[6]
The dynamic of this equilibrium is relatively slow com-
pared to the time-scale of the NMR experiment since three
distinct narrow signals were observed for PdCl2(PPh3)2,
PdCl3(PPh3)Ϫ, and PPh3 (Figure 1). Similar equilibria have
been observed using 31P-NMR spectroscopy by treating
nBu4NBr with PdBr2(PPh3)2 and nBu4NI with PdI2(PPh3)2
in THF (Equation 2, Table 1).
Results and Discussion
Evidence of the Formation of PdX3(PPh3)؊
Complexes by Reaction of X؊ and PdX2(PPh3)2 by
31P-NMR Spectroscopy in THF and DMF
The 31P-NMR spectra of PdCl2(PPh3)2 (3 mmol dmϪ3
)
in THF exhibited a single signal at δ ϭ 23.80. Two new
In DMF, in the absence of any added halide ions, the
31P-NMR spectra of PdCl2(PPh3)2 exhibited three signals.
The major signal at δ ϭ 24.4 is assigned to PdCl2(PPh3)2.
The two other minor (10%) but of same magnitude signals
are assigned to PPh3 (δ ϭ Ϫ5.54 ) and to the dimer
[a]
´
´
Ecole Normale Superieure, Departement de Chimie, CNRS
URA 1679,
24 Rue Lhomond, F-75231 Paris Cedex 5, France
Fax: (internat.) ϩ 33-1/44323325
E-mail: Amatore@ens.fr
Anny.Jutand@ens.fr
Eur. J. Inorg. Chem. 1999, 1081Ϫ1085
WILEY-VCH Verlag GmbH, D-69451 Weinheim, 1999
1434Ϫ1948/99/0707Ϫ1081 $ 17.50ϩ.50/0
1081

FULL PAPER
Table 1. 31P-NMR signals and reduction peak potentials of palladium(II) complexes involved in the equilibrium PdX2(PPh3)2 ϩ XϪ
v
PdX3(PPh3)Ϫ ϩ PPh3, with nBu4Nϩ as the countercation
X
Cl
Br
I
Cl
Br
I
δ[a]
δ[a]
22.57
22.86
29.3
δ[a]
13.34
13.35
30.70
Ep [V][b]
Ϫ1.040
Ϫ1.015
Ϫ1.255
Ϫ1.370
Ep [V][b]
Ϫ0.830
Ep [V][b]
Ϫ0.725
PdX2(PPh3)2
PdX2(PPh3)2
PdX3(PPh3)Ϫ
PdX3(PPh3)Ϫ
THF
DMF
THF
DMF
23.80
24.36
27.42
28.12
Ϫ1.075
Ϫ1.025
[b]
[a]162 MHz versus H3PO4 as external standard. Ϫ
Volt versus SCE at a gold disk electrode at 0.2 V sϪ1 at 25 °C.
Pd2Cl4(PPh3)2 (δ ϭ 33.54) by comparison with authentic valent complex PdIICl3(PPh3)Ϫ. This assumption has been
samples.[9] These results suggest that the equilibrium ac- confirmed by the cyclic voltammetry of an authentic sample
cording to Equation 3[10] operates in DMF and not in THF of [PdCl3(PPh3)Ϫ, nBu4Nϩ][7] performed in THF which ex-
and is frozen on the time-scale of 31P-NMR spectroscopy.
hibited the unique reduction peak R2 (Figure 2b). More-
over, when 10 equivalents of PPh3 were added to
PdCl3(PPh3)Ϫ, the reduction peak R2 of PdCl3(PPh3)Ϫ dis-
appeared and was replaced by the reduction peak R1 of
PdCl2(PPh3)2 (Figure 2b), evidencing the backward reac-
tion of equilibrium 2 (X ϭ Cl).
In DMF, the 31P-NMR spectra of PdBr2(PPh3)2 also ex-
hibited three signals. However, the signal at δ ϭ 25.47, as-
signed to the dimer Pd2Br4(PPh3)2, was detected at traces
level compared to that of PdBr2(PPh3)2 at δ ϭ 22.86. A
solution of PdI2(PPh3)2 exhibited a single signal at δ ϭ
13.40 in DMF suggesting that the dimer Pd2I4(PPh3)2 was
not formed.
The formation of anionic species was also observed by
performing cyclic voltammetry on mixtures of nBu4NBr
and PdBr2(PPh3)2, as well as on mixtures of nBu4NI and
PdI2(PPh3)2 in THF. The reduction peak potentials of the
neutral and anionic complexes are collected in Table 1.
