Reactivity of Electron-Deficient Benzoheterocycle Triosmium Clusters. 5. The Chemistry of Os3(CO)9(μ3-η 2-(C(9)-N)-5,6-benzoquinolyl)(μ-H) and Its Phosphine Derivative

Article abstract of DOI:10.1021/om9902535

The reactivity of Os₃(CO)₉(μ₃-η²-C ₁₃H₈N)(μ-H) (1) is reported. The reaction of 1 with triphenyl phosphine gives Os₃(CO)₉(μ-η²-C₁₃H ₈N)(μ-H)(PPhOs) (2), in which the phosphine is bonded to the same osmium as the C(10) of the heterocyclic ring, isostructural with the quinoline analogue. Thermal or photochemical decarbonylation of 2 gives moderate yields of Os₃(CO)₈(μ₃-η³-C ₁₃H₈N)(μ-H)PPh₃ (3), which has a σ-π-vinyl bonding mode with the C(9)-C(10) double bond and not the expected μ₃-η² electron-deficient bonding mode. The reactivity of 1 with hydride followed by protonation yields the electron-precise Os₃(CO)₉(μ₃-η²-C ₁₃H₉N)-(μ-H)₂ (4), and labeling experiments using D^-/H⁺ indicate direct nucleophilic attack at C(9) by the deuteride followed by protonation at the metal core. The reaction of 1 with lithium isobutyrile nitrile followed by protonation gives Os₃(CO)₉(C₁₃H₉(4-(CH ₃)₂CCN)N)(μ-H) (5), a result of nucleophilic addition across the 3,4 double bond. Reaction with n-butyllithium followed by protonation, on the other hand, gives three products, Os₃(CO)₉(μ₃-η²-C ₁₃-H₈(9-C₄H₉)N)(μ-H)₃ (6), the result of addition at the 9-position, Os₃(CO)₉(μ₃-η²-C ₁₃H₇(6-C₄H₉)N)-(μ-H) (7), and Os₃(CO)₉(μ₃-η²-C ₁₃H₇(5-C₄H₉)N)(μ-H) (8), both the result of nucleophilic substitution for hydrogen. The solid-state structures of 3 and 8 are reported, and the mechanistic implications of these results for the synthetic methodology being developed for these electron-deficient clusters will be discussed.

Full text of DOI:10.1021/om9902535

Organometallics 1999, 18, 3519-3527
3519
Rea ctivity of Electr on -Deficien t Ben zoh eter ocycle
Tr iosm iu m Clu ster s. 5. Th e Ch em istr y of
Os3(CO)9(µ3-η2-(C(9)-N)-5,6-ben zoqu in olyl)(µ-H) a n d Its
P h osp h in e Der iva tive‡
Ryan Smith and Edward Rosenberg*
Department of Chemistry, The University of Montana, Missoula, Montana 59812
Kenneth I. Hardcastle, Venessa Vazquez, and J ay Roh
Department of Chemistry, California State University, Northridge, California 91330
Received April 9, 1999
The reactivity of Os3(CO)9(µ3-η2-C13H8N)(µ-H) (1) is reported. The reaction of 1 with
triphenyl phosphine gives Os3(CO)9(µ-η2-C13H8N)(µ-H)(PPh3) (2), in which the phosphine is
bonded to the same osmium as the C(10) of the heterocyclic ring, isostructural with the
quinoline analogue. Thermal or photochemical decarbonylation of 2 gives moderate yields
of Os3(CO)8(µ3-η3-C13H8N)(µ-H)PPh3 (3), which has a σ-π-vinyl bonding mode with the C(9)-
C(10) double bond and not the expected µ3-η2 electron-deficient bonding mode. The reactivity
of 1 with hydride followed by protonation yields the electron-precise Os3(CO)9(µ3-η2-C13H9N)-
(µ-H)2 (4), and labeling experiments using D-/H+ indicate direct nucleophilic attack at C(9)
by the deuteride followed by protonation at the metal core. The reaction of 1 with lithium
isobutyrile nitrile followed by protonation gives Os3(CO)9(C13H9(4-(CH3)2CCN)N)(µ-H) (5), a
result of nucleophilic addition across the 3,4 double bond. Reaction with n-butyllithium
followed by protonation, on the other hand, gives three products, Os3(CO)9(µ3-η2-C13H8(9-
C4H9)N)(µ-H)2 (6), the result of addition at the 9-position, Os3(CO)9(µ3-η2-C13H7(6-C4H9)N)-
(µ-H) (7), and Os3(CO)9(µ3-η2-C13H7(5-C4H9)N)(µ-H) (8), both the result of nucleophilic
substitution for hydrogen. The solid-state structures of 3 and 8 are reported, and the
mechanistic implications of these results for the synthetic methodology being developed for
these electron-deficient clusters will be discussed.
In tr od u ction
tion either at the metal core or on the carbocyclic ring,
depending on the nature of the nucleophile and on the
substituents on the heterocycle. Thus, in the particular
case of quinoline, amines and phosphines react at the
metal core to give the electron-precise adducts, Os3-
(CO)9(µ-η2-C9H6N)(µ-H)L, whose structures are depend-
ent on the nature of the ligand (eq 2).2 This reactivity
is essentially shut down when the strong electron-
donating amino group is substituted in the 5-position
only (i.e., the 3 and 6 amino derivatives still form
adducts).2 Hydride and a wide range of carbanions, on
the other hand, react regioselectively at the 5-position
to give nucleophilic addition products after acid quench
(eq 3).3 Substitution of the quinoline ring at the 6-posi-
tion is stereoselective whereby only the cis-diastereomer
of the addition product is obtained, while substitution
at the 5-position blocks nucleophilic addition and sec-
ondary attack at the 4-position is realized (eq 4).3 We
are now in the process of investigating the reactivity of
related electron-deficient benzoheterocycle complexes
We have been studying the reactivity of the 46e-
triosmium clusters, Os3(CO)9(µ3-η2-benzoheterocycle)(µ-
H), which result from the reaction of Os3(CO)10(CH3-
CN)2 with a wide range of benzoheterocycles containing
pyridinyl nitrogens (eq l, Figure 1). The interesting
(1) Rosenberg, E.; Arcia, E.; Kolwaite, D. S.; Hardcastle, K. I.;
Ciurash, J .; Duque, R.; Gobetto, R.; Milone, L.; Osella, D.; botta, M.;
Dastru’, W.; Viale, A.; Fiedler, J . Organometallics 1998, 17, 415.
