Hydrogenation of carbon dioxide catalyzed by ruthenium trimethylphosphine complexes: A mechanistic investigation using high-pressure NMR spectroscopy

Article abstract of DOI:10.1021/om900128s

Complex cis-(PMe₃)₄RuCl(OAc) (1) acts as a catalyst for CO₂ hydrogenation into formic acid in the presence of a base and an alcohol cocatalyst. NMR spectroscopy has revealed that 1 exists in solution in equilibrium with fa

Full text of DOI:10.1021/om900128s

5466 Organometallics 2009, 28, 5466–5477
DOI: 10.1021/om900128s
Hydrogenation of Carbon Dioxide Catalyzed by Ruthenium
Trimethylphosphine Complexes: A Mechanistic Investigation Using
High-Pressure NMR Spectroscopy
April D. Getty,‡ Chih-Cheng Tai,† John C. Linehan,*,‡ Philip G. Jessop,*,^,
Marilyn M. Olmstead,† and Arnold L. Rheingold§
‡Chemical Sciences Division, Pacific Northwest National Laboratory, P.O. Box 999 MS: K2-57, Richland,
Washington 99352, †Department of Chemistry, University of California, Davis, California 95616-5295,
§
^ Department of Chemistry, Queen’s University, Kingston, Ontario, Canada K7L 3N6, and University of
Delaware, Newark, Delaware 19716. Previous address: Department of Chemistry, University of
California, Davis, California 95616-5295
Received February 17, 2009
Complex cis-(PMe3)4RuCl(OAc) (1) acts as a catalyst for CO2 hydrogenation into formic acid in the
presence of a base and an alcohol cocatalyst. NMR spectroscopy has revealed that 1 exists in solution in
equilibrium with fac-(PMe3)3RuCl(η2-OAc) (2), [(PMe3)4Ru(η2-OAc)]Cl (3a), and free PMe3. Complex 2
was found to be a poor CO2 hydrogenation catalyst under the conditions of catalysis used for 1. Complex 3a
can be prepared by adding certain alcohols, such as MeOH, EtOH, or C6H5OH, to a solution of 1in CDCl3.
-
-
The chloride ion of 3a was exchanged for the noncoordinating anion BPh4 or B(ArF)4 (B(ArF)4 =
tetrakis(3,5-bis(trifluoromethyl)phenyl)borate) to produce [(PMe3)4Ru(η2-OAc)]BPh4 (3b) and [(PMe3)4-
Ru(η2-OAc)]B(ArF)4 (3c). Complexes 3b and 3c were found to be as efficient as 1 in the catalytic
hydrogenation of CO2 to formic acid in the presence of an alcohol cocatalyst. In contrast to 1, 3b and 3c
continued to show high catalytic activity in the absence of the alcohol cocatalyst. High-pressure NMR
spectroscopy was used to investigate the mechanism of CO2 hydrogenation via 3b,c in the presence of base.
The observations were inconsistent with the previously reported phosphine-loss mechanism; a new
mechanism is proposed involving an unsaturated, cationic ruthenium complex of the form
[(PMe3)4RuH]þ (B) as the active catalyst. The role of the base in this system includes not only trapping
of the formic acid product but also initiation of the catalysis by aiding the conversion of 3b,c to B.
Crystallographically determined structures are reported for complexes 2, 3b, {[(PMe3)3Ru]2(μ-Cl)2(μ-
OAc)}BPh4, and {[(PMe3)3Ru]2(μ-Cl)3}Cl.
Introduction
There have been several reports of the hydrogenation of
carbon dioxide to produce formic acid using homogeneous
RuII catalysts.3-5 For instance, Jessop et al.3 performed an
investigation of CO2 hydrogenation in supercritical CO2 that
Carbon dioxide recovered from power plant stacks or the
atmosphere can be used to produce chemicals and fuels. The
reasons for using CO2 as a starting material are numerous.1
Carbon dioxide is cheap, nontoxic, and abundant and is a totally
renewable feedstock compared to oil or coal. One chemical that
can be catalytically produced from CO2 is formic acid. Formic
acid is used for such varied processes as leather tanning, in dye
baths for dyeing natural and synthetic fibers, as an additive for
cleaning agents, for coagulating latex in the production of natural
rubber, and in the synthesis of the artificial sweetener aspartame.2
(3) (a) Munshi, P.; Main, A. D.; Linehan, J. C.; Tai, C.-C.; Jessop, P.
G. J. Am. Chem. Soc. 2002, 124, 7963. (b) Tai, C.-C.; Pitts, J.; Linehan, J.
C.; Main, A. D.; Munshi, P.; Jessop, P. G. Inorg. Chem. 2002, 41, 1606. (c)
Yin, C.; Xu, Z.; Yang, S.-Y.; Ng, S. M.; Wong, K. Y.; Lin, Z.; Lau, C. P.
Organometallics 2001, 20, 1216. (d) Jessop, P. G.; Hsiao, Y.; Ikariya, T.;
Noyori, R. J. Am. Chem. Soc. 1996, 118, 344. (e) Jessop, P. G.; Hsiao, Y.;
Ikariya, T.; Noyori, R. J. Chem. Soc., Chem. Commun. 1995, 707. (f)
Jessop, P. G.; Ikariya, T.; Noyori, R. Nature 1994, 368, 231. (g) Jessop, P. G.;
Hsiao, Y.; Ikariya, T.; Noyori, R. J. Am. Chem. Soc. 1994, 116, 8851.
(4) (a) Urakawa, A.; Jutz, F.; Laurenczy, G.; Baiker, A. Chem.;Eur.
J. 2007, 13, 3886–3899. (b) Ogo, S.; Kabe, R.; Hayashi, H.; Harada, R.;
Fukuzumi, S. Dalton Trans. 2006, 4657–4663. (c) Ng, S. M.; Yin, C.; Yeung,
C. H.; Chan, T. C.; Lau, C. P. Eur. J. Inorg. Chem. 2004, 1788–1793. (d)
Himeda, Y.; Onozawa-Komatsuzaki, N.; Sugihara, H.; Arakawa, H.; Kasuga,
K. Organometallics 2004, 23, 1480–1483. (e) Kayaki, Y.; Shimokawatoko,
Y.; Ikariya, T. Adv. Synth. Catal. 2003, 345, 175–179. (f) Man, M. L.; Zhou,
Z.; Ng, S. M.; Lau, C. P. J. Chem. Soc., Dalton Trans. 2003, 3727–3735. (g)
*To whom correspondence should be addressed. E-mail: john.
linehan@pnl.gov (J.C.L.); jessop@chem.queensu.ca(P.G.J.).
(1) Arakawa, H.; Aresta, M.; Armor, J. N.; Barteau, M. A.; Beck-
man, E. J.; Bell, A. T.; Bercaw, J. E.; Creutz, C.; Dinjus, E.; Dixon, D. A.;
Domen, K.; DuBois, D. L.; Eckert, J.; Fujita, E.; Gibson, D. H.;
Goddard, W. A.; Goodman, D. W.; Keller, J.; Kubas, G. J.; Kung, H.
H.; Lyons, J. E.; Manzer, L. E.; Marks, T. J.; Morokuma, K.; Nicholas,
K. M.; Periana, R.; Que, L.; Rostrup-Nielson, J.; Sachtler, W. M. H.;
Schmidt, L. D.; Sen, A.; Somorjai, G. A.; Stair, P. C.; Stults, B. R.;
Tumas, W. Chem. Rev. 2001, 101, 953, and references therein.
(2) (a) Kirk, R. E.; Othmer, D. F.; Kroschwitz, J. I.; Howe-Grant, M.
Kirk-Othmer Encyclopedia of Chemical Technology, 4th ed.; Wiley &
Sons: New York, 1991. (b) Reutemann, W.; Kieczka, H. In Ullman’s
Encyclopedia of Industrial Chemistry [Online]; Wiley-VCH: Weinheim,
2002.
ꢀ
ꢀ
Elek, J.; Nadasdi, L.; Papp, G.; Laurenczy, G.; Joo, F. Appl. Catal. A: Gen.
2003, 255, 59–67. (h) Yi, C. S.; Liu, N. Organometallics 1995, 14, 2616. (i)
Koinuma, H.; Kawakami, F.; Kato, H.; Hirai, H. J. Chem. Soc., Chem.
Commun. 1981, 213. (j) Inoue, Y.; Izumida, H.; Sasaki, Y.; Hashimoto, H.
Chem. Lett. 1976, 863. (k) Inoue, Y.; Sasaki, Y.; Hasimoto, H. J. Chem.
Soc., Chem. Commun. 1975, 718. (l) Kokomnikov, I. S.; Lobeeva, T. S.;
Vol'pin, M. E. Bull. Acad. Sci. USSR, Div. Chem. Sci. (Engl. Transl.) 1972,
2263.
r
pubs.acs.org/Organometallics
Published on Web 08/26/2009
2009 American Chemical Society

