Metal-Peptide Frameworks (MPFs): "Bioinspired" metal organic frameworks

Article abstract of DOI:10.1021/ja0762588

Chiral metal-organic frameworks (MOFs) have attracted a growing interest for their potential use in energy technologies, asymmetric catalysis, chiral separation, and on a more basic level, the creation of new topologies in inorganic materials. The current

Full text of DOI:10.1021/ja0762588

Published on Web 02/05/2008
Metal-Peptide Frameworks (MPFs): “Bioinspired” Metal
Organic Frameworks
Alexandre Mantion,*,† Lars Massu¨ger,‡ Pierre Rabu,§ Cornelia Palivan,†
Lynne B. McCusker,‡ and Andreas Taubert*,†,|
Department of Chemistry, Klingelbergstrasse 80, UniVersity of Basel, CH-4056 Basel,
Switzerland, Laboratory of Crystallography, ETH Zurich, CH-8093 Zurich, Switzerland, IPCMS
UMR 7504 CNRS - UniVersite´ Louis Pasteur, 23, rue du Loess, BP 43, F-67034 Strasbourg
CEDEX 2, France, and Institute of Chemistry, UniVersity of Potsdam and Max-Planck-Institute
of Colloids and Interfaces, D-14476 Golm, Germany
Received August 20, 2007; E-mail: ataubert@uni-potsdam.de; a.mantion@unibas.ch
Abstract: Chiral metal-organic frameworks (MOFs) have attracted a growing interest for their potential
use in energy technologies, asymmetric catalysis, chiral separation, and on a more basic level, the creation
of new topologies in inorganic materials. The current paper is the first report on a peptide-based MOF, a
metal peptide framework (MPF), constructed from an oligovaline peptide family developed earlier by our
group (Mantion, A.; et al. Macromol. Biosci. 2007, 7, 208). We have used a simple oligopeptide, Z-(L-
Val)2-L-Glu(OH)-OH, to grow porous copper and calcium MPFs. The MPFs form thanks to the self-
assembling properties of the peptide and specific metal-peptide and metal-ammonia interactions. They
are stable up to ca. 250 °C and have some internal porosity, which makes them a promising prototype for
the further development of MPFs.
lightweight building materials. Other MOFs have useful optical9
and electronic properties10 or have been proposed as transport
Introduction
vessels for biological applications.11
Metal organic frameworks (MOFs) have attracted tremendous
attention over the past few years. This is due to the fact that
they are porous, have large surface areas of over 5000 m2/g,
and have tunable pores sizes and topologies.1-4 MOFs are
constructed from relatively simple organic building blocks,
which act as connectors between transition metals or lanthanides.
Because of their “infinite” connectivity, MOFs can be viewed
as porous inorganic polymers.5 Furthermore, like zeolites and
aluminophosphates, MOFs are crystalline materials. However,
whereas zeolites and other microporous materials are purely
inorganic, MOFs are organic/inorganic hybrids. Despite the
presence of organic building blocks, MOFs are robust materials.
Their low density and high surface area are useful for gas
sorption and energy technologies,6,7 filtration,8 or possibly even
MOFs can be synthesized using a wide variety of methods
from hydrothermal to simple diffusion methods.1,4,12,13 To date,
most ligands used as connectors are rigid and do not carry
additional functionalities, which could modify the properties of
the MOF pores. There are only a few studies on flexible
connectors and connectors that also self-assemble in the absence
of a metal ion. For example, several groups have studied MOFs
obtained from adipic acid, auxiliary bases, and Ca2+, lanthanides,
or 3d transition metals.14-21
Interestingly, there have been no reports on the use of peptides
or peptidomimetics as connectors. This is particularly intriguing
because peptide chemistry offers a virtually unlimited structural
diversity, including additional metal coordinating sites, etc.
Moreover, peptides are intrinsically chiral, which should make
† University of Basel.
(9) Fan, J.; Zhu, H.-F.; Okamura, T.-A.; Sun, W.-Y.; Tang, W.-X.; Ueyeama,
N. N. J. Chem. 2003, 27, 1409.
(10) Ye, Q.; Song, Y.-M.; Wang, G.-X.; Chen, K.; Fu, D.-W.; Chan, W.; Hong,
P.; Zhu, J.-S.; Huang, S. D.; Xiong, R.-G. J. Am. Chem. Soc. 2006, 128,
6554.
(11) Horcajada, P.; Serre, C.; Vallet-Regi, M.; Sebban, M.; Taulelle, F.; Ferey,
G. Angew. Chem., Int. Ed. 2006, 45, 5974.
(12) Kitagawa, S.; Noro, S. Compreh. Coord. Chem. 2004, 7, 231.
(13) Clegg, W. Compreh. Coord. Chem. 2004, 1, 579.
‡ ETH Zurich.
§ IPCMS.
| University of Potsdam.
(1) Yaghi, O. M.; Li, H.; Davis, C.; Richardson, D.; Groy, T. Acc. Chem. Res.
1998, 31, 474.
(2) Barton, T. J.; Bull, L. M.; Klemperer, W. G.; Loy, D. A.; McEnaney, B.;
Misono, M.; Monson, P. A.; Pez, G.; Scherer, G. W.; Vartuli, J. C.; Yaghi,
O. M. Chem. Mater. 1999, 11, 2633.
(3) James, S. L. Chem. Soc. ReV. 2003, 32, 276.
(4) Rowsell, J. L. C.; Yaghi, O. M. Microporous Mesoporous Mater. 2004,
73, 3.
(5) Robin, A. Y.; Fromm, K. M. Coord. Chem. ReV. 2006, 250, 2127.
(6) Rosi, N. L.; Eckert, J.; Eddaoudi, M.; Vodak, D. T.; Kim, J.; O’Keeffe,
M.; Yaghi, O. M. Science 2003, 300, 1127.
(7) Duren, T.; Sarkisov, L.; Yaghi, O. M.; Snurr, R. Q. Langmuir 2004, 20,
2683.
(8) Chen, B.; Liang, C.; Yang, J.; Contreras, D. S.; Clancy, Y. L.; Lobkovsky,
E. B.; Yaghi, O. M.; Dai, S. Angew. Chem., Int. Ed. 2006, 45, 1390.
(14) Borkowski, L. A.; Cahill, C. L. Inorg. Chem. 2003, 42, 7041.
(15) Borkowski, L. A.; Cahill, C. L. Inorg. Chem. Commun. 2004, 7, 725.
