Complex kinetics of simple substitution reactions of Os3(CO)9(μ-C4Ph4) with smaller P-donor nucleophiles

Article abstract of DOI:10.1021/om9906464

The osmacyclopentadiene ring in the triangular Os₃ cluster Os₃(CO)₉-(μ-C₄Ph₄) bridges two of the Os atoms and is found to activate associative attack at the third Os atom (which is in an Os(CO)₄ moiety) by a factor of ～10⁹ compared with reactions of the parent cluster Os₃-(CO)₁₂. The overall reactions in heptane of 17 P-donor nucleophiles with Tolman cone angles θ ≤ 143° lead to substitution at the Os(CO)₄ center in three rapid but kinetically observable stages: (i) reversible attack by a nucleophile to form an adduct in which an Os-Os bond in the cluster has been broken, (ii) loss of CO and formation of a new closed Os₃ cluster, and, usually, (iii) isomerization of the initially formed cluster to form the final substituted product. The dependence of the rates on the σ-basicity and size of the nucleophiles shows that the standard reactivity toward adduct formation is very high, only high nuclearity clusters that contain encapsulated C atoms being known to react faster. Nucleophiles with θ ≤ 120 ± 4° react at rates that are independent of their size and that increase substantially with their σ-basicity. When the cone angles are larger, steric retardation is observed. Equilibrium constants for adduct formation by 14 of the nucleophiles were obtained from the rates of adduct formation and loss of L from the Os₃(CO)₉L(μ-C₄Ph₄) adducts and by two other methods. Loss of CO or L from the adducts becomes slower the greater the net donicity of the ligands but is essentially independent of ligand size. Rates of isomerization of the initial product are not very precise and are not appreciably sensitive to the nature of the ligands. Activation parameters were obtained for reactions involving five of the nucleophiles, and these, together with the magnitudes of the electronic and steric effects, can lead to estimates of the contributions of bond making and bond breaking in the transition states.

Full text of DOI:10.1021/om9906464

5518
Organometallics 1999, 18, 5518-5530
Com p lex Kin etics of “Sim p le” Su bstitu tion Rea ction s of
Os3(CO)9(µ-C4P h 4) w ith Sm a ller P -Don or Nu cleop h iles
Anthony J . Poe¨* and Consuelo Moreno#
Lash Miller Chemical Laboratories, University of Toronto, 80 St. George Street,
Toronto, Ontario, Canada M5S 3H6
Received August 10, 1999
The osmacyclopentadiene ring in the triangular Os3 cluster Os3(CO)9(µ-C4Ph4) bridges two
of the Os atoms and is found to activate associative attack at the third Os atom (which is in
an Os(CO)4 moiety) by a factor of ∼109 compared with reactions of the parent cluster Os3-
(CO)12. The overall reactions in heptane of 17 P-donor nucleophiles with Tolman cone angles
θ e 143° lead to substitution at the Os(CO)4 center in three rapid but kinetically observable
stages: (i) reversible attack by a nucleophile to form an adduct in which an Os-Os bond in
the cluster has been broken, (ii) loss of CO and formation of a new closed Os3 cluster, and,
usually, (iii) isomerization of the initially formed cluster to form the final substituted product.
The dependence of the rates on the σ-basicity and size of the nucleophiles shows that the
standard reactivity toward adduct formation is very high, only high nuclearity clusters that
contain encapsulated C atoms being known to react faster. Nucleophiles with θ e 120 ( 4°
react at rates that are independent of their size and that increase substantially with their
σ-basicity. When the cone angles are larger, steric retardation is observed. Equilibrium
constants for adduct formation by 14 of the nucleophiles were obtained from the rates of
adduct formation and loss of L from the Os3(CO)9L(µ-C4Ph4) adducts and by two other
methods. Loss of CO or L from the adducts becomes slower the greater the net donicity of
the ligands but is essentially independent of ligand size. Rates of isomerization of the initial
product are not very precise and are not appreciably sensitive to the nature of the ligands.
Activation parameters were obtained for reactions involving five of the nucleophiles, and
these, together with the magnitudes of the electronic and steric effects, can lead to estimates
of the contributions of bond making and bond breaking in the transition states.
In tr od u ction
schematically as I, which contains an osmacyclopenta-
dienyl ring. Loss of CO and uptake of a third C2Ph2
The interaction of alkyne substrates with transition
metal carbonyl clusters has attracted a great deal of
interest over the past two decades.1 One of the major
reasons has been the desire to use cluster-bound sub-
strates as model systems to gain an understanding of
reactions between unsaturated organic absorbates on
catalytic metal surfaces.2 Alkyne substrates show a wide
variety of bonding modes toward dinuclear complexes
or small metal carbonyl clusters,2,3 and reactions of C2-
Ph2 with Os3(CO)12 lead to several of these.3-5 Thus,
Os3(CO)11(C2Ph2) is formed initially,5 followed by
Os3(CO)10(µ-C2Ph2)4a,5 and then Os3(CO)9(η2-C2Ph2)(µ-
C2Ph2)5 in which the two C2Ph2 ligands rapidly couple
to form Os3(CO)9(µ-C4Ph4).4b The structure is shown
#
Present address: Departamento de Qu´ımica Inorga´nica, Univer-
molecule lead to Os3(CO)8(η2-C2Ph2)(µ-C4Ph4) and then
C6Ph6.4c,5 On the other hand, reaction with CO leads to
facile fragmentation of cluster I5 to form Os2(CO)6(µ-
sidad Auto´noma de Madrid, Cantoblanco, 28049-Madrid, Spain.
(1) Sappa, E.; Tiripicchio, A.; Braunstein, P. Chem. Rev. 1983, 83,
203-239. Raithby, P. R.; Rosales, M. J . Adv. Inorg. Chem. Radiochem.
1985, 29, 169-247. Gallop, M. A.; J ohnson, B. F. G.; Khattar, R.; Lewis,
J .; Raithby, P. R. J . Organomet. Chem. 1990, 386, 121-137. Compre-
hensive Organometallic Chemistry II; Abel, E. W., Stone, F. G. A.,
Wilkinson, G., Eds.; Pergamon: New York, 1995; Vol. 5, p 772.
(2) Muetterties, E. L. J . Organomet. Chem. 1980, 200, 177-190;
Chem. Soc. Rev. 1982, 11, 283-320; Pure Appl. Chem. 1982, 54, 83-
96. Whyman, R. In Transition Metal Clusters; J ohnson, B. F. G., Ed.;
Wiley: New York, 1980; Chapter 8. Bailey, D. C.; Langer, S. H. Chem.
Rev. 1981, 81, 109-148. Somorjai, G. A. Chem. Soc. Rev. 1984, 113,
321-349.
(4) (a) Tachikawa, M. T.; Shapley, J . R.; Pierpont, C. J . J . Am. Chem.
Soc. 1975, 97, 7172-7174. (b) Ferraris, G.; Gervasio, G. J . Chem. Soc.,
Dalton Trans. 1974, 1813-1817. (c) Vaglio, G. A.; Gambino, O.; Ferrari,
R. P.; Cetini, G. Inorg. Chim. Acta 1973, 7, 193-194. (d) Gambino, O.;
Vaglio, G. A.; Ferrari, R. P.; Cetini, G. J . Organomet. Chem. 1971, 30,
381-386. (e) Dodge, R. P.; Mills, O. S.; Shoemaker, V. Proc. Chem.
Soc. 1963, 380-381. (f) Rushman, P.; van Buuren, G. N.; Shiralian,
M.; Pomeroy, R. K. Organometallics 1983, 2, 693-694.
(3) Lewis, J .; J ohnson, B. F. G. Gazz. Chim. Ital. 1979, 109, 271-
289.
(5) Poe¨, A. J .; Sampson, C. N.; Smith, R. T. J . Am. Chem. Soc. 1986,
108, 5459-5464.
10.1021/om9906464 CCC: $18.00 © 1999 American Chemical Society
Publication on Web 11/25/1999

