Structural and spectroscopic study of the dihydrogen bond in an imine triosmium complex

Article abstract of DOI:10.1021/om010650r

The presence of an intramolecular XH?HM interaction between the imine proton donor and the terminal hydride in H(μ-H)Os₃(CO) ₁₀(HN=CPh₂) has been investigated by X-ray analysis, NMR and IR spectroscopy and theoretical calculations. The localization of the hydrogen atoms in the crystal structure yielded a H?H distance of 1.79(6) A? for this unconventional H?H interaction; theoretical calculations suggested an H?H distance of 1.89 A? in the solid state. A NMR determination of the interproton distance, obtained from the isolation of the selective H,H dipolar contribution to the hydride relaxation time, afforded a value of 2.00 ± 0.05 A?. The difference between NMR and solid state determinations may be explained on the basis of the occurrence, in solution, of a large amplitude oscillatory motion of the imine ligand along the N-Os coordination axis. Further evidence of the presence of the favorable N-H?H-M intramolecular hydrogen bond interaction has been obtained from the red shift of the v(N-H) stretching in H(μ-H)Os₃(CO) ₁₀(HN=CPh₂) with respect to that of the related Os ₃(CO)₁₁(HN=CPh₂) compound. DFT(B3LYP) calculations gave results in agreement with the experimental findings and allowed further insight into the nature of the N-H?H-M dihydrogen bond, pinpointing the electrostatic nature of this interaction and the role of the high polarizability of the Os-H bond.

Full text of DOI:10.1021/om010650r

50
Organometallics 2002, 21, 50-57
Str u ctu r a l a n d Sp ectr oscop ic Stu d y of th e Dih yd r ogen
Bon d in a n Im in e Tr iosm iu m Com p lex
S. Aime,*,† E. Diana,† R. Gobetto,† M. Milanesio,‡ E. Valls,§ and D. Viterbo‡
Dipartimento Chimica I.F.M., Universita` di Torino, Via P. Giuria 7, 10125 Torino, Italy,
Dipartimento di Scienze e Tecnologie Avanzate, Universita` del Piemonte Orientale,
Corso T. Borsalino 54, 15100 Alessandria, Italy, and Unitat de Quimica Inorganica,
Universitat Autonoma de Barcelona, 08193 Bellaterra, Spain
Received J uly 30, 2001
The presence of an intramolecular XH‚‚‚HM interaction between the imine proton donor
and the terminal hydride in H(µ-H)Os3(CO)10(HNdCPh2) has been investigated by X-ray
analysis, NMR and IR spectroscopy and theoretical calculations. The localization of the
hydrogen atoms in the crystal structure yielded a H‚‚‚H distance of 1.79(6) Å for this
“unconventional” H‚‚‚H interaction; theoretical calculations suggested an H‚‚‚H distance of
1.89 Å in the solid state. A NMR determination of the interproton distance, obtained from
the isolation of the selective H,H dipolar contribution to the hydride relaxation time, afforded
a value of 2.00 ( 0.05 Å. The difference between NMR and solid state determinations may
be explained on the basis of the occurrence, in solution, of a large amplitude oscillatory
motion of the imine ligand along the N-Os coordination axis. Further evidence of the
presence of the favorable N-H‚‚‚H-M intramolecular hydrogen bond interaction has been
obtained from the red shift of the ν(N-H) stretching in H(µ-H)Os3(CO)10(HNdCPh2) with
respect to that of the related Os3(CO)11(HNdCPh2) compound. DFT(B3LYP) calculations
gave results in agreement with the experimental findings and allowed further insight into
the nature of the N-H‚‚‚H-M dihydrogen bond, pinpointing the electrostatic nature of this
interaction and the role of the high polarizability of the Os-H bond.
In tr od u ction
as “unconventional” hydrogen bonds. In particular,
Orlova and Scheiner have demonstrated that such
H‚‚‚H interaction is favored in the case of poor and
moderate proton donors H-X and a strong π acceptor
trans ligand.9 It is easy to foresee that a better under-
standing of the role of these kinds of interactions may
contribute to a better design of reagents and catalysts
in organometallic chemistry.
We have suggested that the Os-H‚‚‚H-N interaction
is responsible for driving the stereochemistry in a series
of adducts containing primary and secondary amines
coordinated to a triosmium cluster.10,11 Moreover it has
been found that the addition of aldehydes or ketones to
the ammonia-containing derivative leads to the forma-
tion of terminally coordinated imine ligands which are
stabilized through the same type of interactions.12
Herein we present an in-depth experimental (using
IR, NMR, and X-ray data) and theoretical characteriza-
tion of the Os-H‚‚‚H-N DHB in the trimetallic cluster
H(µ-H)Os3(CO)10(HNdCPh2) (1 in Chart 1), reporting
the first crystal structure, in which Os-H acts as
H-bond acceptor, with an H‚‚‚H distance below 2.0 Å;
this interaction is the shortest H‚‚‚H distance, among
the Os-H‚‚‚H-X contacts, found in the Cambridge
Structural Database (CSD).13 Our findings allow us to
get a more detailed picture of the structural and
The recent observation of a new type of dihydrogen
bond (DHB) between hydrides (M-H) and classical
hydrogen donor groups, such as N-H or O-H, repre-
sents an important step toward the understanding of
the role of weak interactions in determining the stereo-
chemistry and reactivity of organometallic systems.1-5
The M-H‚‚‚H-X (X ) O, N) linkage may be either
intra- or intermolecular, and the H‚‚‚H distance has
been reported to be as small as 1.7 Å.6
Theoretical investigations of the DHB in main group
elements7,8 and in monometallic systems9 provided a
first rationalization scheme to classify these interactions
† Universita` di Torino.
‡ Universita` del Piemonte Orientale.
§ Universitat Autonoma de Barcelona.
(1) (a) Lee, J . C.; Peris, E.; Rheingold, A. L.; Crabtree R. H. J . Am.
Chem. Soc. 1994, 116, 11014-11019. (b) Lee, J . C.; Rheingold, A. L.;
Muler, B.; Pregosin, P. S.; Crabtree, R. H. J . Chem. Soc., Chem.
Commun. 1994, 1021-1022. (c) Park, S.; Ramachandran, R.; Lough,
A. J .; Morris, R. H. J . Chem. Soc., Chem. Commun. 1994, 2201-2202.
(2) Crabtree, R. H.; Siegbahn, P. E. M.; Eisenstein, O.; Rheingold
A. L.; Koetzle, T. F. Acc. Chem. Res. 1996, 29, 348-354, and references
therein.