Article
Organometallics, Vol. 28, No. 18, 2009 5467
included the use of trans-(PMe3)4RuCl2, cis-(PMe3)4RuH2,
or cis-(PMe3)4RuCl(OAc) (1, OAc is acetate) as catalyst
precursor. We have focused on complex 1 as the catalyst
precursor because it is as active as the Ru-dihydride, but is
far more easily prepared and handled. We earlier reported
the effects of certain alcohols and bases on the yield of formic
acid in this system (eq 1).3a
referenced to solvent peaks relative to TMS, 31P{1H} NMR
spectra were referenced externally to 85% H3PO4 (downfield
being positive), and the H NMR spectra were referenced to
2
THF-d8. NMR spectra recorded at atmospheric pressure were
obtained using a standard broadband 5 mm NMR probe, and
NMR spectra recorded at high pressures were conducted using a
10 mm broadband probe with a 10 mm o.d., 3.5 mm i.d. PEEK
(polyether ether ketone) NMR cell run unlocked.6 Elemental
analyses were performed by Atlantic Microlab, Inc.
The complexes (PMe3)4RuCl(OAc) (1),7 cis-(P(OEt)3)4RuCl2,8
8
and cis-(P(OEt)3)4RuH2 were prepared according to literature
procedures. The catalytic runs in the multivial reactor were
performed with equipment described in an earlier paper.9 Pressure
vessels were loaded under dry nitrogen atmosphere.
Safety Warning. Operators of high-pressure equipment such as
that required for these experiments should take proper precau-
tions, including but not limited to the use of blast shields and
pressure-relief mechanisms, to minimize the risk of personal injury.
Synthesis of fac-(PMe3)3RuCl(η2-OAc) (2). In the drybox,
175 mg (0.35 mmol) of cis-(PMe3)4RuCl(OAc) (1) was dissolved
in 6 mL of CH2Cl2, and the solution was filtered through glass
wool layered with Celite to remove a small amount of insoluble
material. The filtrate was dispensed into a glass bomb, removed
from the drybox, and placed on a high-vacuum apparatus. The
CH2Cl2 was removed under reduced pressure, and the bomb was
placed in a sand bath adjusted to ca. 150 ꢀC. The sealed bomb
was periodically opened to vacuum for ca. 30 s every 10-15 min
during the 2 h heating period. After cooling, the bomb was
brought into the drybox and the solid was dissolved in 4 mL of
CH2Cl2. This solution was filtered and the product crystallized
by vapor diffusion with pentane. The yellow crystalline solid
was collected by decanting the mother liquor, and the crystals
were washed with pentane (3 ꢀ 2 mL). Yield: 102 mg (0.24
mmol), 69%. Anal. Calcd for C11H30O2P3ClRu: C, 31.17; H,
7.14. Found: C, 31.15; H, 7.17. 1H NMR (CDCl3): δ 1.40-1.50
(overlapping dd’s, 27H, P(CH3)3), 1.97 (s, 3H, O2CCH3). 31P-
Near-stoichiometric amounts of base are required for the
reaction shown in eq 1, as well as a catalytic amount of
alcohol in order to obtain reasonable rates of catalysis.
Acidic, nonbulky alcohols and triflic acid increase the rate
of hydrogenation an order of magnitude above that which
can be obtained using methanol or water. Similarly, the use
of the more basic DBU (1,8-diazobicyclo[5.4.0]undec-7-ene)
rather than NEt3 increases the rate of reaction by an order of
magnitude. The combination of DBU and an alcohol coca-
talyst leads to a side reaction (eq 2) giving alkyl carbonate
salts;3a whether those salts have a role in the catalysis is
unknown. Determining the role of the alcohol in the hydro-
genation, and being able to obtain reasonable rates of
catalysis without it, would be very beneficial.
Herein we report the solution chemistry of 1 and the use of
high-pressure NMR spectroscopy to further investigate the
mechanism of CO2 hydrogenation catalyzed by complex 1
and its derivatives.
{1H} NMR (CDCl3): δ 22.2 (t, 1P, P(CH3)3 trans to Cl, 2JP-P
=
37.0 Hz), 29.4 (d, 2P, P(CH3)3 trans to η2-OAc, 2JP-P=36.5 Hz).
13C{1H} NMR (CDCl3): δ 18.53 (d, 3C, P(CH3)3 trans to Cl,
1JP-C = 29.34 Hz), 19.60 (d, 6C, P(CH3)3 trans to η2-OAc,
1JP-C = 14.71 Hz), 25.14 (s, 1C, O2CCH3), 186.90 (s, 1C,
O2CCH3).
Experimental Section
X-ray Structure Determination of fac-(PMe3)3RuCl(η2-OAc)
(2). Crystals of 2 suitable for analysis by X-ray diffraction were
obtained by sublimation of 43 mg of finely powdered complex 1
in a Schlenk tube fitted with a coldfinger, under strong vacuum
at 160 ꢀC. Data collection was carried out using a Bruker
SMART 100010 and Mo KR radiation. No absorption correc-
tion was applied. The structure was solved by direct methods
(SHELXS-9710) in the space group Pmn21(31). It could not be
solved in the alternative space group Pmmn(59). It was refined
by full-matrix least-squares on F2 (SHELXL-9710). The struc-
ture is disordered with respect to a crystallographic mirror plane
that passes through Ru1, Cl1, P1, C7, and C8. Thus, the atoms
C1, C2, C3, C4, C5, C6, C40, C50, and C60 all have disordered
counterparts and are refined at 0.50 occupancy. The H atoms
were located on a difference map and fixed for C1, C2, C3,
and C8. Other H atoms were riding. All H atoms had iso-
tropic thermal parameters set to 1.2 times that of the equiva-
lent isotropic thermal parameter of the bonded carbon. All
General Comments. All manipulations were carried out under
N2 or by using high-vacuum techniques, unless otherwise noted.
All solvents, liquid alcohols, triethylamine, and DBU (1,8-
diazobicyclo[5.4.0]undec-7-ene) were degassed by repeated free-
ze-pump-thaw cycles before use. Chloroform, chloroform-d,
and dichloromethane were stirred at least 5 min with basic
alumina (activated by heating at 150 ꢀC under vacuum for
5 h) before use. Tetrahydrofuran was dried over sodium benzo-
phenone ketyl; however THF-d8 was used as received (packaged
in 0.7 mL ampules by Aldrich). Trimethylphosphine was dis-
tilled from sodium prior to use. Hydrogen gas (Oxarc, 5.0
grade), deuterium gas (Spectra Gases deuterium-UHP grade
99.5%), CO2 gas (Scott Specialty Gases, SFC grade 99.993%),
and 13CO2 gas (Aldrich Chemical Co., 99 atom % 13C and 6
atom % 18O) were used as received. Unless otherwise specified,
reagents and solvents were used as purchased from Aldrich
Chemical Co., Strem, or Boulder Scientific Company in the case
of NaB(ArF)4 (ArF=3,5-bis(trifluoromethyl)phenyl).
NMR spectra were recorded using a 7.01 T Varian VXR-300
spectrometer, and chemical shifts (δ) are reported in ppm
referenced to residual proton peaks in the deuterated solvent
and reported relative to TMS. 13C{1H} NMR spectra were
(6) Yonker, C. R.; Linehan, J. C. J. Organomet. Chem. 2002, 650, 249.
(7) Mainz, V. V.; Anderson, R. A. Organometallics 1984, 3, 675.
(8) Peet, W. G.; Gerlach, D. H. Inorg. Synth 1974, 15, 38.
(9) Thomas, C. A.; Bonilla, R. J.; Huang, Y.; Jessop, P. G. Can. J.
Chem. 2001, 79, 719.
(5) For reviews of transition metal-catalyzed homogeneous CO2
hydrogenation, see: (a) Jessop, P. G.; Ikariya, T.; Noyori, R. Chem.
Rev. 1995, 95, 259. (b) Leitner, W. Angew. Chem., Int. Ed. Engl. 1995, 34,
2207. (c) Jessop, P. G.; Joo, F.; Tai, C.-C. Coord. Chem. Rev. 2004, 248,
2425–2442.
(10) Data collection: SMART 1000, v. 5.054. Data reduction:
SAINT, v. 6.22; Bruker Analytical X-ray Instruments, Inc.: Madison,
WI, 2001. Sheldrick, G. M. SHELXS-97, 2001. Sheldrick, G. M.
SHELXTL v. 5.10; Bruker Analytical X-ray Instruments, Inc.: Madison,
WI, 1997.