Substitution Reactions of Os3(CO)9(µ-C4Ph4)
Organometallics, Vol. 18, No. 26, 1999 5519
C4Ph4)4d,e and Os(CO)5.4f Thus, reactions of Os3(CO)12
with C2Ph2 not only lead to products with a wide variety
of types of cluster-alkyne bonding but also illustrate
the facility with which C-C coupling can occur at metal
“nanosurfaces”. The cluster Os3(CO)9(µ-C4Ph4) is a
crucial exemplary intermediate in this series of reac-
tions in that C-C coupling has been involved in its
formation and further C-C coupling occurs after sub-
stitution of another C2Ph2 molecule. Its substitution
reactions are therefore of interest mechanistically, and
also kinetically in being capable of showing the quan-
titative effect of the osmacyclopentadiene moiety on the
reactivity at one of the Os atoms.
tions were carried out with a Hewlett-Packard 8452 diode
array spectrophotometer equipped with a cell holder thermo-
stated by a water bath ((0.1 °C). Faster reactions were
monitored by stopped-flow measurements with a Hi-Tech SF-
51 instrument that was equipped with an SU-40 spectropho-
tometer interfaced with a Hewlett-Packard Series 300 com-
puter. Solutions were prepared as described elsewhere.7
Concentrations of complex were ∼10-4 M, and nucleophiles
were always in pseudo-first-order excess. Rate constants for
slower reactions that showed simple one-stage first-order
dependence on [complex] were analyzed by means of a modi-
fied9 KORE program,10 and slower reactions that occurred via
two successive first-order steps were analyzed by use of an
Enzfit program. Stopped-flow data were processed by using a
Hi-Tech HS-1 Technical Datapro software suite, which pro-
vides a package for complete data acquisition, presentation,
and analysis. A single or double exponential analysis with a
fluctuating end point was successfully used to obtain the
pseudo-first-order rate constants.
The main reactions known to be undergone by this
cluster comprised, until recently,6 slow CO dissociation
to form Os3(CO)8(µ-C4Ph4)4c and associative fragmenta-
tion by CO or PPh3 to form mixtures of binuclear and
mononuclear products.5 For substitution to be ac-
complished via the CO dissociative path the Os3(CO)8(µ-
C4Ph4) had to be formed first and the substituting ligand
added afterward so as to avoid the much more rapid
direct associative fragmentation of the Os3(CO)9(µ-C4-
Ph4) by PPh3 or similar ligands.5,6 However, it has been
reported6 that rapid associative substitution by P-donor
ligands can occur directly into Os3(CO)9(µ-C4Ph4) pro-
vided that the Tolman cone angles, θ, of the P-donors
are e143°.6 The products of reactions with P(OCH2)3-
CEt (etpb) and P(OPh)3 were shown crystallographi-
cally6 to result from substitution at the Os(CO)4 moiety,
and the other 16 Os3(CO)8L(µ-C4Ph4) clusters isolated
can be concluded to have similar structures.6 Qualitative
observation of the course of these reactions suggested
that rapid formation of Os3(CO)9(µ-C4Ph4)-L adducts
occurred before CO loss to form the substituted prod-
ucts. Similarly rapid formation of adducts with M5C-
(CO)15 (M ) Ru, Fe) has been observed,7,8 and we report
here a detailed kinetics study of the reactions of 17
P-donor nucleophiles with Os3(CO)9(µ-C4Ph4), which
turn out frequently to occur in three rather than two
observable stages. Rate constants for the initial forma-
tion of the adducts show a clear and quantitatively
resolvable dependence on the σ-basicity and size of the
nucelophiles, and the rates of other reactions are also
found to vary systematically with the nature of the
ligands involved. The osmacyclopentadienyl moiety is
found to activate associative substitution reactions with
P-n-Bu3 by a dramatic factor of 109 compared with
those of the parent cluster Os3(CO)12.
Resu lts
Th e Ra te Con sta n ts. The cluster Os3(CO)9(µ-C4Ph4)
shows a fairly sharp absorbance maximum at 420 nm
(ꢀ ≈ 8 × 104 M-1 cm-1) and a broad maximum at ca.
570 nm (ꢀ ≈ 3 × 104 M-1 cm-1). At 420 nm the
absorbance always decreased rapidly during the first
stage of reaction and less rapidly during subsequent
stages. However, at ∼570 nm the initial rapid decrease
in absorbance was followed by an increase to a final
value comparable with the initial absorbance, and this
sequence of absorbance changes, and their magnitudes,
made it more convenient to monitor the reactions at this
wavelength. The absorbance bands of the substituted
products were virtually the same as those of the
unsubstituted cluster. The rates of the first steps were
generally sufficiently fast, compared with those of any
subsequent ones, for their observed rate constants (k(1))
to be obtainable by simple first-order analysis of the
absorbance decrease. The subsequent absorbance in-
crease almost always occurred in two stages, and in
cases where the third stage was relatively quite slow
and its absorbance change quite small, single-exponen-
tial analysis of the second stage was possible to give k(2).
However, when the rates of the second and third steps
were more comparable, double-exponential analysis of
the data was necessary and provided values of both k(2)
and k(3). Figure 1a shows an example of the decreasing
absorbance during the first stage of a reaction, and b
and c show how a double-exponential analysis is some-
times needed for the absorbance growth during the
second and third stages. The rate constant k(3) for the
third step is sometimes hard to determine accurately
(especially when the absorbance changes are small)
because of the uncertainty of the final absorbance value.
However, the existence of that step is quite clear, and
the double-exponential analysis provides precise values
for the rate constant of the second step. Rate constants
for reactions with etpb are given in Table 1.
Exp er im en ta l Section
The sources and treatment of reagents and the synthesis of
Os3(CO)9(µ-C4Ph4) were exactly as described elsewhere.6 Sol-
vents were bubbled with Ar for 1 h after distillation and stored
under Ar or were degassed by means of at least three freeze-
pump-thaw cycles after distillation and before use. Infrared
spectra were measured on a Nicolet 10DX FTIR spectropho-
tometer. UV-vis spectroscopy and monitoring of slower reac-
There is an inherent ambiguity11 in the analysis of
kinetic data in terms of two successive exponential
changes in the time-dependent absorbance, At, and this
has to be considered. The rate equation when any
(6) Poe¨, A. J .; Farrar, D. H.; Ramachandran, R.; Moreno, C. Inorg.
Chim. Acta 1998, 274, 82-89. (N.B. there is a typographical error in
the structure of Os3(CO)9(µ-C4Ph4) given in this paper; the number of
CO ligands attached to the Os atom in the osmacyclopentadiene ring
should be three and the number attached to the “diene bonded” Os
atom should be two.)
(7) Poe¨, A. J .; Farrar, D. H.; Zheng Y. J . Am. Chem. Soc. 1994, 116,
6252-6261.
(8) Poe¨, A. J .; Zheng, Y. Inorg. Chim. Acta 1996, 252, 311-318.
(9) Hao, J . Ph.D. Thesis, University of Toronto, 1997.
(10) Swain, C. G.; Swain, M. S.; Berg, L. F. J . Chem. Inf. Comput.
Sci. 1980, 20, 47-51.
(11) Espenson, J . H. Chemical Kinetics and Reaction Mechanisms;
McGraw-Hill: New York, 1981; Chapter 4.

5520 Organometallics, Vol. 18, No. 26, 1999
Poe¨ and Moreno
Ta ble 1. Ra te Con sta n ts for th e Rea ction s of
Os3(CO)9(µ-C4P h 4) w ith Etp b in Hep ta n ea
T, °C
[L], mM
k(1), s-1
104k(2), s-1
104k(3), s-1
7.3
7.3b
7.3b
7.3b
7.3
1.23
1.39
2.08
2.78
2.87
3.47
4.16
4.95
5.56
6.45
6.94
8.13
12.1
15.1
1.22
2.04
2.04
2.76
2.76
3.74
5.14
6.46
8.23
9.76
11.5
12.6
2.17
2.87
4.95
6.45
8.13
12.1
15.1
0.217
220
201
221
318
322
376
348
33
21
32
0.219
7.3b
7.3b
7.3
30
25
0.257
0.284
7.3b
7.3
485
617
501
39
31
7.3b
7.3
0.308
0.355
0.393
7.3
7.3
13.6b
13.6b
13.6
13.6b
13.6
13.6
13.6
13.6
13.6
13.6
13.6
13.6
22.0
22.0
22.0
22.0
22.0
22.0
22.0
360
378
345
510
460
599
754
883
994
1130
1200
1340
502
93
66
92
85
89
65
86
62
68
63
37
33
108
88
80
72
88
82
77
0.379
0.407
0.423
0.458
0.481
0.537
0.58
0.597
0.633
0.812
0.835
0.90
670
1000
1250
1600
1320
1690
0.97
1.05
1.19
1.29
a
Reactions monitored by stopped-flow techniques at 572 nm
b
unless otherwise indicated. Reactions monitored with the H.P.
UV-vis spectrophotometer.
species A (which has an absorbance AA ) A0 when only
A is present) changes to species B (absorbance AB if only
pure B is present) and then C (absorbance AC ) A∞) is
shown in eq 1.
A∞ - At ) R exp(-k(2)t) + â exp(-k(3)t)
(1)
where
R ) {(AB - AA)k(2) + (AA - At)k(3)}/(k(3) - k(2)) (2)
and
â ) (AC - AB)k(2)/(k(3) - k(2))
(3)
The two exponential changes that are being considered
here correspond to the second and third of the three
steps involved in the overall reaction so the two rate
constants are designated as k(2) and k(3). It turns out11
that there are two possible solutions to eq 1, and one
cannot, in the first instance, distinguish which of the
rate constants, k(2) and k(3), actually corresponds to
which stage of the reaction. However, there are tests
that can sometimes be applied11 that enable a distinc-
tion to be made, and these show that the first of the
two exponential increases does indeed correspond to the
second stage in the overall process. The tests can be
demonstrated by making use of eq 4 to calculate the
relative absorbances of B and C.
F igu r e 1. (a) Stopped-flow trace of reaction of P(OMe)3
at 13.6 °C (single-exponential analysis; absorbance change
(∆A) ) 0.006). (b) Stopped-flow trace of reaction with etpb
at 22.0 °C (single-exponential analysis; ∆A ) 0.018). (c)
Same as b but with double-exponential analysis. (These
traces have been scanned in directly from the stopped-flow
printout.)
AB/AC ) 1 - (k(3) -k(2))â/A∞k(2)
(4)