(3) Belkova, N. V.; Shubina, E. S.; Ionidis, A. V.; Epstein, L. M.;
J acobsen, H.; Messmer, A.; Berke, H. Inorg. Chem. 1997, 36, 1522-
1525.
(4) Peris, E.; Lee, J . C.; Rambo, J . R.; Eisenstein, O.; Crabtree, R.
H. J . Am. Chem. Soc. 1995, 117, 3485-3491.
(5) Custelcean, R.; J ackson, J . E. Chem. Rev. 2001, 101, 1963-1980.
(6) Braga, D.; Grepioni, F.; Tedesco, E.; Calhorda, M. J .; Lopes, P.
E. M. New J . Chem. 1999, 219-226, and references therein.
(7) Kulkarni, S. A.; Srivastava, A. K. J . Phys. Chem. A 1999, 103,
2836-2842.
(8) Grabowsky, S. J . J . Phys. Chem. A 2000, 104, 5551-5557.
(9) Orlova, G.; Scheiner, S. J . Phys. Chem. A 1998, 102, 260-269.
(10) Aime, S.; Gobetto, R.; Valls, E. Organometallics 1997, 16, 5140-
5141.
(11) Aime, S.; Ferriz, M.; Gobetto, R.; Valls, E. Organometallics
2000, 19, 707-710.
(12) Aime, S.; Ferriz, M.; Gobetto R.; Valls, E. Organometallics 1999,
18, 2030-2032.
10.1021/om010650r CCC: $22.00 © 2002 American Chemical Society
Publication on Web 12/04/2001

Dihydrogen Bond in an Imine Triosmium Complex
Organometallics, Vol. 21, No. 1, 2002 51
Ta ble 1. Su m m a r y of Cr ysta l Da ta a n d Str u ctu r e
Refin em en t P a r a m eter s for 1
Ch a r t 1
formula
fw
C23 H13NO10Os3
1033.94
cryst syst
triclinic
space group
a
P1h
8.013(1) Å
11.324(1) Å
15.004(2) Å
84.067(2)°
80.878(2)°
77.391(2)°
1308.6(2) Å3
2
b
c
R
â
γ
V
Z
F(000)
density (calc)
temp
932
2.624 g cm-3
293(2) K
diffractometer
radiation (graph monochr.)
abs coeff
cryst size
color
SMART-CCD
0.71073 Å
14.583 mm-1
0.14 × 0.10 × 0.05 mm
orange
electronic determinants of the formation of a DHB on
the surface of a triosmium cluster.
shape
triangular plate
ω
scan method
frame width
time per frame
no. frames
detector-sample distance
θ range
0.30°
30 s
2400
2.95 cm
1.38-30.53
-11 e h e 11, -16 e k e 16,
-21 e l e 21
23 300
7948 [R(int) ) 0.0417]
2.93
Exp er im en ta l Section
Syn th eses a n d NMR Da ta . All solvents were dried and
stored over molecular sieves. (µ-H)2Os3(CO)1014 and Os3(CO)11-
(NCCH3)15 were prepared according to published procedures.
index ranges
no. reflns collected
no. ind reflns
abs corr (SADABS)
av redundancy
max.-min. transmn
refinement method
no. obsd reflns
13CO metal carbonyls were prepared by using as starting
material 13C-enriched (ca. 20%) Os3(CO)12 obtained by direct
exchange of 13CO with Os3(CO)12 in n-octane at 110 °C for 3
days in sealed vials.16 13CO (99% enriched) was purchased from
Isotec (Miamisburg, OH). 1H and 13C NMR spectra were
recorded on a J EOL EX-400 spectrometer operating at 399.65
and 100.24 MHz, respectively. Nonselective inversion recovery
pulse sequences were used to obtain 1H T1 values.17 The
samples for T1 measurements were degassed using freeze-
pump-thaw techniques. The temperature was calibrated with
0.156-0.039
2
full-matrix least-squares on |Fo|
5947 [I>2σ(I)]
no. data/restraints/params
7948/1/346
2
goodness-of-fit on |Fo|
0.880
R indices [I>2σ(I)]
R indices (all data)
largest diff peak and hole
weight (calcd)
R1 ) 0.0237, wR2 ) 0.0471
R1 ) 0.0356, wR2 ) 0.0481
1.307, -1.108 e Å-3
1
a standard methanol H thermometer sample.15 Errors in the
2
w ) 1/[σ2(|Fo| ) + (P)2],
reported T1 values were estimated to be in the range (2%.
The accurate determination of the interproton distances has
been obtained by fitting the experimental T1 vs 1/T and by
applying the reported equations (vide infra).
2
2
P ) (|Fo| + 2|Fc| )/3
IR (CHCl3 solution, carbonyl region): 2096(w), 2077(w,sh),
2066(s), 2044(s), 2019(s,br), 2001(s,br).
IR An a lysis. IR spectra were recorded as CH2Cl2 or CHCl3
solution for the carbonyl stretching region and as KBr pellets
for the other spectral regions on a Bruker FTIR spectropho-
tometer model Equinox 55.
Syn th esis of H(µ-H)Os3(CO)10(HNdCP h 2) (Ch a r t 1). In
a typical reaction 5 mg of (µ-H)2Os3(CO)10 (5.9 × 10-6 mol) in
0.5 mL of CD2Cl2 was transferred into a 5 mm NMR tube, and
an excess of Ph2CdNH was added. The solution color readily
turned from purple to yellow. 1H NMR (CD2Cl2, 400 MHz, 218
Cr ysta l Str u ctu r e An a lysis. Crystals of the syn isomer
(1 in Chart 1) of H(µ-H)Os3(CO)10(HNdCPh2) suitable for X-ray
analysis were obtained by evaporation of a hexane/chloroform
solution. Crystallographic data and details of data collection
and refinement are given in Table 1. Single-crystal diffraction
data were collected at RT on a Bruker SMART CCD area
detector diffractometer, using graphite-monochromatized Mo
KR (λ ) 0.71073 Å) radiation.18
3
K): δ 9.87 (NH), 7.67-7.28 (aromatic), -9.90 (d, J ) 2.8 Hz,
H2), -15.85 (d, 3J ) 2.8 Hz, H3). IR (CH2Cl2 solution, carbonyl
region): 2100(m), 2074(w,sh), 2062(s), 2049(s), 2021(ms), 2004-
(ms), 1983 (m,sh), 1964 (m,sh).