5468 Organometallics, Vol. 28, No. 18, 2009
Getty et al.
non-hydrogen atoms were refined with anisotropic thermal
parameters. Handedness of the crystal was not determined, as
the Flack parameter 0.36(6) indicated the likelihood of racemic
twinning. The largest peak in the final difference map was 2.13 e
Table 1. Crystallographic Data Collection Parameters for
2, 3b, and 4
2
3b
4
-3
10
˚
˚
, 0.97 A from Ru1. Selected details of the crystal data
A
empirical
formula
fw
temp (K)
cryst syst
space group
C
11H30O2-
P3ClRu
C
38H59BO2-
C
44H77BCl2O2-
collection and refinement for 2 are listed in Table 1.
P4Ru
783.61
100(2)
monoclinic
P21/n
P6Ru2
1107.73
100(2)
monoclinic
P21/c
Synthesis of [(PMe3)4Ru(η2-OAc)]X (X = BPh4-, 3b; X =
B(ArF)4-, 3c). In the drybox, 206 mg (0.41 mmol) of cis-(PMe3)4-
RuCl(OAc) (1) was dissolved in 17 mL of CHCl3 and 5.5 mL of
MeOH (24% MeOH/CHCl3 v/v solution). After 20 min of
stirring, a solution of NaBPh4 (141 mg, 0.41 mmol) in 3 mL of
MeOH was added. The slightly cloudy (NaCl) mixture was
removed from the drybox and filtered through glass wool
layered with Celite. The pale yellow filtrate was concentrated
and the product crystallized with pentane. The solid was further
purified by recrystallization from CH2Cl2/pentane at -23 ꢀC to
yield a yellow crystalline solid (3b, 283 mg, 0.36 mmol, 89%
yield). Anal. Calcd for C318H59O2P4BRu: C, 58.23; H, 7.59.
Found: C, 57.79; H, 7.60. H NMR (CDCl3): δ 1.20 (virtual
423.78
91(2)
orthorhombic
Pmn21
˚
a (A)
9.985(2)
9.103(2)
10.243(2)
90
90
90
14.058(11)
18.919(14)
14.875(11)
90
99.348(13)
90
3904(5)
0.597
Mo KR,
0.71073
2.14-26.50
-16 e h e 17
-23 e k e 18
-17 e l e 18
17 598
9.1922(6)
31.816(2)
17.6521(12)
90
91.7780(10)
90
5160.0(6)
0.909
Mo KR,
0.71073
2.22-28.29
-12 e h e 11
-42 e k e 40
-18 e l e 23
32 756
˚
b (A)
˚
c (A)
R (deg)
β (deg)
γ (deg)
3
˚
volume (A )
)
931.0(3)
1.237
μ (mm -1
radiation,
Mo KR,
0.71073
2.24-30.02
-14 e h e 14
-12 e k e12
-11 e l e 14
8228
˚
λ (A)
θ range (deg)
index ranges
2
triplet (vt), 18H, P(CH3)3, JP-H = 4.5 Hz), 1.28 (vt, 18H,
2
5
P(CH3)3, JP-H =3.0 Hz), 1.89 (t, 3H, O2CCH3, JP-H =1.0
Hz), 6.89 (t, 4H, B(p-C6H5)4, 3JH-H=6.9 Hz), 7.04 (t, 8H, B(o-
C6H5)4, JH-H =7.2 Hz), 7.39 (m, 8H, B(m-C6H5)4). 31P{1H}
3
no. of reflns
collected
NMR (CDCl3): δ -1.43 (t, 2P, P(CH3)3, 2JP-P = 29.9 Hz), 21.7
(t, 2P, P(CH3)3, JP-P =29.7 Hz). 13C{1H} NMR (CDCl3): δ
2
final R, Rw
[I > 2σ(I)]
0.0368,
0.0785
0.0631,
0.1498
0.0505,
0.1123
17.02 (vt, 6C, P(CH3)3, 1JP-C=13.2 Hz), 21.45 (vt, 6C, P(CH3)3,
1JP-C =17.5 Hz), 24.77 (s, 1C, O2CCH3), 121.88 (s, 4C, B(p-
C6H5)4), 125.72 (s, 8C, B(m-C6H5)4), 136.56 (s, 8C, B(o-C6H5)4),
cis-(PMe3)4RuH2 (5) was observed using 31P{1H} NMR and 1H
NMR .11 Upon complete conversion of 3b,c to 5, the NMR
probe was cooled to -80 ꢀC before introduction of carbon
dioxide into the cell. The NMR probe was raised to -50 ꢀC after
addition of carbon dioxide, and within an hour conversion of 5
to 6 (cis-(PMe3)4RuH(O2CH)) was observed by NMR spectros-
copy. Production of the [HþDBU][O2CH-] salt was observed
after ca. 3 h at 24 ꢀC, as well as an unidentified ruthenium
product postulated to be {[(PMe3)4Ru]2(μ-OAc)2}X2 (9, X =
BPh4-, B(ArF)4-, or O2CH-).
The NMR cell was then vented to the atmosphere, and the
contents were collected using aliquots of CDCl3 or CH2Cl2 (5 ꢀ
0.5 mL) to rinse the cell. A stream of N2 was used to evaporate
the volatiles and produce a pale yellow oil. The oil was dissolved
in CDCl3, and proton, phosphorus, and carbon (in the cases of
using 13CO2) NMR spectra were recorded.
1
164.46 (q, 4C, B(i-C6H5)4, JB-C = 48.3 Hz), 186.75 (s, 1C,
O2CCH3).
3c was prepared similarly, giving 396 mg (0.30 mmol), 75%
yield. Anal. Calcd for C46H51O2P4F24BRu: C, 41.61; H, 3.87.
1
Found: C, 41.47; H, 3.85. H NMR (CDCl3): δ 1.34 (vt, 18H,
P(CH3)3, 2JP-H=4.5 Hz), 1.37 (vt, 18H, P(CH3)3, 2JP-H=3.3
Hz), 1.91 (t, 3H, O2CCH3, JP-H =1.0 Hz), 7.53 (s, 4H, B(p-
5
C6H3(CF3)2)4, 7.71 (m, 8H, B(o-C6H3(CF3)2)4). 31P{1H} NMR
2
(CDCl3): δ -1.45 (t, 2P, P(CH3)3, JP-P = 30.2 Hz), 21.6 (t,
2P, P(CH3)3, 2JP-P=30.2 Hz). 19F NMR (CDCl3): δ -67.8 (s).
=
1
13C{1H} NMR (THF-d8): δ 16.82 (vt, 6C, P(CH3)3, JP-C
1
12.2 Hz), 21.43 (vt, 6C, P(CH3)3, JP-C = 18.3 Hz), 24.91 (s,
1C, O2CCH3), 122.08 (d, 8C, B(C6H3(CF3)2)4, 1JF-C=275 Hz),
127.47 (s, 4C, B(p-C6H3(CF3)2)4), 130.18 (m, 8C, B(m-C6-
H3(CF3)2)4), 135.73 (s, 8C, B(o-C6H3(CF3)2)4, 163.00 (q,
1
4C, B(i-C6H3(CF3)2)4, JB-C = 48.8 Hz), 187.69 (s, 1C, O2-
NMR Data for 5. 1H NMR at 50 ꢀC (THF-d8): δ -10.0 (br,
RuH), DBU signals obscure signals for PMe3. 31P{1H} NMR at
50 ꢀC (THF-d8): δ -6.75 (t, 2P, P(CH3)3, 2JP-P=26.0 Hz), 0.98
(t, 2P, P(CH3)3, 2JP-P=26.0 Hz). 1H NMR at -80 ꢀC (THF-d8):
δ -10.25 (dt, 2H, RuH, 2JHtransP=52.2 Hz, 2JHcisP=30.0 Hz),
DBU signals obscure signals for PMe3. 31P{1H} NMR at -80 ꢀC
(THF-d8): δ -4.93 (t, 2P, P(CH3)3, 2JP-P = 26.7 Hz), 2.61 (t, 2P,
P(CH3)3, 2JP-P=26.0 Hz).
CCH3).
X-ray Structure Determinations of [(PMe3)4Ru(η2-OAc)]BPh4
(3b) and {[(PMe3)3Ru]2(μ-Cl)2(μ-OAc)}BPh4 (4). Both struc-
tures were determined using a Bruker platform diffractometer
equipped with an APEX detector operated at 100 K using Mo
KR radiation. Diffraction symmetry and systematic absences in
the data were used to determine uniquely the space groups. The
structures were solved by direct methods and completed from
subsequent difference Fourier syntheses. All non-hydrogen
atoms were refined with anisotropic thermal parameters, and
all hydrogen atoms were treated as idealized contributions. All
software used in the collection and analysis of data is contained
in the current libraries provided by Bruker AXS (Madison, WI).
Selected details of the crystal data collection and refinement for
3b and 4 are listed in Table 1.
NMR Data for 6. 1H NMR at -50 ꢀC (THF-d8): δ -8.15 (dq,
2
1H, RuH, JHtransP = 95.4 Hz, JHcisP = 27.8 Hz), 8.11 (s,
2
1H, RuO2CH), DBU signals obscure signals for PMe3. 31P{1H}
NMR at -50 ꢀC (THF-d8): δ -10.66 (m, 1P, P(CH3)3 2trans to
2
H, JP-P = 19.9 Hz), 1.27 (m, 2P, trans P(CH3)3’s, JP-P
=
=
2
26.7 Hz), 21.07 (m, 1P, P(CH3)3 trans to O2CH, JP-P
19.8 Hz).
1
NMR Data for 9. H NMR at 24 ꢀC (THF-d8): δ 2.02 (s,
O2CCH3), DBU signals obscure signals for PMe3. 31P{1H}
General Procedure for Monitoring CO2 Hydrogenation in
THF by High-Pressure NMR Spectroscopy. The ruthenium
compound was weighed into a small vial and brought into the
drybox (ca. 8 mg, 0.010 mmol of 3b, or ca. 16 mg, 0.012 mmol of
3c). Then 500 μL of THF or THF-d8 and ca. 10 equiv of DBU
were added to the vial, and 200 μL of this solution was dispensed
into the PEEK NMR cell, which was protected from the atmo-
sphere through use of tubing/valve assembly. This was attached
to a high-pressure gas manifold. Upon insertion of the PEEK
cell into the NMR probe, 6.7 bar (100 psi) of H2 was introduced,
and the system was heated at 50 ꢀC for ca. 3 h. Formation of
2
NMR at 24 ꢀC (THF-d8): δ 2.07 (t, 2P, P(CH3)3, JP-P
=
29.7 Hz), 17.36 (t, 2P, P(CH3)3, 2JP-P = 30.6 Hz). 1H NMR at
24 ꢀC (CDCl3): δ 1.99 (s, O2CCH3), DBU signals obscure signals
for PMe3. 31P{1H} NMR at 24 ꢀC (CDCl3): δ 0.40 (t, 2P,
2
2
P(CH3)3, JP-P = 29.7 Hz), 16.98 (t, 2P, P(CH3)3, JP-P
=
29.7 Hz).
€
(11) Gusev, D. G.; Hubener, R.; Burger, P.; Orama, O.; Berke, H. J.
Am. Chem. Soc. 1997, 119, 3716.