Substitution Reactions of Os3(CO)9(µ-C4Ph4)
Organometallics, Vol. 18, No. 26, 1999 5521
All the terms on the right-hand side are provided by
the initial solution to eq 1. If the values of k(2) vary
with [L] (as do the values of k(2) in Table 1), the ratios
AB/AC must still be constant at all values of [L] if the
rate constants k(2) and k(3) are correctly assigned to
the formation of B and C, respectively. If those rate
constants are not properly assigned, then the ratio will
show a pronounced dependence on [L]. In that case eq
5 will provide the correct absorbance (AB′) for the
AB′ ) AA + (AB - AA)k(2)/k(3)
(5)
intermediate B′. AB and AB′ cannot both be constant
with varying [L]. Thus the data in Table 1 for the
reactions with etpb at 13.6 °C and [etpb] ) 1.221 and
12.58 mM (k(2)/k(3) ) 4.0 and 40.6, respectively) show
values of AB/AC ) 0.93 and 0.94. At 22.0 °C and [etpb]
) 2.171, 4.947, 8.129, and 15.11 mM (values of k(2)/
k(3) ) 4.6, 12.5, 18.2, and 21.9, respectively) the values
of AB/AC are all constant at 0.96, and the values of AB′
are not at all constant. This shows that the values of
k(2) for etpb in Table 1 are genuinely applicable to the
second stage of the overall reaction.
Even if neither of the rate constants is dependent on
[L], an assignment can be made on the basis of esti-
mates of AB and AB′ and an evaluation of which is the
more reasonable.11 For L ) etpb, where we know that
the assignment is correct, and in the region of [L] where
k(2) is constant, the absorbance increases typically from
A to B by ∼80% of the overall increase from A to C. On
the other hand the values derived for AB′ show that the
absorbance would increase by a factor of more than 10
from A to B′ and decrease by almost as much from B′ to
C. For L ) P(O-i-Pr)3, where the dependence of k(2)
on [L] is not nearly as pronounced as for L ) etpb, the
absorbance increase from A to B is 75% of the overall
change, whereas the increase from A to B′ would be
twice the overall increase, with a large decrease on
formation of C. For L ) PEt3, where k(2) does not
change at all with [L], the absorbance increase from A
to B is about 30% of the total change, whereas for A to
B′ the absorbance would increase by a factor of 2.5
compared to the overall change and decrease by a factor
of 1.5 from B′ to C. The results are almost identical for
L ) P-n-Bu3 and PCy2H. We can therefore conclude
again that the original assignment of the intermediate
to B rather than B′ is correct, and we take this to be
generally true for all these reactions.
F igu r e 2. Plots of 1/k(2) against 1/[L] for reactions with
L ) etpb at 7.3 (b), 13.6 (2), and 22.0 (9) °C.
obtained at 7.6 °C, but unlike the situation with all
other values of k(3), these show a dependence on
[P(OPh)3]. Apart from these unusual features, which
have not been explored further, the systematic varia-
tions of the values of k(1) are in accord with the
systematic variations shown by the other nucleophiles
(see below). With PPh2H the initial decrease in absor-
bance gives very precise values for k(1). However, this
decrease is followed by a further decrease that leads to
rate constants showing a complex dependence on [PPh2H]
that is probably due to the observed partial decomposi-
tion of this nucleophile. This decomposition does not
affect the eventual formation of the substituted prod-
uct,6 but it does affect the kinetics.
The values of k(1) always increase linearly with [L]
according to eq 6 and values of k(1)a and k(1)b were
k(1) ) k(1)a + k(1)b[L]
(6)
obtained by a weighted linear least-squares analysis7,12
and are shown in Table 2. Values of k(2) are also
frequently dependent on [L], but in the way shown in
eq 7.
The rate constants k(1), k(2), and (where available)
k(3) are tabulated in Table 1 and in tables of the
Supplementary Information, and they can be assigned
unambiguously to three successive stages of reaction
beginning with Os3(CO)9(µ-C4Ph4) and ending with
Os3(CO)8L(µ-C4Ph4).6
All but two of the nucleophiles studied behave in the
way described above, but P(OPh)3 and PPh2H behave
differently. At 13.6 and 21.6 °C and with values of
[P(OPh)3] below ∼20 mM, a double-exponential analysis
for the first rapid decrease in absorbance is required
and the rate constants obtained are rather erratic.
However, at higher values of [P(OPh)3] and at all three
temperatures, single-exponential absorbance decreases
are found as for all the other nucleophiles (see below).
Another complication that is unique to this nucleophile
is that what appear to be reliable values for k(3) are
k(2) ) ak(2)lim[L]/{1 + a[L]}
(7)
Values of k(2)lim can be estimated by a weighted linear
least-squares analysis of the dependence of 1/k(2) on
1/[L] (Figure 2) and are also shown in Table 2. Values
of k(3) (Table 2) are always independent of [L] within
the precision of their determination. Rate constants
obtained by monitoring at 382 nm were in quite good
agreement with those based on 570 nm monitoring
apart from the value of k(1)a for P(OMe)Ph2, which is
very high. When the rate constants for the second stage
were of the right magnitude, it was possible to obtain
values from both Hewlett-Packard and stopped-flow
measurements and the agreement was also good. The
(12) Chen, L.; Poe¨, A. J . Can. J . Chem. 1989, 67, 1924-1930

5522 Organometallics, Vol. 18, No. 26, 1999
Poe¨ and Moreno
Ta ble 2. Kin etic Da ta for th e Rea ction s of Os3(CO)9(µ-C4P h 4) w ith L in Hep ta n e, a t 13.6 °C a n d Mon itor ed
a t 570 n m (u n less oth er w ise in d ica ted )
θc
k(1)b
103k(1)a
(s-1
σ(k(1))d
103k(2)lim
(s-1
σ(k(2))d
104k(3)
(s-1
σ(k(3))d
b
no.a
1
L
pKa′
(deg)
(M-1 s-1
)
)
(%)
)
(%)
)
(%)
etpb
-0.30
101
13.5 ( 0.8e
192 ( 6
338 ( 5
3.2(7)
1.6(9)
97 ( 17
141 ( 5
177 ( 26
240 ( 16
284 ( 61
202 ( 38
39 ( 2
71.3 ( 2.9
167 ( 5
26.0 ( 0.66
51 ( 1
12.6(10)
11.3(8)
13.5(10)
3.7(9)
13.6(7)
17(12)
9.9(10)
10.5(12)
3.6(6)
5.2(12)
3.8(12)
14(8)
8.2(10)
1.4(7)
30.1 ( 2.2
77 ( 4
19(7)
22.8 ( 0.7
17(10)
f
37.9 ( 1.0
20.0 ( 0.7
24.0 ( 0.5e
32.3 ( 0.4
44.6 ( 1.3g
18.9 ( 0.5h
21.7 ( 0.7
29.8 ( 0.8f
25.3 ( 0.4
141 ( 2
33.4 ( 1.2
1.84 ( 0.07i
2.64 ( 0.1
3.98 ( 0.26j
31.3 ( 0.5
11.5 ( 0.7
13.4 ( 0.7k
92.4 ( 4.1
40.1 ( 1.2
45.3 ( 1.2
10.2 ( 0.1
726 ( 7
173 ( 8
67 ( 5
120 ( 4
219 ( 14
49 ( 4
81 ( 5
205 ( 9
71 ( 3
41 ( 10
440 ( 20
86 ( 4
104 ( 5
193 ( 20
70 ( 3
21 ( 3
113 ( 5k
∼0
1.1(7)
6.1(11)
4.3(8)
3.3(14)
5.9(8)
7.5(13)
6.8(12)
4.9(11)
4.2(12)
4.8(13)
5.7(13)
4.8(7)
85 ( 5
175 ( 7
40 ( 2
14(7)
11(8)
16(10)
24(11)
11(4)
26(9)
28(12)
10(9)
2
3
P(OMe)3
P(OEt)3
0.83
1.64
107
109
75 ( 6
204 ( 11
117 ( 10
115 ( 8
270 ( 9
95 ( 6
4
P(OMe)2Ph
1.48
120
160 ( 19
39 ( 2
38.5 ( 0.2
5
6
7
8
P(OEt)2Ph
PMe2Ph
PPh2H
1.99
5.07
0.52
121
122
126
128
19(10)
P(OPh)3
-2.79
34.1 ( 2.4
35 ( 5
5.3(6)
23(11)
23(10)
3.0(13)
8.7(9)
7.6(10)
24(7)
4.7(10)
5.4(5)
78 ( 25
9
10
P(O-i-Pr)3
P(OMe)Ph2
3.38
2.09
130
132
3.6(13)
11.1(10)
6.3(11)
12.0(9)
9.6(6)
31.3 ( 0.4
114 ( 3
108 ( 4
∼50
12(11)
106 ( 3k
9 ( 1
11
12
13
14
PEt3
7.96
8.57
8.67
2.35
132
132
132
133
15 ( 1
20 ( 1
11 ( 1
17(8)
8(8)
19(6)
P(n-Pr)3
P(n-Bu)3
P(OEt)Ph2
∼0
4.2 ( 0.4
17.5 ( 2.0
10.0 ( 0.2l
8.8 ( 0.2m
2.7 ( 0.1
5.5 ( 0.2
14 ( 1
24(6)
15(5)
∼0
6.9(13)
2.6(10)
22.4 ( 1.0
5.1(7)
4.3(6)
10(5)
13(28)
17(9)
15
16
PMePh2
4.06
4.60
136
40.5 ( 1.6n
49.5 ( 1.4
68.2 ( 1.4j
21.8 ( 1.1
17.8 ( 1.6
38 ( 6
88.5 ( 6.6
13 ( 4
13.2(12)
7.6(14)
4.7(9)
∼0.1
∼0.4
∼5
PEtPh2
PCy2H
140
143
11.0(10)
24.0 ( 0.8l
31.1 ( 0.9m
∼15
14.2(17)
8.0(7)
17
6.08
15.8 ( 0.8
6 ( 4
13.0(8)
14.2 ( 1.1
23(9)
Nucleophile number. The pKa′ values are taken from ref 14e. c Tolman cone angles. Estimated standard error of measurement of
a
b
d
k (numbers of different concentrations of L are given in parentheses). e At 7.3 °C. f At 22.0 °C. At 21.3 °C. At 7.6 °C. At 7.9 °C. At
g
h
i
j
k
l
m
21.6 °C. Monitored at 382 nm. Monitored with the Hewlett-Packard spectrophotometer. Monitored with the stopped-flow equipment.
At 7.5 °C.l
n
Ta ble 3. Activa tion a n d Th er m od yn a m ic P a r a m eter sa
etpb
P(OEt)3
P(OMe)2Ph
P(OPh)3
PMePh2
q
q
∆H+L
∆S+L
∆H-L
∆S-L
∆H+L
∆S+L
10.0 ( 1.4
-15.4 ( 4.7
14.3 ( 0.4
-9.4 ( 1.2
-4.3 ( 1.5
-6.0 ( 4.9
10.7 ( 3.1
-24 ( 11
6.7 ( 0.4
-28.1 ( 1.4
13.1 ( 1.1
-17.0 ( 3.9
-6.4 ( 1.2
-11.1 ( 4.1
17.1 ( 1.0
-3.9 ( 3.3
18.5 ( 1.2
-3.4 ( 4.3
4.68 ( 0.65
-36.0 ( 2.3
16.4 ( 1.3
-6.1 ( 4.6
-11.7 ( 1.5
-29.9 ( 5.1
22.7 ( 2.1
14.7 ( 7.5
11.2 ( 4.0
-28 ( 14
8.83 ( 0.88
-26 ( 3
5.7 ( 0.5
-30.9 ( 1.6
18.1 ( 1.4
-1.9 ( 4.7
-12.4 ( 1.4
-29.0 ( 5.0
q
7.8 ( 2.7
-36 ( 10
1.0 ( 2.8
10 ( 10
4.8 ( 4.8
-48 ( 17
q
°
°
q
∆H-CO
∆S-CO
q
q
∆Hi
9.5 ( 7.0
q
∆Si
-36 ( 24
a
Enthalpies in kcal mol-1 and entropies in cal K-1 mol-1
.
tabulated standard errors of measurement show that
the values of k(1) are generally very precise, those of
k(2) are somewhat less so, and those of k(3) are often of
relatively poor precision. Reactions of five of the nu-
cleophiles were followed over a range of temperatures,
and activation parameters of good precision were ob-
tained by a weighted least-squares analysis12 and are
reported in Table 3.
E q u ilib r iu m Con st a n t s for t h e F ir st St a ge of
Rea ction . It is very clear that the values of k(1)a are
not constant as would be expected if they corresponded
to rate constants for CO dissociation or rate-determining
isomerization of the cluster to form a reactive interme-
diate.7,8 The form of the absorbance changes observed
here for the substitution reactions of Os3(CO)9(µ-C4Ph4)
is similar to those observed for substitution reactions
of M5C(CO)15 (M ) Ru7 and Fe8). In those cases the first
step is also accompanied by a substantial decrease in
absorbance of the band in the visible-near-UV region
of the electronic spectra. This absorbance decrease is
ascribed to reaction of the P-donor ligand, L, with the
cluster to form an adduct, and for reasons identical with
those adduced for the M5C(CO)15 clusters,7,8 we conclude
that the first stage of reaction here is also adduct
formation. The rate equation, shown in eq 6, shows that
adduct formation is reversible so that the values of k(1)b
in Table 2 can be assigned to the adduct formation
reaction (and be renamed k+L), while k(1)a can be
assigned to the loss of L from the adduct and reforma-
tion of the starting cluster, so that it can be renamed
k-L. The adduct formation reactions of Ru5C(CO)15 show
a detectable, but not very precise, value of k-L only when
L ) etpb or P(OPh)3.7 For Fe5C(CO)15, in addition to
etpb and P(OPh)3, the nucleophiles P(OMe)3 and PPh-