Syn th esis of Os3(CO)11(HNdCP h 2). Os3(CO)11(NCCH3)
(10 mg) dissolved in 1 mL of dichloromethane was allowed to
react with an equimolar amount of benzophenone imine (HNd
CPh2). Complete conversion to Os3(CO)11(HNdCPh2) occurs in
a few minutes, and the compound was characterized by 1H
and 13C data. 1H NMR (CD2Cl2, 400 MHz, 193 K): δ 9.22 (NH)
7.30-7.65 (aromatic). 13C NMR for the carbonyl resonances
(CD2Cl2, 400 MHz, 193 K): δ 188.9 (2) (2J C,C ) 34.6 Hz), 185.2
(2) (2J C,C ) 34.6 Hz), 182.2 (1), 177.9 (2), 175.2 (2), 174.3 (2).
Absorption correction was performed using SADABS.19 The
structure was solved by direct methods (SIR97)20 and refined
by full-matrix least-squares (SHELX97).21 Anisotropic thermal
parameters were assigned to all the non-hydrogen atoms.
(18) Low-temperature data were collected on a larger crystal using
our diffractometer with a scintillation counter, but the data were even
more severely affected by absorption. We then decided to use a much
smaller crystal on a CCD diffractometer where low temperature was
not available, because absorption is a much greater problem than
thermal motion.
(19) Sheldrick, G. M. SADABS; University of Go¨ttingen: Germany,
1996.
(20) Altomare, A.; Burla, M. C.; Camalli, M.; Cascarano, G. L.;
Giacovazzo, C.; Guagliardi, A.; Moliterni, A. G.; Polidori, G.; Spagna,
R. J . Appl. Crystallogr. 1999, 32, 115-1134 (web site: http://
www.ba.cnr.it/IRMEC/SirWare_main.htm).
(13) Allen, F. H.; Kennard, O. Chem. Des. Autom. News 1993, 8 (1),
1 & 31.
(14) Kaesz, H. D. Inorganic Synthesis; Angelici, R. J ., Ed.; J ohn
Wiley & Sons: New York, 1990; Vol. 28, p 238.
(15) Nicholls, J . N.; Vargas, M. D. Inorganic Synthesis; Angelici, R.
J ., Ed.; J ohn Wiley & Sons: New York, 1990; Vol. 28, p 232.
(16) Aime, S.; Milone, L.; Osella, D.; Sappa, E. Inorg. Chim. Acta
1978, 29, 2211-2218.
(17) Martin, M. L.; Delpuech, J . J .; Martin, G. J . In Practical NMR
Spectroscopy; Heyden: London, 1980.

52 Organometallics, Vol. 21, No. 1, 2002
Aime et al.
Hydrogens bonded to the metals were located in a difference
Fourier synthesis, using the structure refined with anisotropic
displacement factors and reflections with 2θ < 40°. Aromatic
and iminic hydrogens were set in calculated positions, after
checking that they were present in the difference Fourier
synthesis. Final atomic coordinates and temperature factors,
together with all the other crystallographic information have
been deposited.22
Th eor etica l Ca lcu la tion s. Ab initio Hartree-Fock23 (HF)
SCF-MO calculations24 on the syn-(1) and anti-(2) isomers
(Chart 1) and on some simple adducts (vide infra) were per-
formed using the Gaussian9825 program. The initial geometry
of 1 was obtained from the X-ray analysis, and that of 2 was
achieved by graphical manipulation of 1. Geometrical optimi-
zations were carried out at Hartree-Fock level, using Stut-
tgart RSC ECP pseudo-potentials and related basis set26 for
the Os atoms, the 6-31+G(d,p)23 basis set for the H‚‚‚H contact
region, and the 3-21G23 basis set for the remaining part of the
molecule. Single-point energy calculations were carried out at
the DFT level by adopting the hybrid B3-LYP functional,
including the exchange part proposed by Becke27 and the
correlation functional proposed by Lee et al.28 Stuttgart RSC
ECP26 for the Os atoms, 6-311++G(2d,2p)23 for the H‚‚‚H
contact region, and 6-31G(d)23 for the remaining part of the
molecule were employed.
tion also under mild experimental conditions. Only
benzophenone imine is stable enough to allow its use
in reactions with metal carbonyls.33
The reaction between the coordinatively unsaturated
(µ-H)2Os3(CO)10 cluster and HNdCPh2 proceeds readily
at room temperature with a dramatic change in color
from violet to pale yellow. This reaction is conveniently
carried out in the NMR tube and turns out to be
sensitive to the type of solvent.
In CD2Cl2 a single product corresponding to H(µ-H)-
Os3(CO)10(HNdCPh2) (1) is obtained, as clearly dem-
onstrated by the 13C NMR spectrum of a 13CO-enriched
sample at 218 K which shows 10 carbonyl resonances
of equal intensity at δ 185.5 (2J C-C ) 34.9 Hz), 184.3
(2J C-C ) 34.9 Hz), 180.3 (2J C-H ) 20.8 Hz), 175.9, 175.2,
174.9, 173.3, 173.1, 172.7 (2J C-H ) 12.8 Hz), and 168.7
(2J C-H ) 10.8 Hz), respectively. A partial assignment
of the carbonyl resonances is straightforward on the
2
2
basis of the J C-C and J C-H couplings. For instance,
the two lowest field resonances showing the carbon-
carbon coupling pattern correspond to the axial carbo-
nyls in the rear Os(CO)4 unit. The carbonyl trans to the
terminal hydride falls at 180.3 ppm, whereas the
resonances at 172.7 and 168.7 ppm correspond to
carbonyls trans to the bridging hydride. When the
temperature is increased, a small broadening and shift
of most of the resonances (more evident for the reso-
nances at δ 180.3 and 172.7) followed by their sharpen-
ing is observed. In analogy with what was previously
observed for other H(µ-H)Os3(CO)L complexes,34 this
behavior can be explained with the exchange of the syn-
and anti-isomers of H(µ-H)Os3(CO)10(HNdCPh2) (1 and
2 in Chart 1). In fact a dynamic process involving the
two hydrides and the two carbonyls which show the
more pronounced broadening is responsible for the
mutual exchange of the syn-isomer with the minor
(nondetectable in CD2Cl2) anti-isomer.