(2) Rosenberg, E.; Bergman, B.; Bar Din, A.; Smith, R.; Gobetto, R.;
Milone, L.; Viale, A.; Dastru’, W. Polyhedron 1998, 17, 2975.
(3) Bergman, B.; Holmquist, R. H.; Smith, R.; Rosenberg, E.;
Hardcastle, K. I.; Visi, M.; Ciurash. J . Am. Chem. Soc. 1998, 120,
12818.
feature of these electron-deficient clusters is that on
reaction with nucleophiles these species undergo reac-
‡ For parts 1-4, see refs 1-4.
10.1021/om9902535 CCC: $18.00 © 1999 American Chemical Society
Publication on Web 07/22/1999

3520 Organometallics, Vol. 18, No. 17, 1999
Smith et al.
F igu r e 1. µ3-η2 electron-deficient benzoheterocycle triosmium complexes.
will have on the course of this substrate’s reactivity
toward nucleophiles.
Resu lts a n d Discu ssion
A. R ea ct ion of Os3(CO)9(µ3-η2-C13H 8N)(µ-H )(1)
w ith Tr ip h en ylp h osp h in e. We have reported the
synthesis and structure of 1 previously.4 The compound
is obtained in sufficient yield (50% based on Os3(CO)12)
to warrant a full investigation of its chemistry. The
reaction of 1 with triphenylphosphine proceeds quan-
titatively to yield a single product, Os3(CO)9(µ-η2-
C13H8N)(µ-H)PPh3 (2), whose structure can be assigned
to an adduct having the phosphine substituted on the
same osmium atom as C(10) of the heterocyclic ring (eq
5).1 This assignment is based on the 13C NMR of a
13CO-enriched sample of 2 whose spectrum in the
carbonyl region shows nine resonances, which can be
assigned on the basis of known trends for this type of
triosmium carbonyl cluster.1,5 Two phosphorus-coupled
resonances at 182.53 (2J 13P-13C ) 4.6 Hz) and 185.64-
toward nucleophiles (Figure 1).4 We report here the
reactions of the 5,6-benzoquinoline complex Os3(CO)9-
(µ3-η2-C13H8)N)(µ-H) (1), where the 5-position of the
quinoline ring system is blocked, raising the question
as to what effect the presence of the third aromatic ring
(4) Abedin, Md. J .; Bergman, B.; Holmquist, R.; Smith, R.; Rosen-
berg, E.; Ciruash, J .; Hardcastle, K.; Roe, J .; Vazquez, V.; Roe, C.;
Kabir, S.; Roy, B.; Alam, S.; Azam, K. A. Coord. Chem. Rev., in
press.

Benzoheterocycle Triosmium Clusters
Organometallics, Vol. 18, No. 17, 1999 3521
(2J 31P-13C ) 6.2 Hz) ppm are observed. The proton-
coupled spectrum reveals that the resonance at 182.53
ppm collapses to a doublet of doublets (2J 1H-13CO )
9.0 Hz), while the more downfield resonance shows an
unresolved hydride-carbonyl coupling. We can thus
assign these resonances to the radial and axial carbon-
yls on the same osmium atom as the phosphorus
respectively based on the observed couplings and their
relative chemical shifts. Two resonances at 178.10
(2J 1H-13C ) 13 Hz) and 177.93 (2J 1H-13CO ) 4.5 Hz)
ppm can be assigned to the radial carbonyl trans and
cis to the hydride which bridges the same edge as the
heterocycle. The placement of the phosphorus on the
same osmium atom as that bound to carbon is based on
the fact that in complexes of this type the carbonyl trans
to the hydride on the nitrogen-bound osmium always
shows a larger coupling constant than the one on the
carbon-bound osmium atom.5,6 This trend has been
observed in a wide range of compounds and for the
quinoline analogue of 2, for which a crystal structure
has been obtained.1 Three of the remaining five reso-
nances at 176.50, 178.02, and 184.43 ppm show no
resolvable hydride or phosphorus coupling but are
slightly broadened at room temperature and sharpen
as the temperature is lowered to 0 °C. We can assign
these resonances to three of the four carbonyls, two
radial and one axial on the Os(CO)4 group in 2, which
are just beginning to undergo tripodal motion on the
NMR scale at +22 °C. This motion is typical of the
Os(CO)4 groups in this type of cluster.5 The two remain-
ing resonances at 177.06 and 184.50 ppm are inter-
changeably assigned to the other axial carbonyl on the
Os(CO)4 group and the one nitrogen-bound osmium
atom. The structure of 2 is identical with its quinoline
analogue.1
F igu r e 2. Solid-state structure of Os3(CO)8(µ3-η2-C13H8N)-
(µ-H)PPh3 (3) showing the calculated position of the hy-
dride.