Article
Organometallics, Vol. 28, No. 18, 2009 5469
Figure 1. Thermal ellipsoid drawing of fac-(PMe3)3RuCl(η2-
OAc) (2) showing 50% probability contours. Hydrogen atoms
have been omitted for clarity.
Figure 2. Thermal ellipsoid drawing of the cation in
[(PMe3)4Ru(η2-OAc)]BPh4 (3b) showing 50% probability con-
tours. Hydrogen atoms and the BPh4- anion have been omitted
for clarity.
Results
Solution Equilibrium of cis-(PMe3)4RuCl(OAc) (1). In
solution, complex 1 is in equilibrium with two other ruthe-
nium-containing species, both of which have been isolated
and characterized by X-ray crystallography.12 The 31P{1H}
NMR spectrum of a solution of 1 in CDCl3 contains signals
for 1 with the expected A2BC splitting pattern3a as well
as signals for free PMe3, fac-(PMe3)3RuCl(η2-OAc) (2), and
[(PMe3)4Ru(η2-OAc)]Cl (3a, δ 22.2, overlap t; δ -1.0, t). The
approximate relative amounts of the ruthenium compounds,
determined by 31P NMR spectroscopy, were 73% 1, 12% 2,
and 10% 3a.13 In the presence of excess PMe3 (>10 equiv) the
31P{1H} NMR spectrum in C6D6 contained only signals
expected for cis-(PMe3)4RuCl(OAc) and free PMe3.
Complex 2 can be prepared by heating or subliming
1 under vacuum, yielding crystals that were analyzed by
X-ray crystallography, NMR spectroscopy, and elemental
analysis. The molecular structure of 2 is shown in Figure 1,
and the summary of crystal data and details of the struc-
ture and its refinement are shown in the Supporting Infor-
mation.14
Addition of methanol to a solution of 2 in CDCl3 leads to
the formation of a new species. The 31P{1H} NMR spectrum
acquired 10 min after addition of methanol showed small
broadened signals for 2 in addition to a sharp intense A2B
pattern at δ 23.8 (d) and 20.4 (t) and two very small broad
signals at δ 24.5 and 21.4 ppm. The proton NMR signals shift
from a single resonance at 1.97 ppm to two singlets at 2.03
and 1.88 ppm in ca. 1:1 ratio for the acetate ligand when
CD3OD is added to 2. The PMe3 protons change from an
overlapping singlet and doublet between 1.49 and 1.40 ppm
to two sets of signals, a doublet at 1.55 ppm (2JP-H = 8.7 Hz)
and a virtual triplet at 1.39 ppm, upon addition of CD3OD.
The spectroscopic data are consistent with a ruthenium
dimer structure {[(PMe3)3Ru]2(μ-Cl)(μ-OAc)2}þ similar to
{[(PMe3)3Ru]2(μ-Cl)2(μ-OAc)}þ (complex 4, vide infra).
Addition of MeOH to a CDCl3 solution of 1 (from 7 to
75% MeOH) results in conversion of 1 to 3a. The signals in
the 31P{1H} NMR spectrum for 1, 2, and free PMe3 dis-
appear, while the signals for 3a increase. In addition to
the signals for 3a, there are very small signals due to
the species formed upon the reaction of 2 with MeOH in
CDCl3 (vide supra). Complex 3a can also be formed in
supercritical CO2 from 1 in the presence of MeOH (ca. 30
equiv). Other methods to prepare 3a include using methanol,
ethanol, or o-cresol in CDCl3 or C6D6 (25% solution), using
nitromethane in CDCl3 (25% solution), or abstracting
the chloride with AgBF4 in CDCl3 (in this case the counter-
The structure of 2 can be best described as a six-coordinate
pseudo-octahedron, with some distortion of the chloride and
phosphines toward the bidentate acetate ligand. The Ru-P
bond length trans to chloride is slightly greater than those
˚
trans to acetate (ca. 0.03 A) and roughly equal to those found
in the Ru-trichloride dimer {[(PMe3)3Ru]2(μ-Cl)3}Cl (see
Supporting Information).
-
ion is BF4 rather than Cl-). The formation of 3a was
(12) Attempted crystallizations of complex 1 from methanol or other
solvents were unsuccessful, generating only small quantities of crystals
of {[(PMe3)3Ru]2(μ-Cl)3}Cl, which were crystallographically character-
ized (see Supporting Information). Crystals of the same dimer were
found in small quantities when trans-(PMe3)4RuCl2 was recrystallized.
The dimer cation has been reported previously (with BF4 counterion):
Statler, J. A.; Wilkinson, G.; Thornton-Pett, M.; Hursthouse, M. B. J.
Chem. Soc., Dalton Trans. 1984, 1731.
(13) The relative amounts of 1, 2, and 3a were obtained by integration
of the 31P{1H} NMR spectra (1 s delay, and assuming the T1 values for
the different Ru-bound phosphines are identical) of four different
samples of 1 dissolved in CDCl3 (the values shown are averages, and
the standard deviation is e1%). In addition to these ruthenium com-
pounds, a small amount of cis-(PMe3)4RuCl2 (ca. 5%) was present in
some of the samples, which was carried over from the synthesis of 1 (see
ref 7).
unaffected by the presence of base. Scheme 1 summarizes
the different methods we have discovered to prepare complex
3, including the preparative scale method we used to obtain
3b and 3c.
Interestingly, using tert-butanol, 2-propanol, 2,6-di-tert-
butylphenol, acetone, or acetonitrile (all ca. 25% solutions in
CDCl3) resulted in little or no conversion of 1 to 3a. This
trend in effectiveness of the alcohols to promote the conver-
sion of 1 to 3 parallels the observed trend in effectiveness of
the alcohols to promote the hydrogenation of CO2. Munshi
et al. reported that the yield of formic acid was best with
C6F5OH, moderate with MeOH or EtOH, and very poor
with t-BuOH or 2,6-tBu2C6H3OH as the added alcohol when
(14) Complete details of the X-ray crystal structure determination are
provided in the Supporting Information.

5470 Organometallics, Vol. 28, No. 18, 2009
Getty et al.
Scheme 1. Various Methods To Prepare [(PMe3)4Ru(η2-OAc)]X (3a, X = Cl-; 3b, X = BPh4-; 3c, X = B(ArF)4
)
-
using complex 1 and Et3N for the CO2 hydrogenation
reaction.3a
The coordination geometry at each ruthenium atom of 4 is
more nearly that of a regular octahedron. There are no
significant differences in the Ru-P distances as a function
of trans relationships. Again, there are no close contacts
between ions in the unit cell.
Close examination of the 31P{1H} NMR spectrum
(acetone-d6) of the mixture of 3b and 4 used for the X-ray
structure determinations reveals that 4 constitutes approxi-
mately 5% of the sample.16
The two ionic ruthenium complexes, [(PMe3)4Ru(η2-
OAc)]BPh4 (3b) and [(PMe3)4Ru(η2-OAc)]B(ArF)4 (3c),
were prepared on a large scale by dissolving 1 in a 25%
solution of MeOH in CHCl3 followed after ca. 20 min by
addition of a solution of either NaBPh4 or NaB(ArF)4,15 in
MeOH (Scheme 1). The crystalline sample of 3b obtained
from a solution of 3b in CH2Cl2 layered with pentane and
stored in the freezer (-23 ꢀC) for three days contained large
block crystals (3b), as well as a small amount of fine needles
(4). These two types of crystals were separated. 3b was
recrystallized again by dissolving the separated sample in a
minimal quantity of CH2Cl2 and cooling to -10 ꢀC. The
structure of 4 was determined without further recrystalliza-
tion.
Catalytic Activity of fac-(PMe3)3RuCl(η2-OAc) (2),
[(PMe3)4Ru(η2-OAc)]X (3b, X = BPh4-; 3c, X = B(ArF)4-),
cis-(P(OEt)3)4RuCl2, and cis-(P(OEt)3)4RuH2. The catalytic
activity of complexes 2, 3b, 3c, cis-(P(OEt)3)4RuCl2, and
cis-(P(OEt)3)4RuH2 toward the hydrogenation of CO2 was
investigated. The catalytic activities of 1 and 2 were com-
pared directly using a multivial reactor.17 In a typical reac-
tion, NEt3 (0.5 mL), the catalyst(3μmol), andMeOH(0.5mL)
were kept at 50 ꢀC under 40 bar of H2 for 1 h prior to the
addition of CO2. Upon addition of CO2 (60 bar), the reaction
was allowed to proceed for 1 h at 50 ꢀC. The formic acid to
amine mole ratio (AAR = mol HCO2H/mol NEt3) generated
with1as the catalyst was 0.50 (0.05 (an average of three runs),
whereas the AAR using 2was only 0.13 (0.01. Under identical
experimental conditions, complex 1 produced almost 4 times
as much formic acid as complex 2.
The catalytic activities of 1, 3b, and 3c were compared
directly using the same multivial reactor described in the
experiment above, with and without alcohol. Each vial in the
reactor contained CHCl3 (0.5 mL), NEt3 (0.5 mL), and the
catalyst (3 μmol), except for one vial, which contained no
catalyst (the control). The experiments involving alcohol
contained C6F5OH (18.4 mg, 0.1 mmol), and the conditions
for temperature, H2, CO2, and reaction times are the same as
The structure of the cation in 3b is shown in Figure 2.
Selected bond lengths and angles are listed in the Supporting
Information.14
The structure of 3b is best described as a distorted octahe-
dral, as opposed to a trigonal-bipyramidal structure in which
the acetate group is viewed as occupying a single coordina-
tion site. The Ru-P bonds for trans-related P(1) and P(2)
˚
average about 0.1 A longer than those trans to the acetate
group. The small volume occupied by the acetate group has
allowed the four phosphine ligands to move toward the
acetate group; for example, the P(1) and P(2) phosphines
are bent toward the acetate group and the P(3)-Ru(1)-P(4)
angle has opened to 96.9ꢀ. In the unit cell, there are no close
contacts made between ions.
The thermal ellipsoid drawing of the cation in 4 is shown in
Figure 3, and the structure is a ruthenium dimer of the form
{[(PMe3)3Ru]2(μ-Cl)2(μ-OAc)}BPh4. The relevant crystallo-
graphic parameters are found in the Supporting Informa-
tion.14
(16) The relative amount of 3b to 4 in this sample was obtained by
integration of the signals for each compound in the 31P{1H} NMR
spectrum taken overnight and with a 30 s relaxation delay.
(17) For description of the reactor, see ref 3b.
(15) B(ArF)4- = tetrakis(3,5-bis(trifluoromethyl)phenyl)borate.

