Synthesis, structural characterization and reactivity of copper(I)-nitrile complexes: Crystal and molecular structure of [Cu(BzCN)4][BF4]

Article abstract of DOI:10.1016/j.poly.2007.04.007

Complexes of general formula [CuL₄][BF₄] (L = benzonitrile - PhCN 2 or phenylacetonitrile - BzCN 3) have been prepared and structurally characterized by NMR spectroscopy and X-ray crystallography. Their structure and reactivity have been compared to the well known [Cu(MeCN)₄][BF₄] (1). The ⁶³Cu line width and the ⁶³Cu chemical shift have been evaluated by varying the temperature and the concentration of the complex 2 in benzonitrile solutions. The phenylacetonitrile solutions of the complex 3 give extremely broad signals which are beyond detection. Accordingly, compound 3 has been studied by ⁶³Cu MAS NMR spectroscopy. The solution NMR data are consistent to the prevalence of dynamic equilibrium between tetra- and low-coordinated species in both complexes. The X-ray structure of 3 revealed that the copper(I) atom sits in a slightly distorted tetrahedral geometry, surrounded by four BzCN ligands.

Full text of DOI:10.1016/j.poly.2007.04.007

Polyhedron 26 (2007) 3871–3875
www.elsevier.com/locate/poly
Synthesis, structural characterization and reactivity of
copper(I)-nitrile complexes: Crystal and molecular structure of
[Cu(BzCN)4][BF4]
a
a,
a
Viviane A.S. Falcomer , Sebastiao S. Lemos *, Guilherme B.C. Martins ,
˜
b
b
b
Gleison A. Casagrande , Robert A. Burrow , Ernesto S. Lang
a
Instituto de Quı´mica, Universidade de Brası´lia, 70904-970 Brası´lia – DF, Brazil
Departamento de Quı´mica, Universidade Federal de Santa Maria, 97105-900 Santa Maria – RS, Brazil
b
Received 3 February 2007; accepted 13 April 2007
Available online 21 April 2007
Abstract
Complexes of general formula [CuL4][BF4] (L = benzonitrile – PhCN 2 or phenylacetonitrile – BzCN 3) have been prepared and
structurally characterized by NMR spectroscopy and X-ray crystallography. Their structure and reactivity have been compared to the
well known [Cu(MeCN)4][BF4] (1). The 63Cu line width and the 63Cu chemical shift have been evaluated by varying the temperature
and the concentration of the complex 2 in benzonitrile solutions. The phenylacetonitrile solutions of the complex 3 give extremely
broad signals which are beyond detection. Accordingly, compound 3 has been studied by 63Cu MAS NMR spectroscopy. The solution
NMR data are consistent to the prevalence of dynamic equilibrium between tetra- and low-coordinated species in both complexes. The
X-ray structure of 3 revealed that the copper(I) atom sits in a slightly distorted tetrahedral geometry, surrounded by four BzCN
ligands.
ꢀ 2007 Elsevier Ltd. All rights reserved.
Keywords: Copper(I)-nitrile complexes; 63Cu NMR spectroscopy; 63Cu MAS NMR spectroscopy; X-ray structure; Dynamic equilibrium
1. Introduction
I = 3/2 but the more abundant 63Cu (69.09%) has been
most applied in NMR spectroscopy. Since it is a quadrupo-
lar nucleus, the 63Cu NMR spectroscopy is in practice
almost restricted to compounds in which the copper(I)
atoms adopt a rigorous tetrahedral geometry, Td. Any
deviation from that geometry will broaden the resonance,
sometimes it becomes beyond detection. For instance, the
resonance due to the compound [Cu(PMe3)4][BF4] will
immediately disappear if it is treated with [Cu(MeCN)4]-
[BF4] due to scrambling low-symmetry products, namely,
[Cu(PMe3)(4ꢀn)(MeCN)n]+ [5]. The effect of varying tem-
perature, solvent, counterion, added Cu(II) or water has
been evaluated in a number of papers regarding cop-
Tetrakis(acetonitrile)copper(I) cation, [Cu(MeCN)4]+,
has been widely used as starting material for the synthesis
of copper(I) complexes, precursor for catalysts in the syn-
thesis of three-membered rings (cyclopropanation and azir-
idination) [1], and as reference for 63Cu NMR spectroscopy
in solution [2]. Surprisingly, much less has been known
about the related benzonitrile- or phenylacetonitrile-cop-
per(I) complexes.