2.05(7) Å); the longest edge (Os(1)-Os(2) ) 2.917(8) Å)
is bridged by the nitrogen atom and the π-bonds to
Os(1) (Os(1)-C(10) ) 2.02(4) and Os(1)-C(9) )
2.0.50(7) Å). It seems reasonable to suggest that these
relatively disparate metal-metal bond lengths are
imposed by the requirements of the σ-π-bonding mode
and the rigidity of the 5,6-benzoquinoline ring. The
doubly bridged edge is intermediate in length (Os(1)-
Os(3) ) 2.844(8) Å). This overall trend in relative bond
lengths is similar to that observed in the more flexible
σ-π-vinyl 5,6-dihydroquinoline analogues of 3.1,3 The
metal-carbon σ and π bond lengths are also shorter on
average in 3 than in that series of complexes (average
bond lengths are 2.14 Å for the σ bond and 2.32 Å for
the π bonds). The average C-C bond lengths in the two
benzenoid rings are normal (1.39(5) Å), and the C(9)-
C(10) is in the range for an isolated metal-coordinated
double bond (1.35(9) Å); but the large errors in these
bond lengths make it difficult to draw any firm conclu-
sions from these data.
Thermolysis or photolysis of 2 leads to the formation
of an electron-precise, σ-π-vinyl complex involving the
C(9)-C(10) double bond, Os3(CO)8(µ3-η3-C13H8N)(µ-H)-
PPh3 (3, eq 6). The assignment of this structure is based
The chemical shift of the hydrogen on C(9) (5.75 ppm)
is considerably to lower field than the vinyl hydrogens
in related 5,6-dihydroquinoline σ-π-vinyl complexes
(∼4.0 ppm), indicating some retention of aromatic
character in the central ring.3 Broad band modulated
phosphorus decoupling of the proton spectrum collapses
the doublet resonance at 5.75 ppm (2J 31P-1H ) 5.6 Hz)
to a singlet, while the other hydrocarbon resonances
remain unchanged. This result indicates that the σ-π-
vinyl structure remains unchanged in solution and puts
an upper limit of 35.2 s-1 (<2πJ ) on the rate of σ-π-
interchange. This is in sharp contrast to the 5,6-
dihydroquinoline complex in which the σ-π interchange
is rapid on the NMR time scale at room temperature.
The quinoline analogue of 2 does not give a µ3-η3 σ-π-
vinyl complex on thermolysis or photolysis. Indeed, only
nonspecific decomposition and phosphine ligand dis-
sociation to give the unsubstituted nona- and decacar-
on a solid-state structural investigation, and the 1H
NMR data indicate that this structure persists in
solution.
The solid-state structure of 3 is shown in Figure 2,
crystal data are given in Table 1, and selected distances
and bond angles are given in Table 2. The structure
consists of an Os3 triangle with three significantly
different metal-metal bond lengths, the shortest being
the edge bridged by the nitrogen atom and σ-bound to
C(10) (Os(2)-Os(3) ) 2.768(8) and Os(3)-C(10) )
(5) Rosenberg, E.; Day, M.; Espitia, D.; Hardcastle, K. I.; Kabir, S.
E.; McPhillips, T.; Gobetto, R.; Milone, L.; Osella, D. Organometallics
1993, 12, 2390.
(6) Rosenberg, E.; Hardcastle, K. I.; Kabir, S. E.; Milone, L.; Gobetto,
R.; Botta, M.; Nishimura, N.; Yin, M. Organometallics 1995, 14, 3068.

3522 Organometallics, Vol. 18, No. 17, 1999
Ta ble 1. Cr ysta l Da ta a n d Str u ctu r e Refin em en t of 3 a n d 7
Smith et al.
3
7
empirical formula
fw
C
39H24NO8Os3P
C26H17NO9Os3
1058.01
1236.16
temperature
wavelength
crystal system
space group
unit cell dimensions
293(2) K
293(2) K
0.71073 Å
monoclinic
P21/c
0.71073 Å
monoclinic
P21/n
a ) 11.144(10) Å
a ) 9.9783(6) Å
b ) 16.148(12) Å, â ) 99.24(9)°
c ) 20.029(14) Å
3558(6) Å3, 4
b ) 12.8751(8) Å, â ) 92.588(1)°
c ) 21.8259(13) Å
2801.1(3) Å3, 4
volume, Z
density (calcd)
abs coeff
2.308 Mg/m3
2.509 Mg/m3
10.789 mm-1
13.626 mm-1
F(000)
2288
1920
crystal size
0.15 × 0.05 × 0.05 mm
0.15 × 0.10 × 0.05 mm
θ range for data collection
limiting indices
no. of reflns colld
no. of indep reflns
abs corr
refinement method
no. of data/restraints/params
goodness-of-fit on F2
final R indices [I > 2σ(I)]
R indices (all data)
extinction coeff
1.63-17.99°
1.84-28.28°
-9 e h e 9, 0 e k e 14, -17 e 1 e 17
7534
-13 e h e 13, -17 e k e 16, -29 e 1 e 28
29149
6537 (Rint ) 0.0612)
SADABS Bruker AXS
full-matrix least-squares on F2
6535/8/350
2444 (Rint ) 0.0000)
none
full-matrix least-squares on F2
2246/82/182
1.398
0.916
R1 ) 0.1198, wR2 ) 0.2473
R1 ) 0.1932, wR2 ) 0.4268
1.35(4)
R1 ) 0.0392, wR2 ) 0.0811
R1 ) 0.0758, wR2 ) 0.0919
0.00031(4)
largest diff peak and hole
1.919 and -1.065 e Å-3
1.702 and -1.429 e Å-3
Ta ble 2. Selected Dista n ces (Å) a n d An gles(d eg)
for 3a
Sch em e 1. Reson a n ce Str u ctu r es for 1
Distances
Os(1)-Os(2)
Os(1)-Os(3)
Os(2)-Os(3)
Os(1)-C(9)
Os(1)-C(10)
Os(1)-P(1)
Os(3)-C(10)
Os(2)-N
2.917(8)
2.844(8)
2.768(8)
2.50(4)