From 31P-NMR spectroscopy, the equilibrium according
to Equation 2 was found to be slow compared to the time-
scale of the NMR experiment. In cyclic voltammetry, the
reduction of PdX2(PPh3)2 at R1 is governed by a CE mecha-
nism.[11] If the equilibrium according to Equation 2 is fast
compared to the time-scale of the cyclic voltammetry, re-
duction of PdX2(PPh3)2 at R1 should result in a shift of the
equilibrium to its left-hand side by the continuous con-
sumption of PdX2(PPh3)2 due to its reduction in the dif-
fusion layer. A dynamic concentration should be measured
which becomes larger as the scan rate is reduced. If not,
thermodynamic equilibrium concentrations can be deter-
mined. The ratio r ϭ i/i0 [i: reduction peak current at R1 of
PdCl2(PPh3)2 in the presence of 100 equivalents of
nBu4NCl; i0: same current but in the absence of added
nBu4NCl] was then plotted versus the logarithm of the scan
rate (Figure 3a).
Determination of the Equilibrium Constant KX by
Cyclic Voltammetry in THF and DMF
As recalled in the introduction, progressive addition of
nBu4NCl to a solution of PdCl2(PPh3)2 (1.5 mmol·dmϪ3 in
THF containing nBu4NBF4 0.3 mol·dmϪ3) provoked a split
of its bielectronic reduction peak R1 at Ep ϭ Ϫ1.04 V versus
SCE into two reduction peaks, R1 and R2 at Ep ϭ Ϫ1.25 V
versus SCE (Figure 2a).[2][3]
This ratio clearly does not depend on the scan rate which
means that within the time-scale of the cyclic voltammetry
investigated here (0.2 < v < 20 V sϪ1), equilibrium 2 is
frozen, so that the thermodynamic concentration of
PdCl2(PPh3)2 in the equilibrium can be determined by the
ratio r ϭ i/i0 ϭ [PdCl2(PPh3)2]equil/[PdCl2(PPh3)2]0. This al-
lows the determination of the equilibrium constant KCl. In
Figure 2. Cyclic voltammetry performed in THF (containing
nBu4NBF4, 0.3 mol·dmϪ3) at a steady gold disk electrode (i.d. ϭ
0.5 mm) with a scan rate of 2 V sϪ1, at 25 °C: (a) PdCl2(PPh3)2
(1.5 mmol·dmϪ3) in the presence of 0 (——), 14 ( Ϫ Ϫ Ϫ ), 85 ( ···· )
equivalents of nBu4NCl; (b) [PdCl3(PPh3)Ϫ, nBu4Nϩ]
(0.8 mmol·dmϪ3): alone (——); in the presence of 10 equivalents
of PPh3 ( ···· )
From the 31P-NMR investigation reported above, it can the present case, the dynamics of equilibrium 2 are not ac-
be inferred that the second reduction peak is not relevant cessible because it would require scan rates much lower
to the reduction of intermediate palladium(I) complexes as than 0.2 V sϪ1, which could not be used since diffusion pro-
earlier proposed[2][3] but to the reduction of the anionic di- cesses would be contaminated by convection.
1082
Eur. J. Inorg. Chem. 1999, 1081Ϫ1085

Formation of Anionic PdX3(PPh3)Ϫ Complexes by Reaction of Halide Ions with PdX2(PPh3)2
FULL PAPER
PdX2(PPh3)2 (i ϰ D1/2 [12]
)
does not vary with the exper-
imental conditions i.e. when the concentration of nBu4NX
and consequently the ionic strength increases. This has been
checked by performing cyclic voltammetry of PdX2(PPh3)2
for X ϭ Cl, Br, I, in the presence of various amounts of
nBu4NX but with a large excess of PPh3 (10 equivalents),
so that the equilibrium according to Equation 2 was com-
pletely shifted to its left-hand side and consequently only
the reduction peak R1 of PdX2(PPh3)2 was observed. Under
these conditions, the reduction peak current
i of
PdX2(PPh3)2, at a constant scan rate, was found to vary
with the concentration of nBu4NX (X ϭ Cl, Br). This is
illustrated in Figure 3b where the ratio rD ϭ i/i0 [i: reduction
peak current at R1 of PdCl2(PPh3)2 in the presence of n
equivalents of nBu4NCl and 10 equivalents of PPh3; i0:
same current but in the absence of added nBu4NCl] was
plotted versus log [nBu4NCl]. The ratio rD differs from
unity at high chloride concentrations. This suggests that the
diffusion coefficient of PdCl2(PPh3)2 is not constant but de-
creases upon increasing the concentration of chloride ions
because of the higher and higher viscosity of the solution.