(16) de Lill, D. T.; Bozzuto, D. J.; Cahill, C. L. Dalton Trans. 2005, 2111.
(17) de Lill, D. T.; Gunning, N. S.; Cahill, C. L. Inorg. Chem. 2005, 44, 258.
(18) Dimos, A.; Tsaousis, D.; Michaelides, A.; Skoulika, S.; Golhen, S.; Ouahab,
L.; Didierjean, C.; Aubry, A. Chem. Mater. 2002, 14, 2616.
(19) Kiritsis, V.; Michaelides, A.; Skoulika, S.; Golhen, S.; Ouahab, L. Inorg.
Chem. 1998, 37, 3407.
(20) Lobo, R. F.; Zones, S. I.; Davis, M. E. J. Inclusion Phenom. 1995, 21, 47.
(21) Kim, Y.; Suh, M.; Jung, D.-Y. Inorg. Chem. 2004, 43, 245.
9
10.1021/ja0762588 CCC: $40.75 © 2008 American Chemical Society
J. AM. CHEM. SOC. 2008, 130, 2517-2526
2517

A R T I C L E S
Mantion et al.
Scheme 1. Oligovaline Peptidesa
solution was allowed to warm up to room temperature, upon which
the solution was stirred for 24 h. The solvents were removed by rotary
evaporation, and the residue was dissolved in 200 mL of chloroform.
The organic solution was washed three times with 100 mL of a mixture
of saturated hydrogen carbonate solution and sodium chloride, once
with aqueous sodium chloride, three times with 100 mL of 10% citric
acid and aqueous sodium chloride, and twice with aqueous sodium
chloride. The organic phase was dried with sodium sulfate, concentrated
to dryness, and the residue was triturated with pentane to give a white
solid (3.4 g, 75%) after drying. IR (neat, cm-1) 3479, 3277, 3063, 2976,
1633, 1533, 1429, 1393, 1371, 1349, 1283, 1247, 1222, 1153, 1088,
1071, 1038, 1028, 1020, 1005, 987, 920, 908, 864, 846, 799, 774, 753,
740, 698, 667. Elemental analysis calculated: C 62.92, H 8.35, N 7.10,
O 21.63; measured: C 62.97, H 8.32, N 7.19. FAB-MS calculated
[M-H]+ ) 592, measured [M-H]+ ) 592. 13C NMR (δ in ppm vs
TMS, d6-DMSO, 100 MHz) 172.25, 171.84, 171.70, 171.51, 156.91,
137.96, 129.16, 128.56, 128.52, 128.45, 81.40, 80.56, 66.19, 61.11,
57.94, 52.58, 31.71, 31.12, 28.56, 28.43, 27.04, 20.11, 20.01, 18.99,
18.96. 1H NMR (δ in ppm vs TMS, d6-DMSO, 400 MHz) 8.23 (d, 0.9
H, H amide), 8.1 (d, 0.1 H, H amide), 7.92 (d, 0.1 H, H amide), 7.72
(d, 0.8 H, H amide), 7.33 (m, 6 H, H phenyl + H carbamate), 5.01 (s,
2 H, benzyl), 4.22 (t, 1 H, R Val-1), 4.12 (m, 1 H, R Val-2), 3.94 (m,
1 H, R Glu), 2.22 (m, 2 H, γ Glu), 1.92 (m, 3 H, â Val-1 + â Val 2
+ â Glu), 1.72 (m, 1 H, â Glu), 1.36 (m, 18 H, tert-butyl ester x 2),
0.83 (m, 12 H, γ Val-1 + γ Val-2)
a 1 is the parent compound used in our earlier studies.29,30 2 is the peptide
ligand used for construction of the MPFs and for magnetic dilution in the
EPR experiments in the current paper.
them prime candidates for the fabrication of chiral MOFs. Chiral
MOFs are, beyond the more fundamental search for new
topologies, highly attractive as asymmetric catalysts, for chiral
recognition, and applications requiring noncentrosymmetric
crystal structures like nonlinear optical devices.22-26
There are essentially three approaches for the preparation of
chiral MOFs: (i) the exploitation of molecules, where the chiral
resolution happens without further assistance,27 (ii) the use of
chiral co-ligands in addition to achiral connectors,23 or (iii) the
use of a chiral connector.25,28 The chiral connector approach is
currently most often used because of its flexibility.22-28
We have recently introduced a family of self-assembling
peptides based on oligovalines. These peptides gel a variety of
solvents, including the inorganic liquid tetraethylorthosilicate
(TEOS). Furthermore, they act as templates for complex titania
and silver/peptide nanostructures.29,30 The current paper shows
that replacing one valine unit with a glutamic acid, Scheme 1,
leads to peptides that can form peptide analogs of MOFs, which
we have termed metal peptide frameworks (MPFs). The peptides
form strong bonds with Cu2+ and Ca2+, and the resulting
crystalline materials precipitate from solution as long needle-
like micrometer-sized particles.
Z-L-Val-L-Val-L-Glu(OH)OH 2. Two grams of Z-L-Val-L-Val-
L-Glu(OtBu)OtBu were reacted with 25 mL of a 95% aqueous
trifluoroacetic acid (TFA) solution for 30 min under argon. Then the
solvent was removed under reduced pressure. The residue was treated
three times with 30 mL of chloroform. After evaporation of the
chloroform, the slightly yellow solid was triturated with diethyl ether,
filtered, and the whitish solid was dried overnight under vacuum,
leaving 1.29 g of white solid 2 (80%). IR (neat, cm-1) 3289, 3074,
2976, 1637, 1530, 1419, 1340, 1293, 1242, 1217, 1179, 1150, 1075,
1043, 1029, 995, 965, 934, 917, 859, 842, 798, 778, 756, 736, 697.
FAB-MS: calculated [M-H]+ ) 480; measured [M-H]+ ) 480.
Elemental analysis calculated: C 57.61, H 6.94, N 8.76, O 26.69;
1
measured: C 56.91, H 6.90, N 8.60. H NMR (δ in ppm vs TMS,
d6-DMSO, 400 MHz) 12.35 (very broad, 2 H, acidic protons), 8.17 (d,
0.94 H, H amide), 8.03 (d, 0.17 H, H amide), 7.90 (d, 0.12 H, H amide),
7.72 (d, 0.9 H, H amide), 7.36 (m, 6 H, H carbamate + H phenyl),
5.01 (s, 2 H, H benzylic), 4.18 (m, 2 H, R Val-1 + R Val-2), 3.92 (m,
1 H, R Glu), 1.94 (m, 3 H, 1H â-Glu + â Val-1 + â Val-2), 1.76 (m,
1 H, 1 H â Glu), 0.82 (m, 14 H, 2 H γ Glu + 6 H γ Val-1 + 6 H γ
Val-2). 13C NMR (δ in ppm vs TMS, d6-DMSO, 100 MHz) 174.59,
173.91, 172.95, 171.82, 156.94, 137.95, 129.99, 129.39, 129.25, 66.20,
61.13, 58.13, 51.98, 31.63, 31.08, 30.80, 27.04, 20.08, 19.96, 19.03,
18.95.