2.02(4)
2.37(3)
2.05(7)
2.01(2)
1.82(4)
C(9)-C(10)
C(9) C(14)
C(10)-C(12)
C(12)-N(1)
C(11)-C(12)
C(13)-C(14)
C(2)-C(3)
C-Ob
1.375(4)
1.375(4)
1.375(4)
1.424(4)
1.375(4)
1.375(4)
1.375(4)
1.20(5)
Os-COb
Angles
60.07(14) C(9)-C(10)-C(12)
62.43(13) C(11)-C(12)-N(1) 130.7(3)
Os(1)-Os(2)-Os(3)
Os(1)-Os(3)-Os(2)
Os(2)-Os(1)-Os(3)
Os(3)-Os(1)-C(9)
Os(3)-Os(1)-C(10)
Os(1)-Os(3)-C(10)
Os(3)-Os(2)-N(1)
Os(1)-Os(2)-N(1)
Os-C-Ob
110.0(4)
57.50(12) C(2)-N(1)-C(12)
105.1(4)
133.0(3)
110.6(3)
112.7(3)
72.l7(2)
52.1(2)
41.2(2)
82.5(2)
85.0(2)
167(6)
N(1)-C(2)-C(13)
C(2)-C(3)-C(4)
C(4)-C(11)-C(12)
C(11)-C(13)-C(14) 111.9(3)
C(15)-C(13)-C(14) 132.1(3)
hydride donors. In the case of 1, the 5-position is
blocked, and according to the contributions to the
electron distribution from the expected resonance struc-
tures, the 9-position of the 5,6-benzoquinoline ring
should be the secondary site of nucleophilic attack
(Scheme 1). Indeed, this is the case for 1, where initial
attack by hydride followed by protonation yields the
electron-precise cluster Os3(CO)9(µ3-η2-C13H9N)(µ-H)2 (4,
eq 7). That initial attack by H- is indeed at the
a
Numbers in parentheses are average standard deviations.
Average values.
b
bonyl precursors of the quinoline complex are obtained.
Thus, phosphine substitution apparently prevents for-
mation of the µ3-η2 electron-deficient bonding mode upon
thermolysis or photolysis of the electron-precise precur-
sors.1 We have previously shown that the reverse
reaction (i.e., carbonylation of the 46e- electron cluster)
results in a migration of the heterocyclic ligand to a
different edge of the metal triangle.1 From microscopic
reversibility, we can infer this process is also occurring
during decarbonylation. It may be that the presence of
the bulky phosphine on the osmium atom bound to C(10)
in 2 (C(8) in the quinoline analogue) prevents this ligand
migration by increasing the metal carbon bond strength
or by steric crowding in the transition state or in the
product.
B. Rea ction of 1 w ith Nu cleop h iles. The electron-
deficient µ3-η2 quinoline triosmium complexes undergo
regiospecific nucleophilic attack at the 5-position with
9-position is verified by the reaction of 1 with D-/H+,
where no deuterium is incorporated into the hydrides
and where the two doublets at 3.89 and 3.22 ppm (2J )

Benzoheterocycle Triosmium Clusters
Sch em e 2. Hyd r id e Atta ck on 1
Organometallics, Vol. 18, No. 17, 1999 3523
two at 3.18 ppm which is strongly coupled (1H COSY)
to the most downfield resonance at 8.89 ppm assigned
to the hydrogen on C(2) and to an aliphatic resonance
at 3.90 ppm assigned to the hydrogen on C(4). The
presence of the isobutyryl group is indicated by the
observation of two diastereotropic methyls at 1.38 and
1.24 ppm. The same coupling pattern between the most
downfield resonance and the aliphatic protons is ob-
served for the nucleophilic addition product obtained
from the reaction of Os3(CO)9(µ3-η2-C9H5(5-Cl)N)(µ-H)
with LiC(CH3)2CN, where the 5-position is blocked as
in 1 and nucleophilic addition takes place across the 3,4
double bond.3
The reaction of 1 with LinBu takes a completely
different course. Three products are obtained in ap-
proximately equal yield: Os3(CO)9(µ3-η2-C13H8(9-(CH2)3-
CH3)N)(µ-H)2 (6) and two green isomers having the
formula Os3(CO)9(µ3-η2-C13H7(5- or 6-(CH2)3CH3)N)(µ-
H) (7 and 8, eq 10). Compound 6 could be readily
15.0 Hz) assigned to the hydrogens on C(9) appear as
sharp singlets of relative intensity of 0.5 each. The
hydride region appears as a singlet at -14.20 ppm and
two singlets at -14.746 and -14.753 (rel int ) 1:0.5:
0.5), which can be understood in terms of the presence
of the two expected isotopomers formed which differ
with respect to the position of one bridging hydride
relative to the deuteride on C(9) (Scheme 2). Although
we cannot rigorously exclude a short-lived dihydride
anion intermediate with preferential deuterium transfer
to C(9), resulting from a thermodynamic isotope effect,
one would expect some incorporation of hydrogen into
the C(9) position, which is not observed. The intermedi-
ate anion (Scheme 2) shows a single hydride resonance
at -14.38 ppm. This pattern of reactivity with D-/H+
is analogous to that observed for the 5,6-dihydroquino-
line complexes (eq 8).
1
identified from its H NMR, which showed two hydride
doublets at -14.28 and -15.02 ppm (2J ) 1.2 Hz), a
doublet of doublets at 2.78 ppm, which is coupled to two
n
of the Bu methylene hydrogens at 1.45 and 1.06 ppm.
Seven aromatic resonances are observed, which are
1
divided into a group of three and a group of four by H
COSY experiments.