Article
Organometallics, Vol. 28, No. 18, 2009 5471
1
CDCl3, and NMR spectra were recorded. The H NMR
spectrum had a large signal at 8.7 ppm (s) corresponding to
the formate group of [HþDBU][O2CH-] and a broad signal
at 11.2 ppm corresponding to the protonated base, HþDBU.
In addition, the spectrum had signals for DBU, PMe3, and
-
OAc- ranging from 1.4 to 3.5 ppm and signals for B(ArF)4
ranging from 7.5 to 7.8 ppm. The only signals in the 31P{1H}
NMR spectrum were two triplets at δ 16.8 and 0.3, corre-
sponding to the new ruthenium product 9.
Since the ruthenium compounds have poor solubility in
the supercritical CO2/H2 mixture, the remainder of the NMR
investigation was performed with THF or THF-d8 as the
solvent.
NMR Investigation of the Reaction of 3b,c with H2 in THF.
In a typical experiment, 3b or 3c was dissolved in THF or
THF-d8 (ca. 0.02 M solutions) in the drybox and ca. 10 equiv
DBU was added. A 200 μL aliquot of this solution was
dispensed into the PEEK NMR tube. Upon insertion of the
PEEK tube into the NMR probe, hydrogen (6.7 bar) was
introduced and NMR spectra were recorded. Figure 4 illus-
trates the 31P{1H} NMR spectra obtained at various times
during the production of formic acid catalyzed by 3c in the
presence of DBU as the base.
Figure 3. Thermal ellipsoid drawing of the cation in
{[(PMe3)3Ru]2(μ-Cl)2(μ-OAc)}BPh4 (4) showing 50% probabil-
ity contours. Hydrogen atoms and the BPh4- anion have been
omitted for clarity.
Table 2. Comparison of Catalytic Activity of 1, 3b, and 3c
alcohol present
(yes/no)
standard deviation
for AAR
catalyst
average AARa
The following NMR descriptions are for 3b, and there are
no significant differences in the NMR spectra for the experi-
ments in which 3c was used as the catalyst. Without heating
above room temperature, no reaction occurs between 3b (or c),
H2, and DBU over the course of a day (Figure 4a). After
only 20 min at 50 ꢀC, a significant amount of the starting
material (3b or 3c) converts to two compounds, each having a
31P{1H} NMR A2B2 splitting pattern (Figure 4b). One of the
compounds was the known species cis-(PMe3)4RuH2 (5),11
which has two triplets at δ 1.0 and -6.7 in the 31P{1H} NMR
spectrum. The other compound, represented by two triplets
at δ 17.1 and 1.9 ppm, was an unidentified Ru-PMe3
compound that never exceeds ca. 10% of the total 31P signal.
In addition to the signals for these ruthenium compounds
and the starting material, there is also a small signal that
corresponds to free PMe3. As the reaction proceeds, the
intensity of the signals for 3b or 3c decrease, while those for
5 increase and those for the unidentified Ru-PMe3 compound
and free PMe3 disappear after ca. 3 h at 50 ꢀC. The 1H NMR
spectrum at this point has a broad signal at -10.0 ppm,
consistent with the reported NMR data for cis-(PMe3)4RuH2
(5). In addition, the H2 dissolved in THF was observed at
4.3 ppm. No conversion of 3b or 3c to 5 occurs in the absence of
base, even after 7 h at 50 ꢀC.
1
3b
3c
1
3b
yes
yes
yes
no
no
no
0.12
0.18
0.19
0.06
0.14
0.16
0.00
0.006
0.001
0.000
0.001
0.018
0.047
0.000
3c
no catalyst
yes
a The average AAR is for two runs (AAR = formic acid to amine
mole ratio).
described above. Table 2 lists the results. As expected,
3b and 3c perform about equally well compared to each
other with or without alcohol present, and both ionic
catalysts perform better than 1 in the absence of alcohol.
With alcohol present, 1 has slightly less catalytic activity than
3b and 3c.
The known compounds cis-(P(OEt)3)4RuCl2 and cis-(P-
8
(OEt)3)4RuH2 have substantial solubility in supercritical
CO2. However, these complexes demonstrated no activity
toward CO2 hydrogenation with or without alcohol.
NMR Investigation of CO2 Hydrogenation by 3b,c in
scCO2. Experiments were performed with compounds 3b
and 3c in which the solids were loaded into the PEEK NMR
tube under nitrogen, along with 50 μL of DBU (0.33 mmol,
ca. 37 equiv for 3b and ca. 35 equiv for 3c) and 15 μL of C6H6
(for shimming purposes). Hydrogen (20 bar) and carbon
dioxide (94 bar) were introduced into the cell, and the probe
was heated at 50 ꢀC. The 31P{1H} NMR spectrum of the
reaction mixture obtained after 30 min of heating contained
small signals at δ 21.0 (t) and 5.8 (t), corresponding to a
new ruthenium-trimethylphosphine complex (9); however,
the S/N ratio of the spectrum was very poor. Unfortunately,
the ruthenium compounds and the [HþDBU][O2CH-] salt
formed during the reaction were not soluble enough in the
supercritical CO2/H2 mixture to obtain good-quality 31P-
{1H} or 1H NMR spectra. After heating the mixture for ca. 2
h, the NMR probe was cooled to room temperature and the
PEEK tube vented to the atmosphere. The contents of the
tube were transferred to a vial, the tube was rinsed several
times with CH2Cl2, and the volatiles were removed under a
flow of nitrogen. The resulting yellow oil was dissolved in
The 1H NMR spectrum of 5 at -80 ꢀC is nearly the same as
that at 50 ꢀC, with slight changes in some of the chemical
shifts, and the Ru-hydride signal at -10.25 ppm has become
a sharp doublet of quartets. The 31P{1H} NMR spectrum
contains only resonances for 5; however the signals are
shifted downfield by ca. 1 ppm compared to the spectrum
recorded at 50 ꢀC.
NMR Investigation of the Reaction of 5 with CO2 and H2.
Upon complete formation of 5, the NMR probe was cooled
to -80 ꢀC and CO2 (27 bar) was added. The NMR spectra
recorded before and after the addition of CO2 (27 bar) to the
PEEK tube at -80 ꢀC were identical. However, raising the
probe temperature to -50 ꢀC results in conversion of 5 to a
species with an A2BC splitting pattern in the phosphorus
NMR spectrum within 1 h (Figure 4c). The phosphorus
NMR data for this new compound (δ21.0, m; 1.3, m; -10.6, m
in a 1:2:1 ratio) is consistent with the NMR data reported

5472 Organometallics, Vol. 28, No. 18, 2009
Getty et al.
Figure 4. 31P{1H} NMR spectra of [(PMe3)4Ru(η2-OAc)]B(ArF)4 (3c) combined with DBU and H2 (6 bar), then CO2 (27 bar), in
THF-d8.
1
for cis-(PMe3)4Ru(O2CH)H (6).3d Changes in the proton
NMR spectrum occur in the hydride and the formate
regions. The signal for the Ru-H was -8.15 ppm (dq), while
the signal corresponding to the Ru-O2CH appeared at 8.11
ppm (s). This experiment was repeated using 13CO2. The
signal corresponding to the Ru-O213CH functionality ap-
peared as a doublet centered at 8.11 ppm (1JC-H = 188 Hz)
in the 1H NMR spectrum. In the 13C{1H} NMR spectrum,
the signal corresponding to the Ru-O213CH group appeared
as a singlet at 174.0 ppm, and this signal became a doublet
(1JC-H = 190 Hz) when the spectrum was recorded with the
proton-decoupler turned off. Similar results were obtained
for 3c; however only one side of the doublet assigned as the
Figure 4d). The H NMR spectrum no longer contained the
signal at -8.15 ppm for the Ru-H group, and no other signals
were observed upfield of 0 ppm. The singlet at 8.11 ppm
corresponding to the Ru-O2CH functionality is no longer
present, and the only signal in this region (8.65 ppm, s) is
assigned to the formate proton of the [HþDBU][O2CH-]
salt.18 The small singlet at 4.3 ppm is no longer present in the
spectrum, indicating that the hydrogen has been consumed.
When 13CO2 was used, the signal for the carbon-bound
formate proton in the 1H NMR spectrum of the
[HþDBU][O213CH-] salt appeared as a doublet centered at
8.5 ppm (1JC-H = 181 Hz). The corresponding 13C{1H} NMR
spectrum contained a singlet at 167.8 ppm corresponding
to the [HþDBU][O213CH-] salt, which became a doublet
(1JC-H = 183 Hz) when the proton-decoupler was turned
off. Since the formate regions of the proton and carbon (in the
cases of using 13CO2) NMR spectra of the final reaction
mixture contained only the corresponding signals for the
[HþDBU][O2CH-] salt, 9 is most likely not a Ru-formate
Ru-O213CH group was observed in the H NMR spectrum
1
when using 13CO2 due to overlap with the B(ArF)4- signal at
7.7 ppm.
The NMR probe temperature was raised to 0 ꢀC, following
which 31P{1H} NMR spectra were recorded at 5 min inter-
vals for ca. 30 min. The intensities of the signals for complex 6
decreased as those corresponding to several intermediates
increased. After 15 min at 0 ꢀC, there were four sets of triplets
corresponding to four Ru-phosphine compounds with A2B2
splitting patterns in approximately equal concentrations
to each other and to 6. Three possible structures consistent
with such patterns are cis-(PMe3)4Ru(O2CH)2 (7), [(PMe3)4-
Ru(η2-O2CH)]O2CH (8), and {[(PMe3)4Ru]2(μ-OAc)2}X2
(9, X = BPh4-, B(ArF)4-, or O2CH-) (see Discussion). In
addition, there were two other sets of signals corresponding
to Ru-phosphine complexes with A2B2 splitting patterns in
lower concentrations.
complex. A possible structure for 9 is a Ru-acetate complex of
the form {[(PMe3)4Ru]2(μ-OAc)2}X2 (X = BPh4-, B(ArF)4
-
,
(18) This assignment is based on the 1H NMR spectrum of the
[HþDBU][O2CH-] salt prepared independently. Into a 50 mL beaker
was dispensed 1.5 mL of DBU (0.010 mol), 10 mL of Et2O, and then 0.48
mL of formic acid (0.012 mol). The cloudy mixture was stirred vigor-
ously for ca. 10 min. Upon letting the mixture stand for 5 min, an oil
separated from the Et2O. The ether was decanted and the oil was washed
with fresh ether (3 ꢀ 5 mL). The beaker was placed under vacuum
overnight to remove residual Et2O and H2O. The resulting white solid
was stored in the drybox (1.41 g, 7.11 mmol, 71% yield). 1H NMR
(THF-d8): δ 1.65-1.73 (m, 6H, DBU), 1.94 (m, 2H, DBU), 2.89 (m, 2H,
DBU), 3.35 (m, 2H, DBU), 3.46 (m, 2H, DBU), 3.52 (m, 2H, DBU), 8.46
(s, 1H, O2CH). 13C{1H} NMR (THF-d8): δ 20.84 (s, 1C, DBU), 25.99 (s,
1C, DBU), 28.03 (s, 1C, DBU), 30.16 (s, 1C, DBU), 32.30 (s, 1C, DBU),
38.91 (s, 1C, DBU), 49.30 (s, 1C, DBU), 54.45 (s, 1C, DBU), 166.87 (s,
1C, DBU), 167.16 (s, 1C, O2CH).
The temperature of the NMR probe was raised to 24 ꢀC,
and within ca. 3 h all of the Ru-phosphine species converged
to a single ruthenium product, 9, with an A2B2 splitting
pattern in the 31P{1H} NMR spectrum (δ 17.3, t; 2.1, t,