Consequently, the equilibrium constant KX was determined
as explained above but by plotting the ratio i/(i0 ϫ rD) in-
stead of i/i0 versus log n (Figure 4). The values of the equi-
librium constant KX (X ϭ Cl, Br, I) are collected in Table 2.
Figure 3. (a) Variation of the ratio r ϭ i/i0 versus the logarithm of
the scan rate (v in V·sϪ1); i: reduction peak current at R1 of
PdCl2(PPh3)2, 1.5 mmol·dmϪ3 in THF (containing nBu4NBF4,
0.3 mol·dmϪ3) in the presence of 100 equivalents of nBu4NCl; i0:
same current but in the absence of added nBu4NCl at 20 °C; the
reduction peak currents were measured at steady gold disk electro-
des (i.d. ϭ 0.5 mm for v < 5 V·sϪ1 and i.d. ϭ 0.125 mm for v > 5
V·sϪ1); (b) variation of the ratio rD ϭ i/i0 versus the logarithm of
the concentration of nBu4NCl; i: reduction peak current at R1 of
PdCl2(PPh3)2, 1.5 mmol·dmϪ3 in THF (containing nBu4NBF4,
0.3 mol·dmϪ3) in the presence of n equivalents of nBu4NCl and 10
equivalents of PPh3; i0: same current but in the absence of added
nBu4NCl; the reduction peak currents were measured at a steady
gold disk electrode (i.d. ϭ 0.5 mm) with a scan rate of 0.2 V·sϪ1
at 20 °C
Table 2. Equilibrium constant KX for the equilibrium PdX2(PPh3)2
ϩ XϪ v PdX3(PPh3)Ϫ ϩ PPh3, with nBu4Nϩ as the countercation
at 25 °C
X
Cl
Br
I
10ϩ3KX
10ϩ3KX
THF
DMF
10 (±1)
2 (±1)
0.9 (±0.1)
n.d.
1.7 (±0.3)
n.d.
From this expression one obtains
r2 Ϫ 2r(1 ϩ nKX/2) ϩ 1 ϭ 0
so that
Figure 4. Determination of the equilibrium constant KCl (see Equa-
tions 2 and 4) in THF at 20 °C; plot of i/(i0 ϫ rD) versus the loga-
rithm of n (equivalents of nBu4NCl); i: reduction peak current at R1
of PdCl2(PPh3)2, 1.5 mmol·dmϪ3 in THF (containing nBu4NBF4,
0.3 mol·dmϪ3) in the presence of various amounts of nBu4NCl; i0:
same current but in the absence of added nBu4NCl; the reduction
peak currents were measured at a steady gold disk electrode (i.d. ϭ
0.5 mm) with a scan rate of 0.2 V·sϪ1; for the definition of rD see
caption of Figure 3b; the solid lines are the theoretical predictions
according to Equation 4
r ϭ 1 ϩ nKX/2 Ϫ {(nKX/2)2 ϩ nKX}1/2
(4)
The ratio r ϭ i/i0 was measured for various amounts of XϪ
[n equivalents/PdII] at a given scan rate. Plotting the exper-
imental value of r versus log n at a given rate constant and
comparison with the theoretical curve should allow the de-
The equilibrium constant KCl has been determined at dif-
termination of KX. However, this calculation is based on ferent temperatures in THF. The equilibrium according to
the fact that the reduction peak current i at R1 is pro- Equation 2 favors production of the anionic species at
portional to the concentration of the species PdX2(PPh3)2 higher temperatures. The Arrhenius plot of lnKCl versus 1/
which is in fact true if the diffusion coefficient D of T (Figure 5) allows the determination of the activation par-
Eur. J. Inorg. Chem. 1999, 1081Ϫ1085
1083

FULL PAPER
ameters from the slope and intercept, so that ∆H0 ϭ 20 kJ complexes, Pt2Cl4(PR3)2.[14] However, PdX2(PPh3)2 remains
molϪ1 and ∆S0 ϭ Ϫ 20 J KϪ1
.
the more stable complex in THF and DMF for moderate
halide ion concentrations. The phosphane substitution is
easier for chloride ions which reveals the higher nucleo-
philic affinity of ClϪ for the palladium(II) center than that
of BrϪ or IϪ. This can be understood as an interaction be-
tween “hard” Lewis acids and bases. The substitution of the
phosphane by the chloride is more efficient in THF than in
DMF probably due to the lower capacity of THF relative
to DMF to solvate chloride ions compared to the larger
anionic species PdX3LϪ.