The graphic analysis was performed using the MOLDRAW29
and the XP30 programs, while the retrieval of the related
structures was from the CSD.13
Rigid energy scans around selected bonds were carried out
using the simple potential functions implemented in MOLD-
RAW.29
Resu lts
Routes to imino metal complexes are usually based
on the modification of previously coordinated ligands
such as nitriles, oximes, and amines.31 In polynuclear
derivatives these reactions lead invariantly to systems
in which the organic ligand occupies a bridging posi-
tion.32 The direct use of imines as ligands is hindered
by the difficulty of avoiding their polymerization reac-
As previously11 observed in related derivatives, when
H(µ-H)Os3(CO)10(HNdCPh2) is dissolved in polar sol-
vents (i.e., acetone) the two isomers are formed in
equimolar amounts. In CD2Cl2 solution, the intramo-
lecular Os-H‚‚‚H-N interaction drives the stereochem-
istry of the adduct, whereas the competition of acetone
to act as proton acceptor leads to an equimolecular
distribution of the two isomers. In acetone the activation
barrier for the exchange between the syn- and anti-
isomer corresponds to ca. 13.9 kcal/mol, as calculated
from the band shape analysis as a function of temper-
ature. In carrying out this determination, the monitor-
(21) Sheldrick G. M. SHELXL-97; University of Go¨ttingen: Ger-
many, 1997 (web site: http://shelx.uni-ac.gwdg.de/SHELX/).
(22) Supporting Information (coordinates of H atoms, anisotropic
thermal parameters, bond distances and angles) has been deposited
by the Cambridge Crystallographic Data Centre, with deposition
number CCDC163107.
(23) Hehre, W. J .; Radom, L.; Schleyer, P.v. R.; Pople, J . A. Ab Initio
Molecular Orbital Theory, 1st ed.; J ohn Wiley & Sons: New York, 1986;
Chapter 4.
(24) Various platforms, among which the IBM SP3 and O3K
supercomputers of CINECA (www.cineca.it), were employed, because
of the large size of the model system.
(25) Frisch, M. J .; Trucks, G. W.; Schlegel, H. B.; Scuseria, G. E.;
Robb, M. A.; Cheeseman, J . R.; Zakrzewski, V. G.; Montgomery, J . A.,
J r.; Stratmann, R. E.; Burant, J . C.; Dapprich, S.; Millam, J . M.;
Daniels, A. D.; Kudin, K. N.; Strain, M. C.; Farkas, O.; Tomasi, J .;
Barone, V.; Cossi, M.; Cammi, R.; Mennucci, B.; Pomelli, C.; Adamo,
C.; Clifford, S.; Ochterski, J .; Petersson, G. A.; Ayala, P. Y.; Cui, Q.;
Morokuma, K.; Malick, D. K.; Rabuck, A. D.; Raghavachari, K.;
Foresman, J . B.; Cioslowski, J .; Ortiz, J . V.; Stefanov, B. B.; Liu, G.;
Liashenko, A.; Piskorz, P.; Komaromi, I.; Gomperts, R.; Martin, R. L.;
Fox, D. J .; Keith, T.; Al-Laham, M. A.; Peng, C. Y.; Nanayakkara, A.;
Gonzalez, C.; Challacombe, M.; Gill, P. M. W.; J ohnson, B. G.; Chen,
W.; Wong, M. W.; Andres, J . L.; Head-Gordon, M.; Replogle, E. S.;
Pople, J . A. Gaussian 98, revision A.7; Gaussian, Inc.: Pittsburgh, PA,
1998.
(26) Andrae, D.; Haeussermann, U.; Dolg, M.; Stoll, H.; Preuss, H.
Theor. Chim. Acta 1990, 77, 123. STUTTGART ECPs are available at
the address: http://www.theochem.uni-stuttgart.de/.
(27) Becke, A. D. J . Chem. Phys. 1993, 98, 5648-5652.
(28) Lee, C.; Yang, W.; Parr, R. G. Phys. Rev. B 1988, 37, 785-789.
(29) Ugliengo, P.; Viterbo, D.; Chiari, G. Z. Kristallogr. 1993, 207,
9-23 (web site: http://www.ch.unito.it/ifm/fisica/moldraw/moldraw.
html).
(31) (a) Mehrota, R. C. In Comprehensive Organometallic Chemistry;
Wilkinson, G., Gillard, R. D., McCleverty, J . A., Eds.; Pergamon:
Oxford, U.K., 1987; Vol. 2, pp 279-281. (b) Prenzler, D. P.; Hockless,
D. C. R.; Heath, G. A. Inorg. Chem. 1997, 36, 5845-5849. (c) Francisco,
L. W.; White, P. S.; Templeton, J . L. Organometallics 1996, 15, 5127-
5136. (d) Balley, P. J .; Grant, K. J .; Pace, S.; Parsons, S.; Stewart, L.
J . J . Chem. Soc., Dalton Trans. 1997, 4263-4270. (e) Esteruelas, M.
A.; Lahoz, F. J .; Lo´pez, A. M.; On˜ate, E.; Oro, L. A. Organometallics
1995, 14, 2496-2500.
(32) (a) Farmery, K.; Kilner, M.; Midcalf, C. J . Chem. Soc. A 1970,
2279-2285. (b) Kilner, M. Midcalf, C. J . Chem. Soc., Dalton Trans.
1974, 1620-1626. (c) Andrews, M. A.; Van Burkirk; G. Knobler, C.
B.; Kaesz, H. D. J . Am. Chem. Soc. 1979, 101, 7245-7254. (d) Adams,
R. D.; Katahira, D. A.; Yang, L. W. J . Organomet. Chem. 1981, 219,
241-249. (e) Kohler, J . U.; Lewis, J .; Raithby, P. R.; Rennie, M. A.
Organometallics 1997, 16, 3851-3854.
(33) (a) Andreu, P. L.; Cabeza, J .A.; del Rio, I.; Riera, V.; Bois, C.
Organometallics 1996, 15, 3004-3010. (b) Cabeza, J . A.; del Rio, I.;
Grepioni, F.; Riera, V. Organometallics 2000, 19, 4643-4646.
(34) Keister J . B.; Shapley, J . R. Inorg. Chem. 1982, 21, 3304-3310.
(30) Sheldrick G. M. SHELXTL/ IRIS; Bruker Analytical X-Ray
Systems, Inc.: Madison, WI, 1990.