Experimental
Materials. Amino acids were purchased from Bachem AG (Buben-
dorf, Switzerland), and all other chemicals were purchased from Fluka
(Buchs, Switzerland). All chemicals were used as received.
Ligand Synthesis. The synthesis of the precursor peptide ZVVOH
and of peptide 1 has been described elsewhere.29
Z-L-Val-L-Val-L-Glu(OtBu)OtBu. Isobutyl chloroformate (1.4 mL,
10.3 mmol) was added to a salt/ice cooled solution of Z-L-Val-L-Val-
OH (3.0 g, 8.6 mmol) and 4-methylmorpholine (1.7 mL, 14.6 mmol)
in 25 mL of acetonitrile under argon. The reaction mixture was stirred
for 2 min. Then, 3.0 g (10.3 mmol) of L-glutamic acid di-tert-butyl
ester hydrochloride and 1.7 mL (14.6 mmol) of 4-methylaminomor-
pholine were added. After 15 min, the cooling was removed and the
Complex Synthesis. The metal complexes were prepared by
dissolving peptide 2 in a 60/40 (v/v) ethanol/water mixture. The pH of
the solution was adjusted to pH 8 with diluted aqueous ammonia. In a
typical experiment, 56 µL of a 3 mM aqueous copper(II) or calcium
nitrate solution were added to the ligand solution at room temperature
under stirring. After copper or calcium salt addition, the pH was
readjusted to 8. The copper complex (MPF-9) immediately precipitated,
and the calcium complex (MPF-2) precipitated after 1 h at 80 °C. To
complete the reaction, the reaction mixture was further reacted for 48
h at 80 °C in a closed vessel. Then, the solids were centrifuged, washed
with ethanol, and dried under vacuum overnight. Various metal/peptide
ratios were used during the synthesis, but all precipitates had a 1:1
metal/peptide stoichiometry. Yields were above 99%. Elemental analysis
(MPF-2, C23H39CaN3O12) calculated: C 46.85, H 6.67, N 7.13, O 32.56,
Ca 6.80; measured: C 47.80, H 6.29, N 7.57. Elemental analysis
(MPF-9 corresponding to the dibasic form of 2 plus 1 Cu2+ plus 2
(22) Kesanli, B.; Lin, W.-B. Coord. Chem. ReV. 2003, 246, 305.
(23) Bradshaw, D.; Claridge, J. B.; Cussen, E. J.; Prior, T. J.; Rosseinsky, M.
J. Acc. Chem. Res. 2005, 38, 273.
(24) Zang, S.; Su, Y.; Li, Y.; Zhu, H.; Meng, Q. Inorg. Chem. 2006, 45, 2972.
(25) Lin, W. J. Solid State Chem. 2005, 178, 2488.
(26) Sun, D.; Ke, Y.; Collins, D. J.; Lorigan, G. A.; Zhou, H.-C. Inorg. Chem.
2007, 46, 2725.
(27) Gao, E.-Q.; Yue, Y.-F.; Bai, S.-Q.; He, Z.; Yan, C.-H. J. Am. Chem. Soc.
2004, 126, 1419.
(28) Wu, C.-D.; Hu, A.; Zhang, L.; Lin, W. J. Am. Chem. Soc. 2005, 127, 8940.
(29) Mantion, A.; Taubert, A. Macromol. Biosci. 2007, 7, 208.
(30) Mantion, A.; Guex, G.; Foelske, A.; Mirolo, L.; Fromm, K. M.; Painsi,
M.; Taubert, A. Soft Matter, 2008, DOI: 10.1039)b712826f.
9
2518 J. AM. CHEM. SOC. VOL. 130, NO. 8, 2008

Metal−Peptide Frameworks
A R T I C L E S
Table 1. X-Ray Powder Diffraction Data Collection
synchrotron facility
wavelength
SNBL (Station B) at ESRF
0.50007 Å
diffraction geometry
analyzer crystal
sample
2θ range
step size
Debye-Scherrer
Si 111
rotating 1.0 mm capillary
1.020-30.501°
0.003° 2θ
time per step
1.0 s
NH3 plus 2 H2O: C23H41CuN5O10), calculated: C 45.20, H 6.76, N
11.46, Cu 10.40, O 26.18; measured: C 45.98, H 6.38, N 10.39.
Spectroscopy. 1H and 13C NMR spectra were recorded on an Avance
400 MHz NMR spectrometer. Infrared spectra were obtained from the
neat samples on a Shimadzu FTIR 8300 with a Golden Gate ATR unit.
Spectra were recorded from 300 to 4500 cm-1 with a resolution of 1
cm-1. FAB-MS spectra were taken on a Finnigan MAT 312.
Microscopy. Scanning electron microscopy was done on a Philips
XL30 FEG ESEM operated at 10 kV. Samples were sputtered with
gold or platinum prior to imaging.
X-Ray Diffraction. Powder diffraction patterns were measured on
the Swiss-Norwegian Beamlines (SNBL) at the European Synchrotron
Radiation Facility (ESRF) in Grenoble. Details of the data collection
are given in Table 1.
Powder diffraction data analysis and interpretation. Powder
diffraction patterns were indexed using Topas Academic.31,32 Le Bail
parameters extraction were performed using Fullprof.33,34 Structures of
the complexes were generated using FOX.35,36 Crystal was used to
correct some hydrogen positions.37 The Final complex structure was
refined using XRS-82.38
Voids Volume Evaluation. After refinement and before evaluating
the voids size and position, nonrefined hydrogen atoms were added
using the hydrogen addition functionality in Crystals37 using an average
C-H distance of 1 Å. Plotting and structure analysis was performed
using Platon39 and CrystalMaker (CrystalMaker Software Ltd).
Thermal Analysis. Thermogravimetric analysis (TGA) was done
with a Mettler Toledo TGA/SDTA 851e from 25 to 550 °C with a
heating rate of 10 °C/min in N2.