Substitution Reactions of Os3(CO)9(µ-C4Ph4)
Organometallics, Vol. 18, No. 26, 1999 5523
Ta ble 4. Equ ilibr iu m Con sta n ts (M-1) for Ad d u ct
F or m a tion a t 13.6 °C (u n less oth er w ise in d ica ted )
K+L′′
(100Aadduct/Ainitial)
no.a
1
L
K+L
K+L
′
etpb
70 ( 7b
188 ( 55b
104 ( 50b,c
167 ( 43
91 ( 9b,c
116 ( 42d
48 ( 14
287 ( 43 (59 ( 20)b
68 ( 4
143 ( 14 (46 ( 12)
50 ( 2d
116 ( 9
177 ( 24 (76 ( 22)c
167 ( 19 (17 ( 8)
413 ( 31 (2 ( 1)b
497 ( 46 (28 ( 7)
410 ( 43 (29 ( 8)e
2
3
P(OMe)3
P(OEt)3
360 ( 35b
215 ( 20
205 ( 20e
513 ( 90b
440 ( 45
320 ( 35e
4
P(OMe)2Ph 354 ( 40f
268 ( 20
470 ( 110f 443 ( 21 (4-8))f
230 ( 13
158 ( 42d
362 ( 54
371 ( 25 (1.5-6.5))
261 ( 15 (21 ( 4)d
661 ( 9 (16 ( 1)
145 ( 10d
5
6
7
8
P(OEt)2Ph
PMe2Ph
PPh2H
360 ( 80
3700 ( 1000
76 ( 6
P(OPh)3
21.4 ( 1.8g
11 ( 1g
26.3 ( 1.5 (1-5)g
25.5 ( 1.3 16.8 ( 4.1 14.2 ( 0.8 (0-6)
21.0 ( 3.6h
P(O-i-Pr)3 448 ( 26
11 ( 5h
647 ( 37
12.1 ( 1.4 (21 ( 10)h
823 ( 10 (16 ( 1)
630 ( 12 (11 ( 1)
9
10 P(OMe)Ph2 564 ( 114
11 PEt3
12 P(n-Pr)3
13 P(n-Bu)3
v. large
v. large
v. large
F igu r e 3. Plot of f/[L] against 1 - f, for L ) P(OEt)2Ph,
where f ) (A0 - Aeq)/A0 as in eq 8.
14 P(OEt)Ph2 458 ( 26
791 ( 38 (11 ( 2)
3400 ( 340 (9 ( 1)i
2700 ( 770 (12 ( 4)
1300 ( 160 (25 ( 2)h
1540 ( 110 (6 ( 1)
3300 ( 450 (9 ( 1)
15 PMePh2
2300 ( 300i
1340 ( 250
775 ( 75h
cluster exists in the form of the adduct. Plots of (A0 -
Aeq)/A0[L] against Aeq/A0 are found to be linear (e.g.,
Figure 3), so that values of K+L′′ and A∞/A0 can be
obtained from the gradients and the ratio intercepts/
gradients, respectively. Their values are given in Table
4.
16 PEtPh2
17 PCy2H
1900 ( 650
5000 ( 3000
a
b
Nucleophile number. At 7.3 °C. c These values were obtained
by analysis with omission of data at low [L]. At 22.0 °C. e At 21.3
d
°C. f At 7.6 °C. At 7.9 °C. At 21.6 °C. At 7.5 °C.
g
h
i
System a tic Dep en d en ce of th e Ra te Con sta n ts
on th e Electr on ic a n d Ster ic P r op er ties of th e
Nu cleop h iles. Th e Kin etics of Ad d u ct F or m a tion .
Systematic studies of adduct formation by P-donor
ligands with metal carbonyl clusters are quite rare.7,8,13
However, data for a large number of associative reac-
tions of metal carbonyl clusters are available,14 and it
has been concluded7,14a,c that adduct formation could be
an initial step in these reactions but that further
reaction of the adducts is too fast for them to be
observable. All the data for these reactions are success-
fully analyzable according to eq 913,14c,e or related
equations.15 These partition the rates between electronic
(OMe)2 show significant values of k-L as well, although
they are not at all precise either.8 In the reactions
described here, quite precise values of k-L are known
for virtually all the adduct formation reactions. The
ratios k+L/k-L therefore provide values that can be
assigned to the equilibrium constants, K+L, for forma-
tion of the adducts, and these are reported in Table 4.
The equilibrium constants can also be obtained8 from
the dependence on [L] of the rate constants, k(2), for
the second stage of the overall reaction according to eq
7. Because the second-stage reactions are generally
much slower than the first, the reacting species during
the second stages are equilibrium mixtures containing
Os3(CO)9(µ-C4Ph4) and the adducts Os3(CO)9L(µ-C4Ph4).
It follows8 that the values of a in eq 7 should be equal
to the equilibrium constants for adduct formation, and
they are listed as K+L′ in Table 4.
A third way of estimating the equilibrium constants
is available from the dependence on [L] of the absor-
bance changes during adduct formation. The computer
output of the analysis of absorbance changes by the
stopped-flow equipment provides values of A0, the
absorbance of the initial cluster Os3(CO)9(µ-C4Ph4), and
Aeq, the absorbance of the equilibrium mixtures of the
original cluster and the adducts Os3(CO)9L(µ-C4Ph4). It
can easily be shown that, in the simplest case, the
decrease in absorbance during this first stage should
depend on [L] according to eq 8, where K+L′′ is the
log k+L ) R + â(pKa′ + 4) + γ(θth - θ)λ
(9)
effects due to different nucleophile σ-basicities and steric
effects due to different nucleophile sizes. In eq 9 the
σ-basicity parameter pKa′ is related either to pKa
values16 or to Bartik’s ø values17 depending on whether
(13) (a) Hudson, R. H. E.; Poe¨, A. J . Organometallics 1995, 14, 3238-
3248. (b) Neubrand, A.; Poe¨, A. J .; van Eldik, R. Organometallics 1995,
14, 3249-3258.
(14) (a) Poe¨, A. J . In Metal Clusters; Moskovits, M., Ed.; Wiley-
Interscience: New York, 1986; Chapter 4. (b) Darensbourg, D. J . In
The Chemistry of Metal Cluster Complexes; Shriver, D. F., Kaesz, H.
D., Adams, R. D., Eds.; VCH: New York, 1990; Chapter 4. (c) Poe¨, A.
J .; Farrar, D. F.; Zheng, Y. J . Am. Chem. Soc. 1992, 114, 5146-5152.
(d) Poe¨ A. J . In Mechanisms of Inorganic and Organometallic Reactions;
Twigg, M. V., Ed.; Plenum: New York, 1994; Vol. 8, Chapter 10. (e)
Poe¨, A. J .; Chen, L. Coord. Chem. Rev. 1995, 143, 265-295.
(15) Liu, H.-Y.; Eriks, K.; Prock, A.; Giering, W. P. Organometallics
1990, 9, 1758-1766.
(A0 - Aeq)/A0[L] ) -K+L′′A∞/A0 + K+L′′Aeq/A0 (8)
(16) Henderson, W. A.; Streuli, C. A. J . Am. Chem. Soc. 1960, 82,
5791-5800. Allman, T.; Goel, R. G. Can. J . Chem. 1982, 60, 716-722.
(17) Bartik, T.; Himmler, T.; Schulte, H.-G.; Seevogel, K. J . Orga-
nomet. Chem. 1984, 272, 29-41.
equilibrium constant derived in this way and A∞ is the
unknown absorbance when [L] is very large and all the

