Phosphine ligand exchange in tetrakis(trimethylphosphine)(hydrido)osmium anilides, phenoxides, and thiophenoxides. Examples of anion dissociation and of labilization by ligand π-base effects

Article abstract of DOI:10.1021/om990850r

The series of phenoxide complexes cis-L₄Os(H)OC₆H₄Z (L = PMe₃; Z = H, OMe, CF₃, NH₂, CN), anilides cis-L₄Os(H)NHC₆H₄Z (Z = H, OMe, CF₃), and thiophenoxides cis-L₄Os(H)-SC₆H₄Z (Z = H, OMe) have been prepared by treatment of fac-L₃Os(H)(η²-CH₂PMe₂) with the corresponding neutral arene. ¹H NMR spectra of coordinated phenoxides and thiophenoxides show rapid phenyl group rotation, while that of the anilides is slow. Rates and stereochemistry of substitution of P(CH₃)₃ (L) by P(CD₃)₃ (L′) were determined by ³¹P NMR in benzene at 80°C. Anilides substitute first only in the mutually trans sites (α sites) with cleanly first-order kinetics and subsequently into the site trans to the anilide (site c). The latter shows non-first-order behavior that is accurately modeled by iterative kinetics calculations using a mechanism of L dissociation only from the a sites presumably to give a quasi-trigonal-bipyramidal intermediate stabilized by π-electron donation from the anilide lone pair. A three-point Hammett plot against σ_p yielding ρ = -1.8 is consistent with transition state stabilization by π-donation. Association of L′ occurs only into site α, but subsequent substitution allows the resident L′ to move into site c. Thiophenoxides exhibit substitution rates and stereochemical patterns very similar to those of the anilides and are believed to proceed by the same mechanism. Phenoxide complexes incorporate L′ into all sites, with each site incorporating phosphine independently of the others since exchange rates at all sites are first order. A Hammett plot of exchange rates against sigma minus (σ^-) is somewhat scattered (R² = 0.96) but exhibits a positive slope ρ^- = +0.36. Phenoxide dissociation is postulated, but the fact that substantial concentrations of intermediates partially substituted in all positions is seen during the reaction is inconsistent with rate-determining phenoxide dissociation. An ionization preequilibrium followed by slower phosphine exchange steps in the ion pair is postulated. Treatment of L₄Os(H)(OC₆H₄CN) with excess L in propylene carbonate, DMSO-d₆, DMF, or CD₃NO₂ at 80°C all resulted in conversions to [L₅OsH][OAr]. These results suggest that low steady-state concentrations of [L₅OsH][OAr] ion pairs in benzene are possible, consistent with an ion pair mechanism for ligand exchange.

Full text of DOI:10.1021/om990850r

1166
Organometallics 2000, 19, 1166-1174
P h osp h in e Liga n d Exch a n ge in
Tetr a k is(tr im eth ylp h osp h in e)(h yd r id o)osm iu m An ilid es,
P h en oxid es, a n d Th iop h en oxid es. Exa m p les of An ion
Dissocia tion a n d of La biliza tion by Liga n d π-Ba se Effects
Thomas C. Flood,* J ohn K. Lim, Mark A. Deming, and Walter Keung
Department of Chemistry, University of Southern California,
Los Angeles, California 90089-0744
Received October 22, 1999
The series of phenoxide complexes cis-L4Os(H)OC6H4Z (L ) PMe3; Z ) H, OMe, CF3, NH2,
CN), anilides cis-L4Os(H)NHC6H4Z (Z ) H, OMe, CF3), and thiophenoxides cis-L4Os(H)-
SC6H4Z (Z ) H, OMe) have been prepared by treatment of fac-L3Os(H)(η2-CH2PMe2) with
1
the corresponding neutral arene. H NMR spectra of coordinated phenoxides and thiophen-
oxides show rapid phenyl group rotation, while that of the anilides is slow. Rates and
stereochemistry of substitution of P(CH3)3 (L) by P(CD3)3 (L′) were determined by 31P NMR
in benzene at 80 °C. Anilides substitute first only in the mutually trans sites (a sites) with
cleanly first-order kinetics and subsequently into the site trans to the anilide (site c). The
latter shows non-first-order behavior that is accurately modeled by iterative kinetics
calculations using a mechanism of L dissociation only from the a sites presumably to give
a quasi-trigonal-bipyramidal intermediate stabilized by π-electron donation from the anilide
lone pair. A three-point Hammett plot against σp yielding F ) -1.8 is consistent with
transition state stabilization by π-donation. Association of L′ occurs only into site a, but
subsequent substitution allows the resident L′ to move into site c. Thiophenoxides exhibit
substitution rates and stereochemical patterns very similar to those of the anilides and are
believed to proceed by the same mechanism. Phenoxide complexes incorporate L′ into all
sites, with each site incorporating phosphine independently of the others since exchange
rates at all sites are first order. A Hammett plot of exchange rates against sigma minus
(σ-) is somewhat scattered (R2 ) 0.96) but exhibits a positive slope F- ) +0.36. Phenoxide
dissociation is postulated, but the fact that substantial concentrations of intermediates
partially substituted in all positions is seen during the reaction is inconsistent with rate-
determining phenoxide dissociation. An ionization preequilibrium followed by slower
phosphine exchange steps in the ion pair is postulated. Treatment of L4Os(H)(OC6H4CN)
with excess L in propylene carbonate, DMSO-d6, DMF, or CD3NO2 at 80 °C all resulted in
conversions to [L5OsH][OAr]. These results suggest that low steady-state concentrations of
[L5OsH][OAr] ion pairs in benzene are possible, consistent with an ion pair mechanism for
ligand exchange.
In tr od u ction
and co-workers developed an understanding of cis
labilization in low-valent carbonyl complexes of the Mn3
and Cr4 groups. In the last 15 years a number of papers
have appeared that deal with cis labilization of ligands
and stabilization of unsaturation by π-donor ligands
with late organotransition metal systems, and it is
probably appropriate to say that organometallic chem-
ists are now generally aware of these effects.5,6 Never-
theless, the potential usefulness of π-donor ligands for
Catalysis of reactions of organic substrates by soluble
metal complexes usually depends on the existence or
transient generation of coordinative unsaturation. Be-
cause of the kinetic inertness of third-row platinum
group metals, they were long considered to be unlikely
candidates for homogeneous catalysis. Crabtree’s dis-
covery that iridium can serve as the basis for some of
the most active soluble hydrogenation catalysts known1
called attention to the potential of third-row metals for
efficient catalysis. Nevertheless, the general inertness
of heavy metal complexes remains a pervasive issue for
which any general new approaches could be helpful.
Cis labilization effects of π-lone-pair-donor ligands
have been well-known for many years in the context of
classical coordination chemsitry.2 In the 1970s Brown
(2) (a) Garrick, F. J . Nature 1937, 139, 507. (b) Basolo, F.; Pearson,
R. G. Mechanisms of Inorganic Reactions: A Study of Metal Complexes
in Solution, 2nd ed.; Wiley: New York, NY, 1967; pp 177-193 and
261-265. (c) Tobe, M. L. Acc. Chem. Res. 1770, 3, 377-385. (d) Pearson,
R. G. J . Chem. Educ. 1978, 55, 720-721. (e) Tobe, M. L. Adv. Inorg.
Bioinorg. Mech. 1983, 2, 1-94.
(3) (a) Atwood, J . D.; Brown, T. L. J . Am. Chem. Soc. 1975, 97,
3380-3385. (b) Lichtenberger, D. L.; Brown, T. L. J . Am. Chem. Soc.
1978, 100, 366-373, and references therein.
(4) Atwood, J . D.; Brown, T. L. J . Am. Chem. Soc. 1976, 98, 3160-
3166.
(1) Crabtree, R. H. Acc. Chem. Res. 1979, 12, 331-338.
10.1021/om990850r CCC: $19.00 © 2000 American Chemical Society
Publication on Web 02/19/2000

Phosphine Ligand Exchange
Sch em e 1a
Organometallics, Vol. 19, No. 6, 2000 1167
L4Os(H)Me (2) undergoes PMe3 dissociation at useful
rates above 100 °C (t1/2 ≈ 4 days at 105 °C),9 and L3-
Os(H)(η2-CH2PMe2) (1) exchanges above 140 °C.10 On
the other hand, L4Os(H)(OH) (5) undergoes hydro-
genolysis as shown in Scheme 1 at 80 °C,8 and heating
of 5 with L′ at 80 °C revealed a kinetically competent
rate of exchange of L/L′ in 5 at that temperature (t1/2
≈
2 h). Given that methyl and hydroxyl groups in 2 and
5, respectively, are similar in size, clearly the hydroxyl
group is causing a strong labilization of the phosphine
ligands on osmium.