SQUID Measurements. The magnetic properties of MPF-9 were
investigated on neat powder samples with a Quantum Design MPMS-
XL SQUID magnetometer between 1.8 and 300 K. The magnetic field
strength was varied from -5 to +5 T. All magnetic data were corrected
for diamagnetism of the sample and the sample holder. To evaluate
the antiferromagnetic coupling between the neighboring Cu(II) centers,
we analyzed the susceptibility by using a model of Heisenberg S ) 1/2
spin chains. Writing the exchange Hamiltonian Hˆ ) -J∑i,jSˆiSˆj, we used
the expression of the susceptibility øT given in the literature.41
Figure 1. SEM images of (a) MPF-2 and (b) MPF-9.
complexes were measured as powders at 75 and 293 K. The complexes
were magnetically diluted with 2 to a maximum extent of 10% of Cu-
(II). Spectra were recorded using 100 kHz modulation frequency and
a microwave power of 0.5 to 2 mW. The diphenylpicrylhydrazyl
(DPPH) radical was used as a field marker (g ) 2.0036). The spectra
for the Cu(II) species were simulated using Symfonia software (Bruker),
which is based on a third-order perturbation theory treatment and
assumes Gaussian shapes for the line width tensors. The isotopes were
considered in their natural abundance for each spectrum.
Results and Discussion
Figure 1 shows representative scanning electron microscopy
(SEM) images of the calcium complex MPF-2 and the copper
complex MPF-9. Both samples have a fiber-like morphology
with lengths ranging up to several micrometers and widths of
ca. 200 nm. The calcium based MPF-2 crystals are somewhat
shorter but have a more regular morphology with well developed
crystal faces and edges. The copper-based MPF-9 appears less
crystalline and seems to consist of smaller “blobs” that make
up the rodlike particle.
NµB2g2(0.25 + 0.0750y + 0.15047y2)
k(1 + 0.9931y + 0.34427y2 + 3.0313y3)
øT )
(1)
In expression 1, y ) |J|/kT, where J is the exchange interaction between
electron spin moments Si.
Electron Paramagnetic Resonance Spectroscopy. EPR spectros-
copy was done on an EPR Bruker ElexSys E500 operating at 9.6 GHz
spectrometer equipped with an Oxford Instruments cryostat. All
Figure 2 shows thermogravimetric analysis (TGA) data. The
pure ligand 2 exhibits a single sharp weight loss at 272 °C.
There is no indication of the presence of water or other volatile
substances, and no indication of decarboxylation. TGA curves
of MPF-2 show four steps. The first one, at 35 °C, is relatively
small (5%) and is caused by solvent or water evaporation. A
clear weight loss of 19% is then observed at 262 °C. This weight
loss is attributed to the loss of four molecules of water strongly
bound in the crystal lattice or to solvent trapped in the structure.
The third step is less clear and has an inflection point at
(31) Cheary, W.; Coelho, A. J. Appl. Cryst. 1992, 25, 109.
(32) Coelho, A. TOPAS; Academic 4.0 ed., 2006.
(33) Rodriguez-Carvajal, J. Phys. B 1993, 192, 55.
(34) http://www-llb.cea.fr/fullweb/fp2k/fp2k.htm, August 2006.
(35) Favre-Nicolin, V. R.; Cerny, R. J. Appl. Cryst. 2002, 35, 734.
(36) Fox, Free Objects for Crystallography, http://objcryst.sourceforge.net.
(37) Betteridge, P. W.; Carruthers, J. R.; Cooper, R. I.; Prout, K.; Watkin, D. J.
J. Appl. Cryst. 2003, 36, 1487.
(38) Baerlocher, C. The X-ray RietVeld System (XRS-82); ETH Zu¨rich: Zu¨rich,
1982.
(39) Spek, A. L. J.Appl.Cryst. 2003, 36, 7.
(40) Kueppers, H.; Liebau, F.; Spek, A. L. J. Appl. Cryst. 2006, 39, 338.
(41) Hatfield, W. E. J. Appl. Phys. 1981, 52, 1985.
9
J. AM. CHEM. SOC. VOL. 130, NO. 8, 2008 2519

A R T I C L E S
Mantion et al.
Figure 2. TGA curves of the pure peptide and the MPFs.
Figure 4. EPR spectra and spectrum simulations of MPF-9 powder
measured at 75 and 293 K. (a) Spectrum measured at 75 K, (b) spectrum
simulation, (c) spectrum measured at 293 K, and (d) spectrum simulation.
the â-sheet is conserved after reaction with calcium and copper,
respectively, because the three bands at 1688, 1628, and 1533
cm-1 (MPF-2) and 1688, 1633, and 1533 cm-1 (MPF-9),
indicative of the â-sheet, are still visible in the crystalline
precipitates.
Figure 3. IR spectra of MPF-2 (top), MPF-9 (center), and pure peptide
(bottom).
Besides the conservation of the â-sheet, IR also shows
differences between the pure peptide and the metal complexes
MPF-2 and MPF-9. After reaction of the ligand with the metal
ions, the CdO stretch vibration of the glutamic acid at 1705
cm-1 disappears. This is in line with the fact that the ligand
interacts with the metal ion via its acid moiety. Finally,
differences between the peak positions of the asymmetric νCdO
and the symmetric νCdO vibrations (1533 cm-1, difference of
116 cm-1 in MPF-2, and 1564 cm-1, difference of 182 cm-1 in
MPF-9) suggest that a carboxylic acid acts as a bridging unit
between metal centers in MPF-2 and as an asymmetric mono-
dentate ligand in MPF-9.44,45
Figure 4 shows EPR spectra of MPF-9 powders measured at
75 and 293 K. Both spectra have an anisotropic overall shape,
but because the signals are broad, the hyperfine pattern of Cu-
(II) is invisible. Therefore, simulation of the spectra was done
using an orthorhombic gyromagnetic tensor, because a slight
orthorhombic character of the gyromagnetic tensor has been
previously obtained for other systems where Cu(II) was bonded
to glutamic acid.46 The hyperfine tensor could not be obtained
due to the large linewidths. Table 2 summarizes the results from
the simulation.
340 °C and a loss of 27%. The last weight loss of 14% is
observed at 429 °C. The total mass loss indicates (assuming
that Ca is present as CaO after TGA) that one Ca ion is bonded
to one ligand molecule.