The reaction of 1 with Li(CH3)2CCN yields one major
product, Os3(CO)9(µ3-η2-C13H9(4-(CH3)2CCN)N)(µ-H) (5),
in good yield (eq 9, 58%). The structure of the product
Compounds 7 and 8 each showed a single hydride
resonance at -13.55 and -12.97 ppm, respectively. Both
show a singlet aromatic resonance at 8.11 and 8.67 ppm,
which we can assign to the hydrogen on C(9). Compound
7 shows an additional singlet at 7.90 ppm, a set of two
aromatic resonances, and a set of three aromatic
resonances by 1H COSY experiments. Compound 8
shows two sets of three aromatic resonances. These data
support a structure for compound 7 where the nBu group
1
is based on H NMR, infrared, and elemental analysis.
1
The key features of the H NMR data that define the
structure of 5 are (1) four strongly coupled (1H COSY)
aromatic resonances assigned to the hydrogens on C(5)-
C(8); (2) a singlet resonance at 8.08 ppm assigned to
the proton on C(9); (3) a multiplet of relative intensity

3524 Organometallics, Vol. 18, No. 17, 1999
Smith et al.
derivative based on the fact that both 7 and 8 have
resonances at 7.90 and 8.25 ppm, respectively, which
we can assign to C(8) in both compounds on the basis
of 1H COSY NMR. Additionally, C(5) would be the
expected site of nucleophilic attack if the electron
deficiency is primarily communicated to the positions
ortho and para to the carbon atom involved in electron-
deficient bonding to the metal core (Scheme 1). Although
we cannot be certain of the mechanism by which 7 and
8 are formed, it seems reasonable to propose that initial
nucleophilic attack at th 5- and 6-positions, respectively,
is followed by hydride abstraction during the proton
quench. We have previously noted spontaneous rearo-
matization after acid quench in the reactions of carban-
ions with these types of complexes.3
Con clu sion s
The reactions of 1 reported here, which are sum-
marized in Scheme 3, reveal several special features of
the 5,6-dibenzoquinoline ring system relative to its
quinoline analogue. First, whereas the attempted de-
carbonylation of Os3(CO)9(µ-η2-C9H6N)(µ-H)PPh3 yields
only Os3(CO)n(µm-η2C9H6N)(µ-H) (n ) 9 or 10, m ) 2 or
3) and nonspecific decomposition, the thermolysis of 2
yields 3, where the formation of two benzenoid aromatic
rings allows isolation of a stable σ-π-vinyl complex. In
the case of the quinoline complex, formation of such a
bonding mode would require disruption of this benzo-
heterocycle’s aromaticity.
F igu r e 3. Solid-state structure of (µ3-η2-C13H7(6-nBu)N)-
(µ-H) (7) showing the calculated position of the hydride.
Ta ble 3. Selected Dista n ces (Å) a n d Bon d An gles
(d eg) for 7a
Distances
Os(1)-Os(2)
Os(1)-Os(3)
Os(2)-Os(3)
Os(1)-C(10)
Os(3)-C(10)
Os(2)-N(1)
Os-COb
2.781(1)
2.776(1)
2.794(1)
2.25(1)
2.31(1)
2.18(1)
1.90(1)
C(9)-C(10)
C(10)-C(12)
C(9)-C(14)
C(12)-N(1)
C(2)-N(1)
C(6)-C(15)b
C-Ob
1.37(1)
1.45(1)
1.42(1)
1.37(1)
1.33(1)
1.52(1)
1.13(1)
Similarly, in the case of nucleophilic attack on 1,
attack at the 9-position by hydride or LinBu is probably
also driven, at least in part, by the formation of the two
benzenoid aromatic centers. This is our first observation
of nucleophilic attack at what would be C(7) in the case
of the quinoline analogues. Our initial thought was that
this position is sterically blocked, but the results
reported here seem to suggest that initial attack at the
5-position by small nucleophiles may be preferred for
electronic reasons in the case of the quinoline complexes.
There is, of course, a significant steric component to the
regiochemistry observed since the bulkier LiC(CH3)2-
CN attacks only at C(4). We have no explanation at
present as to why secondary attack by LinBu on 1 is at
C(5) and C(6) rather than C(4). It is tempting to suggest
that the electron deficiency felt at C(13) is transferred
to these positions, but this does not explain why LiC-
(CH3)2CN does not attack at C(6), which is not sterically
encumbered. Studies of the reactivity of related benzo-
heterocycles and calculations aimed at revealing the
details of electron distributions in these complexes are
currently under way in our laboratories.
Angles
Os(1)-Os(2)-Os(3)
Os(1)-Os(3)-Os(2)
Os(2)-Os(1)-Os(3)
Os(1)-Os(3)-C(10)
Os(3)-Os(1)-C(10)
Os(1)-Os(2)-N(1)
Os(3)-Os(2)-N(1)
Os-C-Ob
59.74(1) C(9)-C(10)-C(12)
59.91(1) C(12)-N(1)-C(2)
60.36(1) C(10)-C(9)-C(14)
116.(1)
120.(1)
125.(1)
51.5(2)
53.3(2)
85.2(2)
84.2(2)
177(1)
C(10)-C(12)-C(11) 122.(1)
N(1)-C(12)-C(11)
N(1)-C(2)-C(3)
C(12)-C(11)-C(4)
C(2)-C(3)-C(4)
120.(1)
121.(1)
119.(1)
120.(1)
a
Numbers in parentheses are average standard deviations.
Average values.
b
has substituted for hydrogen on either C(6) or C(7) and
for compound 8, substitution on C(5) or C(8).
We were able to obtain crystals for 7 which were
suitable for single-crystal X-ray diffraction study. The
solid-state structure of 7 is shown in Figure 3, crystal
data are given in Table 1, and selected distances and
bond angles are given in Table 3. The structure consists
of an essentially equilateral triangle of osmium atoms.