Motivated by the paucity of specific data concerning
π-donor-induced dissociation, by our hydrogenolysis
results, and by the potential generality and usefulness
of π-basic ligands to facilitate generation of unsatura-
tion, we undertook an examination of the labilization
effect of π-donor ligands on phosphine ligands in cis-
L4Os(H)X complexes. In addition to generating some
useful quantitative comparisons of atom effects and
substituent effects on π-basicity, we have uncovered an
interesting alternative mechanism in this system that
draws attention to the fact that in substitution reactions
the relatively weak σ bonding of oxygen to heavy metals
can result in ionization mechanisms dominating over
π-donor-induced dissociation.11
a
L ) PMe3.
the generation of coordinative unsaturation in organo-
metallics is probably still generally underappreciated.
There are few published quantitative investigations into
the capability of different π-basic ligands to induce
dissociation, particularly for non-carbonyl-containing
complexes. Indeed, we are aware of only one other
investigation that quantitatively compares the relative
π-donicities of O, S, and N in the same system.5c
Our interest in π-donor effects began a number of
years ago as we examined an acid-catalyzed hydro-
genolysis of the Os-C bond in fac-L3Os(H)(η2-CH2PMe2)
(1, L ) PMe3) and in L4Os(H)Me (2) that forms L4OsH2
(3) and involves [L4OsH3]+ as an acid.7 Intermediacy of
coordinatively unsaturated [L4OsH]+, presumably in
equilibrium with a solvated or weak-anion-coordinated
L4Os(H)X, was also inferred, where [L4OsH]+ is gener-
ated by the protonation-induced reductive elimination
Resu lts a n d Discu ssion
Syn th esis. Most late transition metal hydroxide,
alkoxide, and amide complexes have been prepared by
simple ligand metathesis (eq 1) using sodium or lithium
8
of the C-H bond from osmium. We also found that
LnM-X + M′-Y f LnM-Y + M′-X
(1)
water catalyzes a similar reaction, although it proceeds
by a different mechanism and yields L3OsH4 (4) and L
in addition to L4OsH2 (3), as illustrated for L4Os(H)Me
(2) in Scheme 1. Residual water on oven-dried (110 °C)
NMR tubes is sufficient to give very slow rates of
hydrogenolysis. Intermediacy of L4Os(H)(OH) (5) was
inferred from the data. Since 5 is saturated, efficacy of
the catalytic cycle in Scheme 1 requires generation of
an unsaturated derivative. We have previously found
that P(CD3)3 (henceforth L′) is an excellent label with
which to measure rates and stereochemistry of PMe3
dissociation from metal complexes.9 It was shown that
alkoxide or amide salts, especially square-planar species
for which associative substitution is usually fast. How-
ever, this technique was not effective with the inert Os-
(II) complexes of interest to us. It happened that the
availability of fac-L3Os(H)(η2-CH2PMe2) (1)12 offered a
general and effective direct route to cis-L4Os(H)X by
protonolysis (eq 2). Such acid cleavages of metal-carbon
(5) (a) Mayer, J . M. Comments Inorg. Chem. 1988, 8, 125-135. (b)
Amarasekera, J .; Rauchfuss, T. B.; Wilson, S. R. J . Chem. Soc., Chem.
Commun. 1989, 14-16. (c) Bryndza, H. E.; Domaille, P. J .; Paciello,
R. A.; Bercaw, J . E. Organometallics 1989, 8, 379-385. (d) Dewey, M.
A.; Gladysz, J . A. Organometallics 1990, 9, 1351-1352. (e) Lunder, D.
M.; Lobkovsky, E. B.; Streib, W. E.; Caulton, K. G. J . Am. Chem. Soc.
1991, 113, 1837-1838. (f) Hoffman, N. W.; Prokopuk, N.; Robbins, M.
J .; J ones, C. M.; Doherty, N. M. Inorg. Chem. 1991, 30, 4177-4181.
(g) Hubbard, J . L.; McVicar, W. K. Inorg. Chem. 1992, 31, 910-913.
(h) Saura-Llamas, J .; Gladysz, J . A. J . Am. Chem. Soc. 1992, 114,
2136-2144. (i) Darensbourg, D. J .; Klausmeyer, K. K.; Mueller, B. L.;
Reibenspies, J . H. Angew. Chem., Int. Ed. Engl. 1992, 31, 1503-1504.
(j) J ohnson, T. J .; Huffman, J . C.; Caulton, K. G. J . Am. Chem. Soc.
1992, 114, 2725-2726. (k) Poulton, J . T.; Folting, K.; Streib, W. E.;
Caulton, K. G. Inorg. Chem. 1992, 31, 3190-3191.
(6) Some reviews of late metal complexes bearing π-donor ligands:
(a) Mehrotra, R. C.; Agarwal, S. K.; Singh, Y. P. Coord. Chem. Rev.
1985, 68, 101-130. (b) Bryndza, H. E.; Tam, W. Chem. Rev. 1988, 88,
1163-1188. (c) Fryzuk, M. D.; Montgomery, C. D. Coord. Chem. Rev.
1989, 95, 1-40. (d) Doherty, N. M.; Hoffman, N. W. Chem. Rev. 1991,
91, 553-573.
bonds have been known for many years and have been
particularly useful for generation of complexes of weakly
coordinating anions.13 All of the heteroatom complexes
listed in eq 2 were prepared by reacting highly basic 1
with slightly less than 1 equiv of the appropriate acid
(7) Desrosiers, P. J .; Shinomoto, R. S.; Deming, M. A.; Flood, T. C.
Organometallics 1989, 8, 2861-2865.
(8) Desrosiers, P. J . Ph.D. Thesis, University of Southern California,
1985. Deming, M. A. Ph.D. Thesis, University of Southern California,
1992.
(9) Desrosiers, P. J .; Shinomoto, R. S.; Flood, T. C. J . Am. Chem.
Soc. 1986, 108, 7964-7970.
(10) Shinomoto, R. S.; Desrosiers, P. J .; Harper, T. G. P.; Flood, T.
C. J . Am. Chem. Soc. 1990, 112, 704-713.

1168 Organometallics, Vol. 19, No. 6, 2000
Flood et al.
(phenol, aniline, or thiophenol) in benzene or THF
solvent followed by heating of the solution (for the
anilides) in a sealed tube. The severity of conditions
required is inversely proportional to the acidity of the
substrate in that phenols and thiophenols react with 1
during the time of mixing, while anilines require heat-
ing for several days at 110 °C. Among the anilines, the
rate parallels the effect of the para substituent on its
acidity. For example, at 110 °C formation of cis-L4Os-
(H)(p-HNC6H4OMe) required 80 h and cis-L4Os(H)(p-
HNC6H4CF3) 40 h to complete. The products are ther-
molytically inert, with no changes in their 1H and 31P
NMR spectra after several days at 110 °C.
observed at ca. 100 °C. Four distinct aromatic proton
resonances are also evident in the spectra of 7b and 7c
at ambient temperature. This restricted rotation is
likely of steric origin since we assume that the presence
of the N-H group substantially restricts the ability of
the Os-N-Ar bond angle to deform toward linearity
compared to Os-O-Ar and Os-S-Ar, so the anilide
phenyl ring would be pushed more tightly against the
phosphines.
P h osp h in e Liga n d Exch a n ge Stu d ies. 31P NMR
Sp ectr a l P a tter n Up on P (CH3)3/P (CD3)3 Exch a n ge.
The lability of phosphine toward dissociation and the
resulting substitution pattern in 6-8 were examined
by the use of deuterium-labeled phosphine P(CD3)3 (L′).
Employing a large excess of L′ ensures that any dis-
sociation event leads to incorporation of the label in
place of the normal L, and so the rate of dissociation is
clearly measurable. If dissociation of L is preferred from
a specific site, then the principle of microscopic revers-
ibility requires that the labeled ligand L′ reenter the
same site, thus revealing the preference. These substi-
tutional principles have been well discussed by Atwood
and Brown.14,15 The essential experiment in the cis-L4-
OsXY system has been previously described in more
detail,9 and additional examples in osmium,10 iridium,16
and ruthenium5c chemistry have been given. A major
advantage of using P(CH3)3/P(CD3)3 labeling is that
there is a 3 ppm difference in the 31P NMR chemical
shifts of the free ligands L (δ -60.2) and L′ (δ -63.2)
and usually ca. a 2 ppm difference when they are
coordinated, with M-L′ also at higher field than M-L.