Copper is one of the elements that is amenable to NMR
spectroscopy and its properties have been summarized else-
where [3,4]. Both natural isotopes possess nuclear spin
per(I)-acetonitrile complexes [2,6–10].
ꢀ
Although the complexes [Cu(MeCN)4]X (X ¼ BF4
;
PF6ꢀ; ClO4ꢀ; CF3SO ꢀ) have been studied by 63Cu
*
Corresponding author. Tel.: +55 61 3307 2918; fax: +55 61 3273 4149.
3
NMR in a number of seminal reports [8–10], much less is
E-mail address: sslemos@unb.br (S.S. Lemos).
0277-5387/$ - see front matter ꢀ 2007 Elsevier Ltd. All rights reserved.
doi:10.1016/j.poly.2007.04.007

3872
V.A.S. Falcomer et al. / Polyhedron 26 (2007) 3871–3875
known for their analogues [Cu(PhCN)4][BF4] (2) and
[Cu(BzCN)4][BF4] (3). In fact, to the best of our knowl-
edge, no report has been given regarding the characterization
of the complex 3. We started our studies motivated by the
possibility of using such compounds as starting materials
for the preparation of new Cu(I) complexes. In addition,
the evaluation of the factors that can affect detection of
copper by NMR spectroscopy on these related systems will
enlarge the application of this tool for the study of cop-
per(I)-containing materials.
HBF4 40% (0.16 mL, 1.0 mmol) was added at once. The
red suspension was stirred until the color has discharged
(approximately 1 h). After that, the product started to crys-
tallize from the medium and the crystallization process was
completed by adding toluene (1:1 v/v). The microcrystal-
line solid was filtered off, washed with ethyl ether and vac-
uum dried. Yield 405 mg (65%). M.p. 90–92 ꢁC. Anal. Calc.
for C32H28BCuF4N4: C, 62.10; H, 4.56; N, 9.05; Cu, 10.27.
Found: C, 62.61; H, 4.50; N, 8.84; Cu, 9.48%. NMR (ace-
tone-d6)1H d 7.31–7.16 (m, 20H, CH arom.), 3.95 (s, 8H,
CH2); 13C{1H} d 130.8, 129.2, 128.3, 128.0, 118.5 (broad,
CN), 23.0; 11B{1H} d 2.2; 19F d ꢀ146.7. Selected IR bands
(Nujol mull): 2284 (m CN), 1050 cmꢀ1 (br, masym BF4ꢀ).
The complex [Cu(PhCN)4][BF4] (2) was prepared as
described above but in argon atmosphere to prevent its
decomposition.
2. Experimental
2.1. General
Solvents and starting materials were used as received
from the suppliers. Preparation of complex 2 was con-
ducted using standard Schlenk techniques, under argon
atmosphere.
1
NMR (acetone-d6) H d 7.8–7.6 (m); 13C{1H} d 132.3,
130.7, 127.9, 109.3 (broad, CN); 11B{1H} d 0.2; 19F d
ꢀ147.1.
2.2. Instrumentation
2.4. Crystallographic determination of 3
Elemental analyses (C, H, N) were performed on a
Fisions MOD EA 1108 analyser. The infrared spectra
(4000–400 cmꢀ1) were recorded on BOMEM BM 100
FT-IR spectrometer. NMR spectra were recorded on a
Varian MERCURY plus spectrometer 7.05 T, operating
at 300.07 MHz for 1H, 96.27 MHz for 11B, 75.46 MHz
for 13C, 282.31 MHz for 19F and 79.58 MHz for 63Cu.
Suitable crystal of 3 for X-ray diffraction studies was
grown by slow diffusion of ethyl ether into its BzCN
solution.
Crystallographic measurements were made on a Bruker
Kappa X8 Apex II CCD area detector with graphite mono-
˚
chromatized Mo Ka radiation (k = 0.71073 A) at 100(1) K.
Raw image data were processed with SAINT and SADABS [12]
to produce the final intensity data corrected for polariza-
tion, Lorentz and absorption effects. The structure was
solved by direct methods (SHELXS-97) and additional atoms
were located in the difference Fourier map and refined on
F2 (SHELXL-97) [13]. Hydrogen atoms were refined using a
riding model.