The 5,6-benzoquinoline ring is bound to the metal core
in a µ3-η2 fashion, and the osmium ligand bond lengths
(Os(1)-C(10) ) 2.25(1) Å, Os(3)-C(10) ) 2.31(1) Å,
Os(2)-N(1) ) 2.18(1) Å) are very similar to those found
in 1 (Os(1)-C(10) ) 2.25(1), Os(3)-C(10) ) 2.32(1) Å,
Exp er im en ta l Section
Ma ter ia ls. All reactions were performed in an atmosphere
of prepurified nitrogen, but were worked up in air. Acetonitrile
and methylene chloride were distilled from calcium hydride
before use, and tetrahydrofuran was distilled from benzophe-
none ketyl. Osmium carbonyl was purchased from Strem
Chemical and used as received. 5,6-Benzoquinoline, isobuty-
ronitrile, lithium triethylborohydride, trifluoroacetic acid,
triphenylphosphine, diisopropylamine, and LinBu were pur-
chased from Aldrich Chemical and used as received. Compound
1 and LiC(CH3)2CN were prepared according to literature
n
Os(2)-N(1) ) 2.14(1) Å).4 The Bu group is located on
C(6) and is disordered in the structure. The C(14)-C(6)
bond length is a normal C-C single bond at 1.52(1) Å,
and the C(5)-C(6) and C(6)-C(7) bonds are typical of
aromatic C-C bonds at 1.37(1) and 1.41(1) Å.
Although we cannot definitely assign the structure
of 8, we tentatively assign it to a C(5)-nBu-substituted

Benzoheterocycle Triosmium Clusters
Organometallics, Vol. 18, No. 17, 1999 3525
Sch em e 3. Su m m a r y of th e Rea ctivity of 1
procedures.3,4 Os3(CO)10(CH3CN)2 was prepared according to
literature procedures.7
to remove excess triphenylphosphine as a fore run, and then
concentrated to ∼3 mL. To the orange-yellow solution was
added ∼3 mL hexanes and the product recrystallized at -20
°C to yield 24 mg (100%) of Os3(CO)9(µ-η2-C13H8N)(µ-H)PPh3
(2).
Sp ectr a . 1H and 13C NMR were recorded on a Varian Unity
Plus 400 at 400 and 100 MHz, respectively. Chemical shifts
are reported downfield positive with respect to tetramethyl-
silane and, the magnitudes of coupling constants are reported
for only those that are relevant to the structural identification
of the compounds reported. Infrared spectra were recorded on
a Perkin-Elmer 1600 FT-IR spectrometer. Elemental analyses
were performed by Schwarzkopf Microanalytical Laboratories,
Woodside, NY.
Sp ectr oscop ic a n d An a lytica l Da ta for 2. Anal. Calcd
for C40H24NO9Os3P: C, 38.0; H, 1.91; N, 1.10. Found: C, 37.54,
H, 2.07; N, 1.04. IR (ν CO) in hexane: 2086(w), 2044(s),
2027(s), 2010(s), 1992(m), 1976(w), 1968(w), 1960(w), 1944-
1
(w) cm-1. H NMR in CDCl3: δ 9.44 (d, H(2)), 8.91 (d, H(4)),
8.31 (d, H(8)), 7.92 (s, H(9)), 7.41 (dd, H(6)), 7.36 (dd, H(7)),
7.10-7.20 (m, 15 H), 7.16 (dd, H(3)), 7.10 (d, H(5)), -11.63 (d,
J 31P-1H)15.6 Hz, hydride). 13C NMR (CDCl3): δ 185.64
(2J 31P-13C ) 6.2 Hz), 184.50, 184.43, 182.53 (2J 13P-13C ) 4.6
Hz), 178.10 (2J 1H-13C ) 13 Hz), 178.02, 177.93 (2J 1H-13CO
) 4.5 Hz), 177.06 176.50. 31P NMR in CDCl3: δ 12.37 relative
to external H3PO4 in CDCl3.
Syn th esis of Os3(CO)9(µ-η2-C13H8N)(µ-H)P P h 3 (2). Com-
pound 1 (21 mg, 0.02 mmol) was dissolved in 15 mL of CH2Cl2
under nitrogen, and triphenylphosphine (10 mg, 0.04 mmol)
in 2 mL of CH2Cl2 was injected into the solution by syringe.
After 5 min the solution turned from dark green to orange-
yellow. The solution was run through a short silica gel column,
Syn th esis of Os3(CO)8(µ3-η3-C13H8N)(µ-H)P P h 3 (3). Com-
pound 2 (97 mg 0.07 mmol) was dissolved in 35 mL of CH2Cl2
in a quartz reaction vessel, and the yellow-orange solution was
(7) Lewis, J .; Dyson, P. J .; Alexander, B. J .; J ohnson, B. F. G.;
Martin, C. M.; Nairn, J . G. M.; Parsini, E. J . Chem. Soc., Dalton Trans.
1993, 981.

3526 Organometallics, Vol. 18, No. 17, 1999
Smith et al.
photolyzed under a slow bubble of N2 in a Rayonet photo-
chemical reactor equipped with 3000 Å lamps. After 1 h and
15 min, the reaction solution was evaporated to dryness, taken
up in a minimum amount of methylene chloride, and purified
by thin-layer chromatography using 30:70 CH2Cl2/hexane as
eluent. Two major bands were eluted in addition to minor
amounts of 1 and its decacarbonyl precursor. The faster
moving yellow band contained unreacted 2 (51 mg), and the
slower moving brown band contained 3 (29 mg, 63% based on
consumed 2). Longer photolysis times led to lower yields owing
to the photosensitivity of 3.
23 mg of 6 (23%), and two green bands, 23 mg of 7 (23%) and
22 mg of 8 (22%), were isolated and identified.