This makes the determination of the labeling results
relatively straightforward. The mutually trans LaLa are
Sp ectr a . NMR spectroscopy is naturally the tech-
nique of choice for characterizing and following the
chemistry of complexes 6-8, with 31P NMR the most
informative regarding PMe3 (L) lability. The 31P{1H}
NMR spectra of cis-L4Os(H)(XC6H4Z) compounds exhibit
an A2MX, A2XY, or A2BX pattern, where the resonances
for the mutually trans phosphines (site a, hereafter
designated La) are invariably at lowest field. The
chemical shifts are characteristic of the coordinating
heteroatom X, but they vary little with the para sub-
stituent Z for a given X, differences being less than (1
ppm for each phosphine site. Phenoxides 6 have the
lowest field chemical shifts, with average δ at -36.5,
-41.8, and -50.3 for mutually trans L (La), L trans to
OsH (Lb), and L cis to OsH (Lc), respectively (see eq 2
for site labels). 31P resonances for anilides 7 appear at
higher field, with average δ at -41.8 (La), -50.6 (Lb),
and -52 (Lc). Thiophenoxides 8 appear at higher field
still, with the relative chemical shift of the two mutually
cis L reversed in order from that of 6 and 7: average δ
at -50 (La), -51.2 (Lc), and -54.1 (Lb). In general, the
two mutually cis L are distinguished by a selective off-
resonance decoupling experiment that takes advantage
of the fact that the phosphine located trans to the
hydride (Lb) is strongly coupled with it (2J PH 65-80 Hz),
2
strongly coupled with J PP ≈ 270 Hz. Normally this
coupling is not visible since the two P have the same
chemical shift, but a single substitution of one of the
trans phosphines gives L and L′ resonances that differ
2
by ca. 2 ppm. The large J PP thus results in an AB
2
while L cis to the hydride (Lc) has a smaller J PH (20-
pattern for the mutually trans LaLa′ phosphines span-
ning ca. 5 ppm (at 146 MHz) and superimposed on the
disappearing LaLa resonance and the growing La′La′
resonance. This region of the 31P spectrum is shown in
Figure 1 for (L/L′)4Os(H)(OC6H4OMe) (6b). In addition,
incorporation of L′ causes a small downfield isotopic
shift (ca. 5 Hz at 146 MHz for 31P) in the 31P resonances
of any other L that is cis to it. This allows one to tell
whether any exchange in the mutually cis sites (Lb and/
or Lc) occurs in addition to La exchange. Figure 2, for
example, shows the LaLa doublet of doublets at δ -42.0
ppm and the more intense multiplet centered at δ -42.6
ppm of the A half of the LaLa′ AB pattern of partially
L/L′ exchanged L4Os(H)(NHC6H5) (7a ). The multiplet
at δ -42.6 ppm is seen to have a second doublet-of-
doublet pattern of ca. 90% the intensity of the first at
ca. 5 Hz lower field corresponding to part of the sample
that is doubly exchanged, i.e., (LaLa′)(Lb)(Lc′)Os(H)-
(NHC6H5). The absence of any doubling of the δ -42.0
ppm peak clearly indicates the absence of any (La)2(Lb)-
(Lc′)Os(H)(NHC6H5) or (La)2(Lb′)(Lc)Os(H)(NHC6H5).
1
25 Hz). Shifting of the H Rf decoupling band to lower
frequency decouples any carbon-bonded hydrogen from
P, but leaves the OsH coupled, although with a slightly
2
diminished J PH compared to that measured in the
proton spectrum.
The Os-H 1H NMR resonances, which appear as
doublets of quartets, also progress to higher field across
the series, with average δ for 6 at -7.9, 7 at -8.5, and
8 at -9.5 ppm. The NH2 resonance of uncoordinated
anilines, a broad singlet at ca. 2.7 ppm, becomes an
anilide NH doublet (3J PH ≈ 6 Hz) centered from 2.3 ppm
(Z ) H) to 1.9 ppm (Z ) OCH3) in the coordinated,
deprotonated anilide of L4Os(H)NHC6H4Z. In contrast
to phenoxides 6 and thiophenoxides 8, anilides 7 exhibit
restricted rotation of the aromatic ring on the NMR time
scale. Thus, all five aromatic resonances are apparent
1
in the H NMR spectrum of cis-L4Os(H)(HNC6H5) (7a ),
and in variable-temperature spectra, coalescence is
(11) The synthesis and some aspects of the chemistry of the
ruthenium analogues of our phenoxides and anilides, (Me3P)4Ru(H)-
(XC6H4Z) (X ) NH, O), have been reported. (a) Osakada, K.; Ohshiro,
K.; Yamamoto, A. Organometallics 1991, 10, 404-410. (b) Hartwig, J .
F.; Andersen, R. A.; Bergman, R. G. Organometallics 1991, 10, 1875-
1887.
(14) Atwood, J . D.; Brown, T. L. J . Am. Chem. Soc. 1976, 98, 3155-
3159.
(12) Werner, H.; Gotzig, J . Organometallics 1983, 2, 547-549.
(13) (a) Beck, W.; Sunkel, K. Chem. Rev. 1988, 88, 1405-1421. (b)
Strauss, S. H. Chem. Rev. 1993, 93, 927-942.
(15) Brown, T. L. Inorg. Chem. 1989, 28, 3229-3230.
(16) Harper, T. G. P.; Desrosiers, P. J .; Flood, T. C. Organometallics
1990, 9, 2523-2528.

Phosphine Ligand Exchange
Organometallics, Vol. 19, No. 6, 2000 1169
Ta ble 1. Ra tes of Dia sa p p ea r a n ce of P (CH3)3 (L)
13P {1H} Reson a n ces a s Th ey Ar e Rep la ced by
P (CD3)3 (L′) in Sp ecified Sites of cis-L4Os(H)(X) in
C6D6 a t 80 °C
a
cmpd
X
107 × ka
107 × kb
107 × kc
6d
6a
6b
6c
6e
7b
7a
7c
8b
8a
OC6H4-NH2
OC6H4-H
14.8
17.4
20.2
37.8
51.1
5860
2110
225
4.69
4.86
4.70
9.74
14.0
slow
slow
slow
slow
slow
1.53
6.29
5.05
12.2
17.2
OC6H4-OMe
OC6H4-CF3
OC6H4-CN
NHC6H4-OMe
NHC6H4-H
NHC6H4-CF3
SC6H4-OMe
SC6H5
(1300)b
(410)b
(50)b
(210)b
(130)b
1997
481
a
Constant for rate of first substitution into mutually trans
b
positions. Behavior is not first order (see text). Rate constant is
an approximation to first-order behavior given for comparison.
F igu r e 1. Partial 31P{1H} NMR spectrum of ca. half-
P(CH3)3/P(CD3)3-exchanged (L/L′)4Os(H)OC6H4OMe (6b ).
Shown is the region of the mutually trans ligand reso-
nances LaLa, LaLa′, and La′La′. See the text for interpreta-
tions.
independent Lb and/or Lc dissociation is occurring. This
turns out to be a useful phenomenon, as described
below.
First-order rate constants for L/L′ exchange in indi-
vidual sites are shown in Table 1. In the case of the
anilides (7) and sulfides (8), the Lc resonance disappears
in a kinetically more complicated way (see below), so
the site Lc constants, given for comparison, are from an
approximate (but incorrect) fit to a first-order plot.
An ilid es: L4Os(H)(NHC6H4Z) (7). Phosphine ex-
change in the anilide compounds appears to proceed via
phosphine dissociation that is facilitated by the π-basic-
ity of the anilido group. Dissociation occurs at the
mutually trans phosphine a sites only, and then label
subsequently migrates into the c site (cis to the hydride).
The basis of this conclusion can be seen in Figure 2,
where the resonance at δ 42.0 ppm of the unexchanged
LaLa site of 7a is a clean doublet of doublets with no
downfield doubling from isotopic labeling in the c site;
that is, (LaLa)(Lb)(Lc′)Os(H)(NHC6H5) is not present and
none is detectable even when the LaLa δ 42.0 resonance
is nearly gone. This is in contrast to the previously
discussed LaLa′ multiplet of 7a at δ 42.6 ppm, which is
clearly doubled, indicating the presence of a mixture of
(LaLa′)(Lb)(Lc)Os(H)(NHC6H5) and (LaLa′)(Lb)(Lc′)Os(H)-
(NHC6H5).