1
The H, 11B{1H}, 13C{1H}, and 19F NMR spectra were
obtained in acetone-d6 as solvent at room temperature.
Chemical shifts (d) are given in ppm relative to SiMe4
1
(internal reference for H and 13C), neat BF3 Æ OEt2 (exter-
nal reference for 11B) and CCl3F (external reference for
19F), respectively. The solution NMR studies were per-
formed using broad band probe heads equipped with a var-
iable-temperature unity precise to 0.5 ꢁC. 63Cu NMR
spectra were acquired at 300 K (unless otherwise stated)
in a 5 mm tube (o.d.) and were externally referenced to a
0.1 mol/L solution of 1 in CD3CN by using a capillary
(d = 0). A single pulse sequence was used. Typical NMR
parameters were the following: spectroscopic window
55866 Hz, pulse 10.7 ls (p/2 pulse length), acquisition time
0.293 s, 2048 scans, FT size 65536 points. An exponential
function Lb = 10 has been applied before Fourier trans-
form. The 63Cu line width at half-height, Dm1/2, was deter-
mined by fitting it to Gaussian line shape using the VNMR
6.1C software. The Dm1/2 for 1 was determined to be 530 Hz
in accord with previous results [10].
Crystal data for 3 C32H28BCuF4N4: M = 618.93, color-
less plate, 0.09 · 0.43 · 0.44 mm3, monoclinic, P2(1)/n (no.
˚
˚
14 alternate), a = 9.9620(11) A, b = 29.881(3) A, c =
3
˚
˚
10.2236(12) A, b = 90.363(5)ꢁ, V = 3043.3(6) A , Z = 4,
Dcalc = 1.351 g cmꢀ3, F000 = 1272, 2hmax = 30.48ꢁ, 53060
reflection collected, 9250 independent, Rint = 4.58%. Final
goodness-of-fit = 1.036, R1 [I > 2r(I)] = 0.0367, wR2 [all
data] = 0.096, 380 parameters, Dqmin = ꢀ0.52, Dqmax
=
ꢀ3
˚
0.71 e A
.
3. Results and discussion
3.1. Solution 63Cu NMR
3.1.1. Concentration effect
2.3. Preparation of the complexes
A stock solution of 2 0.50 mol/L in benzonitrile, PhCN,
was prepared and used to make the desired solutions for
63Cu NMR studies. A dependence of the line width at
half-height (Dm1/2) on the concentration of 2 has been
noticed (Fig. 1). It is worth mentioning that a 1.0 mol/L
solution of 2 shows no 63Cu NMR signal i.e. the bandwidth
is so broad that it is beyond detection.
The copper-nitrile complexes have been obtained based
on the reported method for the preparation of 1 [11]. Typ-
ically, the complex [Cu(BzCN)4][BF4] (3) was prepared as
follows: Cu2O (72 mg, 0.5 mmol) was suspended in phenyl-
acetonitrile – BzCN (5 mL). Under continuous stirring,

V.A.S. Falcomer et al. / Polyhedron 26 (2007) 3871–3875
3873
Fig. 3. Dependence of the 63Cu line width on the temperature of a
0.05 mol/L solution of [Cu(PhCN)4][BF4] (2) in benzonitrile solution.
Fig. 1. Dependence of line width of the 63Cu resonance on the
concentration of [Cu(PhCN)4][BF4] (2) in benzonitrile solution.
3.1.2. Temperature effect
The effect of the temperature on the 63Cu chemical shift
has been known to depend on the complex studied. For 1, a
negative effect of ꢀ0.75 ppm/K has been observed with the
temperature increase [3].
A larger effect of about
ꢀ1.0 ppm/K has been observed for 2 within the tempera-
ture range of 268–343 K (Fig. 2). As can be seen in
Fig. 3, a line width minimum is observed at 313 K.
600
500
400
ppm
300
200
3.2. Solid-state 63Cu NMR
Fig. 4. 63Cu MAS NMR spectrum of complex [Cu(BzCN)4][BF4] (3)
referenced against static CuCl (d = 0) [20]. Spectroscopic window
100 kHz, pulse width 5.5 ls, acquisition time 0.05 s, 256 scans, FT size
32768, spin rate 4 kHz. An exponential function Lb = 10 has been applied
before Fourier transform. The two arrows point to spinning sidebands.