Article
Organometallics, Vol. 28, No. 18, 2009 5473
2
or O2CH-). The proton and phosphorus NMR spectra de-
scribed above are identical to the ones obtained by performing
the catalysis reaction with 3b,c in supercritical CO2 (vide
supra). In addition, complex 9 is obtained by releasing
the pressure on the PEEK NMR cell after conversion of 3b,c
to 5, without the addition of CO2, further supporting our
assignment of 9 as a Ru-acetate complex rather than a Ru-
formate.
When CO2 was added to the PEEK cell containing the
ruthenium-dihydride (5) prepared in situ, and the tempera-
ture was kept at 50 ꢀC, complex 6 was never observed by
phosphorus NMR, as 5 was rapidly converted to 9
and [HþDBU][O2CH-] was produced (the reaction was
complete within 1 h). Free PMe3 was observed in the
phosphorus NMR spectra recorded during this conversion
period (in either THF or CO2 solvent). In addition, a small
amount of trimethylphosphine oxide was observed as a
singlet at δ 33.1.
After the reaction mixture was left overnight at room
temperature to confirm that no further changes were occur-
ring, the PEEK NMR cell was vented to the atmosphere. The
contents of the cell were removed by pipet, the cell was
washed with five 0.5 mL aliquots of CH2Cl2 or CHCl3, and
the volatiles were removed from the collected solution under
a flow of nitrogen. The pale yellow oily residue was dissolved
in CDCl3, and NMR spectra were recorded. The 31P{1H}
NMR spectrum contains two large triplets at δ 17.0 and 0.40
(9) and a small singlet at δ 39.2 (OPMe3). This spectrum is
slightly different than the one collected before the PEEK cell
was vented to the atmosphere. The 31P{1H} NMR spectrum
recorded while the solution was still under CO2 pressure
contained the two triplets with a separation (Δδ) of 15.3 ppm
(in THF). However, the two triplets in the spectrum obtained
after release of the CO2 and washes with CDCl3 have a Δδ of
16.6 ppm (in CDCl3). The difference in the Δδ values for
these spectra is small, and there are no significant differences
in the proton or carbon NMR spectra taken in THF or THF-
d8 before venting the cell versus the spectra taken in CDCl3
after venting and workup in air; therefore, we conclude that 9
remains unchanged. The signals for DBU and PMe3 range
from 1.3 to 3.3 ppm in the 1H NMR spectrum, and the signal
corresponding to OAc- is a singlet at 1.99 ppm. The broad
signal that was once at ca. 11 ppm (HþDBU) moved upfield
to ca. 4 ppm. The carbon-bound formate proton of the
[HþDBU][O2CH-] salt appears as a singlet at 8.75 ppm;
however, this signal appeared as a doublet centered at 8.78
ppm (1JC-H = 188 Hz) in the experiments where 13CO2 was
used. For these experiments, carbon NMR spectra were
recorded and the formate signal appears as a singlet at
169.4 ppm, whereas it appears as a doublet centered at
169.4 ppm (1JC-H = 190 Hz) when the proton-decoupler is
turned off.
NMR tube were removed and analyzed in CH2Cl2 by H
NMR. The 31P{1H} NMR spectra recorded during this
experiment were identical to those described above when
using H2 rather than D2.
NMR Investigation of the Reactivity of 9. The CDCl3 was
evaporated from the product mixture obtained originally
from the reaction of 3b with H2, DBU, and CO2 in THF, and
the resulting pale yellow oily residue was redissolved in 0.5
mL of THF-d8 under an inert atmosphere. DBU was added
(15 μL, 0.1 mmol), 200 μL of this solution was dispensed into
the PEEK NMR cell, and the cell was then sealed. Hydrogen
(6.7 bar) was added to the cell once it was inserted into the
NMR probe, and the probe was heated to 50 ꢀC while
collecting NMR spectra to observe the progress of the
reaction. Complex 9 was converted to cis-(PMe3)4RuH2 (5)
within ca. 2 h. The NMR probe was cooled to -80 ꢀC, CO2
was added (10 bar), and the probe was raised to -50 ꢀC. The
reaction was monitored by 31P{1H} NMR spectroscopy for
1 h until all of 5 had been converted to cis-(PMe3)4RuH-
(CO2H) (6). Finally, the probe was raised to room tempera-
ture, and the ruthenium product 9 was observed once again
by 1H and 31P{1H} NMR spectroscopy.
Reaction of 3b with d9-PMe3. The exchange of phosphine
ligands of 3b was confirmed by running a series of experi-
ments with d9-PMe3 and 3b containing protonated PMe3
in THF. In all cases, the 31P{1H} signals of the free
phosphines contained both deuterated and protonated
PMe3.
NMR Investigation of CO2 Hydrogenation in the Presence
of Excess PMe3. A solution containing 3b, 10 equiv of DBU,
and 20 equiv of PMe3 in THF-d8 was pressurized with 6.7 bar
of H2 in a PEEK NMR cell and heated to 50 ꢀC. Significant
conversion of the starting material occurred within 30 min to
give two new Ru-phosphine complexes as observed by 31P-
{1H} NMR spectroscopy, the expected cis-(PMe3)4RuH2
complex (5) and the other assigned as [(PMe3)5RuH]BPh4
(10) (-9.0 ppm, d, 4P; -22.3 ppm, quintet, 1P; 2JPP = 26 Hz)
based on the known compound [(PMe3)5RuH]PF6 (-8.5
ppm, d; -21.9 ppm, quintet; 2JPP = 26 Hz in acetone-d6).19
The Ru-H signal for 10 appeared at -11.5 ppm (dq) in the 1H
NMR spectrum (-11.3 ppm according to the literature19).
Complexes 5 and 10 appeared to reach equilibrium after ca.
5 h at 50 ꢀC and were in a 1:1 ratio. The probe temperature
was lowered to -80 ꢀC, and CO2 (27 bar) was added to the
cell. No reactivity was observed until the NMR probe
temperature was raised from -80 to -50 ꢀC. Within 15 min,
a significant amount of cis-(PMe3)4RuH2 (5) had con-
verted to cis-(PMe3)4RuH(O2CH) (6) as expected; however,
very little of complex 10 appeared to react. Over the course of
ca. 3 h at -50 ꢀC, the concentration of complex 10 dimin-
ished appreciably but remained observable in the 31P{1H}
NMR spectrum. Upon warming to 24 ꢀC, the expected
ruthenium product, 9, was the only Ru-phosphine complex
observed.
The NMR experiment was also performed using D2
instead of H2. The ruthenium-bound deuterium atoms of
5-d2 appear as a multiplet at -9.9 ppm at -80 ꢀC in the 2H
NMR spectrum (THF). After addition of CO2 at -80 ꢀC,
raising the probe to -50 ꢀC for 1 h, then lowering the
temperature back to -80 ꢀC in order to acquire the 2H
NMR spectrum without further reactivity, the Ru-D signal
of 6-d2 appears as a multiplet at -8.2 ppm and the Ru-O2CD
signal appears as a singlet at 8.6 ppm. After sitting overnight
at 24 ꢀC, the sample’s 2H NMR spectrum shows a signal at
8.1 ppm for the O2CD of the [HþDBU][O2CD-] salt. This
signal moved to 8.6 ppm when the contents of the PEEK
The contents of the NMR cell were dissolved in CDCl3
after the gases were vented. The 31P{1H} NMR spectrum
showed both complex 9 and complex 10 in a ratio of 1.4:1
(9:10). The 1H NMR spectrum contained signals for
[HþDBU][O2CH-] and 9, as well as a Ru-H signal corre-
sponding to 10.
(19) (a) Werner, H.; Werner, R. J. Organomet. Chem. 1981, 209, C60–
C64. (b) Kayaki, Y.; Ikeda, H.; Tsurumaki, J.-I.; Shimizu, I.; Yamamoto, A.
Bull. Chem. Soc. Jpn. 2008, 81, 1053–1061.