Sp ectr oscop ic a n d An a lytica l Da ta for 6. Anal. Calcd
for C26H19NO0Os3: C, 29.46; H, 1.81; N, 1.32. Found: C, 29.66;
H, 1.98; N, 1.52. Ir (ν CO) in hexane: 2096(m), 2072(s), 2049-
(s), 2010(s), 1998(s), 1976(m), 1969(w) cm-1 1H NMR in
.
CDCl3: δ 8.17 (d, H(2)), 7.86 (d, H(4)), 7.50 (d, H(8)), 7.27 (m,
2H, H(6) and H(7)), 7.03 (d, H(5)), 6.56 (dd, H(3)), 2.73 (dd,
H(9)), 1.48 (m, 1H), 1.20-0.72 (m, 5H), 0.67 (t, CH3), -14.30
(d, 2J 1H-1H ) 1.2 Hz, hydride), -15.04 (d, 2J 1H-1H ) 1.2 Hz,
hydride).
Sp ectr oscop ic a n d An a lytica l Da ta for 7. Anal. Calcd
for C26H17NO9Os3: C, 29.52; H, 1.62; N, 1.32. Found: C, 29.78;
H, 2.00; N, 1.24. IR (ν CO) in hexane: 2111(w), 2076(m),
Sp ectr oscop ic a n d An a lytica l Da ta for 3. Anal. Calcd
for C39H24O8Os3NP: C, 37.9; H, 1.95; N, 1.13. Found: C, 38.35;
H, 2.15, N, 1.25. IR (ν CO) in hexane: 2078(m), 2042(s),
2026(s), 2010(s), 1994(m), 1986(w), 1968(w), 1960(w) cm-1. 1H
NMR in CDCl3: δ 8.95 (d, H(2)), 8.53 (d, H(4)), 8.14 (d, H(8)),
7.58 (dd, H(7)), 7.29, 7.42 (m, 15 H), 7.18 (dd, H(7)), 7.05 (dd,
2049(s), 2025(s), 1997(m), 1976(w) cm-1 1H NMR in CDCl3:
.
δ 9.06 (d, H(2)), 8.71(d, H(4)), 8.11 (s, H(5)), 7.91 (d, H(8)), 7.91
(s, H(9)), 7.56 (d, H(7)), 7.17 (dd, H(3), 2.81 (dd, 2H), 1.70 (m,
2H), 1.41 (m, 2H), 0.94 (t, CH3), -13.55 (s, hydride).
3
H(3)), 6.61 (d, H(5)), 5.75 (d, J 31P-1H)5.6 H, H(9)), -13.69
2
(d, J 31P-1H)8.8 Hz, hydride).
Sp ectr oscop ic a n d An a lytica l Da ta for 8. Anal. Calcd
for C26H17NO9Os3: C, 29.52; H, 1.62; N, 1.32. Found: C, 29.77;
H, 1.76; N, 1.32. IR (ν CO) in hexane: 2096(w), 2077(m),
Th e Syn th esis of Os3(CO)9(µ3-η2-C13H9N)(µ-H)2(4). Com-
pound 1 (51.5 mg, 0.051 mmol) was dissolved in 15 mL of CH2-
Cl2 under N2, and 50 µL of 1.0 M solution of Et3BXLi (X ) H
or D) was added dropwise to the stirred solution by syringe.
The mixture was allowed to stir for 20 min, and CF3SO3H (4.9
µL, 0.06 mmol) was added. The yellow-orange solution was
stirred for 5 min and then rotary evaporated, and the residue
was taken up in minimum CH2Cl2 and separated by prepara-
tive TLC using 35:65 CH2Cl2/hexanes. The fastest moving
yellow band yielded 28 mg (82% based on consumed 1), and
the slower moving green band contained 17 mg of 1.
2050(s), 2026(s), 1997(m), 1976(w) cm-1 1H NMR in CDCl3:
.
δ 9.12 (d, H(2)), 8.75 (d, H(4)), 8.68 (s, H(5)), 8.25 (d, H(8)),
7.93 (dd, H(7)), 7.57 (d, H(6)), 7.21 (dd, H(3)), 3.07 (t, 2H), 1.70
(m, 2H), 1.49 (m, 2H), 0.97 (t, CH3), -12.97 (s, hydride).
X-r a y Str u ctu r e Deter m in a tion of 3 a n d 7. Crystals of
each compound for X-ray examination were obtained from
saturated solutions in hexane/dichloromethane solvent sys-
tems at -20 °C. Suitable crystals of each were mounted on
glass fibers, placed in a goniometer head on the diffractometer,
and centered optically. For 3, unit cell parameters and an
orientation matrix for data collection were obtained by using
the centering program in the CAD4 system; crystals of 7 were
all too small for the CAD4, so data were collected on the
Bruker SMART CCD system at the Center for Molecular
Structure (CMolS) at California State University Fullerton.
Details of the crystal data for each compound are given in
Table 1. For 3, the actual scan range was calculated by scan
width - scan range + 0.35 tan θ, and backgrounds were
measured by using the moving-crystal moving-counter tech-
nique at the beginning and end of each scan. Two representa-
tive reflections were monitored every 2 h as a check on
instrument and crystal stability. For 7, frames were collected
using the SMART8 software, and the data were processed with
the program SAINT9 at CMolS. Lorentz, polarization, and
decay corrections were applied to each set of raw crystal data,
and the data for 7 were also corrected for absorption using
SADABS.10 The weighting scheme used during refinement for
7 was 1/σ2, based on counting statistics; for 3, an adjusted
weighting scheme in SHELXL was employed. Each of the
structures was solved by the Patterson method using SHELX-
TL/PC,11 which revealed the positions of the osmium atoms.