While the solution of 3 could not be studied by 63Cu
NMR spectroscopy since its resonance is extremely broad,
the 63Cu MAS NMR spectrum of microcrystalline 3
(Fig. 4) shows a resonance whose line width is comparable
to that of the standard CuCl.
3.3. Solution structure by 63Cu spectroscopy
It has been known that detectable narrow lines for 63Cu
NMR are usually obtained if the coordination sphere
around Cu(I) is rigorously tetrahedral. These line widths
are governed essentially by the quadrupolar relaxation
rate, (1/T2)Q [7], which depends on the symmetry. The (1/
T2)Q depends, in addition to some other parameters, on
the reorientational correlation time, sR, which in turns
depends on the viscosity of the solvent, gS, and the radius
of the complex, rc, (Stokes–Einstein model).
sR / gSr3c
ð1Þ
In complex 2, an increase in the 63Cu line width occurs with
the increase of the complex concentration. This result is in
line with previous reports concerning the complexes [Cu
Fig. 2. Dependence of the 63Cu chemical shift on the temperature of a
0.05 mol/L solution of [Cu(PhCN)4][BF4] (2) in benzonitrile solution.

3874
V.A.S. Falcomer et al. / Polyhedron 26 (2007) 3871–3875
(MeCN)4]X, where X ¼ BF4ꢀ; ClO4ꢀ in acetonitrile as sol-
vent [2,8,10]. It is also known that a dramatic increase of
the line width occurs with an increase of the mole ratio ben-
zonitrile/acetonitrile [7]. This phenomenon has been inter-
preted as a consequence of an increase of the viscosity of
the medium as the concentration of the complex increases
(Eq. 1). Attempts to acquire 63Cu NMR spectra of 2 in
(CD3)2CO were unsuccessful. It is concluded that the li-
gand benzonitrile tends to dissociate giving the three-coor-
dinate, or even two-coordinate, species as depicted below
(Eq. 2). When benzonitrile is employed as solvent such dy-
namic equilibrium favors the tetra-coordinated species and
so the signal can be detected.
C32
C31
N3
C22
C12
C21
N2
N1
Cu1
N4
C41
C42
C11
F4
F1
B1
½CuðPhCNÞ4ꢁþ ¼ ½CuðPhCNÞð4ꢀnÞꢁþ þ nPhCN
ð2Þ
F3
F2
Complex 2 exhibited an anomalous profile in the variable-
temperature 63Cu spectra above 213 K, since decreasing of
the line width would be expected with increasing tempera-
ture for quadrupolar relaxation. This same profile has been
observed for [Cu(MeCN)4]X although their line width min-
imum is located at about 268 K [2]. Such anomaly has been
interpreted based on ligand exchange becoming more se-
vere with increasing temperature.
Fig. 5. ORTEP [21] plot of [Cu(BzCN)4][BF4] (3) with the atom
numbering. Thermal ellipsoids are scaled at 50%. Selected bond distances
˚
(A) and bond angles (ꢁ): Cu(1)–N(1), 1.9828(12); Cu(1)–N(2), 1.9869(12);
Cu(1)–N(3), 1.9815(12); Cu(1)–N(4), 1.9838(12); N(3)–Cu(1)–N(1),
102.14(5); N(3)–Cu(1)–N(4), 114.74(5); N(1)–Cu(1)–N(4), 111.90(5);
N(3)–Cu(1)–N(2), 110.78(5); N(1)–Cu(1)–N(2), 114.89(5); N(4)–Cu(1)–
N(2), 102.85(5); C(11)–N(1)–Cu(1), 170.74(12); C(21)–N(2)–Cu(1),
170.72(12); C(31)–N(3)–Cu(1), 169.69(12); C(41)–N(4)–Cu(1), 168.37(12).