MPF-9 behaves similarly, except that there are only three
thermal events. The first weight loss takes place at 91 °C, which
is due to solvent evaporation. This weight loss of 6% is assigned
to one equivalent of ethanol or two equivalents of adsorbed
water. The second event occurs at 160 °C and is assigned to
the loss of crystallographically bound solvent, most likely
solvent molecules coordinated to the metal center. A final weight
loss of 66% is observed at 260 °C. These data are consistent
with a ratio of one copper to one ligand molecule, where the
copper is possibly surrounded by one or two water or ammonia
molecules (see elemental analysis and EPR results below, which
suggest that two molecules of ammonia are bonded to the copper
center) in the coordination sphere.
Figure 3 shows representative IR spectra of the pure peptide
2, MPF-2, and MPF-9. The ligand bands at 3066, 1705, 1390,
1290, and 918 cm-1 can be assigned to O-H stretch, CdO
stretch, C-O stretch, and O-H bending vibrations, respectively.
Bands at 3280, 1637, 1528, and 1221 cm-1 can be assigned to
the amide A, amide I, amide II, and amide III vibrations,
respectively. The band at 1690 cm-1 can be attributed to the
CdO vibration of the carbamate moiety. As in our previous
work,29 the amide band positions are indicative of a â-sheet
structure. However, also as in our earlier work, IR does not
reveal whether the organization of the â-sheet is parallel or
antiparallel, because the diagnostic bands are obscured by
vibrations of the N-benzylcarboxy group.42,43 IR indicates that
The presence of the half-field EPR signal shows that spin-
spin interactions are present in MPF-9. This indicates, in
accordance with the very low solubility of MPF-9 in organic
(42) Palumbo, M.; Da Rin, S.; Bonora, G. M.; Toniolo, C. Makromol. Chem.
1976, 177, 1477.
(43) Baron, M. H.; De Loze, C.; Toniolo, C.; Fasman, G. D. Biopolymers 1979,
18, 411.
(44) Deacon, G. B.; Phillips, R. J. Coord. Chem. ReV. 1980, 33, 227.
(45) Barth, A. P. Prog. Biol. Mol. Biol. 2000, 74, 141.
(46) Brondino, C. D.; Casado, N. M. C.; Passeggi, M. C. G.; Calvo, R. Inorg.
Chem. 1993, 32, 2078.
9
2520 J. AM. CHEM. SOC. VOL. 130, NO. 8, 2008

Metal−Peptide Frameworks
A R T I C L E S
Table 2. EPR Parameters Obtained by Simulating the
Experimental Spectrum of MPF-9
g
Γ/G
complex
T/K
gxx
gyy
gzz
giso
A
Γxx
Γyy
Γzz
MPF-9
75 2.082 2.113 2.235 2.143
293 2.083 2.126 2.252 2.154
-
-
55 80 62
65 75 68
solvents, that there is a network of Cu(II) ions connected by
bridging moieties. In most Cu complexes where the metal ions
are bonded to two amino acid molecules this is realized by
carboxylate as well as by hydrogen bond bridges. Both paths
are in principle able to transfer spin polarization and therefore
provide the dominant pathways for super-exchange.46
Figure 5. Curie-Weiss Plot of MPF-9.
For two S ) 1/2 centers with gyromagnetic values near 2 and
an electron-electron distance r > 4.5 Å, the integrated intensity
of the half-field signal is determined only by the interspin
distance r.47 By assuming in this simple model that the
anisotropic exchange is negligible, the integrated intensity ratio
between the ∆MS ) (1 and ∆MS ) (2 transitions has been
used to estimate a Cu(II)-Cu(II) distance of 5.75 Å. Because
the gyromagnetic tensor is anisotropic, the anisotropic exchange
interaction is not negligible, and therefore, the calculated
electron-electron distance has an error of ca. 10%.
The values for the gyromagnetic tensor suggest that the first
coordination sphere around the copper ion is distorted and
involves, besides the oxygen atoms from the carboxylate groups,
two nitrogen atoms. These two atoms are most likely from
ammonia used in the synthesis of MPF-9. The distortion around
the copper center is suspected to arise from the extended
coordination ring of the glutamic acid moiety. This extended
cycle is different from the more conventional 5-coordination
ring in Cu(II)-L-arginine complexes where the 2N2O square
planar geometry is preserved.48 Moreover, these values indicate
that the main contribution to the ground state wave function is
given by the d(x2-y2) orbital.
Figure 6. Magnetization versus field curve for MPF-9 recorded at 1.8 K.
The EPR spectrum measured at 293 K exhibits the same
shape, although the signal intensity slightly decreases. To
simulate this spectrum, an octahedral gyromagnetic tensor and
large linewidths, as in the case of low temperature measure-
ments, were used. The values of the gyromagnetic tensor differ
slightly from those obtained for the low temperature measure-
ments but did not show a fluxional behavior of the copper ions.
The absence of any hyperfine structure has been previously
observed for ferromagnetic chain complexes involving carboxy-
late bridges. There, carboxylate groups involving a copper-apical
oxygen bond are more effective in transferring spin polarization
than hydrogen bonds.49 However, the relatively large electron-
electron distance suggests that, in the case of MPF-9, the
bridging does not involve carboxylate, but hydrogen bonds,
which stabilize the hybrid system by forming long copper-
copper 1-D chains. Moreover, the observed spectra are clearly
different from spectra observed for copper(II) complexes of
linear-chain fatty acids.50
Figure 7. Temperature variation of the magnetic susceptibility, ø (O) and
of the øT product (0) of MPF-9 recorded in a 5000 Oe field. The full line
corresponds to the fit carried out with the expression for Heisenberg S )
1/2 chains.
MPF-9, and Figure 6 shows the corresponding saturation curve
recorded at 1.8 K. The ø-1 dependence of T in the Curie-Weiss
plot and the saturation curve clearly show that the structure is
composed of S ) 1/2 Cu(II) ions and constitutes 1D Heisenberg
1
S ) /2 chains.51 The Curie-Weiss law is well-followed, and
the fittings leads to C ) 0.437 K cm3 mol-1 and θ ) -0.30 K.
Figure 7 shows the corresponding øT vs T curve. ø is almost
constant down to 50 K and is well fitted using the Hattfield
1
model of 1D Heisenberg S ) /2 chains.41 A small hump at
50-60 K, not observed in the susceptibility vs T curve, was
revealed when plotting the øT product variation. This can be
attributed either to very small amount of impurity or to the
presence of oxygen, the signal of which is apparent due to the
low mass of sample used for the measurement (3.2 mg). This
anomaly however does not prevent interpreting the global
Besides EPR, magnetic measurements were used to further
characterize MPF-9. Figure 5 shows a Curie-Weiss plot of
(47) Eaton, S. S.; More, K. M.; Sawant, B. M.; Eaton, G. R. J. Am. Chem. Soc.