Dihydrogen Bond in an Imine Triosmium Complex
Organometallics, Vol. 21, No. 1, 2002 53
Ta ble 2. Selected Bon d Len gth s (Å) a n d An gles
(d eg) for Com p lex 1
Distances
Os(1)-N(1)
Os(1)-Os(3)
Os(1)-Os(2)
Os(1)-H(3)
Os(2)-Os(3)
Os(2)-H(2)
Os(2)-H(3)
N(1)-H(1)
2.150(4)
2.9077(4)
2.9958(3)
1.67(4)
2.8804(3)
1.56(5)
1.78(4)
0.958(19)
Angles
N(1)-Os(1)-Os(3)
N(1)-Os(1)-Os(2)
Os(3)-Os(1)-Os(2)
N(1)-Os(1)-H(3)
Os(3)-Os(1)-H(3)
Os(2)-Os(1)-H(3)
Os(3)-Os(2)-Os(1)
Os(3)-Os(2)-H(2)
Os(1)-Os(2)-H(2)
Os(3)-Os(2)-H(3)
Os(1)-Os(2)-H(3)
H(2)-Os(2)-H(3)
Os(2)-Os(3)-Os(1)
Os(1)-N(1)-H(1)
91.47(10)
85.29(10)
58.38(1)
82.0(16)
89.3(16)
31.0(15)
59.28(1)
76.6(18)
73.3(19)
88.2(14)
28.9(14)
77(2)
F igu r e 1. Drawing of the molecular structure of compound
1, showing the adopted labeling scheme, with displacement
ellipsoids drawn at the 30% probability level.
62.34(1)
111(3)
ing of the coalescence process of the imine protons of
the two isomers at δ 10.59 and 10.83, respectively
(rather than the corresponding hydride resonances),
proves to be particularly useful since the smaller inner
separation allows the observation of the fully collapsed
spectrum.
The H-H distance in the Os-H‚‚‚H-N linkage has
been first evaluated in solution by extracting the dipolar
contribution to the relaxation rate of the terminal
hydride arising from the imine proton. This procedure
consists of comparing the relaxation rates of the protio-
and deuterio-isotopomers.31,32 The H(µ-H)Os3(CO)10-
(2HNdCPh2) isotopomer was prepared by addition of
deuterated benzophenone imine to a CD2Cl2 solution of
causes some distortions in the Os3(CO)10 moiety, with
respect to the pseudo-C3-symmetry of Os3(CO)12. The
angles between the Os1-Os2-Os3 plane and the car-
bonyl groups C12 and C13 linked to Os2 are Os1-Os2-
Os3/Os2-C12-C13 ) 14.3°, Os1-Os2-Os3/Os2-C12
) 5.4°, and Os1-Os2-Os3/Os2-C13 ) 12.6°, respec-
tively. The corresponding angles at Os1 and Os3 are
much closer to zero (0.2-1.2°). Of the two phenyl groups
of the benzophenonimine ligand, one is almost parallel
to the Os1-Os2-Os3 plane (“equatorial”), while the
other is almost perpendicular (“axial”).
The most relevant structural information is indeed
the short H‚‚‚H distance in the Os-H- - -H-N linkage
[1.79(6) Å]. It could be argued that a dihydrogen
distance measured from X-ray data may be rather
inaccurate, but in our complex the H1 position is
dictated by the stereochemistry of the imine group and
the electron-rich H2 is clearly detectable from the
difference Fourier map. Some theoretical calculations
(see Theoretical Calculations section below) were per-
formed to confirm such a short H- - -H distance. The
angular distortions of the ligands on Os2 contribute to
the shortening of the distance between the interacting
hydrogen atoms.
(µ-H)2Os3(CO)10. The interatomic H‚‚‚H distance can
H-H
then be obtained from 1/(T1,D
)
according to the
following equations:
1
T1,D
1
1
H-H
H-H
)
)
-
× 1.0625 (1)
[
]
T1(protio) T1(deuterio)
4
3γH p2τc
1
T1,D
1
4
+
(2)
6
2
1 + 4τc2 ω0
2
2
[
]
10rH-H 1 + τc ω0
Particularly reliable values of the molecular correla-
tion time (τc) are obtained when the T1 of the hydride
resonance displays its minimum value. At the operating
frequency of 400 MHz (ω0), T1 minimum (0.47 s) is
obtained at ca. 208 K. At this temperature τc ) 0.62/ω0
(in rad s-1); thus τc ) 2.45 × 10-10 rad s-1. When this
value is introduced in eq 2, an rH,H distance of 2.00 (
0.05 Å is obtained.
Also the overall crystal packing (Figure 2) shows
interesting features as it is driven by stacking interac-
tions between “axial” phenyl groups (phenyl-phenyl
distance 3.7 Å) and by six C-H‚‚‚OC intermolecular
contacts37 (average and minimun H‚‚‚O distance of 2.83
and 2.66 Å, respectively). Such a distance suggests that
these contacts can be ascribed to vdW interactions, since
the sum of vdW radii of H and C is 2.87 Å.
X-r a y Cr ysta l Str u ctu r e of 1. The molecular struc-
ture of 1 is shown in Figure 1, and selected bond lengths
and angles are presented in Table 2.
E st im a t e of t h e DH B Bon d in g E n er gy b y IR
An a lysis. The stretching vibration of the N-H group
The syn disposition of the hydride H2 and of the imine
is confirmed. The observed Os2-H2 distance of 1.56(5)
Å is one the smallest among those reported from X-ray
data on related derivatives in the Cambridge Database.
More reliable Os-H bond distances, ranging from 1.5735
to 1.6736 Å, can be retrieved from neutron data. The
presence of an axial hydrogen instead of a bulkier ligand
(35) Maltby, P. A.; Sclaf, M.; Steinbeck; M.; Lough, A. J .; Morris, R.
H.; Klooster, W. T.; Koetzle; T. F.; Sristava, R. C. J . Am. Chem. Soc.
1996, 118, 5396-5407.
(36) (a) Hart, D. W.; Bau, R.; Koetzle, T. F. J . Am. Chem. Soc. 1977,
99, 7557-7564. (b) J ohnson, T. J .; Albinati, A.; Koetzle, T. F.; Ricci,
J .; Eisenstein, O.; Huffman, J . C.; Caulton, K. G. Inorg. Chem. 1994,
33, 4966-4976.