Conversely, we have reported that the affinity of ClϪ for
palladium(0) centers is less than that of BrϪ or IϪ which
can be rationalized as an interaction between “soft” cen-
ters.[3] Those results evidence the ability of halide ions to
coordinate both palladium(0) and palladium(II) complexes
and thus probably to influence further reactions involving
such complexes. We have already reported the consequences
of the ligation of palladium(0) complexes by halide[3Ϫ5] or
acetate ions[5][15] on the mechanism of palladium-catalyzed
reactions. Although palladium(II) dihalide complexes are
less involved as promoters of catalytic reactions than pal-
ladium(0) complexes, arylpalladium(II) halide complexes
are postulated to be crucial intermediates in palladium(0)-
catalyzed reactions involving aryl halides. Reactions of hal-
ide ions XЈϪ with ArPdX(PPh3)2 complexes do not result
in a substitution of one phosphane ligand but rather substi-
tution of the halide XϪ to form neutral ArPdXЈ(PPh3)2
complexes.[16] In fact, we have established that
ArPdX(PPh3)2 complexes are not reactive intermediates in
cross-coupling[3][5] or Heck[5][15] reactions and the effect of
chloride or acetate anions on the course of these reactions
is to produce new anionic or neutral complexes such as
ArPdX(Cl)(PPh3)2Ϫ, ArPdX(OAc)(PPh3)2Ϫ, ArPd(OAc)-
(PPh3)2 where the two phosphane ligands remain attached
to the palladium(II) center
Figure 5. Arrhenius plot for the equilibrium PdCl2(PPh3)2 ϩ ClϪ
v PdCl3(PPh3)Ϫ ϩ PPh3 (with nBu4N as counter-cation) in THF
containing nBu4NBF4 (0.3 mol·dmϪ3); variation of ln KCl versus
1/T
The small value of ∆S0 is indicative of a reaction in which
there is no variation of the numbers of species in solution.
Miscellaneous Reactions
All reactions investigated above concern the substitution
of the phosphane ligand by the same halide XϪ as those
present in PdX2(PPh3)2. Reactions are more complicated
when different halides are considered. For example: when
30 equivalents of nBu4NI were added to a solution of
PdCl2(PPh3)2 (2.4 mmol·dmϪ3) in THF, PPh3 was detected
in the 31P-NMR spectrum as a narrow signal indicative of
the substitution of the phosphane ligand by one iodide ion
which generates anionic species: PdCl2I(PPh3)Ϫ.
PdCl2(PPh3)2 ϩ IϪ Ǟ PdCl2I(PPh3)Ϫ ϩ PPh3
(5)
However, this complex could not be characterized since
an unresolved broad signal was present at δ ϭ 17.5 suggest-
ing that scrambling reactions occurred by equilibria such as
in Equation 6 and/or 7.
Experimental Section
General: 31P-NMR spectra were recorded with a Bruker W 400
spectrometer (162 MHz) with H3PO4 as an external reference. Cyc-
lic voltammetry was performed with a home-made potentiostat and
a waveform generator Tacussel GSTP4 or EGG. The cyclic voltam-
mograms were recorded with a Nicolet 301 oscilloscope.
Under these conditions, several neutral or anionic com-
plexes may be formed: PdICl(PPh3)2, PdI2(PPh3)2,
PdCl2I(PPh3)Ϫ, PdClI2(PPh3)Ϫ, PdI3(PPh3)Ϫ involved in
different and simultaneous equilibria, responsible for the Chemicals: DMF was distilled from calcium hydride under vacuum
unresolved broad signal observed in the 31P-NMR spec-
and kept under argon. Commercial nBu4NCl, nBu4NBr, nBu4NI
(Acros) were used after crystallisation. They were melted and dried
trum.