1983, 105, 6560.
(48) Ohata, N.; Masuda, H.; Yamauchi, O. Inorg. Chim. Acta 1999, 286, 37.
(49) Colacio, E.; Ghazi, M.; Kivekas, R.; Moreno, J. M. Inorg. Chem. 2000,
39, 2882.
(50) Sharrock, P.; Dartiguenave, M.; Dartiguenave, Y. Bioinorg. Chem. 1978,
9, 3.
(51) Carlin, R. L. Magnetochemistry; Springer: Berlin-Heidelberg, 1986.
9
J. AM. CHEM. SOC. VOL. 130, NO. 8, 2008 2521

A R T I C L E S
Mantion et al.
Figure 8. Synchrotron powder diffraction patterns of MPF-9 and MPF-2 (λ ) 0.50007 Å). In both cases, the first peak has been cut at approximately 25%
of its full height to show more detail at higher angles.
Table 3. Unit Cells for MPF-2 and MPF-9
sample
a (Å)
b (Å)
c (Å)
R
(deg)
â
(deg)
γ
(deg)
space group
MPF-2
MPF-9
13.6714
19.2610
9.7210
4.8885
29.9315
30.5096
90
90
94.826
92.524
90
90
monoclinic P
monoclinic C2
Figure 9. Crystal structure of MPF-9 generated by FOX and before Rietveld
refinement, including hydrogen atoms.
Scheme 2. Chemical Structure Used as Input for FOX Analysis
Figure 10. Crystal structure of MPF-9 after refinement of the model
generated by FOX, including the water molecule.
Both values are of the same order of magnitude and are in
good agreement with the EPR values, as is the small Curie-
Weiss constant θ. A small decrease in the øT product θ is
observed from 0.437 K cm3 mol-1 at 295 K to 0.283 K cm3
mol-1 at 1.8 K. The decrease of øT and the negative sign of θ
confirm the existence of small antiferromagnetic exchange
behavior of MPF-9. Fitting of the curve leads to g ) 2.164(1)
and J/k ) -1.38(1) K. The negative J value is characteristic
for a weak antiferromagnetic coupling, as already observed from
the EPR data.52
(52) Rodriguez-Fortea, A.; Alemany, P.; Alvarez, S.; Ruiz, E. Chem.-Eur. J.
2001, 7, 627.
9
2522 J. AM. CHEM. SOC. VOL. 130, NO. 8, 2008

Metal−Peptide Frameworks
A R T I C L E S
Figure 11. Observed (top), calculated (middle), and difference (bottom) profiles for the Rietveld refinement of MPF-9. The inset has been scaled up by a
factor of 5 to show more detail.
Table 4. Rietveld Refinement of MPF-9
interactions.53 The weak J value confirms that the magnetization
exchange occurs through nitrogen atoms and H-bonds and not
unit cell
space group
a (Å)
b (Å)
c (Å)
â (deg)
refinement
number of observations
number of contributing reflections
number of geometric restraints
number of positional parameters
non-H
H
number of profile parameters
RF
Rwp
Rexp
C2
through carboxylate-mediated interactions like in amino-acid
copper(II) complexes.54 As the behavior of complexes with a
single carboxylate bridge indicates that J vanishes at distances
over 3 Å because the direct interaction predominates over the
superexchange mechanism, we exclude a single carboxylate
bridge network in the case of MPF-9.52
Because neither MPF-2 nor MPF-9 crystallize as single
crystals, powder diffraction methods had to be used for structure
solution. The X-ray powder diffraction patterns for MPF-2 and
MPF-9 are shown in Figure 8. They were indexed using Topas
academic31,32 and the results are shown in Table 3.
The quality of the powder diffraction pattern for MPF-2 did
not allow structure solution. However, the powder pattern does
suggest a lamellar structure. The peaks at 2θ values of 0.9068,
1.92, 2.88, and 3.88° (d-spacings of 29.85, 14.91, 9.95, and 7.46
Å) can be indexed as 001, 002, 003, and 004. This indicates a
lamellar structure perpendicular to the c direction, which in turn
suggests a â-sheet structure.
The diffraction pattern of MPF-9 was of better quality, and
a crystal structure model could be obtained using the direct-
space global-optimization algorithms implemented in the com-
puter program FOX.35,36 The chemical information used as input
for these optimizations (connectivity, bond distances and angles)
is shown in Scheme 2. Le Bail extraction was performed to
determine Cagliotti parameters and the η ratio with Fullprof
using a Pseudo-Voigt profile.33,34 The best of the several copper
complex structures generated by FOX is shown in Figure 9.
Elemental and thermogravimetric analyses indicated that there
is at least one water molecule per copper ion present in the
structure, but to keep the model as simple as possible, this was
19.2665(10)
4.8848(2)
30.5146(21)
92.5(1)
4723
1107
207
113
57
11
0.103
0.266
0.217
not included in the FOX simulations. Some of the torsion angles
were fixed or restricted to a certain range to limit the possibility
of unrealistic torsion angles. For the optimization, data to sin
θ/λ ) 0.20 (11.5° 2θ) were used. To allow for the steric
requirements of the hydrogen atoms, they were included in the
model submitted to Fox.
To complete and refine the model from FOX, a Rietveld
refinement using the program XRS-8237 was undertaken. The
complete diffraction pattern was used, and bond distance and
angle restraints were included to keep the model chemically
sensible. H atoms were added where their positions could be
calculated from the connected atoms, but for the ammonia
molecules and methyl groups, where the positions of the H
atoms are not well defined, the occupancy factor for the
corresponding N and C atoms were simply increased to include
the three electrons from the attached H atoms (e.g., CH3 was
modeled as 1.5 C) and the atomic displacement factor increased
slightly. During the course of the refinement, the weight of the
geometric restraints with respect to the diffraction data was
reduced progressively, and in the final stages a factor of 1.0
(53) Zou, J.-Z.; Liu, Q.; Xu, Z.; You, X.-Z.; Huang, X.-Y. Polyhedron 1998,
17, 1863.
(54) Gerard, M. F.; Aiassa, C.; Casado, N. M. C.; Santana, R. C.; Perec, M.;
Rapp, R. E.; Calvo, R. J. Phys. Chem. Solids 2007, 68, 1533.