Further evidence for the requirement for the presence
of La′ prior to Lc′ incorporation comes from interpreta-
tion of the rates of exchange at specific sites. Figure 3
shows the time dependence of the integrals of the LaLa
and Lc phosphine resonances, where symbols are the
experimental data, and the lines are calculated using
an iterative kinetics modeling program developed by
Weigert.17 It is assumed for the calculation that the
intermediate is a distorted trigonal bipyramid with an
equatorial amido ligand and that dissociation occurs
only from the a sites. These suppositions are illustrated
in Scheme 2. Starting 9a dissociates L from site a with
simultaneous reorganization of the two remaining equa-
torial L to form a distorted trigonal bipyramid, 10a . This
distortion assumes that the amido lone pair of electrons
is oriented so as to lie in the equatorial plane as shown
in structure 11. CPK space-filling models suggest there
should be a steric preference for the amide aryl group
F igu r e 2. Partial 31P{1H} NMR spectrum of partially-(L/
L′)-exchanged L4Os(H)NHC6H5 (7a ). Shown is the LaLa
resonance and the most intense of the four LaLa′ AB
resonances. See the text for interpretations.
An important complication in these phosphine ex-
change reactions is that a five-coordinate intermediate
may have a square-pyramidal or trigonal-bipyramidal
structure or any structure between these, and/or it may
be stereochemically labile by nondissociative reorgani-
zation (pseudorotation, more below). Thus, it is possible
that mutually cis sites b and/or c could acquire label by
partial reorganization or more extensive pseudorotation
of the intermediate arising from LaLa dissociation, even
if no Lb or Lc dissociation would occur. However, if
dissociation of Lb and Lc did not occur, then the only
way that L′ could appear in either of those sites would
be to first have exchange into site a to form LaLa′ and
then to have the remaining La dissociate, whereupon
the coordinated La′ could migrate to site b or c in the
intermediate. An important point here is that since free
L′ is in large excess, only L′ will coordinate to the
pentacoordinate intermediate to form product. As a
result, no L′ can appear in sites b or c unless it also
appears in at least one of the a sites as well, unless site
b or c itself dissociates ligand at a competitive rate.
Stated another way, no isotopically shifted unlabeled
LaLa resonance will appear ca. 5 Hz downfield unless
(17) (a) “Gear”, copyright 1995 by F. J . Weigert. (b) Stabler R. N.;
Chesick, J . Int. J . Kinet. 1978, 10, 461-469. (c) Weigert, F. J . Comput.
Chem. 1987, 11, 273.

1170 Organometallics, Vol. 19, No. 6, 2000
Flood et al.
F igu r e 4. Log of the rate constant for replacement of the
first L by L′ at site LaLa in L4Os(H)(NHC6H4Z) (6, Z ) OMe,
H, CF3) plotted against the Hammett sigma (σ). The slope
(F) is -1.8.
F igu r e 3. Plot of the relative 31P NMR integrals of
unlabeled L in sites LaLa and Lc vs time in the exchange
of L for L′ into L4Os(H)(NHC6H4OMe) (7b). Symbols are
data, and lines are calculated (see text).
9b. Dissociation of the other La now forms 10b, the
labeled equivalent of 10a , and entry of L′ from behind
the page into the cis position as shown in Scheme 2
forms 9c and from the front gives 9d . Use of this model
in the iterative kinetic calculations gave a good visual
fit to the non-first-order concentration dependence of
disappearance of the Lc 31P resonance shown in Figure
3. None of the processes shown in Scheme 1 lead to label
incorporation into position b, so other much slower
fluxionality of the unsaturated intermediates or perhaps
a much slower direct dissociation from site b accounts
for the eventual complete exchange of all positions in
the amido complexes.
Sch em e 2
Figure 4 shows a Hammett plot of the para-substi-
tuted amido series with a good (albeit 3-point) linear
relationship (Fp ) -1.8), where the electron-donating
substituent increases the rate of phosphine ligand
dissociation. The relative rates of total phosphine dis-
sociation are greater for the amides than for the
phenoxides, consistent with the greater basicity of
nitrogen (Table 1). The basicity of nitrogen also miti-
gates against dissociation of anilide anion, which would
lead to the type of ion pair mechanism by which the
phenolate complexes appear to react (see below).
Th iop h en oxid es: L4Os(H)(SC6H4Z) (8). Phosphine
ligand exchange at 80 °C in 8a and 8b occurs at the a
and c sites, although the exchange at the former is
faster, with the rates of exchange comparable to those
of the anilides. Quantifying the exchange kinetics is
more difficult than with the phenoxides or amides
because the 31P NMR site a resonances overlap slightly
with the Lc resonance. Only the two derivatives 8a and
8b were prepared and rates measured, but the ratio of
8 site a:b:c exchange rates (3:slow:1) is essentially the
same as for anilides 7 (4:slow:1), suggesting a similar
mechanism for ligand labilization. Examination of the
LaLa triplet resonance of the A2MX pattern in the 31P-
{1H} NMR spectra shows no isotopic doubling, so
exchange into the c site does not occur independently.
Thus, an a phosphine must be substituted before label
is incorporated into the c site, and the most probable
interpretation is the same as that just described for the
anilides.
to be oriented toward the hydride in starting materials
7, with the result of aligning the nitrogen lone pair with
the mutually trans phosphine ligand axis as depicted
in structure 11. Presumably such an alignment should
direct π-symmetry overlap of the lone pair with the
developing vacant osmium orbital aligned along the
P-Os-P axis upon dissociation of one of the mutually
trans phosphine ligands, thus stabilizing the transition
state for its dissociation. This combination of sterically
based orientation and single-faced π-bonding account
for the stereospecificity of substitution in the anilido
complexes. The picture of heteroatom π-donation sta-
bilizing the dissociation transition state and the dis-
sociated intermediate and the geometry change illus-
trated in structure 11 above are both amply corroborated
by theoretical calculations at several levels.3b,18 Now,
when L′ coordinates with 10a , the principle of micro-
scopic reversibility requires that it enter cis to X to form
The π-bonding nature of metal thiolates is well-
known.19 One of the lone pairs is in an 3sp2 orbital, and
the other is in an essentially 3p orbital. Since an sp2
orbital is more electronegative than a p orbital, the sp2
electron pair is less basic than the p pair. Furthermore,
(18) (a) Davy, R. D.; Hall, M. B. Inorg. Chem. 1989, 28, 3524-3529.
(b) Riehl, J .-F.; J ean, Y.; Eisenstein, O.; Pelissier, M. Organometallics
1992, 11, 729-737

Phosphine Ligand Exchange
Organometallics, Vol. 19, No. 6, 2000 1171
the sp2 pair is pointed away from the metal, while the
p pair is aligned optimally for π-interaction with the
metal. The phenyl group should prefer to orient itself
toward the hydride site just as would be supposed for
the anilides or phenoxides, and so the p electron pair
would be aligned along the mutually trans phosphine
P-M-P axis, accounting for the stereochemical prefer-
ence for their dissociation. Thus the mechanism of
labilization and stereospecificity are believed to be the
same as those discussed above for the amide complexes.
In view of the fact that complexes 6 undergo L/L′
exchange by phenoxide ionization (see below), it is
interesting to consider why the thiophenoxide complexes
should exchange via a lone pair-induced phosphine
dissociation instead. The heterolytic cleavage of the Os-
OPh and Os-SPh bonds can be considered as the sum
of the specific bond dissociation energy (SBDE) for
homolysis, the ionization potential (IP) of the osmium
radical in the given solvent, and the electron affinity
(EA) of the OPh and SPh radicals in the given solvent.
From the literature,20 the gas-phase EA of phenoxy
radical equals 52.0 kcal/mol and that of thiophenoxy
radical is 57 kcal/mol. From other literature data21 one
can calculate the aqueous phase EA, both of which are
close to 134 kcal/mol. In organic solvents one may
estimate that the difference in EA is therefore between
0 and 5 kcal/mol. A lower limit for the difference in
homolytic SBDE of Cp*Ru(PMe3)2-X (X ) SH, OH) has
been estimated to be 18.5 kcal/mol.22 Assuming that the
heats of solution of PhOH, PhSH, PhO•, PhS•, and H•
are all small in benzene or THF and that their differ-
ences are therefore negligible, and assuming that the
difference in bond energies of Os-S and Os-O is at
least as large as for ruthenium (presumably it would
be larger), one finds that the lower limit for the
difference in ionization energy is about 13 kcal/mol.