Remaining non-hydrogen atoms were found by successive
refinement and ∆F syntheses. The hydrides bridging the
Os(1) and Os(3) atoms in each molecule were positioned using
HYDEX,12 and other hydrogen atoms were placed in their
expected chemical positions using the HFIX command in
SHELXL/PC and were included as riding atoms. All non-
hydrogen atoms were refined anisotropically in 7, but in 3,
the crystal was very small and of poor diffraction quality, so
the carbon, oxygen, and nitrogen atoms were given separate,
group isotropic temperature factors, which were refined, and
in addition, general restraints to carbonyl and ring atom
distances were added in order to refine the structure. Data
processing for 3 was carried out on PC’s using SCAD4.13
Sp ectr oscop ic a n d An a lytica l Da ta for 4. Anal. Calcd
for C22H11O9Os3N: C, 26.32; H, 1.10; N, 1.40. Found: C, 25.86;
H, 1.04; N, 1.37. Ir (ν CO) in hexane: 2096(m), 2073(s), 2045-
(s), 2016(s), 2002(m), 1992(m), 1978(w), 1970(w) cm-1. 1H NMR
in CDCl3: δ 8.21 (d, H(2)), 7.91 (d, H(4)), 7.48 (d, H(8)), 7.29
(m, 2H, H(6) and H (7)), 7.07 (d, H(5)), 6,63 (dd, H(3)), 3.90 (d,
2
J 1H-1H )14.4 Hz, H(9)), 3.22 (d, J 1H-1H ) 14.4 Hz,
H1C(9)), -14.20 (s, hydride), -14.75 (s, hydride).
Th e Syn th esis of Os3(CO)9(µ3-η2-C13H9(4-C(CH3)2CN)N)-
(µ-H) (5). Compound 1 (95 mg, 0.09 mmol) was dissolved in
15 mL of THF and cooled to -78 °C. To this solution was added
1.7 mL (0.25 mmol) of a freshly prepared solution of LiC(CH3)2-
CN in THF. The resulting orange-brown solution was stirred
for 15 min, warmed to room temperature, and then stirred
for an additional 15 min. The solution was again cooled to -78
°C, and 15 µL of trifluoroacetic acid was added dropwise. The
solution was then warmed to room temperature, rotary
evaporated to dryness, taken up in CH2Cl2, filtered, and then
purified by preparative TLC using 40:60 CH2Cl2/hexane as
eluent. One major band was isolated in addition to minor
amounts of 1 and two other unidentified minor yellow bands,
which yielded 59 mg (58%) of 5.
Sp ectr oscop ic a n d An a lytica l Da ta for 5. Anal. Calcd
for C26H16N2O9Os: C, 29.15; H, 1.51; N, 2.61. Found: C, 28.91;
H, 2.21; N, 2.38. IR (ν CO) in hexane: 2077(m), 2047(s), 2023-
(s), 1992(br), 1975(w), 1060(w), 1944(w). 1H NMR in CDCl3:
8.89 (m, H(2)), 8.08 (s, H(9)), 8.04 (d, H(5)), 7.90 (dd, H(7)),
7.84 (d, H(8)), 7.59 (dd, H(6)), 3.91 (dd, H(4)), 3.18 (M, 2H,
H(3), H(3)1), 1.38 (s, CH3), 1.24 (s, CH3), -14.20 (s, hydride).
Syn th esis of Os3(CO)9(µ3-η2-C13H8(9-n -Bu )N)(µ-H)2(6)
a n d Os3(CO)9(µ3-η2-C13H7(5- or 6-n -Bu )N)(µ-H) (7 a n d 8).
Compound 1 (105 mg, 0.10 mmol) was dissolved in 10 mL of
THF and cooled to -78 °C. LinBu (220 µL of 1.45 M in hexane,
0.32 mmol) was slowly added to the solution with stirring.
After stirring for 1 h at -78 °C and 1 h at 0 °C, 25 µL of
trifluoroacetic acid (0.32 mL) was slowly added with stirring,
and then the reaction mixture was rotary evaporated, taken
up in CH2Cl2, filtered, and separated by preparative TLC with
30:70 CH2Cl2/hexane to yield three product bands in addition
to a small amount of unreacted 1. A fast moving yellow band,
(8) SMART V4.210; Bruker AXS: Madison, WI 53719, 1996.
(9) SAINT V4.05; Bruker AXS: Madison, WI 53719, 1996.
(10) Sheldrick, G. SADABS; University of Goettingen, Germany,
1996.
(11) SHELXTL/ PC, V5.03; Bruker AXS: Madison, WI 53719, 1997.
(12) Orpen, A. G. J . J . Chem. Soc., Dalton Trans. 1980, 2509.

Benzoheterocycle Triosmium Clusters
Organometallics, Vol. 18, No. 17, 1999 3527
Neutral atom scattering factors and values of ∆f ′ and ∆f′′ were
from Volume C of the International Tables for Crystal-
lography.14 For both crystals, structure solution, refinement,
and preparation of figures and tables for publication were
carried out on PC’s by using SHELXTL-PC.10
Ack n ow led gm en t. We gratefully acknowledge the
support of the National Science Foundation (CHE96
25367) for support of this research.
Su p p or tin g In for m a tion Ava ila ble: Complete distances
and bond angles, atomic coordinates, anisotropic thermal
parameters, and hydrogen coordinates for 3 and 7, Tables
4-11. This material is available free of charge via the Internet
at http://pubs.acs.org.
(13) SCAD4; Bruker AXS: Madison, WI 35719.
(14) Internatinal Tables for X-ray Crystallography, Vol. C; Kluwer
Academic Publishers: Dordrecht, 1992; Tables 6.1.1.4, pp 500-502,
and 4.2.6.8, pp 219-222.
OM9902535