9
J. AM. CHEM. SOC. VOL. 130, NO. 8, 2008 2523

A R T I C L E S
Mantion et al.
Figure 12. Packing scheme of the crystal. View slightly off the crystallographic b-axis. Hydrogen atoms have been removed for clarity.
was applied. A water molecule was located in a difference
electron density map and added to the structural model as an O
atom with an occupancy of 1.25 (to account for the two
unlocated H atoms). To keep the number of parameters to a
minimum, the atomic displacement parameters were fixed at
typical values and not refined. Refinement of this model
converged with the R-values RF ) 0.103, Rp ) 0.211 and Rwp
) 0.266 (Rexp ) 0.217). The crystal structure including the water
molecule is shown Figure 10. The profile fit is shown in Figure
11. Further details of the refinement are given in Table 4. The
atom numbering convention is given in the Supporting Informa-
tion, Figure S1. Atomic positions are given in Table S1, selected
interatomic distances are given in Table S2, and bond angles
are given in Table S3. Platon’s CALC Void40 function log
journal is given in Table S4.
The copper ion is coordinated to two oxygens of the
carboxylic acid and to two ammonia molecules introduced
during the synthesis in a distorted square planar geometry. The
Figure 13. â-sheet-like structure and hydrogen bonding in the peptide
backbone of MPF-9. Hydrogen atoms have been removed for clarity and
N‚‚‚O interactions are indicated by a dotted line.
Cu-N distances of 1.98(3) and 1.99(3) Å are in accordance
with the generally observed values of 1.90-2.02 Å. Similarly,
the Cu-O distances of 1.98(3) and 1.99(3) Å correspond to
those observed in related systems.55-58 A global view of the
crystal structure of MPF-9 is shown in Figure 12, and a
magnified view along the crystallographic a-axis in Figure 13.
Both figures show that there are multiple interactions present
in MPF-9. The water molecule forms H-bonds with ammonia
molecules of neighboring Cu complexes along the [100]
direction (N‚‚‚O 2.88(5) and 3.02(4) Å) to create a helical
H-bonding network along the [010] direction (Figure 14 and
Supporting Information Figure S2). The other ammonia mol-
ecule forms H-bonds with the L-glutamic acid residue of the
neighboring Cu complex along the [001] direction (N‚‚‚O 2.87-
(4) Å). There are two further H-bonds linking the peptides along
the [010] direction (N‚‚‚O 2.84(4) Å and N‚‚‚O 3.01(4) Å). This
dense hydrogen-bonding network accounts for the insolubility
of this material in both water and organic solvents. The structure
is, within the limitations of the method, very well-defined. It is
also consistent with the results of the elemental analysis, TGA,
EPR, and magnetic measurements.
The structure can be described in terms of two 2D substruc-
tures, which are interconnected via hydrogen bonding. However,
there is no extended Cu(II) 3D network. The first substructure
is composed of a double layer of copper ions in the ab plane
that are diagonally displaced with respect to one another. Along
[010], the copper(II) centers are separated by a distance of 4.9
Å and along [100], by 6.2 Å and 10.0 Å. Such a structure is in
line with the magnetic properties of the material, which exhibits
a weak antiferromagnetism and thus long Cu(II)-Cu(II) dis-
tances. Because of the spacing of the Cu(II) centers, the
magnetic transfer cannot be based on carboxylate pathways, and
must therefore be based on hydrogen bonding. Despite the
existence of hydrogen bonds between adjacent complexes, there
is no magnetic coupling between them, because the large number
of atoms involved in the hydrogen-bonding network limits the
magnetic interaction. The large distance of 6.2 Å between two
columns of copper (II) as well as the packing of the ligands
prevents any magnetic interactions between them and is in line
with the modeling of the magnetic properties as a Heisenberg
(55) Devereux, M.; McCann, M.; Cronin, J. F.; Ferguson, G.; McKee, V.
Polyhedron 1999, 18, 2141.
(56) Chawla, S. K.; Na¨ttinen, K.; Rissanen, K.; Yakhmi, J. V. Polyhedron 2004,
23, 3007.
(57) Rodriguez-Martin, Y.; Ruiz-Pe´rez, C.; Sanchiz, J.; Lloret, F.; Julve, M.
Inorg. Chim. Acta 2001, 318, 159.
1
(58) Li, J.; Zeng, H.; Chen, J.; Wang, Q.; Wu, X. Chem. Commun. 1997, 1213.
S ) /2 1D chain.
9
2524 J. AM. CHEM. SOC. VOL. 130, NO. 8, 2008

Metal−Peptide Frameworks
A R T I C L E S
Figure 14. Hydrogen-bonding network in the ac plane (left) and a perpendicular view of the helical linkage between water and ammonia molecules along
[010] (right). The hydrogen atoms have been removed for clarity.
Figure 15. Packing diagram of MPF-9. Yellow balls indicate the voids.
The second double layer in the ab plane is that of the peptide
with its long molecule axis oriented roughly along [001]. In
addition to the H-bonding between the peptides along [010],
the phenyl rings form a column along [010]. The centers of
gravity of the individual phenyl residues are 4.9 Å apart,
indicating some π-π-stacking interaction. The intercolumn
distances are in the same range and the rings form a zigzag
arrangement, which implies that π-π stacking is responsible
for both the internal and external column stabilization.
The hydrogen-bonding scheme adopted by the peptides in
the crystal is shown in Figure 13. There is a â-sheet-like
interaction along the crystallographic b-axis. The presence of a
â-sheet bonding scheme has also been shown by IR spectros-
copy (Figure 3), but the crystal structure further shows that not
all of the peptide is part of the â-sheet. Only the carbamate and
the L-valine-L-valine peptide bond are involved. To maximize
the number of hydrogen bonds, the â-sheet is distorted slightly
to accommodate the rather large side chains of the valine units.
This distortion enables the L-glutamic acid residue to adapt to
the metal center coordination geometry. The crystallographic
results confirm those of the IR spectroscopy, and also reveal a
more complex bonding scheme than simple â-sheets. Further-
more, the crystal structure shows that the â-sheet orientation,
which could not be determined from IR, is parallel and not
antiparallel along the b-axis.
9
J. AM. CHEM. SOC. VOL. 130, NO. 8, 2008 2525

A R T I C L E S
Mantion et al.