Thus, very probably the intrinsic Os-S bond strength
slows down the intervention of an ion pair mechanism,
and the greater π-basicity of sulfur toward osmium
accelerates the L dissociation path compared to oxygen,
accounting for the mechanistic preference of 8 versus
6.
F igu r e 5. Plot of the relative 31P NMR integrals of
unlabeled L in sites LaLa, Lb, and Lc vs time in the exchange
of L for L′ into L4Os(H)(OC6H4OMe) (6b). Symbols are data,
and lines are calculated for a first-order decay.
a substantial portion of the 31P resonance of the
unlabeled mutually trans LaLa in L4Os(H)(OAr) is
shifted ca. 5 Hz downfield. As shown in Figure 1 for 6b,
the doublet-of-doublets for unexchanged LaLa at δ -35.9
ppm is doubled again to a total of eight peaks by the
deuterium isotope effect of Lb′ or Lc′ incorporation. As
Lb′ or Lc′ incorporation progresses, the isotopically
shifted LaLa multiplet becomes more dominant. Thus,
independent exchange of L′ into either sites Lb or Lc or
both is occurring without necessary prior La dissociation.
Late in the exchange reaction a third set of peaks begins
to form within the LaLa′ resonance, probably corre-
sponding to the additive cis isotope effect in (LaLa′)(Lb′)-
(Lc′)Os(H)(OC6H4OMe), but no tripling of the LaLa
resonance occurs, indicating that no detectable (LaLa)-
(Lb′)(Lc′)Os(H)(OC6H4OMe) forms. This observation is
reasonable on the basis of probability. A second indica-
tion that substitutions at sites b and c do not require
prior site a exchange comes form the fact that L/L′
exchange rates at all sites are first order, as illustrated
for L4Os(H)(OC6H4OMe), 6b, in Figure 5. As shown for
the anilido complexes above, substitutions that require
prior labeling in other parts of the molecule tend not to
exhibit first-order behavior.
A plot of the log of the sum of rate constants for
disappearance of L resonances corresponding to sites
LaLa, Lb, and Lc in L4Os(H)(OC6H4Z) (6, Z ) NH2, OMe,
H, CF3, CN) versus the Hammett sigma (σp) is some-
what scattered (R2 ) 0.93) with Fp ) +0.42. There is a
slight improvement on plotting the exchange rate data
against sigma minus (σ-): F- ) +0.36 (R2 ) 0.96)
(Figure 6). In any event, it is clear that the magnitude
of the slope is small. The observations that the correla-
tion is marginally better with σ- and the slope is
positive suggest that phenoxide character exists in the
transition state of the rate-determining step of phos-
phine exchange. The most consistent interpretation is
that phenoxide dissociation is occurring, although the
stereochemical data are inconsistent with rate-deter-
mining phenoxide dissociation (more below). The small
size of F- is puzzling since an ionization preequilibrium
followed by slower phosphine exchange steps should
show the full equilibrium effect of the phenoxide stabili-
ties on the slope of the Hammett plot. If the ionization
mechanism does operate, then it must be that the size
of the substituent effects is somehow attenuated in ion
P h en oxid es: L4Os(H)(OC6H4Z) (6). On heating
each para-substituted phenoxide compound at 80 °C
with free L′ in benzene-d6, labeled phosphine is incor-
porated into all three sites, with each site apparently
incorporating phosphine independently of the others.
The conclusion that some sites are not exchanged first
as a prerequisite for others to substitute comes from two
lines of evidence. First, the 31P NMR spectra show the
isotopic doubling phenomenon discussed above. That is,
(19) (a) Ashby, M. T.; Enemark, J . H. J . Am. Chem. Soc. 1986, 108,
730-733. (b) Ashby, M. T.; Enemark, J . H.; Lichtenberger, D. L. Inorg.
Chem. 1988, 27, 191-197. (c) Ashby, M. T. Comments Inorg. Chem.
1990, 10, 297-313.
(20) Handbook of Chemistry and Physics, 80th ed.; Lide, D. R., Ed.;
CRC Press: Boca Raton, FL, 1999; pp 10-153.
(21) Some literature data (kcal/mol): (E ) O, S); pKa of PhEHa
)
10.0, 6.5; aqueous ∆Hionization of PhEH ) 5.6,b 2 (estimated); SBDE of
PhE-Hc ) 86.5, 83.3; gas-phase IP of H• ) 313.6;d ∆Hhydration of H(+)
(g)
)
) -261.e From the last two numbers, the aqueous phase IP of H•
53, and from these data the numbers in the text can be calculated. (a)
Bordwell, F. G. Acc. Chem. Res. 1988, 21, 456-462. (b) Isaacs, N. S.
Physical Organic Chemistry, 2nd ed.; Longman Scientific and Techni-
cal: Essex, UK, 1995; p 239. (c) Handbook of Chemistry and Physics,
80th ed.; op cit.; pp 9-65. (d) Ibid., pp 10-162. (e) Atkins, P. W.
Physical Chemistry, 5th ed.; W. H. Freeman: New York, 1994; p 88.
(22) Bryndza, H. E.; Fong, L. K.; Paciello, R. A.; Tam, W.; Bercaw,
J . E. J . Am. Chem. Soc. 1987, 109, 1444-1456.

1172 Organometallics, Vol. 19, No. 6, 2000
Flood et al.
Sch em e 3
F igu r e 6. Log of the sum of rate constants at sites LaLa,
Lb, and Lc for replacement of L by L′ in L4Os(H)(OC6H4Z)
(6, Z ) NH2, OMe, H, CF3, CN) plotted against the
Hammett sigma minus (σ-): F- ) +0.36.
pairs in benzene relative to free phenoxide ions in water.
Substitution reactions of organometallic complexes that
involve dissociation of OH, OR, or OAr have been
postulated by others.23
ion pair, 12 f 9, is much faster than for L′ uptake, 12
f 13, guaranteeing a fast preequilibrium; (3) interme-
diate 12 is a square pyramid; (4) coordination of L′ to
form 13 and dissociation of L/L′ from 13 occur only in
the radial plane; (5) the only fluxionality of intermediate
12 is the hydride shift equilibration 12c a 12d , which
is extremely rapid; (6) L′ is in large excess so that only
it coordinates. Assumption (5) is reasonable since ambi-
ent-temperature 31P NMR spectra of L4OsH(OTf) reveal
rapid averaging of the Lb and Lc resonances even in a
solvent as coordinating as THF, while the LaLa reso-
nance remains sharp.7 It has also been shown that the
equilibrium of eq 3 is established in less than 4 h at 80
The rates of label incorporation into individual sites
b and c are directly measurable, but the precise rate
constant for incorporation into the two a sites is not as
straightforward. The LaLa doublet of doublets contains
the intensity of two phosphorus nuclei, but a single L′
substitution in site a moves the intensity of both away
from the LaLa chemical shift and into the La and La′
resonances of the second-order AB resonance set. Thus,
the disappearance of the LaLa resonance is exactly the
rate of the first substitution into the a site. However,
since there are two identical sites from which L can
dissociate, the rate constant for disappearance of the
LaLa resonance would be twice that of a corresponding
single site substitution in a mechanism that proceeds
by substitution one site at a time. If all sites had the
same rate constant for phosphine dissociation and the
intermediate was a rigid square pyramid, for example,
then the ratio of rates for disappearance of phosphorus
resonances LaLa (excluding LaLa′)/Lb/Lc would be 2:1:1.
In contrast, if an ion pair mechanism was operating
with rate-determining ionization (ki ) followed by fast
L/L′ exchange, every ionization would exchange all four
sites and kL L ) kL ) kL ) ki and LaLa/Lb/Lc would be
7
°C, so the phosphine dissociation step is kinetically
competent for the present mechanism. Using these
assumptions and five rate constants (see Experimental
Section), calculated rate ratios were kL L /kL /kL
)
a
a
b
c
1:0.36:0.36. Obviously, this is not a unique solution for
this complex system, but it is one model and one solution
that is sufficiently close to the experimentally observed
data to corroborate the reasonableness of the ionic
mechanism.
a
a
b
c
1:1:1. In addition, no buildup of any partially substituted
intermediates would be seen if ionization was fully rate
determining. In fact, this is contrary to observation since
substantial concentrations of intermediates partially
substituted in all positions are seen during the reaction.