5474 Organometallics, Vol. 28, No. 18, 2009
Scheme 2. Proposed Mechanism for Generation of 5 from 3b,c, H2, and a Base (DBU)
Getty et al.
Discussion
[(depe)2RuCl(η2-H2)]þ as the major product. Adding 1 equiv
of a base (OR- or Et3N) results in the formation of
[(depe)2RuH(η2-H2)]þ, and further addition of base gives
(depe)2RuH2.
Mechanism of CO2 Hydrogenation Catalyzed by the RuII
Compounds 3b,c, 5, and 10. Generation of cis-(PMe3)4RuH2
(5) from 3b,c. Scheme 2 outlines our proposed mechanism for
the formation of the CO2 hydrogenation catalyst precursor,
5, from the [(PMe3)4Ru(η2-OAc)]þ cation.
There have also been several studies related to the acidity
of metal-bound dihydrogen,11,21 and the pKa of
[(PMe3)4RuH(η2-H2)]þ (intermediate C in Scheme 2) has
recently been reported to be 16.6 on the THF scale and 11.2
on the aqueous scale.21a Although we propose that inter-
mediate B is formed by intramolecular deprotonation of A
by OAc-, it is also possible that intermediate B is formed by
intermolecular deprotonation of intermediate A by DBU
followed by dissociation of acetate. The proposed mechan-
ism in Scheme 2 is consistent with the fact that no reaction
occurs between 3b,c, H2, and CO2 in the absence of base. The
base is required not only to trap the formic acid product that
is formed during this catalytic process, but also to generate
the RuII-dihydride complex, 5.
Mechanism of Insertion of CO2 into the Ru-H Bond of 5.
The insertion of CO2 into one of the Ru-H bonds of 5
to generate cis-(PMe3)4RuH(O2CH) (6) was observed even
at -50 ꢀC. It therefore has a very low activation energy.
There are several known examples of stoichiometric inser-
tion of CO2 into RuII-H bonds.22 For complex 5, we
can envisage three plausible mechanisms for the insertion:
(a) direct concerted reaction of the Ru-H bond with CO2
via “σ-bond metathesis”23 without prior coordination
of CO2 to the complex, (b) loss of a phosphine ligand
followed by CO2 coordination and then insertion, or
(c) effective loss of a hydride (by protonation and then
H2 loss). Each of these three possibilities will now be
assessed.
While the PMe3 ligands of 3b,c are relatively labile, as
demonstrated by the exchange reaction with d9-PMe3 at
room temperature, we believe a pathway involving phos-
phine loss is not the predominant mechanism for the con-
version of 3b,c to 5. The rate of 5 formation in the presence of
excess PMe3 was not significantly slower than in the absence
of added PMe3, and a significant rate decrease should have
been observed if phosphine loss from 3b,c was required in
making the Ru-dihydride, 5. This result is consistent with the
results published by Jessop et al. for the hydrogenation of
CO2 with various RuII-phosphine complexes in supercritical
carbon dioxide in which the presence of excess phosphine
(6-20 equiv) did not affect the yield of formic acid obtained
after 30 min.3b,3d,3g,9 Instead of phosphine loss from 3b,c, we
propose that hydrogen binds to the ruthenium center at a site
made available by the opening of the acetate chelate to form
intermediate A. The η2-dihydrogen ligand is relatively acidic
and can be deprotonated by the relatively basic acetate
ligand to give the unsaturated RuII-hydride intermediate B
and acetic acid, which is trapped by the DBU. Intermediate B
has a vacant site that would allow another dihydrogen
molecule to bind to the ruthenium center and produce the
RuII-η2-dihydrogen-hydride intermediate C. The reversible
binding and loss of H2 through a mechanism similar to B and
C has been reported.11 The DBU present in the solution
could deprotonate the η2-dihydrogen ligand to form cis-(P-
Me3)4RuH2 (5) and the [HþDBU][X-] salt.
There are numerous examples of syntheses of RuII-hy-
dride complexes using H2 and a base.20 For instance, Cap-
pellani et al. reported the first example of two successive
heterolytic cleavages of hydrogen where the Ru-η2-dihydro-
gen intermediates were identified, and the proposed mechan-
ism is similar to the one shown in Scheme 2.20a The
cis-(depe)2RuCl2 (depe = Et2PCH2CH2PEt2) complex re-
acts with 1 equiv of NaBPh4 and 1 atm of H2 in THF to form
(21) (a) Abdur-Rashid, K.; Fong, T. P.; Greaves, B.; Gusev, D. G.;
Hinman, J. G.; Landau, S. E.; Lough, A. J.; Morris, R. H. J. Am. Chem.
Soc. 2000, 122, 9155. For reviews, see: (b) Jessop, P. G.; Morris, R. H.
Coord. Chem. Rev. 1992, 121, 155. (c) Heinekey, D. M.; Oldham, W. J., Jr.
Chem. Rev. 1993, 93, 913.
(22) (a) Komiya, S.; Yamamoto, A. J. Organomet. Chem. 1972, 46,
C58. (b) Kolomnikov, I. S.; Lobeeva, T. S.; Vol'pin, M. E. Bull. Acad. Sci.
USSR, Div. Chem. Sci. (Engl. Transl.) 1972, 2263. (c) Kolomnikov, I. S.;
Gusev, A. I.; Aleksandrov, G. G.; Lobeeva, T. S.; Struchkov, Yu. T.; Vol'pin,
M. E. J. Organomet. Chem. 1973, 59, 349. (d) Jia, G.; Meek, D. W. Inorg.
Chem. 1991, 30, 1953. (e) Whittlesey, M. K.; Perutz, R. N.; Moore, M. H.
Organometallics 1996, 15, 5166. (f) Reference 3d. (g) Yin, C.; Xu, Z.; Yang,
S.-Y.; Ng, S. M.; Wong, K. Y.; Lin, Z.; Lau, C. P. Organometallics 2001, 20,
1216.
(20) (a) Cappellani, E. P.; Maltby, P. A.; Morris, R. H.; Schweitzer,
C. T.; Steele, M. R. Inorg. Chem. 1989, 28, 4437. (b) Bautista, M. T.;
Cappellani, E. P.; Drouin, S. D.; Morris, R. H.; Schweitzer, C. T.; Sella, A.;
Zubkowski, J. J. Am. Chem. Soc. 1991, 113, 4876. (c) Mezzetti, A.; Del
Zotto, A.; Rigo, P.; Farnetti, E. J. Chem. Soc., Dalton Trans. 1991, 1525. (d)
Frediani, P.; Faggi, C.; Salvini, A.; Bianchi, M.; Piacenti, F. Inorg. Chim.
Acta 1998, 272, 141.
(23) For examples of CO2 insertion into the Ru-H bond in which the
proposed mechanism is direct attack of the Ru-H moiety on CO2, see
refs 22e-22g. For reviews discussing the mechanisms of CO2 insertion
into M-H bonds (M = transition metal) in relation to transition metal-
catalyzed homogeneous CO2 hydrogenation, see ref 5.

Article
The first mechanism, CO2 insertion without prior coordi-
Organometallics, Vol. 28, No. 18, 2009 5475
Catalytic Production of Formic Acid. The insertion of CO2
into a Ru-H bond of 5 forms a key part of the mechanism
for the hydrogenation of CO2 catalyzed by this complex.
While the published mechanism3d,8 shows CO2 insertion
after phosphine loss, that mechanism is no longer considered
likely, as discussed above. The new “hydride-loss” mechan-
ism for CO2 insertion should then be incorporated into the
proposed catalytic cycle for CO2 hydrogenation (as shown in
Scheme 3).
The generation of the active catalyst B from the ruthe-
nium-dihydride, 5, must be possible because the NMR
experiments were designed such that 5 was allowed to form
from 3b,c in situ before CO2 was added to the system.
However, in the catalytic experiments 3b,c likely reacts with
H2 and CO2 in the presence of a base to produce formic acid
without generating 5. Instead, once intermediate B forms, it
can react with CO2 to generate E, then the CO2 can insert into
the Ru-H bond to give the intermediate F, which reacts with
H2 to produce D, and D undergoes an intramolecular
deprotonation of the η2-dihydrogen ligand by the formate
ligand to release formic acid and regenerate B (Scheme 3).
Thus, under many of the reactor-type conditions used to
study the yield of formic acid produced via complexes 1 or
3b,c, compounds 5 and 6 may not be intermediates in the
mechanism.
Transient species B, E, F, and D in the proposed cycle
are not detected spectroscopically, but are in equilibrium
with more stable coordinatively saturated complexes. Due
to the transient nature of these intermediates, none of them
were detected during the NMR experiments. This mechan-
ism involves a cationic ruthenium(II) species as the
active catalyst, which is similar to the proposed mechanism
for CO2 hydrogenation via the cationic rhodium complex
[(PMe2Ph)3RhH2(solv)]BF4 (solv = H2O or THF).27
During the experiments in which excess PMe3 was present,
CO2 was added to the 1:1 mixture of cis-(PMe3)4RuH2 (5)
and [(PMe3)5RuH]BPh4 (10) at -50 ꢀC. Complex 10 ap-
peared to react more slowly than 5 and did not completely
convert to 6. This lower reactivity of 10 with CO2 is con-
sistent with our proposal that insertion of CO2 into a Ru-H
bond requires an empty site at the Ru. In the catalytic
experiments described here, and in those performed in
supercritical CO2, the concentration of added phosphine
(20 equiv) is not large enough to successfully compete for B
with the large concentration of CO2 (the mole fraction of
CO2 in THF at 27 bar is greater than 10%). Only in a
situation where the added phosphine concentration is closer
to that of CO2 would formation of [(PMe3)5RuHþ] be
expected with an apparent decrease in the rate of CO2
hydrogenation. Therefore we would expect that the forma-
tion of [(PMe3)5RuH]þ would inhibit the insertion of CO2 by
the hydride-loss mechanism only if the concentration of
PMe3 is very high, especially if it is higher than the concen-
tration of CO2.
nation, was evaluated by Musashi and Sakaki using DFT
calculations on the (PH3)4RuH2 complex as a model for 5.24
Their calculations indicated that the activation energy, Ea,
for CO2 insertion into the Ru-H bond of the six-coordinate
(PH3)4RuH2 complex was much higher (29.3 kcal/mol) than
the Ea for CO2 insertion into the Ru-H bond of the five-
coordinate (PH3)3RuH2 complex that is generated by phos-
phine dissociation (10.3 kcal/mol). The conclusion was that
generation of an empty site is necessary for the CO2 insertion
reaction. Our low-temperature spectroscopic measurements
confirm that the reaction must have a very low activation
energy. We therefore consider CO2 insertion without an
empty site an unlikely mechanism.
The second plausible mechanism is phosphine loss followed
by CO2 coordination and then insertion. This is the mechan-
ism that was first proposed for this catalytic system.3d How-
ever, Jessop et al. reported previously that adding 20 equiv of
PMe3 to cis-(PMe3)4RuH2 in supercritical carbon dioxide for
the CO2 hydrogenation reactionhad essentially no effect; even
a very large amount of added PMe3 (13 000 equiv, enough for
a solvent effect to appear) only slowed the rate of reaction and
did not stop it.3d We also found (vide supra) that the insertion
of CO2 into the Ru-H bond of 5 is not significantly inhibited,
even at -50 ꢀC, by the presence of a 20-fold excess of PMe3.
Thus we conclude that CO2 insertion into the Ru-H bond of
electron-rich complexes (such as 5) does not involve loss of a
phosphine ligand.
The third plausible mechanism is generation of an open
site by effective loss of a hydride. This could take place by
protonation of one of the hydride ligands of 5 by protonated
DBU, generating a RuII-η2-dihydrogen-hydride intermedi-
ate (C) that would lose H2 to make a site available for CO2
(the sequence 5 f C f B in Scheme 2 and 5 f B f E f F f 6
in Scheme 3). This mechanism is consistent with Musashi and
Sakaki’s conclusion that generation of an empty site is
required for CO2 insertion, but, unfortunately, those authors
did not evaluate this “hydride-loss” route for the generation
of the empty site. We know that such a proton transfer from
HþDBU to 5 is feasible because [(PMe3)4Ru(η2-H2)H]þ (pKa
16.6 on the THF scale)20a and HþDBU (pKa 16.8 on the same
scale)25 are reported to have almost identical acidity. The
analogous reaction of 5 with [NH4]PF6 to give [(PMe3)4-
Ru(NH3)H]þ is known in the literature.26
The hydride-loss pathway, too, could be inhibited by the
presence of excess PMe3, by the conversion of 5 via B to
[(PMe3)5RuH]þ. To what extent, then, would one expect
added PMe3 to inhibit this pathway? Fortunately, the equi-
librium between 5 and [(PMe3)5RuH]þ seems to favor 5
except at very high PMe3 concentrations, whereas the equi-
librium between [(PMe3)3RuH2] and 5 favors 5 even in the
absence of added PMe3. Therefore one would expect that
small amounts of added PMe3 would not significantly inhibit
the hydride-loss pathway nearly as readily as they would the
phosphine-loss pathway.
In addition, we propose that the formic acid is released
from the six-coordinate intermediate D (Scheme 3) via a σ-
bond metathesis pathway, rather than reductive elimination
of HO2CH from a ruthenium-hydrido-formate complex
followed by H2 oxidative addition to regenerate 5. This is
in agreement with the calculations presented by Musashi
and Sakaki as the lowest energy mechanism for formic
acid release.
For these reasons, we conclude that the hydride-loss
mechanism is consistent with the observations and Musashi
and Sakaki’s conclusion that an empty site is required. The
other two mechanisms are considered much less likely.
(24) Musashi, Y.; Sakaki, S. J. Am. Chem. Soc. 2000, 122, 3867.
(25) Kaljurand, I.; Kutt, A.; Soovali, L.; Rodima, T.; Maemets, V.;
Leito, I.; Koppel, I. A. J. Org. Chem. 2005, 70, 1019–1028.
(26) Rappert, T.; Yamamoto, A. Organometallics 1994, 12, 4984–
4993.
(27) Tsai, J.-C.; Nicholas, K. M. J. Am. Chem. Soc. 1992, 114, 5117.