(37) Braga, D.; Grepioni, F.; Desiraju, G. T. Chem. Rev. 1998, 98,
1375. See in particular section V-D and references therein.

54 Organometallics, Vol. 21, No. 1, 2002
Aime et al.
F igu r e 2. Crystal packing of H(µ-H)Os3(CO)10(HNdCPh2)
(1).
F igu r e 3. Comparison of the IR spectra, in the NH
stretching region, of H(µ-H)Os3(CO)10(HNdCPh2) (1) and
of Os3(CO)11(HNdCPh2).
of the coordinated imine ligand may be used as an
indicator of the strength of the DHB present in H(µ-
H)Os3(CO)10(HNdCPh2). J ogansen38 proposed, for linear
hydrogen bonds in organic molecules, the following
empirical relation to obtain the value of the bonding
enthalpy:
Ta ble 3. Geom etr ica l F ea tu r es (d ista n ces in Å a n d
An gles in d eg) of th e H1‚‚‚H2 In ter a ction in 1a
solid state
solution
1 X-ray
structure
H-OPT
gas-phase
modelb
NMR data
2.00(5)
modelc
H1‚‚‚H2
1.79(6)
2.48(5)
1.56(5)
127(4)
134(4)
73(2)
1.89
2.68
1.67
137
119
84
1.93
2.69
1.67
130
113
85
-∆H ) 18∆νNH/(∆νNH + 720) kcal/mol
N‚‚‚H2
Os2-H2
N-H1‚‚‚H2
Os2-H2‚‚‚H1
Os1-Os2-H2
where ∆νNH is the shift in stretching frequency due to
the hydrogen bond. It was demonstrated that this
relation may be used also for transition metal hy-
drides.39
a
The structural results from X-ray and NMR experimental data
b
are compared with those from theoretical calculations. H-OPT:
To apply this procedure, it is necessary to know νNH
of a reference compound with no DHB. To this purpose,
Os3(CO)11(HNdCPh2) (3) has been prepared (see Ex-
perimental Section) from the reaction of the “lightly
stabilized” Os3(CO)11(NCCH3) with benzophenone imi-
ne. Its structure may be envisaged as derived from
Os3(CO)12 upon replacement of an axial carbonyl by the
imine ligand. The synthesis and the spectroscopic data
for this compound are reported in the Experimental
Section.
On comparing the IR spectra of 1 and 3 in the 3000-
3400 cm-1 region, it is evident that ν(NH) in 1 is shifted
to lower wavenumbers (red shift), and it is increased in
intensity in the presence of the DHB (Figure 3); this is
the typical behavior of a “normal” hydrogen bond.
Introducing the observed ∆νNH value of 80.9 cm-1 into
eq 3, ∆H is evaluated to correspond to ca. 2.25 kcal/
mol. In principle this approach is valid only for inter-
molecular interactions when the angle between the NH
and the MH is close to 180°. In the case of H(µ-H)Os3-
(CO)10(HNdCPh2) this interaction occurs in a five-
membered ring, and then it is likely that ring strain
will affect ∆H. Nevertheless the obtained value is
reasonable and in keeping with theoretical calculations
(see Discussion).
only H1, H2, and H3 positions optimized. c Gas-phase: full
geometry optimization, except Os-Os distances.
MO calculations on the isolated molecule of isomers 1
and 2. At first, only the positions of H1, H2, and H3 in
1 were optimized (H-OPT model in Table 3), while the
rest of the structure was fixed at the X-ray geometry.
Then all geometric parameters of isomers 1 and 2,
except for the Os-Os bond lengths, were optimized (gas-
phase model in Table 3). The “H-OPT” approach can be
considered as a good model for the study of the DHB in
the solid state, while the gas-phase model can be
considered as a suitable approximation of the molecules
of 1 and 2 in a relatively nonpolar solvent such as
CD2Cl2.
Calculations with the H-OPT model gave a H1‚‚‚H2
distance of 1.89 Å, which is within 2σ with respect to
the X-ray distance. Nevertheless, this calculated dis-
tance is larger than the X-ray one and seems more
reasonable for a hydrogen bond presenting a ∆νNH value
of 80.9 cm-1. Therefore, from now on, the calculated
H1‚‚‚H2 H-OPT distance of 1.89 Å will be considered
as the most reasonable estimate of this H‚‚‚H contact
in the solid state.
Calculations with the H-OPT model gave an
Os2-H2 distance of 1.67 Å (close to that from neutron
diffraction.35,36), which is larger (but within 2σ) with
respect to the X-ray distance. This lengthening alone
cannot explain the different H1‚‚‚H2 distance in the
X-ray data and in the H-OPT model. In fact, this
difference is mainly due to the different values (∆ > 5σ
in Table 3) of the Os1-Os2-H2 angle in the X-ray
structure and in the H-OPT model. As a result of this
Th eor etica l Ca lcu la tion s. The positions of the
hydrogen atoms involved in the DHB and that of the
bridging hydrogen (located with some uncertainty by the
X-ray analysis) were checked by performing ab initio
(38) Iogansen, A. V. Hydrogen Bond; Nauka: Moskow, 1981; p 112.
(39) Shubina, E. S.; Belkova, N. V.; Krylov, A. N.; Vorontsov, E. V.;
Epstein, L. M.; Gusev, D. G.; Niederman, M.; Berke, H. J . Am. Chem.
Soc. 1996, 118, 1105-1112.

Dihydrogen Bond in an Imine Triosmium Complex
Organometallics, Vol. 21, No. 1, 2002 55
F igu r e 4. Energy barrier for the rotation of the imine group around the Os1-N bond in the angular range -50° to 110°.
angular difference, the Os2-H2 bond is more bent
toward the imine group in the X-ray structure than in
the H-OPT model. A significant angular distortion of
the easily polarizable Os-H bond (see later) is not
surprising.
The difference (0.1 Å) between the H‚‚‚H distances in
the Os-H‚‚‚H-N linkage measured in the solid state
[1.89 Å] and in solution [2.00 Å] cannot be only due to
the occurrence of crystal packing effects or to an error
in the determination of the H1 and H2 positions, as
pointed out before. Also theoretical calculations with
geometrical optimization (GP model) suggest a smaller
H‚‚‚H distance (1.93 Å), with respect to the NMR data.