5524 Organometallics, Vol. 18, No. 26, 1999
Poe¨ and Moreno
which is essentially identical with the value obtained
from nucleophiles 1-6. However, line B lies 0.40 ( 0.04
below line A.
(iii) Nucleophiles 16 and 17 (θ ) 140° and 143°) lie
clearly below line B, and these results indicate that
there is a steric threshold between 122° and 130°, above
which steric effects are operative.
(iv) There is a final group of points that do not fit with
the trends shown by the other groups and that have to
be considered. The nucleophiles P-n-Pr3 and P-n-Bu3
(both with Tolman cone angles of 132°) lie well below
line B, their rate constants being a factor of ∼2 lower
than that for the isosteric nucleophile PEt3, which fits
well to that line. Inspection of the kinetic data for these
nucleophiles does not suggest that experimental error
is the cause of this discrepancy, and we do not believe
that minor variations in the cone angles could account
for it. Further, ligand 15 (PMePh2) has a rate constant
∼2 times higher than would be expected compared with
that for the almost identical nucleophile 16 (PEtPh2), a
nucleophile that lies below line B, as it should with its
cone angle of 140°. We are therefore forced to omit these
sets of data from further analysis. Finally, the nucleo-
phile PPh2H lies significantly above line A, whereas its
Tolman cone angle of 126° suggests that, if anything, it
should lie below it. This deviation will be considered
later.
Analysis of the fit to eq 9 of the set of data for the 13
remaining nucleophiles was carried out according to the
procedure described previously,13a,14c and the results are
shown in Table 5. The value of â is essentially the same
as those found separately from either the group of small
nucleophiles or the group of near-isosteric nucleophiles
(θ ≈ 130°) and the goodness of fit, and the uncertainties
in the derived parameters are distinctly better than
those reported for other clusters.8,13a Even when data
for all the 17 nucleophiles are included in the analysis,
the fit is reasonable, the effects of the three or four
deviant nucleophiles being largely damped by those of
the 13 conforming ones. The values of θth and γ remain
the same, but â is reduced slightly to 0.11 ( 0.02, R2
decreases to 0.669, and the rmsd increases to 0.203.
However, these relatively small changes leave unaltered
any of the chemical conclusions that can be drawn.
One problem with the method of analysis used is that
it gives no measure of the uncertainty in the steric
threshold since the best set of parameters is obtained
by variation of θth until R2 is maximized and the rmsd
is minimized.14c An alternative method is to treat the
data for the smaller (θ e 122°) and larger (θ g 128°)
nucleophiles separately. The smaller nucleophiles give
a value of â ) 0.160 (see above) and log k+L - â(pKa′ +
4) ) R ) 0.58 ( 0.05 at 13.6 °C. When plotted against
θ, the intersection, θth ) 121°, of this horizontal part of
the steric profile with the sloping part obtained from
the fit of the larger nucleophiles to eq 10 can be
F igu r e 4. Dependence of log k+L on pKa′. Line A is drawn
through data for nucleophiles 1-6 with θ e 122°. Line B
is drawn through data for nucleophiles 8-11 and 14 with
cone angles close to 130°. To draw the line, the data were
corrected for the small differences of the cone angles from
130° by using γ ) -0.04 deg-1 (see text). The data
denominated by 9 were not used in any of the analyses.
the latter include an unwanted π-acidity contribution
or not. The parameter θ is simply Tolman’s ligand cone
angle,18 while θth is a cone angle above which steric
effects become significant and λ, the switching factor,
becomes unity instead of zero.15 The value of R is a
measure of a standard reactivity (SR) chosen as the log
k value for a nucleophile with θ < θth and pKa′ )
-4.13a,14e No significant contributions to the rates by aryl
effects,19 or due to nucleophile π-acidity,20 have been
detected so far.
The values of k+L at 13.6 °C in Table 2 have been
successfully fitted to eq 9 as follows. Figure 4 shows the
dependence of log k+L on pKa′, and it is possible to group
the nucleophiles into four categories.
(i) Nucleophiles 1-6 have cone angles that vary from
101° to 123° and form a group where log k+L depends
linearly on pKa′ irrespective of the cone angles (line A).
The gradient, â, is found to be 0.160 ( 0.037 from an
unweighted linear least-squares analysis (rmsd22
)
0.121 and R2 ) 0.823).
(ii) Nucleophiles 8-11 and 14 all have cone angles
close to 130°, and their values of log k+L fall on another
straight line, B, after small adjustments are made to a
common cone angle of 130° according to the value of γ
(-0.04 deg-1) obtained subsequently. The gradient of
line B is 0.160 ( 0.013 (rmsd ) 0.074 and R2 ) 0.982),
(18) Tolman, C. A. Chem. Rev. 1977, 77, 313-348.
(19) Wilson, M. R.; Woska, D. C.; Prock, A.; Giering, W. P. Orga-
mometallics 1993, 12, 1742-1752.
log k+L ) R + â(pKa′ + 4) + γθ
(10)
(20) Giering et al.21 believe they can quantitify π-acidity effects when
the P-donors are present as ligands, but when applied to these
reactions involving P-donors as nucleophiles, their analysis suggests
that π-acidity weakens nucleophile strength. This is contrary to the
reasonable expectation that π-acidity would strengthen the bonding
in such transition states.
obtained, and its uncertainty of ∼(4° can be obtained
as well by allowing appropriately for the covariance in
the uncertainties of the horizontal and sloping parts of
the steric profile. An important point to note here is that
the average adjustment of θ that is required to produce
a perfect fit to the sloping part of the steric profile is
(21) Giering, W. P. Personal communication.
(22) rmsd ) root-mean-square deviation.

Substitution Reactions of Os3(CO)9(µ-C4Ph4)
Organometallics, Vol. 18, No. 26, 1999 5525
Ta ble 5. Electr on ic a n d Ster ic P a r a m eter s for Rea ction s of Os3(CO)9(µ-C4P h 4) a t 13.6 °C
(u n less oth er w ise in d ica ted )
rate constant
Na
θth (deg)
SRb,c
â
γ (deg-1
)
rmsd (R2)
k+L
k+L
13
13
122
121(4)
0.95(9)
+0.152(15)
+0.160(37)e
+0.160f
-0.042(6)
0.109 (0.920)
0.121 (0.823)
0.009 (0.904)
0.169 (0.869)
d
0.90(22)e
-0.040(8)f
k-L
11
15
15
e101
e101
e101
-
-
-0.57(7)
0g
(-0.10)h
k-CO
k-CO
-0.44(9)
(-0.084)h
0.31(15)
0i
0.303 (0.637)
0.287 (0.674)
-0.013(11)j
(0.062)h
a
b
Number of ligands used in analysis. The value of the standard reactivity has been adjusted to 25 °C by using the activation enthalpy
for etpb (Table 3), i.e., by adding 0.31. c Numbers in parentheses for SR and the other parameters are standard deviations. This analysis
was carried out in two separate steps, i.e., for six nucleophiles with θ e 122° eq 9 was used with λ ) 0, and for seven nucleophiles with
θ g 128°, eq 9 was used with λ ) 1 and θth ) 0 (see text). e From data for the smaller nucleophiles. f From data for the larger nucleophiles;
d
g
value of â taken as the same as from data for smaller nucleophiles. Set to zero. No improvement in the analysis is observed when the
h
term in γ is included. The italicized numbers in parentheses are dimensionless values of â obtained by using the relationship between
δ and pKa′′. The corresponding nonitalicized numbers have units of ppm-1(see text). Set to zero. This analysis allowed for the term γθ
i
j
in eq 9.
A difficulty in assessing analyses such as these
obviously has to do with the appropriateness of the
Tolman cone angles as a measure of relative nucleophile
size. Although adjustments to Tolman cone angles are
sometimes made in order to improve fits of specific sets
of kinetic and thermodynamic data,23either the adjust-
ments are unjustified24 or their effects are insignificant.
The modest increase proposed23 for the cone angle of
PEt3 to 136° is not enough to alter the results ap-
preciably, while the enormous proposed increases23 of
21° and 25° in the cone angles of P(OMe)3 and P(OEt)3,
respectively, are not supported by analysis24 of a wide
range of data. Analysis of some published data in the
way described here does suggest24 that cone angles for
these phosphites could justifiably be increased by 10°,
but that would not bring them up above the steric
threshold observed in this study.
Apart from these and other spasmodic cone angle
modifications, a major choice that is available is between
Tolman’s cone angles and the ligand repulsion energies,
ER, derived by Brown et al. by molecular mechanics
calculations on Cr(CO)5L, etc.25 As found with analyses
of other sets of data,7,8,13a,14c use of ER values leads to a
fit that is significantly inferior to that found above.
F igu r e 5. Electronic profile for adduct formation, i.e.,
dependence of log k+L - 0.04(θ - 121) on pKa′ (R2 ) 0.944).
The data denominated by 9 were not used in the analysis
of the fit.
Loss of L fr om th e Ad d u ct. These reactions involve
the loss of the P-donor ligands from the adducts, and
since the adducts have fully formed Os-P bonds with
probable contributions from ligand π-acidity, an elec-
tronic parameter that expresses that contribution should
be involved in any analysis of the systematics. Unfor-
tunately no parameters that provide an independent
measure of the π-acidity for individual P-donor ligands
are available so one has to make do with parameters
such as δ, the 13C chemical shifts in Ni(CO)3L relative
to that in Ni(CO)4,26 or ø.17 Both of these are affected
by π-acidity, although to slightly different extents, and
their use as electronic parameters assumes that any
measured rate constants have the same sensitivity to
the relative contributions of σ-basicity and π-acidity as
only (1.4° and this shows the very satisfactory nature
of the cone angles involved.
The complete data can be represented by the elec-
tronic profile shown in Figure 5, where the values of
log k+L have been adjusted by adding the small correc-
tion 0.40(θth - 121°) to allow for the steric effects of the
larger nucleophiles. The values â ) 0.159 ( 0.012, rmsd
) 0.104, and R2 ) 0.944 show that the data are very
well represented by this analysis, only ca. 6% of the
variation in the values of log k+L remaining unacount-
able for on the basis of these simple electronic and steric
effects. Inevitably, nucleophiles 12, 13, and 15 still do
not fit the profile, but the data for PPh2H (point 7) lie
as close to the line as those for several other nucleo-
philes when no correction is made for its size; that is,
its cone angle can be concluded to be closer to 121° than
to 126°. This is not a large adjustment for a P-donor
containing the very polarizable H atom. The overall fit
of 14 of the 17 nucleophiles to eq 9 is therefore excellent.
(23) For example: Stahl, L.; Ernst, R. D. J . Am. Chem. Soc. 1987,
109, 5656-5680.
(24) Chen, L. Ph.D. Thesis, University of Toronto, Canada, 1991.
(25) Brown, T. L. Inorg. Chem. 1992, 31, 1286-1294. Choi, M.-G.;
Brown, T. L. Inorg. Chem. 1993, 32, 5603-5610. Brown, T. L.; Lee, K.
J . Coord. Chem. Rev. 1993, 128, 89-116.
(26) Bodner, G. M.; May, M. P.; McKinney, L. E. Inorg. Chem. 1980,
19, 1951-1958.