The observed relative rates of substitution of label into
each of the three sites kL L /kL /kL , averaged over the
An ion pair mechanism would be expected to show a
dependence on solvent polarity, so several other solvents
were investigated. Treatment of L4Os(H)(OC6H4CN)
(6e) with excess L in propylene carbonate (PPC, ꢀ )
64.4), DMSO-d6 (ꢀ ) 46.7), DMF (ꢀ ) 36.7), and CD3-
NO2 (ꢀ ) 35.9) over a period of hours at 80 °C all
resulted in high-yield conversions to [L5OsH][OAr], with
31P{1H} NMR resonances at approximately δ -56 (d,
4P, J PP ) 17 Hz) and -61 (pent, 1P).24 Solvent effects
on the chemical shifts were small. Use of excess L′ led
to a complicated spectrum for [(L/L′)5OsH][OAr] because
of the L/L′ chemical shift difference. Even o-C6H4Cl2 (ꢀ
) 9.9) afforded a mixture of 22% of L4Os(H)(OAr) and
78% of [L5OsH][OAr] from 6e and excess L after heating
at 80 °C for 23 h. In THF (ꢀ ) 7.6) and in C6D6 (ꢀ ) 2.3)
at 80 °C the salt was not formed to a great extent, but
a
a
b
c
five complexes 6a -6e, are approximately 1/0.27/0.27,
but the individual relative b and c rates vary between
0.10 and 0.36 with no obvious trends as a function of
para substituent.
The kinetics program was used to calculate time
versus concentration data for a mechanistic model, part
of which is shown in Scheme 3. The assumptions were
as follows: (1) the first step is ionization of phenoxide,
9 f 12; (2) the rate constant for reassociation of the
(23) (a) Rees, W. M.; Churchill, M. R.; Fettinger, J . C.; Atwood, J .
D. Organometallics 1985, 4, 2179-2185. (b) Milstein, D.; Calabrese,
J . C.; Williams, I. D. J . Am. Chem. Soc. 1986, 108, 6387-6389. (c)
Simpson, R. D.; Bergman, R. G. Organometallics 1993, 12, 781-796.
(24) Ermer, S. P.; Shinomoto, R. S.; Deming, M. A.; Flood, T. C.
Organometallics 1989, 8, 1377-1378.

Phosphine Ligand Exchange
Organometallics, Vol. 19, No. 6, 2000 1173
techniques. Manipulations were carried out using an inert
atmosphere glovebox, Schlenk apparatus, or vacuum line
techniques with argon or dinitrogen which was passed through
an oxygen scrubber (BASF Ridox R3-11) and water absorbent
(Mallinckrodt Aquasorb). Solvents were reagent grade with
further purification as follows. Hydrocarbon solvents (alkane
and aromatic) were stirred over concentrated sulfuric acid
(repeated with fresh acid until the acid phase remained
colorless), stirred over calcium hydride (24 h), distilled onto
sodium benzophenone, stirred at reflux until dark blue, or
preferably, deep purple, and distilled from this immediately
before use. THF and diethyl ether from fresh vendors’ contain-
ers were dried over calcium hydride (24 h) and were treated
the same as the hydrocarbon solvents. Deuterated solvents in
vendors’ sealed glass ampules were transferred and stored in
the glovebox over highly activated (6 h at 450 °C in vacuo)
type 4A molecular sieves in a gastight, screw-cap vial and used
without further purification. NMR reference scales used are
all δ with respect to TMS for 1H and 13C, 85% H3PO4 for 31P,
and CFCl3 for 19F NMR. All of the aromatic substrates (para-
substituted phenols, anilines, and thiols) were commercial.
Solids were sublimed and liquids were distilled prior to use.
Elemental analyses were carried out by Galbraith Laborato-
ries, Inc. (Knoxville, TN).
a very small doublet was observed in both solvents at δ
-55.9 ppm, which corresponds to the resonance of the
four radial phosphines in [L5OsH]+: the axial pentet
was apparently too weak to be observed in the noise.
Nevertheless, it is a reasonable supposition that the ion
pair [L5OsH][OAr] was present in low steady-state
concentrations in benzene and THF.
It is possible that a component of the overall substitu-
tion reaction might proceed via phosphine dissociation.
Although one would expect the same stereospecific
behavior by this mechanism that is seen in the anilides
and thiophenoxides, it is at least possible that the
phenoxy complexes might exhibit less stereospecificity
in substitution by L′. For example, the π-basicity of the
phenoxy group is substantially less than that of the
anilides and probably the thiophenoxides, so there
might be less rate acceleration and less control of the
stereochemistry by the phenoxy groups. A phenoxide-
assisted phosphine dissociation mechanism should show
a negative Hammett F, and so a component of this
mechanism in combination with positive F- of the
postulated ionization mechanism could help to account
for the small size of F- discussed above. Unfortunately,
the possibility of a contribution by the phosphine
dissociation mechanism cannot be evaluated from the
extant data.
L4Os(H)(OC6H5) (6a ). Cold THF (-20 °C, ca. 3 mL) was
added to a mixture of L3Os(H)(η2-CH2PMe2)12 (100 mg, 0.2
mmol, for ease in handling cooled to -20 °C prior to weighing)
and phenol (17 mg, 0.18 mmol) in a Schlenk flask. The solution
was allowed to warm to room temperature, and the solvent
was removed under vacuum. The residue was washed at -78
°C with hexanes (5 × 10 mL) to remove residual 1, and the
residue was dried under vacuum. The bright white, powdery
residue amounted to 85 mg, 80% yield. 1H NMR (C6D6): δ
-7.80 (dq, 1H, J PH ) 79, 23 Hz, OsH), 1.22 (d, 9H, J ) 7 Hz,
P(CH3)3), 1.25 (d, 9H, J ) 8 Hz, P(CH3)3), 1.29 (vt, 18H, N )
7 Hz, trans-P(CH3)3), 6.68 (t, 1H, J ) 8 Hz, ArHp), 7.21 (d,
2H, ArHo), 7.41 (t, 2H, J ) 8 Hz, ArHm). 13C{1H} NMR (C6D6):
δ 20.73 (d, J PC ) 21 Hz), 23.10 (dvt, J PC ) 3.5, NPC ) 32 Hz),
28.24 (d, J PC ) 31 Hz), 111.73 (s), 120.41 (d, 4J PC ) 2.0), 129.17
Con clu sion s
Osmium complexes cis-L4Os(H)(XC6H4Z) are easily
prepared from fac-L3Os(H)(η2-CH2PMe2) (which is not
so easily prepared) and the corresponding neutral arene.
The coordinating heteroatom group X ) O, S, or NH
has a large effect on the phosphine substitution chem-
istry of the molecule. In the case of thiophenoxide and
anilide ligands, heteroatom lone pair π-donation stabi-
lizes the transition state for L dissociation only from
the mutually trans sites. The data are inconsistent with
either a square-pyramidal intermediate that is rigid
during its short lifetime or with a five-coordinate
intermediate that is fully fluxional. Most consistent is
a quasi-trigonal-bipyramidal intermediate with one L
rigidly trans to the hydride, where the molecule is
stabilized by π-electron donation from the anilide or
sulfide lone pair in the trigonal plane.
Exchange data for the phenoxide complexes are most
consistent with a rapidly reversible ionization followed
by slower exchange of L via presumed ion pairs [L5OsH]-
[OAr]. Dissociation of L from the cation [L5OsH]+ would
be the overall rate-determining step in the exchange.
The relatively high basicity of nitrogen prevents anilide
ionization and results in strong π-donation, accounting
for the mechanistic preference of 7. Phenoxide’s lower
basicity and the relatively weak Os-O bond presumably
result in an ionic pathway for 6. In the case of thiophen-
oxide complexes, 8, the preference for the L dissociation
exchange path over the ionization path is apparently a
result of the greater Os-S bond strength combined with
its larger π-basicity compared to oxygen. The former
retards ionization, and the latter accelerates phosphine
dissociation.
3
(s), 171.05 (d, J PC ) 6.3 Hz). 31P{1H} NMR (C6D6): δ -36.16
(dd, 2P, J PP ) 19, 16 Hz), -41.74 (dt, 1P, J PP ) 5, 19 Hz, trans
to OsH), -49.94 (dt, 1P, J PP ) 5, 16 Hz, cis to OsH). Anal.
Calcd for C18H42OP4Os: C, 36.73; H, 7.19. Found: C, 37.03;
H, 7.16.