On the other hand the difference cannot be associated
with inaccuracies of the NMR procedure, as the reli-
ability of interatomic distances obtained by using dipo-
lar relaxation data in a number of structural determi-
nations has been proved by a number of studies.42 To
account for the occurrence of such a difference, we
should consider that the H‚‚‚H distance obtained from
the NMR measurements represents an average value
between exchanging structures; indeed a “static” struc-
ture could not be achieved even at the lowest attained
temperature (208 K). One may therefore envisage an
important role of the large amplitude motion of the
imine ligand (without a complete rotation) which causes
the H‚‚‚H distance to span a rather large range of
values. Indeed an energy scan around the Os1-N bond
(Figure 4), carried out using MOLDRAW,29 shows that
there is a rather wide low-energy angular range. Within
this range the H1‚‚‚H2 distance varies between 1.79 and
1.99 Å.
The geometry optimization of 1 and 2 was performed
with constraints only on the Os-Os distances (gas-
phase in Table 3) and permitted us to further investi-
gate the geometrical differences observed for 1 in
solution (NMR data) and in the solid state (X-ray data)
and to estimate the strength of the DHB. The structure
of 1, after geometry optimization of the X-ray structure,
presents bond distances and angles, involving the DHB,
closer to the standard values reported in the literature.
In particular the Os2-H2 distance of 1.67 Å is within
the range suggested by neutron data.35,36 Nevertheless
the substituents at Os2 remain tilted with respect to
the Os1-Os2-Os3 plane.
Single-point calculations at the B3-LYP27,28 level on
the gas-phase geometry of 1 and 2 suggested that the
DHB energy is weak (2.6 kcal/mol); they also permitted
the estimation the Mullikan charges on H1 (+0.21) and
H2 (-0.59), which suggest that an electrostatic compo-
nent is important in this DHB.
Discu ssion
H‚‚‚H Dista n ce in th e Dih yd r ogen In ter a ction .
In the solid state the H‚‚‚H distance (1.89 Å, as
estimated from the H-OPT model) is less than the sum
of the vdW radii and proves that the N-H‚‚‚H-Os
contact is a genuine intramolecular interaction. A search
in the CSD13 showed that this value is among the
shortest ones reported for this type of DHB in transition
metal hydrides and the shortest one involving Os-H.
In fact, many examples of compounds with an Os-H
can be found in the CSD,40 but no Os-H‚‚‚H-X interac-
tions in which the H‚‚‚H distance is shorter than 2.25
Å were found. It is worth noting that the even weaker
Os-H‚‚‚H-C contacts are far more common41 in the
CSD, probably because CH-containing ligands are more
commonly found in organometallic chemistry with re-
spect to NH- or OH-containing ligands.
Likely the conclusion that the NMR distance reflects
a dynamic state of the DHB interaction has to be
extended to all Os-H‚‚‚H-N distances previously re-
ported for related H(µ-H)Os3(CO)10(amine)10 and H(µ-
H)Os3(CO)10(imine)11,12 systems.
Str en gth of th e Dih yd r ogen In ter a ction . The
strength of this intramolecular DHB is about 2.5 kcal/
mol according to IR data and theoretical calculations,
a value typical of very weak hydrogen bonds. The
(40) Braga, D.; Grepioni, F.; Tedesco, E.; Biradha, K.; Desiraju, G.
R. Organometallics 1996, 15, 2692-2699.
(41) Seven structures (CSD13 codes: BAKWIT, BOKCAF, CIDW-
ER11, DACWUZ, DCTPHO, DEQBOQ, HELWOK) present OsH‚‚‚HC
contacts shorter than 2.0 Å, while no OsH‚‚‚HX (X ) O, N) with H‚‚‚H
shorter than 2.25 Å can be found in the CSD.
(42) (a) Kasaka, K. A.; Imoto, T.; Hatano, H. Chem. Phys. Lett. 1973,
21, 398-400. (b) Aime, S.; Botta, M.; Gobetto, R.; Osella, D. Inorg.
Chem. 1987, 26, 2551-2552. (c) Aime, S.; Cisero, M.; Gobetto, R.;
Osella, D.; Arce, A. J . Inorg. Chem. 1991, 30, 1614-1617.
Relaxation time (T1) NMR data in CD2Cl2 solution
indicate a longer DHB with respect to the solid state.

56 Organometallics, Vol. 21, No. 1, 2002
Aime et al.
F igu r e 5. Relevant calculated bond lengths (in Å) for some “normally” H-bonded adducts (7, 8) and for the related free
molecules (4, 5, 6).
calculated values of the interaction energy and of the
H‚‚‚H distance for 1 are in keeping with the trend of
the energy versus distance plot obtained from high-
level MO calculations on a series of main group DHB
complexes.8 Only a limited number of examples of
M-H‚‚‚H-X (M: any transition metal, X: O or N)
interactions showing H‚‚‚H distances shorter than the
sum of the vdW radii can be found in the CSD,13 and
this fact may support the weak nature of these inter-
and intramolecular X-H‚‚‚H-M interactions.
Nevertheless, these weak interactions are capable of
affecting the relative yield of 1 (with DHB) and 2 (with
no DHB) in solvents of different polarity, as reported
above and extensively described in references 11 and
12 for related compounds.
Na tu r e of th e Dih yd r ogen In ter a ction . Electro-
static interactions, similar to those observed in “normal”
X-H‚‚‚X (X ) O, N) hydrogen bonds (NHB), are
expected to play an important role in an interaction
between the positively charged iminic proton (H1) and
a negatively charged hydride (H2). The charge distribu-
tion analysis in 1 obtained from B3-LYP calculations
indicates Mullikan charges of opposite sign on H2
(-0.59) and on H1 (+0.21), confirming a relevant
electrostatic contribution to the DHB. These calculated
charges, although possibly affected by some error,
support the view that in 1 H1 is a typical electron-
depleted hydrogen apt to act as hydrogen bond donor,
while H2 is an electron-rich hydrogen, apt to accept an
H-bond. A charge rearrangement seems to be induced
by DHB, as suggested by the smaller negative charge
on H2 found in 2 (-0.34).