5526 Organometallics, Vol. 18, No. 26, 1999
Poe¨ and Moreno
F igu r e 6. Dependence of log k-L on the net donor capacity
of L as given by δ(13CO) (R2 ) 0.87). The data denominated
by 9 were not used in the analysis of the fit.
F igu r e 7. Dependence of log k-CO on the net donor
capacity of L as given by δ(13CO) (R2 ) 0.64). The data
denominated by 9 were not used in the analysis of the fit.
adducts Os3(CO)9L(µ-C4Ph4) so that k(2) can be renamed
k-CO. The values of log k-CO, like those of log k-L, show
a linear decrease with increasing δ (Figure 7), though
the scatter is quite pronounced. Nevertheless, the data
can be fitted to eq 11 moderately well (Table 5) provided
the data for the one distinctly deviant ligand (P(OPh)3)
are omitted. The value of γ is only marginally signifi-
cant, but the goodness of fit is not improved when it is
set to zero, although the precision of â (in units of
ppm-1) is improved. Examination of the correlation
between the values of δ and pKa′′ (R2 ) 0.968) for the
ligands involved in these reactions allows a dimension-
less value of â to be obtained for comparative purposes.
Th e F in a l Sta ge. The third stages of the reactions
are not kinetically so well-defined. The rather ap-
proximate values of log k(3) are constant at ∼2.01 (rmsd
) 0.12) for six ligands with θ e 130°. No values of k(3)
were obtained for ligands with cone angles of 122°, 126°,
and 128°, but this could have been due more to the
absence of sufficiently large absorbance changes than
to genuinely small (or large) rate constants. When θ >
130°, the rate constants seem to be significantly lower,
but no regular trend is observable. As shall be seen
below, this final stage can be assigned to an isomeriza-
tion of the initial, kinetic product of CO loss and Os3
ring closure. The rate constants k(3) can therefore be
renamed ki.
δ or ø. Any major deviations from this assumption would
show up in deviant behavior by the π-acids compared
with those that are σ-bases only. Figure 6 shows the
dependence of log k-L on δ, and it can be seen that, with
the exception of data for two ligands PPh2H and P(OPh)3
(no. 8), the values of log k-L decrease linearly with
increasing δ, no evidence for any steric effects being
evident. The great majority of the data are therefore in
accord with eq 1127 with γ ) 0. No advantage is gained
log k-L ) R + âδ + γθ
(11)
by including the possibility of a steric effect in the
analysis, and the values of â and the “goodness of fit”
parameters are shown in Table 5. The value of â has
units of ppm-1 and is not immediately comparable with
the values of â obtained for adduct formation, when pKa′
is used as the σ-donor electronic parameter and â is
dimensionless. However, the values of δ and pKa′′ for
the relevant ligands correlate moderately well (R2
)
0.764). Values of pKa′′ for π-acid ligands as well as for
pure σ-donors can be obtained from ø in exactly the
same way as pKa′ values.13a,14c (pKa′ and pKa′′ values
are identical for pure σ-donors.) They are therefore
equivalent to ø but lead to dimensionless values of â
when used as the electronic parameter. They could
therefore be used in the analysis of the log k-L data,
but the fit is not as good as when δ is used. Neverthe-
less, the dimensionless value of â that is obtained (Table
5) can be compared with corresponding values for the
other reactions.
Discu ssion
(1) Th e Stoich iom etr ic Mech a n ism s. The reac-
tions studied here lead eventually to the monosubsti-
tuted clusters Os3(CO)8L(µ-C4Ph4), where the 17 P-do-
nor ligands all have θ e 143°. The crystallographic
structures of two of the products (L ) etpb and P(OPh)3)
show6 that the P-donor substituent has displaced a CO
ligand from the Os (CO)4 moiety in the cluster and that
it is in a position trans to the Os atom in the osmacy-
clopentadiene ring. The close similarity of the FTIR
Loss of CO fr om th e Ad d u ct. The absorbance
changes during the second stage of the reactions are
closely similar to those that have been ascribed to loss
of CO from the adducts M5C(CO)15L (M ) Ru or Fe).7,12
We therefore ascribe them to the loss of CO from the
(27) Hudson, R. H. E.; Poe¨, A. J . Inorg. Chim. Acta 1997, 259, 257-
263.

Substitution Reactions of Os3(CO)9(µ-C4Ph4)
Organometallics, Vol. 18, No. 26, 1999 5527
spectra of the other 15 products to those of the two
crystallographically characterized ones strongly sug-
gests that their structures are the same. The formation
of these products, however, does not proceed via a
simple single step but via at least two kinetically
detectable steps as shown by eqs 12-14.
(See, for example, Table 2 and Figure 2.) Perhaps
related to this is the fact that even a small contribution
to k+L by a unimolecular path (not dependent on [L])
can reduce K+L′ substantially. Thus, for reactions with
etpb at 13.6 °C, if a path such as this has a rate constant
of only 0.01 s-1 (i.e., ∼25% of the lowest measured value
of k(2)), the data are consistent with K+L ) K+L′ ) 70
k+L
M-1 31
This contribution should occur independently of
.
Os3(CO)9(µ-C4Ph4) + L w x
k-L
which ligand is involved, but this does not seem to be
the case and cannot be so when K+L′ < K+L. The values
of K+L′′ depend quite strongly on the term A0 - Aeq in
eq 8. This can be quite imprecise at low values of [L]
either because the difference is small or because there
are systematic errors in A0. The values of A∞/A0 that
are given by this method are usually quite small, but
they also vary randomly even for the same reaction at
different temperatures. Since this is unlikely to be a real
effect, it seems that there must indeed be systematic
errors in this method. However, these cannot account
for all the values of K+L′′ that are higher than the others
because A∞/A0 must always be g0. Thus, for reaction
with P(OEt)3 at 13.6 °C, even if A∞/A0 ) 0, K+L′′ would
still be g320 M-1. On balance we believe that the values
of K+L are the best estimates of the equilibrium con-
stants for adduct formation.
However, the values of A∞/A0, even if only approxi-
mate, are important for another reason. In the case of
the reactions of the clusters M5C(CO)15 (M ) Ru and
Fe),7,8 supporting evidence for the formation of adducts
in the first stage of reaction was the substantial
decrease in absorbance of the bands characteristic of
metal-metal interactions and the existence of known
and isolated adducts. Adduct formation is able to occur
without violation of electron counting rules because it
is accompanied by opening up of the closed cluster by
breaking one M-M linkage, and it was assumed that
this was the explanation for the decrease in absorbance.
The systems studied here provide a quantitative mea-
surement of the extent of such an absorbance decrease,
and Table 5 shows that the decrease is generally quite
large and the final absorbance is often negligible.32
Os3(CO)9L(µ-C4Ph4) (12)
k-co
Os3(CO)9L(µ-C4Ph4)
8
[Os3(CO)8L(µ-C4Ph4)]A + CO (13)
ki
[Os3(CO)8L(µ-C4Ph4)]A
98
[Os3(CO)8L(µ-C4Ph4)]B (14)
Ad d u ct F or m a tion (Eq 12). The fact that k(1)a in
eq 6 is not constant for all the nucleophiles shows that
this rate constant cannot be assigned to a CO-dissocia-
tive process or to rate-determining isomerization.7,8
Rather it has to be assigned to the loss of L from an
adduct Os3(CO)9L(µ-C4Ph4) formed by reversible addi-
tion of L to the original cluster. Thus k(1)a can be
designated as k-L, and the addition process is charac-
terized by k(1)b ) k+L. The significant contribution to
k(1) of the terms k(1)a contrasts with the behavior
observed with the clusters M5C(CO)15 (M ) Ru7 and
Fe8). The reactions of the carbonyls (µ-H)2Os3(CO)10 with
13a
28
P(OPh)3
and Fe2(CO)6(NO)(µ-AsMe2) with SbPh3
also show detectable rate constants for loss of ligand
from the adduct, but they are exceptional among the
larger group of ligands studied.29 The cluster Os3(CO)9(µ-
C4Ph4) is unique, therefore, in forming weak enough
adducts for the rate constants for loss of the added
ligand to be determined with considerable precision for
a wide range of P-donors.
The equilibrium constants, K+L, for adduct formation
can be obtained quite precisely from the ratios k+L/k-L
,
values of which are given in Table 4. However, values
of the equilibrium constants can also be obtained from
the extent of the decrease in the absorbance, during the
first step of the substitution, as a function of [L] and
are given as K+L′′ in Table 4. This method also leads to
ratios of the molar absorbances of reactant cluster and
adduct. A third way of obtaining the equilibrium
constants is from the dependence of k(2) on [L] over a
time scale where the adduct formation equilibrium is
set up relatively rapidly, and these values are given as
K+L′ in Table 4.
Th e Secon d a n d Th ir d Step s (Eqs 13 a n d 14). The
second step involves a virtually complete recovery of the
absorbance in most cases. This supports the conclusion,
based on the kinetics, that this step involves dissociative
loss of CO and re-formation of the closed cluster that
now contains a P-donor substituent.34
The isolated and well-characterized final products
have the same stoichiometry as the products of CO
dissociation. For this reason the third observed step (eq
14) has to be assigned to an isomerization process in
which the kinetically formed substituted cluster product
rearranges to what must be the most stable form as
The precision of the values of K+L is often quite good,
but it inevitably decreases as the intercept of the plot
of k(1) against [L] (eq 6) becomes smaller and less well-
defined and as K+L becomes larger. The precision of K+L
is usually fairly low and depends very much on how
many values of k(2) were obtained at low [L] where k(2)
was much less than k(2)lim. Sometimes the values of K+L
depend very strongly on one or two points at low [L] so
that any error in that point has a large systematic effect.
′
(30) Poe¨, A. J .; Sampson, C. N.; Smith, R. T.; Zheng, Y. J . Am. Chem.
Soc. 1993, 115, 3174-3181.
(31) A much more important and unexpected unimolecular path has
been detected by this sort of discrepancy.30
′
(32) This evidence relates to another explanation for adduct forma-
tion. This involves some form of “slippage” of the osmacyclopentadiene
moiety, analogous to that first proposed for cyclopentadiene com-
plexes.33 However, this process would not account for the substantial
absorbance decrease in the way that cluster opening does since all that
would happen during adduct formation would be the replacement of
one donor electron pair from the diene unit by another from the
nucleophile.
(28) J ackson, R. A.; Kanluen, R.; Poe¨, A. J . Inorg. Chem. 1984, 23,
523-527.
(29) The cluster (µ-H)2Os3(CO)10 undergoes reversible adduct forma-
tion with CO, and precise rate constants for the reverse reaction have
been obtained.30
(33) Schuster-Woldan, H. G.; Basolo, F. J . Am. Chem. Soc. 1966,
88, 1657-1663. Basolo, F. Polyhedron 1990, 9, 1503-1535.