L4Os(H)(OC6H4OMe) (6b). The same procedure as with
6a was used with 22 mg (0.18 mmol) of p-methoxyphenol and
1
100 mg (0.2 mmol) of 1; yield 75 mg, 67%. H NMR (C6D6): δ
-7.82 (dq, 1H, J PH ) 79, 24 Hz, OsH), 1.24 (d, 9H, J ) 8 Hz,
P(CH3)3), 1.26 (d, 9H, J ) 7 Hz, P(CH3)3), 1.31 (vt, 18H, N )
6 Hz, trans-P(CH3)3), 3.60 (s, OCH3), 7.05 (d, 2H, J ) 10 Hz),
7.13 (d, 2H). 13C{1H} NMR (C6D6): δ 20.71 (d, J PC ) 22 Hz),
23.15 (dvt, J PC ) 3.5, NPC ) 31 Hz), 28.35 (d, J PC ) 31 Hz),
4
56.06 (OCH3), 115.24 (s), 119.53 (d, J PC ) 1.9), 148.37 (s),
165.54 (d, 3J PC ) 4.9 Hz). 31P{1H} NMR (C6D6): δ -36.40 (dd,
2P, J PP ) 19, 16 Hz), -42.20 (dt, 1P, J PP ) 6, 19 Hz, trans to
OsH), -50.92 (dt, 1P, J PP ) 6, 16 Hz, cis to OsH). Anal. Calcd
for C19H44O2P4Os: C, 36.89; H, 7.17. Found: C, 36.53; H, 7.10.
L4Os(H)(OC6H4CF 3) (6c). The same procedure as with 6a
was used with 29 mg (0.18 mmol) of p-trifluoromethylphenol
and 100 mg (0.2 mmol) of 1, yielding 89 mg, 72%. 1H NMR
(C6D6): δ -7.99 (dq, 1H, J PH ) 79, 24 Hz, OsH), 1.18 (m, 36H,
P(CH3)3)), 7.06 (d, 2H, J ) 8 Hz), 7.64 (d, 2H). 13C{1H} NMR
(C6D6): δ 20.72 (d, J PC ) 22 Hz), 23.02 (dvt, J PC ) 2.3, NPC
)
1
33 Hz), 27.95 (d, J PC ) 32 Hz), 112.58 (q, J CF ) 31 Hz, CF3),
2
116.82 (s), 119.95 (s), 126.63 (q, J CF ) 3.0), 174.25 (d, 3J )
5.0 Hz). 31P{1H} NMR (C6D6): δ -36.66 (dd, 2P, J PP ) 19, 16
Hz), -41.73 (dt, 1P, J PP ) 6, 19 Hz, trans to OsH), -49.94 (dt,
1P, J PP ) 6, 16 Hz, cis to OsH). Anal. Calcd for C19H41OP4F3-
Os: C, 34.75; H, 6.29. Found: C, 34.02; H, 6.31.
Exp er im en ta l Section
L4Os(H)(OC6H4NH2) (6d ). 6d was prepared using 19 mg
(0.18 mmol) of p-aminophenol and 100 mg (0.2 mmol) of 1;
yield 55 mg, 61%. 1H NMR (C6D6): δ -7.81 (dq, 1H, J PH ) 80,
24 Hz, OsH), 1.24 (d, 9H, J ) 8 Hz, P(CH3)3), 1.27 (d, 9H, J )
Gen er a l Com m en ts. Compounds of the type (Me3P)4OsXY,
especially when X/Y include hydrocarbyls and/or hydrides, are
highly air sensitive and require careful attention to anaerobic

1174 Organometallics, Vol. 19, No. 6, 2000
Flood et al.
7 Hz, P(CH3)3), 1.33 (vt, 18H, N ) 6 Hz, trans-P(CH3)3), 2.62
(s, NH2), 6.70 (d, 2H, J ) 9 Hz), 7.06 (d, 2H). 13C{1H} NMR
17 Hz, cis to OsH). 19F{1H} NMR (C6D6): δ -57.88 (s). Anal.
Calcd for C19H42NF3P4Os: C, 34.81; H, 6.46. Found: C, 35.15;
H, 6.20.
(C6D6): δ 20.71 (d, J PC ) 22 Hz), 23.12 (dvt, J PC ) 2.4, NPC
)
)
4
31 Hz), 28.41 (d, J PC ) 31 Hz), 117.90 (s), 119.73 (d, J PC
L4Os(H)(SC6H5) (8a ). The same procedure as for 7a was
used with 100 mg (0.2 mmol) of 1 and 20 mg (0.18 mmol) of
thiophenol in ca. 3 mL of THF, except that no heating was
necessary. After the sample warmed to room temperature the
31P NMR revealed complete reaction. The pale green, powdery
residue weighed 85 mg, 78%. 1H NMR (C6D6): δ -9.50 (dq,
1H, J PH ) 64, 23 Hz, OsH), 1.27 (d, 9H, J ) 6 Hz, P(CH3)3),
1.30 (d, 9H, J ) 7 Hz, P(CH3)3), 1.43 (vt, 18H, N ) 6 Hz, trans-
P(CH3)3), 6.88 (t, 1H, J ) 7 Hz), 7.20 (t, 2H, J ) 7 Hz), 8.27
(d, 2H, J ) 7 Hz). 13C{1H} NMR (C6D6): δ 22.83 (d, J PC ) 24
Hz), 23.61 (dvt, J PC ) 2.5, NPC ) 35 Hz), 28.58 (dd, J PC ) 27,
1.6), 132.50 (s), 164.52 (d, J PC ) 5.7 Hz). 31P{1H} NMR
(C6D6): δ -35.96 (dd, 2P, J PP ) 19, 16 Hz), -41.89 (dt, 1P,
J PP ) 6, 19 Hz, trans to OsH), -50.74 (dt, 1P, J PP ) 6, 16 Hz,
cis to OsH).
3
L4Os(H)(OC6H4CN) (6e). 6e was prepared using 21 mg
(0.18 mmol) of p-cyanophenol and 100 mg (0.2 mmol) of 1; yield
1
92 mg, 83%. H NMR (C6D6): δ -8.15 (dq, 1H, J PH ) 78, 23
Hz, OsH), 1.13 (m, 36H, P(CH3)3)), 6.87 (d, 2H, J ) 8 Hz), 7.40
(d, 2H). 13C{1H} NMR (C6D6): δ 20.68 (d, J PC ) 21 Hz), 22.95
(dvt, J PC ) 2.7, NPC ) 32 Hz), 27.85 (d, J PC ) 32 Hz), 120.65
(d, 4J ) 2.6), 133.64 (s), C-CN, C-O and CtN not found. 31P-
{1H} NMR (C6D6): δ -36.92 (dd, 2P, J PP ) 19, 17 Hz), -41.75
(dt, 1P, J PP ) 6, 19 Hz, trans to OsH), -49.75 (dt, 1P, J PP ) 6,
17 Hz, cis to OsH).
4
3
2.8 Hz), 119.82 (s), 131.57 (d, J PC ) 3.1), 150.43 (d, J PC
)
10), one C not found (under solvent?). 31P{1H} NMR (C6D6): δ
-49.61 (dd, 2P, J PP ) 18, 13 Hz), -50.86 (dt, 1P, J PP ) 13, 18
Hz, trans to OsH), -53.83 (dt, 1P, J PP ) 13, 18 Hz, cis to OsH).
Anal. Calcd for C19H42SP4Os: C, 35.75; H, 7.00. Found: C,
35.78; H, 6.98.
L4Os(H)(HNC6H5) (7a ). In the glovebox, 100 mg (0.2 mmol)
of 1 (cooled to -20 °C prior to weighing for ease of handling)
and aniline (17 mg, 0.18 mmol) were added to a 9 mm NMR
tube fused to a 14/20 ground-glass joint. Approximately 3 mL
of THF was added, and the tube was removed from the
glovebox, sealed on the vacuum line, and heated in an oil bath
at 110 °C. The reaction, monitored by 31P NMR, was complete
after 60 h. The tube was broken in the glovebox, and the
contents were transferred to a Schlenk flask. Solvent was
removed under vacuum, and the residue was washed with
hexanes (5 × 10 mL) at -78 °C to separate any residual 1.
L4Os(H)(SC6H4OMe) (8b). The same procedure as for 8a
was used with 100 mg (0.2 mmol) of 1 and 22.4 µL (0.18 mmol)
of p-methoxythiophenol in ca. 3 mL of THF; yield 91 mg, 80%.