The hydrogen bonding between the water and coordinated
ammonia molecules that connects copper centers in adjacent
columns is shown in Figure 14. This hydrogen-bonding scheme
is an effective pathway for transferring magnetization through
hydrogen bonding.46,59 The structure presented here is a new
mode of coordination for the well-known glutamic acid
ligand,46,49,60-62 where there is no magnetic exchange through
carboxylate centers, although here the glutamic acids are not
involved in a peptide-like structure. However, the gain in energy
given by the formation of â-sheets via hydrophobic interaction,
hydrogen bonding, and extensive π-π stacking of benzyl rings
is highly favorable and compensates the effect of the nonclas-
sical bonding scheme.
MPF-2 and MPF-9 are just the first examples of MPFs. They
demonstrate that new peptide-based materials can be generated
using rather simple building blocks and low strength, nonco-
valent interactions like hydrogen bonding63,64 and π-π stacking
of benzyl rings,65,66 along with metal complexation. A structur-
ally interesting point is the role of the amines located around
the copper(II). They not only complete the coordination sphere
but also constitute a way of ensuring the stabilization and weak
magnetization transfer in a metal column using hydrogen
bonding. Moreover, π-π stacking is very useful in stabilizing
and organizing the complexes.
Conclusion
The porosity of MPF-9 was analyzed using Platon’s voids
calculation functionality and plotting.40 Figure 15 shows the
pores in MPF-9 that are accessible to small solvent or gas
molecules. The total void volume is 387 Å3 for a total unit cell
volume of 2867 Å3. This results in a total void volume of just
above 13% of the cell volume. MPF-9 only has chiral channel-
like pores parallel to the [010] direction, but two ranges of sizes
can be distinguished. The larger pores have a volume of around
101 Å3. This is large enough to accommodate methanol, ethanol,
or possibly benzene if the linear molecular shape of the latter
three and the possible π-π interactions are considered. The
smaller voids are around 20-37 Å3. These pores are large
enough to accommodate hydrogen and carbon dioxide, or even
water (in the pores at the larger end of the spectrum). However,
the small connections between the 20 Å3 voids prevent them
from forming an extended channel along the b-axis. As a result,
MPF-9 is a candidate for gas adsorption and storage, for
purification, or for chiral separation.
This paper presents the first example of metal-peptide
frameworks, “bioinspired” analogs to the well-known MOFs.
Metal-oxygen and metal-nitrogen bonding and various types
of hydrogen bonding along with π-π-stacking are responsible
for the formation of a porous network structure, which is stable
up to ca. 250 °C. Despite the lack of single crystals, the crystal
structure of the copper complex MPF-9 could be determined
from powder diffraction data using the FOX program. The
refined crystal structure is consistent with the spectroscopic
information.
The 1D channel architecture with two different pore sizes
makes MPF-9 an interesting material for gas storage. Further
applications include gas or chiral separation or catalysis. Finally,
the organization of the Cu(II) ions in a 2D layer pattern yields
a material with an interesting paramagnetic character. The
current paper therefore clearly shows that small peptides are
viable candidates for the preparation of complex metal-organic
materials with defined topologies. MPFs thus further broaden
the more general field of MOFs and entactic metal systems,67
which are interesting for a wide range of technical applications.
In summary, metal-peptide frameworks are a new and
interesting variety of metal-organic frameworks. The coordina-
tion mode of MPF-9 differs from other R-ω dicarboxylic acids.
There is no bridging of the carboxylic acid moiety between the
metal centers as seen in refs 14-21. The current paper therefore
suggests that peptides with different topologies should, like
regular organic dicarboxylic acids, allow for the fabrication of
MPFs with different topologies with specifically designed
(chiral) inner pores. The ligand acts as a chelator, similar to
smaller dicarboxylic acids like malonates.55,56 This is rather
surprising but can be rationalized via the overall energy gain
from the formation of a highly hydrogen-bonded network with
π-π-stacking. This energy gain is presumably higher than the
energy gained by forming an extensive carboxylate network.
Acknowledgment. We thank Prof. E.C. Constable for access
to his IR spectrometer, M. Du¨ggelin and D. Mathys for SEM
measurements, Prof. K. Fromm for access to her TGA, Prof.
M. Meuwly for the access to one of his Linux clusters, F.
Schmid for FOX and IT support, Dr. R. Cˇerny´ and Dr. V. Favre-
Nicolin for FOX support, J. Rohl´ıcˇek providing a computer-
grid capable FOX version, and Dr. Markus Neuburger for useful
discussions. Experimental assistance from the staff at SNBL is
gratefully acknowledged. The Swiss National Science Founda-
tion, the NCCR Nanosciences, the University of Potsdam, and
the Max-Planck-Institut of Colloids and Interfaces are acknowl-
edged for financial support. A.T. thanks the Holcim Stiftung
Wissen for a Habilitation Fellowship.
Furthermore, the synthesis conditions are convenient and there
is no need for autoclave, microwave, or slow diffusion synthesis
protocols, which are often used for MOF’s.5 The ease of
synthesis is an important point for any industrial application.
Although the overall cost of MPFs is somewhat higher than
that of conventional MOFs, the ability to include chiral
information and possibly other chemical information like metal
coordinating sites a priori should alleviate issues of higher cost.
Supporting Information Available: Figures S1 and S2 and
Tables S1-4. This material is available free of charge via the
Internet at http://pubs.acs.org.
JA0762588
(59) Levstein, P. R.; Calvo, R. Inorg. Chem. 1990, 29, 1581.
(60) Steren, C. A.; Gennaro, A. M.; Levstein, A. M.; Calvo, P. R. Condens.
Matter 1989, 1, 637.
(63) Beatty, A. M. Cryst. Eng. Commun. 2001, 243.
(64) Katsura, N.-K. Chem. Soc. ReV. 2005, 34, 109.
(61) Colacio, E.; Dominguez-Vera, J. M.; Costes, J. P.; Kivekas, R.; Laurent,
J. P.; Ruiz, J. Inorg. Chem. 1992, 31, 774.
(65) Roesky, H. W.; Andruh, M. Coord. Chem. ReV. 2003, 236, 91.
(66) Ye, B.-H.; Tong, M.-L.; Chen, X.-M. Coord. Chem. ReV. 2005, 249, 545.
(67) Comba, P. Coord. Chem. ReV. 2000, 200-202, 217.
(62) Pushita, W. J.; Odani, A.; Yamauchi, O. J. Inorg. Biochem. 1999, 73, 203.
9
2526 J. AM. CHEM. SOC. VOL. 130, NO. 8, 2008