In the case of 3, no signal could be detected either in the
phenylacetonitrile solution 0.05–0.2 mol/L range or in ace-
tone-d6 as solutions in the 0.005–0.25 mol/L range, in spite
of much effort devoted to this task. Since compound 3 is
much more inert in the air compared to 2, its reluctance to
give NMR signal should not be attributed to the presence
of Cu2+. The white solid 3 has been kept in a stopped vessel
for about two years with no apparent change although it has
been opened a couple of times. We hypothesized that this
behavior is a consequence of dynamic equilibrium taking
place in solution between the tetra- three- and two-coordi-
nated species, with exchange rate in the order of the NMR
time scale. Even MeCN, which is less crowding compared
to BzCN, can form low-coordinated copper(I) complexes,
demonstrated recently by the X-ray structure of two-coordi-
nated [Cu(MeCN)2][B(C6F5)4] [14]. In order to confirm our
hypothesis we set out to study 3 by solid-state 63Cu NMR
spectroscopy and X-ray crystallography. Accordingly,
63Cu MAS NMR spectrum has been acquired for the micro-
crystalline solid 3 and it turned out to give a reasonably
sharp resonance, namely, Dm1/2 1.06 kHz (Fig. 4) which is
comparable to the reference’s line width obtained at the
same speed. The NMR data of 3 suggest that the copper
atom is tetra-coordinated in the solid-state while in solution
an equilibrium between tetra- and lower coordinated species
is operative. In fact, the X-ray diffraction study confirmed
that the copper(I) atoms is tetra-coordinated by four nitrile
ligands in a slightly distorted tetrahedral geometry (Fig. 5).
3.4. X-ray structure of 3
The X-ray crystallographic study of the complex
[Cu(BzCN)4][BF4] (3) reveals a slightly distorted tetrahe-
dral copper(I) center with four BzCN ligands coordinated
via their nitrogen atoms. The Cu–N distances, which are
˚
˚
almost equal varying from 1.9815(12) A to 1.9869(12) A,
are close to the average found for similar compounds,
˚
1.99(3) A for 47 compounds found in the Cambridge Struc-
tural Database [15]. The angle between the planes N1–
Cu1–N4 and N2–Cu1–N3 is 81.42(3)ꢁ showing a slightly
flattening of the tetrahedron; the N–Cu–N bond angles
range from 102.85(5)ꢁ to 114.89(5)ꢁ. The Cu–N–C angles
fall between 168.37(12)ꢁ and 170.74(12)ꢁ, which is normal
compared to the average for the similar compounds,
172(4)ꢁ. The geometry of the BzCN ligands is not remark-
able. No p–p interactions are seen in the structure of 3,
unlike those of similar [Cu(NCPh)4]+ compounds [16–18]
where the Cu–N–C angles are much more distorted from
the average and the geometry around the copper(I) atom
is more flattened. There are non-classic hydrogen bonding
interactions between one hydrogen atom of each methylene
Table 1
˚
Non-classic hydrogen bonds for 3, [Cu(BzCN)4][BF4] (A and ꢁ)
D–Hꢂ ꢂ ꢂA
d(D–H)
d(Hꢂ ꢂ ꢂA)
d(Dꢂ ꢂ ꢂA)
3.1777(19)
3.248(2)
3.2010(19)
3.1891(19)
\(DHA)
C(12)–H(12B)ꢂ ꢂ ꢂF(1)
C(42)–H(42B)ꢂ ꢂ ꢂF(4)
C(32)–H(32B)ꢂ ꢂ ꢂF(3)a
C(22)–H(22B)ꢂ ꢂ ꢂF(2)a
0.99
0.99
0.99
0.99
2.62
2.67
2.61
2.59
115.7
117.7
118.3
119.2
1
The H and 13C{1H} NMR of 2 and 3 spectra show no
indication of free nitrile in acetone-d6 solution at room
temperature. It can be concluded that one only see aver-
aged signals in such conditions, consistent with the pro-
posed equilibrium depicted in (2).
Symmetry transformations used to generate equivalent atoms.
a
x, y, z ꢀ 1.

V.A.S. Falcomer et al. / Polyhedron 26 (2007) 3871–3875
3875
group of the BzCN ligands and the fluorine atoms of the
BF4 anion, Table 1, to form a continuous chain along
the crystallographic z direction.
(+44) 1223-336-033; or e-mail: deposit@ccdc.cam.ac.uk.
Supplementary data associated with this article can be
found, in the online version, at doi:10.1016/j.poly.2007.
04.007.
3.5. Chemical properties of [Cu(PhCN)4][BF4] (2) and
[Cu(BzCN)4][BF4] (3)
References
[1] C. Borriello, M.E. Cucciolito, A. Panunzi, F. Ruffo, Tetrahedron:
Asymmetry 12 (2001) 2467.
[2] J.K. Irangu, R.B. Jordan, Inorg. Chem. 42 (2003) 3934, and
references cited therein.