1H NMR (C6D6): δ -9.50 (dq, 1H, J PH ) 64, 23 Hz, OsH), 1.26
(d, 9H, J ) 6 Hz, P(CH3)3), 1.31 (d, 9H, J ) 8 Hz, P(CH3)3),
1.40 (vt, 18H, N ) 6 Hz, trans-P(CH3)3), 3.59 (s, OCH3), 7.19
(d, 2H, J ) 7 Hz), 8.10 (d, 2H). 13C{1H} NMR (C6D6): δ 22.27
(d, J PC ) 20 Hz), 22.64 (dvt, J PC ) 2.5, NPC ) 35 Hz), 27.52
(dd, J PC ) 27, 2.8 Hz), 55.95 (s, OCH3), 119.19 (s), 130.52 (d,
3
4J PC ) 3.1), 149.47 (d, J PC ) 10), one C not found (under
1
The bright yellow, powdery residue weighed 89 mg, 84%. H
solvent?). 31P{1H} NMR (C6D6): δ -50.37 (t, 2P, J PP ) 18 Hz),
-51.54 (dt, 1P, J PP ) 12, 18 Hz, trans to OsH), -54.42 (dt,
1P, J PP ) 12, 18 Hz, cis to OsH). Anal. Calcd for C19H42OSP4-
Os: C, 35.95; H, 6.99. Found: C, 35.80; H, 6.83.
NMR (C6D6): δ -8.67 (dq, 1H, J PH ) 74, 23 Hz, OsH), 1.07 (d,
9H, J ) 6 Hz, P(CH3)3), 1.26 (d, 9H, J ) 7 Hz, P(CH3)3), 1.30
(vt, 18H, N ) 6 Hz, trans-P(CH3)3), 2.32 (brd, 1H, J PH ) 6,
NH), 6.57 (d, 1H, J ) 8 Hz), 6.91 (dd, 1H, J ) 8, 3.5 Hz), 7.19
(dd, 1H, J ) 6, 8 Hz), 7.31 (d, 1H, J ) 8 Hz), 7.63 (dd, 1H, J
) 6, 3.5 Hz). 13C{1H} NMR (C6D6): δ 21.80 (d, J PC ) 21 Hz),
23.42 (dvt, J PC ) 2.3, NPC ) 31 Hz), 28.45 (d, J PC ) 29 Hz),
108.15 (s), 115.14 (s), 117.02 (d, 4J PC ) 3.8), 129.12 (s), 129.51
P (CD3)3-P (CH3)3 Exch a n ge Kin etics. A typical phos-
phine exchange experiment was as follows. Approximately 10
mg of L4Os(H)(XC6H4Z) was weighed in the drybox and added
to a 5 mm NMR tube fused to a vacuum valve. Benzene-d6
(0.3 mL) was added to the tube to form a clear solution that
ranged from colorless (phenoxides), to yellow (amides), to pale
green (thiophenoxides). The sample was freeze-pump-thaw
degassed three times, and at least 20 equiv per osmium of L′
(P(CD3)3) was condensed into the tube. Kinetic data were
obtained from 31P NMR spectra using an inverse-gated, proton-
decoupling pulse sequence used to minimize nuclear Over-
hauser enhancements. The average phosphorus T1 value for
these compounds is 10-12 s, so a relaxation delay of 55 s was
used. Integrations against standard materials revealed errors
of less than 5%. Progress of the reaction was monitored either
by continuous data acquisition in the NMR probe at the
designated temperature or by heating the sample in a constant-
temperature bath with periodic measurement of the 31P{1H}
NMR spectra at room temperature. R2 values from the kinetic
plots were usually greater than 0.99.
3
(s), 160.64 (d, J PC ) 6 Hz). 31P{1H} NMR (C6D6): δ -42.15
(dd, 2P, J PP ) 18, 16 Hz), -51.02 (dt, 1P, J PP ) 9, 18 Hz, trans
to OsH), -52.81 (dt, 1P, J PP ) 9, 16 Hz, cis to OsH). Anal.
Calcd for C18H43NP4Os: C, 36.79; H, 7.38. Found: C, 36.59;
H, 7.25.
L4Os(H)(HNC6H4OMe) (7b). The same procedure as for
7a was used with 100 mg (0.2 mmol) of 1 and 22 mg (0.18
mmol) of p-methoxyaniline in ca. 3 mL of THF at 110 °C for
80 h; yield 93.4 mg, 84%. 1H NMR (C6D6): δ -8.33 (dq, 1H,
J PH ) 75, 23 Hz, OsH), 1.09 (d, 9H, J ) 6 Hz, P(CH3)3), 1.28
(d, 9H, J ) 7 Hz, P(CH3)3), 1.33 (vt, 18H, N ) 6 Hz, trans-
P(CH3)3), 1.87 (brd, 1H, 7 Hz, NH), 3.61 (s, OCH3), 6.49 (brs,
1H), 6.95 (brs, 1H), 7.08 (brs, 1H), 7.21 (brs, 1H). 13C{1H} NMR
(C6D6): δ 21.81 (d, J PC ) 22 Hz), 23.40 (dvt, J PC ) 3.6, NPC
)
34 Hz), 28.57 (d, J PC ) 29 Hz), 55.27 (OCH3), 114.56 (s), 115.01
(s), 116.20 (s), 120.94 (s), 141.54 (s), 152.81 (s). 31P{1H} NMR
(C6D6): δ -41.51 (dd, 2P, J PP ) 18, 16 Hz), -50.34 (dt, 1P,
J PP ) 9, 18 Hz, trans to OsH), -52.38 (dt, 1P, J PP ) 9, 16 Hz,
cis to OsH). Anal. Calcd for C19H45ONP4Os: C, 36.95; H, 7.34.
Found: C, 37.21; H, 7.35.
Kin etic Mod elin g of L/L′ Exch a n ge. The Weigert pro-
gram17 was used to iteratively calculate time/concentration
curves for particular exchange schemes. For example, the
model for the exchange of L4Os(H)(NHCHOMe) given in the
results section interconverts 16 isotopic substitutional isomers
of 9 (including enantiomers), 16 of 12, and 10 of 13, using 89
equations and 5 distinct rate constants, some multiplied by
statistical factors. The concentration numbers for the isoto-
pomers of 9 were converted to phosphorus integral values, and
the rate constants kL , kL , and kL were calculated. For the
L4Os(H)(HNC6H4CF 3) (7c). The same procedure as for 7a
was used with 100 mg (0.2 mmol) of 1 and 22.8 µL (0.18 mmol)
of p-trifluoromethylaniline in ca. 3 mL of THF at 110 °C for
40 h; yield 100 mg, 85%. 1H NMR (C6D6): δ -8.61 (dq, 1H,
J PH ) 75, 23 Hz, OsH), 1.01 (d, 9H, J ) 6 Hz, P(CH3)3), 1.19
(vt, 18H, N ) 6 Hz, trans-P(CH3)3), 1.20 (d, 9H, J ) 7 Hz,
P(CH3)3), 2.96 (brd, 1H, 6 Hz, NH), 6.27 (d, 1H, J ) 9 Hz),
7.20 (d, 1H, J ) 8 Hz), 7.37 (d, 1H, J ) 8 Hz), 7.62 (d, 1H, J
) 9 Hz). 13C{1H} NMR (C6D6): δ 21.64 (d, J PC ) 22 Hz), 23.34
(dvt, J PC ) 3.2, NPC ) 31 Hz), 28.07 (d, J PC ) 29 Hz), 107.88
L
a
b
c
rate constants kion ) a4 × 10-4, kassoc ) 1 × 106, kL′ ) 1 × 104,
k-L ) 1 × 10-3, and kisom ) 1 × 108, calculated rates were kL
L
a
a
) 1.99 × 10-6 and kL ) kL ) 7.11 × 10-7; that is, LaLa/Lb/Lc
b
c
) 1:0.36:0.36.
1
(q, J CF ) 31 Hz, CF3), 113.99 (s), 114.3 (s), 116.06 (s), 126.13
Ack n ow led gm en t. This work was supported by the
National Science Foundation.
3
3
(d, J PC ) 2.5 Hz), 126.93 (d, J PC ) 3.8 Hz), 163.22 (brs). 31P-
{1H} NMR (C6D6): δ -41.77 (dd, 2P, J PP ) 18, 17 Hz), -50.45
(dt, 1P, J PP ) 9, 18 Hz, trans to OsH), -51.72 (dt, 1P, J PP ) 9,
OM990850R