5528 Organometallics, Vol. 18, No. 26, 1999
Poe¨ and Moreno
Sch em e 1a
given by the crystallographic studies. Although there
is no evidence as to how this occurs in this system, the
small absorbance change generally observed is quite
consistent with isomerization. Similar processes that in-
volve small absorbance changes in the electronic spectra
have been postulated for reactions of the adducts (µ-H)-
(H)Os3(CO)10L initially formed from (µ-H)(H)Os3(CO)10.13b
Detailed NMR studies have shown that such adducts
exist in several different forms.35 Axial-equatorial
isomerization processes have also been shown to occur
with kinetically formed (µ-H)3Re3(CO)11L clusters.36
(2) Th e In tim a te Mech a n ism s. The details of the
stoichiometric processes described in eq 12-14 can most
simply be discussed in terms of Scheme 1. The structure
of Os3(CO)9(µ-C4Ph4)4b,6 can be represented by II. The
carbonyl ligands on Os(3) are arranged so as to complete
an approximately octahedral coordination when the Os-
Os bonds are included, and they are represented simply
by letters for convenience.
The Os(1)-Os(3) bond is the longest of the three Os-
Os bonds4b and is 0.06 Å longer than the average Os-
Os bond distance in Os3(CO)12.37 The Os(1)-Os(2)-
Os(3) angle is also the largest, and it seems reasonable
to conclude that it is the Os(1)-Os(3) bond that breaks
when the adduct is formed. To create an empty coordi-
nation site on Os(3), the Os(1)-Os(3) bond has to break
heterolytically with the electron pair moving to Os(1).
However, this simple charge separation is unlikely to
occur in the nonpolar heptane solvent used here, and
some way of dissipating the charge has to be envisaged.
In the case of associative substition of Os3(CO)12 it is
believed to be accomplished by transformation of two
terminal CO ligands into bridging ones.39
Inspection of the Os3(CO)9(µ-C4Ph4) structure4b,6 sug-
gests, however, that although bridge formation between
Os(3) and Os(2) is sterically quite possible, the region
between Os(2) and Os(1) is very congested. Instead of
Os(2)-Os(1) bridging, the dissipation of charge could
be brought about by means of a structure represented
by III in Scheme 1. This suggestion is supported by the
fact that several closely analogous structures are known
that contain OsfOs bonds where formal donation of a
lone pair occurs from an otherwise six-coordinate d6 Os
atom, or from a five-coordinate d8 Os atom, to another
Os atom.4a,c,38
As shown, the scheme implies that the nucleophile
approaches Os(3) in the plane of the Os3 cluster and
between CO ligands (c) and (d), and it is CO(c) that has
been forced into a bridging position. Exact reversal of
this process will obviously lead to the known re-
formation of the original cluster, but loss of CO(d) will
allow the reaction to proceed. CO(c) resumes its position
trans to the re-formed Os(3)-Os(1) bond, but L ends
up trans to Os(2). This is not the final structure of the
a
For convenience the Os atoms are numbered here with
superscripts, in contrast with the numbering Os(1), etc., in the
text.
product, so isomerization is necessary to enable the final
position of L to be trans to Os(1).
Although speculative, this scheme has the virtue of
explaining the essential features of the reactions in a
simple and reasonable way. More complex versions
could be offered that might explain more detailed
aspects of the data. Thus, other “flight paths” along
which the nucleophile can approach the Os(CO)4 moiety
are possible, and these could lead to various different
isomers of the adduct. There is also a real possibility
that isomerization of the adduct would occur before CO
loss. Intervention of one or more processes of that sort
could provide a mechanistic explanation for the differ-
ences between the equilibrium constants K+L and K+L′′.
If adduct formation does not occur in a single step, then
eq 8 would be an oversimplification of the situation.
There would be more than one equilibrium constant
involved, as well as extra molar extinction coefficients,
and different apparent equilibrium constants would
inevitably result.
(34) The question of whether this CO dissociation is reversible or
not cannot be answered with certainty. The presence of CO leads to a
competing direct reaction with the starting complex, and the effect of
CO on the rates of the reaction in eq 13 would be difficult to study.
(35) Aime, S.; Gobetto, R.; Valls, E. Inorg. Chim. Acta 1998, 275-
276, 521-527.
(36) Beringhelli, T.; D’Alfonso, G.; Monoja, A. P.; Freni, M. Inorg.
Chem. 1991, 30, 2757-2763.
(37) Churchill, M. R.; DeBoer, B. G. Inorg. Chem. 1977, 16, 878-
884.
(38) Davis, H. B.; Einstein, F. W. B.; Glavina, P. G.; J ones, T.;
Pomeroy, R. K. Organometallics 1989, 8, 1030-1039, and references
therein.