under vacuum. PdCl2(PPh3)2, PdBr2(PPh3)2, PdI2(PPh3)2, was pre-
pared according to described procedures.[17]
Discussion and Conclusion
Electrochemical Set Up and Electrochemical Procedure for Cyclic
Voltammetry: Experiments were carried out in a three-electrode cell
Halide ions can substitute a phosphane ligand in
PdX2(PPh3)2 complexes to afford anionic species
PdX3(PPh3)Ϫ by an equilibrium. Similar anionic
NiCl3(PR3)Ϫ (R ϭ isopropyl) complexes have been reported
to be formed by phosphane substitution[13] while
PtCl3(PR3)Ϫ complexes are usually synthesized by allowing
connected to a Schlenk line. The counter electrode was a platinum
wire of ca. 1 cm2 apparent surface area; the reference was a satu-
rated calomel electrode (Tacussel) separated from the solution by
a bridge filled with 3 mL of THF or DMF containing nBu4NBF4
(0.3 mol·dmϪ3).
General Procedure: 12 mL of THF or DMF containing nBu4NBF4
chloride ions to react with phosphane-deficient dimeric (0.3 mol·dmϪ3) was poured into the cell followed by PdCl2(PPh3)2
1084
Eur. J. Inorg. Chem. 1999, 1081Ϫ1085

Formation of Anionic PdX3(PPh3)Ϫ Complexes by Reaction of Halide Ions with PdX2(PPh3)2
FULL PAPER
31P-NMR spectrum at δ ϭ 27.42 similar to the signal of the
complex formed in Equation 1.
(16.8 mg, 0.024 mmol) and various amounts of nBu4NCl. Cyclic
voltammetry was performed at steady gold disk electrodes with
various scan rates.
[8]
R. J. Goodfellow, P. L. Goggin, D. A. Duddell, J. Chem. Soc.
A 1968, 504Ϫ506.
[9]
J. Chatt, L. M. Venanzi, J. Chem. Soc. 1957, 2351Ϫ2356.
R. Bardi, A. M. Piazzesi, A. Del Pra, G. Cavinato, L. Toniolo,
Inorg. Chim. Acta 1983, 75, 15Ϫ19.
[10]
[11]
[12]
A. J. Bard, L. R. Faulkner, Electrochemical Methods, Wiley,
New York, 1980, p. p. 443Ϫ449.
[1]
E. I. Negishi, T. Takahashi, K. Akiyoshi, J. Chem. Soc., Chem.
Commun. 1986, 1338Ϫ1339.
A. J. Bard, L. R. Faulkner, Electrochemical Methods, Wiley,
New York, 1980, p. 213Ϫ248.
[2]
C. Amatore, M. Azzabi, A. Jutand, J. Organomet. Chem. 1989,
363, C41ϪC45.
[13] [13a] G. G. Eberhardt, W. P. Griffin, J. Catal. 1970, 16, 245Ϫ253.
[3]
[13b]
C. Amatore, M. Azzabi, A. Jutand, J. Am. Chem. Soc. 1991,
Ϫ
Y. Chauvin, unpublished results, personal communi-
113, 1670Ϫ1677.
cation, 1996.
[4]
[14]
[15]
[16]
[17]
C. Amatore, A. Jutand, A. Suarez, J. Am. Chem. Soc. 1993,
P. L. Goggin, R. J. Goodfellow, S. R. Haddock, B. F. Taylor, I.
R. H. Marshall, J. Chem. Soc., Dalton Trans. 1976, 459Ϫ467.
115, 9531Ϫ9541.
[5]
´
For a review see: C. Amatore, A. Jutand, J. Organomet. Chem.
C. Amatore, E. Carre, A. Jutand, M. A. MЈBarki, G. Meyer,
1999, 573, 254Ϫ278.
Organometallics 1995, 14, 5605Ϫ5614.
M. C. Pilon, V. V. Grushin, Organometallics 1998, 17,
1774Ϫ1781.
F. R. Hartley, Organomet. Chem. Rev., Sect. A 1970, 6, 119.
Received January 7, 1999
[6]
L. Mottier, part of Ph.D. thesis, University Paris VII, 1995.
[7]
[PdCl3(PPh3)Ϫ, nBu4Nϩ] has been synthesized independently by
treating Pd2Cl4(PPh3)2 with two equivalents of nBu4NCl, in
THF according to
a
reported procedure:[8] Pd2Cl4-
(PPh3)2 ϩ 2 ClϪ Ǟ 2 PdCl3(PPh3)Ϫ. It was characterized by its
[I99007]
Eur. J. Inorg. Chem. 1999, 1081Ϫ1085
1085