5476 Organometallics, Vol. 28, No. 18, 2009
Getty et al.
Scheme 3. Proposed Mechanism for CO2 Hydrogenation into Formic Acid via 5 in the Presence of a Base (DBU)
The lack of reactivity of cis-(P(OEt)3)4RuH2 toward CO2
insertion could be a result of the electron-poor nature of this
complex that does not allow for protonation of the Ru-H
moiety by DBUHþ. The phosphite ligands do not appear to
be labile,28 thus the open site required at the electron-poor
ruthenium center cannot be made available for CO2 coordi-
nation when using cis-(P(OEt)3)4RuH2.
Scheme 4. Proposed Termination of the Catalytic Cycle and
Structure of the Ruthenium Product {[(PMe3)4Ru]2(μ-OAc)2}X2
(9, X = BPh4-, B(ArF)4-, or O2CH-)
Fate of the Catalyst after Termination of the Catalysis. The
termination of the catalysis occurs when the H2 and DBU
(the limiting reagents in the NMR experiments) are con-
sumed. In the absence of H2, the catalysis should stall at
species F (Scheme 3), which is likely in equilibrium with the
coordinatively saturated species [(PMe3)4Ru(η2-O2CH)]X
(X = BPh4-, B(ArF)4-, or O2CH-). The 31P{1H} NMR
spectrum of the ruthenium-containing product after cataly-
sis has an A2B2 splitting pattern for the phosphines, which is
consistent with the [(PMe3)4Ru(η2-O2CH)]X structure but
1
has no additional formate signals in the H NMR or 13C
NMR spectra other than those for [HþDBU][O2CH-]. To
explain this, we suggest the Ru-coordinated formate ligand
(28) Gerlach et al. have reported the reaction chemistry of various six-
coordinate RuII-dihydride complexes containing phosphine or phos-
phite ligands ( Gerlach, D. H.; Peet, W. G.; Muetterties, E. L. J. Am.
Chem. Soc. 1972, 94, 4545). They observe that exchange reactions are much
more facile at the RuII-phosphine complexes compared to the RuII-phosphite
compounds. We obtained similar results during an attempt to prepare 5 from
cis-(P(OEt)3)4RuH2 and PMe3. Since PMe3 is a much better donor compared
to P(OEt)3, exchange of the phosphites for the phosphines at the RuII center
was expected to occur. However, combinations of either a PMe3/hexane
of [(PMe3)4Ru(η2-O2CH)]X exchanges with the more basic
acetate anion of the [HþDBU][OAc-] salt to generate the
ruthenium dimer {[(PMe3)4Ru]2(μ-OAc)2}X2 (9) and
[HþDBU][O2CH-] (Scheme 4).
We do not observe the Ru-acetate complex 3b,c as the
ruthenium product of the CO2 hydrogenation reaction, but
the NMR spectra at the termination of the reaction indicate
solution of cis-(P(OEt)3)4RuH2 or
a neat PMe3 solution of cis-(P-
(OEt)3)4RuH2 and heating for four days at 85 ꢀC resulted in no changes in
the appearance of the solutions or the NMR spectra.

Article
Organometallics, Vol. 28, No. 18, 2009 5477
the ruthenium product (9) is very similar to 3. Complex 9 is
stable to the atmosphere and high-vacuum, as are 3b and 3c,
and 9 can be obtained by releasing the pressure on the PEEK
cell upon formation of the Ru-dihydride (5) when no CO2
has been added to the system. Thus, we suggest that the
ruthenium product (9) is the ruthenium-acetate dimer shown
in Scheme 4. The difference in energy between 3 (kinetic
product) and 9 (thermodynamic product) is expected to be
very small, and 9 may be slightly lower in energy due to less
strain in the eight-membered Ru-OAc-Ru-OAc ring ver-
sus the four-membered Ru-OAc ring in 3. Acetate ligands
are known to act as bridges between two ruthenium cen-
ters,29 and we observed such a bridge in the crystal structure
of 4 (see Figure 3). Our attempts to isolate and characterize 9
have been unsuccessful. However, we have observed that the
product mixture containing 9 will facilitate CO2 hydrogena-
tion when fresh DBU, H2, and CO2 are added. The NMR
spectra obtained before catalysis show that the ruthenium-
dihydride (5) is formed initially after ca. 2 h at 50 ꢀC in the
absence of CO2, and the ruthenium-hydrido-formate (6) is
generated at low temperature upon addition of CO2, just as if
complex 3b,c had been used as the starting material. These
results are consistent with our proposal that the final ruthe-
nium-containing product, 9, is very similar to 3.
Using high-pressure NMR spectroscopy and 3b,c as
precatalysts for CO2 hydrogenation in the presence of a
base (DBU), we have determined that the likely active
catalyst is an unsaturated, cationic ruthenium complex of
the form [(PMe3)4RuH]þ (B). The active catalyst can be
generated via cis-(PMe3)4RuH2 (5) by reaction of 3b,c with
H2 and DBU at elevated temperature. The formation of B
does not require phosphine dissociation from 3b,c or 5. No
reaction occurs without the DBU present, indicating it is
involved not only in trapping the formic acid product but
also in the generation of the catalyst precursor 5 when 3b,c
reacts with H2 in the absence of CO2 or in the generation of
the catalyst B when 3b,c reacts with a mixture of H2 and
CO2. We suggest that CO2 insertion into the Ru-H bond
involves precoordination of CO2 to the metal. The cis-(P-
(OEt)3)4RuCl2 and cis-(P(OEt)3)4RuH2 complexes were
completely inactive toward CO2 hydrogenation. Likely
reasons are the electron-withdrawing nature of the phos-
phite ligands and their inability to dissociate, causing the
Ru-phosphite complexes to be unable to create an open site
by either phosphite dissociation or protonation of a hy-
dride.
Termination of the catalytic cycle occurs when the H2 and
DBU (the limiting reagents) have been consumed, and a new
Ru-phosphine product (9) is generated. Complex 9 has not
been isolated and characterized; however, the product mix-
ture containing 9 can facilitate CO2 hydrogenation when
fresh DBU, H2, and CO2 are added. We propose that 9 is very
similar to 3 and is a ruthenium(II)-acetate dimer of the form
{[(PMe3)4Ru]2(η2-OAc)2}2þ that is produced by exchange of
acetate for the formate ligand in [(PMe3)4Ru(η2-O2CH)]þ
followed by dimerization.
Conclusions
One role of the alcohol cocatalyst in the CO2 hydrogena-
tion to formic acid system using cis-(PMe3)4RuCl(OAc) (1)
and a base is to facilitate the isomerization of 1 to
[(PMe3)4Ru(η2-OAc)]Cl (3a). The chloride ion of 3a was
exchanged for noncoordinating ions to produce the air-
stable catalyst precursors [(PMe3)4Ru(η2-OAc)]BPh4 (3b)
and [(PMe3)4Ru(η2-OAc)]B(ArF)4 (3c). These complexes
proved to perform as well as 1 in the presence of an alcohol
cocatalyst and, more importantly, in the absence of added
alcohol. Removing the alcohol cocatalyst from this system
eliminates the side reaction between base, CO2, and alcohol
to form alkyl carbonate salts.
Acknowledgment is made by A.D.G and J.C.L. for
support from the Director, Office of Science, Office of
Basic Energy Sciences, Chemical Sciences Division of the
U.S. Department of Energy, under Contract DE-ACO6-
76RLO 1830. Battelle operates the Pacific Northwest
National Laboratory for the Department of Energy. P.
G.J. and C.-C.T. gratefully acknowledge support from
the Office of Science, Office of Basic Energy Sciences,
Chemical Sciences Division of the U.S. Department of
Energy, under grant number DE-FG03-99ER14986.
(29) For examples of complexes that have been characterized by
X-ray crystallography containing at least two ruthenium centers bridged
by OAc-, see: (a) Collin, J.-P.; Joouaiti, A.; Sauvage, J.-P.; Kaska, W. C.;
McLoughlin, M. A.; Keder, N. L.; Harrision, W. T. A.; Stucky, G. D.
Inorg. Chem. 1990, 29, 2238. (b) Binamira-Soriaga, E.; Keder, N. L.; Kaska,
ꢀ
W. C. Inorg. Chem. 1990, 29, 3167. (c) Barral, M. C.; Jimenez-Aparicio, R.;
Kramolowsky, R.; Wagner, I. Polyhedron 1993, 12, 903. (d) Mintert, M.;
Sheldrick, W. S. Inorg. Chim. Acta 1995, 236, 13. (e) Tanase, T.; Yamada, Y.;
Tanaka, K.; Miyazu, T.; Kato, M.; Lee, K.; Sugihara, Y.; Mori, W.; Ichimura,
A.; Kinoshita, I.; Yamamoto, Y.; Haga, M.; Sasaki, Y.; Yano, S. Inorg. Chem.
1996, 35, 6230. (f) Ota, K.; Sasaki, H.; Matsui, T.; Hamaguchi, T.; Yamaguchi,
T.; Ito, T.; Kido, H.; Kubiak, C. P. Inorg. Chem. 1999, 38, 4070.
Supporting Information Available: Thermal ellipsoid draw-
ings for the X-ray crystal structure of {[(PMe3)3Ru]2(μ-Cl)3}Cl,
and the detailed crystallographic data for that complex, plus 2,
3b, and 4 are provided. This material is available free of charge
via the Internet at http://pubs.acs.org.