Experimental IR data confirm a red shift of ν(N-H)
in 1 with respect to Os3(CO)11(HNdCPh2), and this fact
suggests, within the applicability of the J ogansen
formula,38 that DHBs involving transition metal hy-
drides behave as “normal” hydrogen bonds. On the
contrary, DHB differs from normal H bonds because of
a larger contribution of the polarization component, as
suggested by different theoretical works,7 in which the
Kitaura-Morokuma43 energy decomposition technique
was employed. The large polarizability of the Os2-H2
bond was experimentally confirmed by the significant
red shift of the M-H stretching in the IR spectra, when
DHB is formed, reported by Shubina et al.39 Indeed the
calculated (gas-phase model, see Theoretical Calcula-
tions section for details) Os2-H2 distance is longer in
1 with respect to 2 (see Table 3).
To get more insight into the comparison between DHB
and conventional hydrogen bonds, calculations at the
HF/6-31G(d,p) level, with full geometry optimization,
were undertaken on adducts 7 and 8 formed between
benzophenoneimine (4) and acetone (5) or 2-propanol
(6). In Figure 5 the relevant distances for the donor and
acceptor moieties are reported as well as the corre-
sponding ones obtained for the free molecules. No
significant change is observed for the N-H bond length
in the free or H-bonded benzophenoneimine; an analo-
gous invariance of the N-H distance was found on
passing from 2 to 1. Conversely significant changes in
the carbon-oxygen bond lengths have been detected in
the acceptor molecule. In the case of acetone, the
formation of the adduct 7 leads to a shortening of the
CdO bond as the hydrogen-bonding interaction removes
charge from π*-orbitals. When the acceptor group is the
OH group of 2-propanol, the expected elongation of the
C-O bond in 8 occurs. This lengthening of +8 mÅ is
very similar to that calculated for Os2-H on going from
2 to 1 (+6 mÅ). We can then see that the medium
strength hydrogen bond in 8 has the same lengthening
effect on the acceptor moiety of the weak DHB in 1. It
can therefore be inferred that the contribution of the
polarizability of the metal hydrogen bond plays an
important role9 in the case of the interaction involving
this group. It is worth recalling that the polarization of
the M-H bond may be reversed, as it can act as H bond-
acceptor (this work) or -donor,40 depending upon the
surrounding environment.
In sigh ts in to th e Rea ctivity of 1. When a CD2Cl2
solution of 1 is allowed to react with DNdCPh2 at room
temperature for few minutes, there is the disappearance
of both hydride resonances. The full deuteration of the
terminal and bridging sites is further confirmed by the
2
observation of the hydride resonances in the D NMR
spectrum at 213 K. This observation implies the revers-
ible addition/dissociation of the imine ligand, coupled
to the distribution of deuterium at both hydride posi-
tions. One may envisage a multistep process to account
for the observed behavior. As depicted in Chart 2, the
intramolecular DHB may evolve to the reversible for-
(43) Kitaura, K.; Morokuma, K. Int. J . Quantum Chem. 1976, 10,
325-340.

Dihydrogen Bond in an Imine Triosmium Complex
Organometallics, Vol. 21, No. 1, 2002 57
Ch a r t 2
the weak C-H‚‚‚OC contacts, to the approaching of the
two hydrogens involved in the DHB.
• because of the prevailingly electrostatic nature of
the DHB, the value of the X-ray distance between the
centers of mass of the electron density of the two
hydrogen atoms is expected to be smaller than the
distance between the two nuclei.
Experimental IR data and theoretical calculations
agree in suggesting a DHB interaction energy of about
2.5 kcal/mol, which is in keeping with that obtained
from high-level MO calculations on a series of main
group DHB complexes.8 Despite its weakness, the DHB
is responsible for driving the thermal transformation
of 1 into the µ-η2-NdCPh2-containing derivative, likely
through an intermediate species containing a molecular
hydrogen.
As far as the nature of the studied DHB is concerned,
one may conclude the following.
(a) A prevailing electrostatic component is present,
similar to that observed in “normal” hydrogen bonds.
(b) The red shift of the ν(N-H) stretching and the
validity of the J ogansen formula38 also suggest a
“normal” behavior of this DHB.
(c) The relatively large polarization component,9
related to the high polarizability of the M-H bond,
seems to be the most typical feature of this “unconven-
tional” dihydrogen bond. This is not surprising because
the H of M-H cannot exploit other interaction mecha-
nisms, because no π-systems or lone pairs are available.
Finally from our study it can be concluded that, when
stronger “normal” hydrogen bonds are not present,
weaker “unconventional” interactions, such as DHB,
become important in driving the stereochemistry of the
resulting products.
mation of a “nonclassical” hydride intermediate in which
the rotation of the H-D molecule along its coordination
axis allows the H/D exchange on the amine ligand to
take place. Next, the deuteration of the hydride sites
occurs through the scrambling of terminal and bridging
hydrides as recalled above. Interestingly the intermedi-
ate containing the coordinated molecular hydrogen is
the key step for the thermally activated transformation
of 1 into HOs3(CO)10(µ-η2-NHdCPh2), as recently re-
ported by Cabeza et al.33b On this ground the formation
of the latter product is driven by the elimination of the
H2 molecule which results from the formal coupling of
H+ and H-.
Con clu sion s
The geometric and energetic features of the DHB in
1 have been characterized by means of several experi-
mental and theoretical techniques. Combined X-ray,
NMR, and theoretical analyses indicate that the H‚‚‚H
distance in 1 ranges from about 1.9 Å in the solid state
(1.79 Å in the crystal structure and 1.89 Å in the more
reliable H-OPT model from theoretical calculations) to
1.93 Å in the gas phase (theoretical calculation with
geometrical optimization) and to 2.00 Å in solution
(NMR data). This difference can be ascribed to the
following factors:
Ack n ow led gm en t. This work was carried out with
financial support from the E.U.-TMR Program (Project
“Metal Clusters in Catalysis and Organic Synthesis”
contract no. FMRX CT96 0091). Financial support from
the Italian MURST (COFIN 2000) and CNR is grate-
fully acknowledged. INSTM is acknowledged for provid-
ing a privileged access to the CINECA supercomputing
facility. M.M. thanks the “Fondazione G. Donegani” for
a grant. M. Causa` of the University of Piemonte
Orientale and P. Ugliengo of the University of Torino
are thanked for their suggestions on the theoretical
calculations.
• easy torsional rotation of the imine group around
the N-Os1 bond (see Figure 4) in solution;
• deformation vibrations of the three-osmium cluster
which are larger in solution;
Su p p or tin g In for m a tion Ava ila ble: This material is
available free of charge via the Internet at http://pubs.acs.org.
• contribution of the crystal packing forces, such as
OM010650R