Substitution Reactions of Os3(CO)9(µ-C4Ph4)
Organometallics, Vol. 18, No. 26, 1999 5529
En er getics of th e Rea ction s. The most dramatic
feature of the energetics of these reactions is the fact
that the value of k+L for reaction with P-n-Bu3 at 13.6
°C is over 109 times faster than the associative reaction
of P-n-Bu3 with the parent cluster Os3(CO)12 at that
temperature.39 The µ-C4Ph4 moiety clearly has an
enormously labilizing effect that is transmitted to the
Os(CO)4 unit from which the CO is displaced. This can
be related to the length of the Os(1)-Os(3) bond and
the openness of the Os(1)-Os(2)-Os(3) bond angle, as
discussed above, but some of the lability might be
ascribable to electronic effects induced by the osmacy-
clopentadiene moiety. These might facilitate the electron
redistribution that must accompany adduct formation.
However, what the osmacyclopentadiene moiety almost
certainly does not do is labilize the CO ligands on Os(1)
and Os(2) since we have shown that Os2(CO)6(µ-C4Ph4)
is exceedingly inert to substitution.
Beyond this singular behavior lies the fact that the
systematic dependence of rates of various reactions on
the electronic and steric properties of the nucleophiles
is important because it can provide insight into the
intimate energetic natures of the reactions. The param-
eters in Table 5 quantify that dependence for the
reactions discussed here. Not all the ligands involved
conform quantitatively to the simple systematics defined
by the relevant equations, but despite a lack of ready
explanations for the deviants, the systematic behavior
is followed by the great majority of ligands and is, we
believe, significant and useful.
As far as adduct formation is concerned, the standard
reactivity (SR; given by R in eq 9)) for Os3(CO)9(µ-C4-
Ph4) puts it near the top of the ranking of metal carbonyl
clusters. Only Ru5C(CO)15 (SR ) 2.86 ( 0.18)7,13a and
Ru6C(CO)17 (SR ) 1.51 ( 0.26)13a,14c show appreciably
higher values, and these are distinguished by the
presence of the encapsulated C atoms. It is slightly
larger than that of Rh4(CO)9(µ-HC(PPh2)3) (0.64 (
0.01),14e where the tripod ligand presumably introduces
steric strain into the cluster, and it is considerably
larger than that of the still very labile, coordinatively
unsaturated (µ-H)2Os3(CO)10 (-0.53 ( 0.13).13a The
distinguishing features noted are important because
many “undistinguished” clusters such as Rh6(CO)16 or
Os6(CO)18, for instance, are not particularly labile.
Beyond that, these labile clusters are too different for
any detailed conclusions to be drawn, and functional-
ization (by introduction of substituents into the clusters,
for example40) is obviously needed to provide sets of data
that can be more meaningfully compared.
The electronic selectivity, as defined by â in eq 9, is
quite low.14e,40 Values of â can be related to the degree
of bond making as well as to other bonding changes in
the transition states.13b,14c,e The latter may be discour-
aged by increased nucleophile basicity and so lead to
lower â values, but it is also possible that the low
selectivity is simply the result of the very high intrinsic
reactivity of this cluster and that â is mainly a function
of bond making.14e The greater the intrinsic ability of
the cluster to open out spontaneously, the less it will
require the assistance of bond making by the nucleo-
phile to reach the transition state and the less selective
the cluster will be.
Similarly, the onset of steric effects occurs at a rather
intermediate cone angle,14e,40 indicating that ligand-
ligand repulsions in the transition states are only
moderate, as would be expected if metal-nucleophile
bond making itself is only moderate. The flexibility of
the transition state, when the steric threshold is ex-
ceeded, is also quite high,14e,40 as indicated by the fact
that γ has quite a small negative value. All this gives
the picture of a rather loose transition state with
substantial Os-Os bond breaking but relatively little
Os-P bond making, a picture that indicates consider-
able internal stress in the ground-state cluster itself.
The data for loss of the P-donors (eq 13) from the
adducts are unique, only scattered and imprecise data
having been reported elsewhere.7,8,13a The concept of
standard reactivity is not applicable to dissociative
reactions such as these, but the actual rate constants
are very high, matching those of adduct formation.
Apart from this, the loss of L is characterized simply
by a negative value of â, as expected, since Os-L bond
breaking is likely to become more difficult the greater
the basicity of L and the greater the strength of the
Os-L bond. This effect must overcome any effect due
to the increasing electrophilicity of the Os atom as L
leaves, an increase that would favor re-formation of the
Os-Os bond as the cluster closes up again; that is,
Os-L bond breaking is more important than Os-Os
bond making. The absence of any detectable steric
effects is also as expected if steric acceleration, due to
increasing ligand size, is balanced by steric problems
associated with the re-formation of the Os-Os bond.
Inverse steric effects are sometimes seen when forma-
tion, or strengthening, of metal-metal bonds is more
important than metal-ligand bond breaking.12
Because of the use of δ as the electronic parameter,
the units of â for loss of L are ppm-1. However, when â
is made dimensionless (as described above and reported
in Table 5), it is about as negative as the value of â for
adduct formation is positive. In other words, the ener-
getic advantages enjoyed by the more basic ligands on
going to the transition state for adduct formation is
about the same as those experienced on going from the
partially formed Os-P bond in the transition state to
the fully formed Os-P bond in the adduct. In this sense
the transition states are energetically about halfway
between the reactants and the adduct products.
As is the case for loss of L, the rate constants for loss
of CO are very large, and the absolute reactivity is
therefore very high. Although the data in Figure 7 are
quite badly scattered,42 the value of â is clearly negative,
as is that for CO loss from the adducts (µ-H)(H)Os3-
(CO)10L.27 These negative values contrast with the posi-
tive values (up to 0.61 ppm-1) found for Ru3(CO)12-nLn
(n ) 1-3)41 and which indicate transition-state stabi-
lization by better donors, rather than the ground-state
stabilization expected because of stronger Ru-CO bond
(41) Brodie, N. M. H.; Poe¨, A. J . Can. J . Chem. 1995, 73, 1187-
1195.
(42) This can be ascribed to the problem, already alluded to, in
analyzing the data in terms of eq 7. Not only can the values of pKa′ be
strongly affected by data at low [L] (Table 4) but so can the values of
k(2)lim (i.e., k-CO) (Table 2). These data can be affected because at low
[L] the first step will be slower and can overlap with the second.
(39) Poe¨, A. J .; Sekhar, V. Inorg. Chem. 1985, 24, 4376-4380.
(40) Farrar, D. H.; Hao, J .; Mourad, O.; Poe¨, A. J . Organometallics
1997, 16, 5015-5022.

5530 Organometallics, Vol. 18, No. 26, 1999
Poe¨ and Moreno
strengths; that is, the stronger electron donors in some
way compensate the cluster for the loss of the CO ligand
and/or the metal-P bonds become shorter and stronger
because of the release of steric strain. However, when
CO loss is accompanied by metal-metal bond formation,
as in the reactions of the adducts, electronic destabiliza-
tion of the transition state may occur because the re-
formation of the metal-metal bond is inhibited by a
greater electron density on the metal to which the
P-donor is attached and which acts as a electrophile.
The almost complete absence of any steric effects on
the loss of CO shows, as is the case for loss of L, that
relief of steric strain as a ligand leaves is balanced by
repulsions experienced during Os-Os bond re-forma-
tion.
Os-Os bond between the Os atom in the osmacyclo-
pentadiene ring and in the Os(CO)4 group that breaks.
Reversible addition of the nucleophile to the Os(CO)4
unit is followed by CO dissociation from that unit with
concomitant re-formation of the Os-Os bond. The
kinetic product is usually an unstable isomer of the final
product so that isomerization is frequently a third
kinetically observable step.
(iii) Equilibrium constants for adduct formation can
be obtained from the kinetic data for both the first and
second stages and from the extent of absorbance change
in the first stage as a function of [L]. The values
obtained by the different methods are not always the
same. This may be due to experimental problems, but
it is also possible that undetected isomerization of the
adducts before CO loss in the second stage could
contribute to this.
The absence of any substantial ligand dependence of
the rates of isomerization in the final stage of substitu-
tion also suggests that any effects must be small and/
or must balance out in some relatively simple way.
The activation parameters in Table 3 for adduct
formation do not show any particularly obvious trends
(iv) The diene unit in the cluster exerts an enormous
labilizing effect on substitution at the Os(CO)4 moiety,
as shown by the 109 greater rate of substitution by P-n-
Bu3 into this cluster compared with Os3(CO)12.
q
(v) The systematic dependence of the rates of adduct
formation on the σ-donicity and size of 14 of the nu-
cleophiles fit closely to the well-established equation log
k+L ) R + â(pKa′+ 4) + γ(θth - θ)λ.14c,e Size effects do
not become apparent (i.e., λ ) 0) until θ ) θth ) 120 (
4°. The steric effect at larger cone angles (when λ ) 1)
and the always exerted electronic effect are both mod-
erately small and indicate that the cluster is intrinsi-
cally very reactive. The standard reactivity, SR ) R, is
only exceeded by that of Ru6C(CO)17 and Ru5C(CO)15.
(vi) The dissociative loss of L and CO from the adducts
is essentially independent of the size of L but is retarded
by greater net donicity of L. The final isomerization step
is little affected by the size or basicity of L.
(vii) The quantitative sensitivity of the rates to the
donicity and, where appropriate, the size of L allows
consideration of the contributions of bond making and
bond breaking in the transition states, and this can be
supplemented by evidence from the activation param-
eters obtained for five of the ligands. It can be concluded,
for example, that Os-P bond making and Os-Os bond
breaking are both significant in the transition states for
adduct formation and that bond breaking and Os-Os
bond making are both significant in the transition states
for dissociative loss of L or CO from the adducts. It is
also possible to show that the formation of the transition
states on the way to the adducts is as sensitive to the
nature of the nucleophiles as is the subsequent comple-
tion of Os-P bond formation. In that sense the transi-
tion states for adduct formation can be considered to
be approximately halfway from reactant to product.
with pKa′ or θ but they all show the low values of ∆H+L
and negative ∆S+Lq expected for bimolecular reactions.
They also show a rough compensation effect43 when
∆H+L is plotted against ∆S+Lq, but the data are not
q
precise enough to determine if a true isokinetic tem-
perature exists.43 The values of ∆H-Lq are also quite low,
and this probably reflects the significant contribution
of Os-Os bond making that offsets the enthalpy re-
quired for bond breaking. All the entropies are negative,
q
but they become decreasingly so as ∆H-L increases,
probably because of the steadily decreasing importance
of Os-Os bond making that occurs coincidentally with
loss of L. Combination of the activation parameters for
addition and loss of L provide thermodynamic param-
eters for the adduct formation equilibria. Most of the
equilibria are favored by enthalpy and opposed by
entropy effects, as expected from the stoichiometry, but
P(OPh)3 is exceptional, as it is in some of the linear free
energy relationships discussed above. The activation
parameters for loss of CO do not show any perceptibly
regular relationship to those for loss of L, and one of
the pairs shows a high enthalpy and quite favorable
entropy, suggestive of low contribution from Os-Os
bond making. The isomerization reactions vary from
having quite a high and unfavorable enthalpy, with a
compensatingly favorable entropy, to having a low
enthalpy and very unfavorable entropy.
Su m m a r y
(i) The cluster Os3(CO)9(µ-C4Ph4) (I) undergoes very
rapid substitution at the Os(CO)4 moiety, and the
kinetics of reaction with 17 smaller P-donor nucleophiles
(θ e 143°) have been monitored mainly by stopped-flow
methods.
(ii) Reactions occur in at least two stages, and the
absorbance changes and the kinetics of the first stage
indicate that it involves formation of an adduct,
Os3(CO)9L(µ-C4Ph4). The structures of the parent cluster
and of two of the final products suggest that it is the
Ack n ow led gm en t. The support of the Natural
Science and Engineering Research Council, Ottawa, is
gratefully acknowledged, as is The Ministry of Educa-
tion and Science, Madrid, for the award of a fellowship
(to C.M.) that enabled her to visit the University of
Toronto and carry out the work described here. Profes-
sor D. H. Farrar is also thanked for helpful discussions.
Su p p or tin g In for m a tion Ava ila ble: Tables of k(1), k(2),
and k(3) at various values of [L] and temperature are given
in Tables 6-21. This material is available free of charge via
the Internet at http://pubs.acs.org.
(43) Krug, R. R.; Hunter, W. G.; Greiger, R. A. J . Phys. Chem. 1976,
80, 2335-2341 and 2341-2351. Linert, W.; J ameson, R. F. Chem. Soc.
Rev. 1989, 18, 477-505. Linert, W. Chem. Soc. Rev. 1994, 23, 429-
438. Hao, J .; Poe¨, A. J . Transition Met. Chem. 1998, 23, 739-747.
OM9906464