[3] P. Granger, in: P.S. Pregosin (Ed.), Transition Metal Nuclear
Magnetic Resonance, vol. 13, Elsevier, Amsterdan, 1991, p. 265.
[4] J. Malito, Annu. Rep. NMR Spectrosc. 38 (1999) 265.
[5] J.R. Black, W. Levason, M.D. Spicer, M. Webster, J. Chem. Soc.,
Dalton Trans. (1993) 3129.
Crystalline 3 can be handled on a bench top since it is
more air-stable copper(I) complex compared to 1 or 2. It
is also stable in BzCN solution. Apparently, the benzyl
groups block more effectively the access of dioxygen mole-
cule compared to methyl groups, for example. Compound
3 seems to be a more convenient starting material for the
preparation of copper(I) complexes since it is: (i) easily pre-
pared, (ii) safely handled, and (iii) labile regarding ligand
displacement. For instance, the reaction of 3 with 4,6-dim-
ethylpyrimidine-2-thione (dmpymtH) proceeds in phenyl-
acetonitrile/toluene mixture (1:1), under reflux, with
displacement of all BzCN molecules by the heterocyclic
ligand giving the already known complex [Cu(dm-
pymtH)3]2[BF4]2 Æ 2H2O [19]. The lability of 3 is also sup-
ported by the fact that when it was treated with a
coordinating anion containing salt, i.e., KSCN, pure
CuSCN precipitated, confirmed by comparison of its infra-
red spectrum with that of an authentic sample.
[6] D.S. Gill, L. Rodehuser, J.-J. Delpuech, J. Chem. Soc., Faraday
¨
Trans. 86 (1990) 2847.
[7] D.S. Gill, L. Rodehuser, P. Rubini, J.-J. Delpuech, J. Chem. Soc.,
¨
Faraday Trans. 91 (1995) 2307.
[8] U. Ochsenbein, C.W. Schla¨pfer, Helv. Chim. Acta 63 (1980) 1926.
[9] O. Lutz, H. Oehler, P. Kroneck, Z. Phys. A: Hadrons Nucl. 288
(1978) 17.
[10] P. Kroneck, J. Kodweiss, O. Lutz, A. Nolle, D. Zepf, Z. Naturforsch.
37a (1982) 186.
[11] G.J. Kubas, B. Monzyk, A.L. Crumbliss, in: R.J. Angelici (Ed.),
Inorg. Synth. 28 (1991) 68.
[12] Bruker AXS Inc., SAINT (version 7.06A) and SADABS (version 2.10),
Madison, Wisconsin, USA, 2005.
[13] G.M. Sheldrick, SHELXS97 and SHELXL97, University of Go¨ttingen,
Go¨ttingen, Germany, 1997.
Acknowledgement
[14] H.-C. Liang, E. Kim, C.D. Incarvito, A.L. Rheingold, K.D. Karlin,
Inorg. Chem. 41 (2002) 2209.
[15] F.H. Allen, Acta Crystallogr., Sect. B 58 (2002) 380.
[16] S. Brownstein, N.F. Han, E. Gabe, Y. Le Page, Can. J. Chem. 67
(1989) 2222.
Financial support from FINEP CT-INFRA (970/01) is
gratefully acknowledged.
Appendix A. Supplementary material
[17] G.A. Bowmaker, D.S. Gill, B.W. Skelton, N. Somers, A.H. White, Z.
Naturforsch. B 59 (2004) 1307.
[18] J.M. Knaust, D.A. Knight, S.W. Keller, J. Chem. Crystallogr. 33
(2003) 813.
[19] V.A.S. Falcomer, S.S. Lemos, A.A. Batista, J. Ellena, E.E. Castel-
lano, Inorg. Chim. Acta 359 (2006) 1064.
[20] S. Sakida, N. Kato, Y. Kawamoto, Mater. Res. Bull. 37 (2002) 2263.
[21] L.J. Farrugia, J. Appl. Cryst. 30 (1997) 565.
CCDC 632661 contains the supplementary crystallo-
graphic data for this paper. These data can be obtained free
of charge via http://www.ccdc.cam.ac.uk/conts/retriev-
ing.html, or from the Cambridge Crystallographic Data
Centre, 12 Union Road, Cambridge CB2 1EZ, UK; fax: