Spectroscopic characterization of alumina-supported bis(allyl)iridium complexes: Site-isolation, reactivity, and decomposition studies

Article abstract of DOI:10.1021/ic9021036

The covalent attachment of tris(allyl)iridium to partially dehydroxylated γ-alumina is found to proceed via surface hdroxyl group protonation of one allyl ligand to form an immobilized bis(allyl)iridium moiety, (=AIO)lr(allyl)₂, as characterized by CP-MAS ¹³C NMR, inductively coupled plasma-mass spectrometry, and lr L₃ edge X-ray absorption spectroscopy. Extended X-ray absorption fine-structure (EXAFS) measurements taken on unsupported lr(allyl)₃ and several associated tertiary phosphine addition complexes suggest that the η³-allyl ligands generally account for an Ir-C coordination number of 2 rather than 3, with an average Ir-C distance of 2.16 A. Using this knowledge, combined EXAFS and X-ray absorption near-edge structure studies reveal that a small amount of Iro is also formed upon reaction of lr(allyl)₃ with the surface. It was found that the addition of either 2,6-dimethylphenyl isocyanide or carbon monoxide to the supported complex allows spectroscopic identification of the supported bis(allyl)iridium complexes, (=AIO)lr(allyl)₂(CNAr) [Ar=2,6-(CH₃)₂C₆H₄] and (=AIO)lr(allyl)₂(CO)₂, respectively. Although samples of the supported bis(allyl)iridium complex are active for the dehydrogenation of cyclohexane to benzene at temperatures between 180 and 220 °C, in situ temperature-programmed reaction XAFS and continuous-flow reactor studies suggest that Iro nanoparticles, rather than a well-defined lr³⁺ complex, are responsible for the observed activity.

Full text of DOI:10.1021/ic9021036

Inorg. Chem. 2010, 49, 2247–2258 2247
DOI: 10.1021/ic9021036
Spectroscopic Characterization of Alumina-Supported Bis(allyl)iridium
Complexes: Site-Isolation, Reactivity, and Decomposition Studies
Ryan J. Trovitch,† Neng Guo,‡ Michael T. Janicke,† Hongbo Li,† Christopher L. Marshall,‡ Jeffrey T. Miller,‡
Alfred P. Sattelberger,*,§ Kevin D. John,*,† and R. Thomas Baker*,†,^
†Chemistry Division, Los Alamos National Laboratory, Los Alamos, New Mexico 87545, ‡Chemical Sciences
and Engineering Division, Argonne National Laboratory, Argonne, Illinois 60439, §Energy Sciences and
Engineering Directorate, Argonne National Laboratory, Argonne, Illinois 60439, and ^Centre for Catalysis
Research and Innovation and Department of Chemistry, University of Ottawa, Ottawa, Ontario K1N 6N5,
Canada
Received October 23, 2009
The covalent attachment of tris(allyl)iridium to partially dehydroxylated γ-alumina is found to proceed via surface
hydroxyl group protonation of one allyl ligand to form an immobilized bis(allyl)iridium moiety, (dAlO)Ir(allyl)2, as
characterized by CP-MAS 13C NMR, inductively coupled plasma-mass spectrometry, and Ir L3 edge X-ray absorption
spectroscopy. Extended X-ray absorption fine-structure (EXAFS) measurements taken on unsupported Ir(allyl)3 and
several associated tertiary phosphine addition complexes suggest that the η3-allyl ligands generally account for an
˚
Ir-C coordination number of 2 rather than 3, with an average Ir-C distance of 2.16 A. Using this knowledge,
combined EXAFS and X-ray absorption near-edge structure studies reveal that a small amount of Ir0 is also formed
upon reaction of Ir(allyl)3 with the surface. It was found that the addition of either 2,6-dimethylphenyl isocyanide or
carbon monoxide to the supported complex allows spectroscopic identification of the supported bis(allyl)iridium
complexes, (dAlO)Ir(allyl)2(CNAr) [Ar = 2,6-(CH3)2C6H4] and (dAlO)Ir(allyl)2(CO)2, respectively. Although samples
of thesupported bis(allyl)iridium complexare activefor thedehydrogenation of cyclohexane tobenzene at temperatures
between 180 and 220 ꢀC, in situ temperature-programmed reaction XAFS and continuous-flow reactor studies suggest
that Ir0 nanoparticles, rather than a well-defined Ir3þ complex, are responsible for the observed activity.
Introduction
through modification of the ligand environment.5 At the
interface between homogeneous and heterogeneous catalysis,
efforts to graft well-defined organometallic precatalysts onto
solid surfaces have multiplied rapidly over the past decade.6
This “heterogenization” of homogeneous catalysts can im-
prove their recyclability and stability, by limiting bimolecular
decomposition pathways, and enable tuning of the ligand
environment either before or after site isolation of the
organometallic moiety.
There are several well-investigated methods of covalently
tethering an organometallic complex to a surface; however,
these approaches can usually be placed into one of two
categories. The first category involves binding of the metal
complex to an electron-donating functional group, such as a
phosphine or N-heterocyclic carbene, which is already at-
tached to the surface. Notably, this methodology has proven
While both homogeneous1 and heterogeneous2 catalysts
have been invaluable to a wide array of industrial processes,
they each possess specific advantages that warrant their
application in a given transformation. For example, hetero-
geneous catalysts are generally easy to recycle3 and tend to be
fairly robust, making them ideal for transformations that
require elevated temperatures.2,4 On the other hand, homo-
geneous catalysts are often easier to analyze spectroscopically
and can be readily tuned for increased selectivity or activity
*To whom correspondence should be addressed. E-mail: asattelberger@
anl.gov (A.P.S.), kjohn@lanl.gov (K.D.J.), rbaker@uottawa.ca (R.T.B.).
(1) Cornils, B.; Herrmann, W. A. J. Catal. 2003, 216, 23–31.
(2) Weissermel, K.; Arpe, H.-J. Industrial Organic Chemistry; Wiley-VCH:
Weinheim, Germany, 2003.
(3) Thomas, J. M.; Raja, R. Top. Catal. 2006, 40, 3–17.
(4) Guzman, J.; Gates, B. C. Dalton Trans. 2003, 3303–3318.
(5) (a) Hegedus, L. S. Transition Metals in the Synthesis of Complex
Organic Molecules; University Science Books: Mill Valley, CA, 1999. (b)
Parshall, G. W.; Ittel, S. D. Homogeneous Catalysis: The Applications and
Chemistry of Catalysis by Soluble Transition Metal Complexes; John Wiley &
(6) (a) Gates, B. C. In Catalyst Preparation; Regalbuto, J. R., Ed.; Taylor &
Francis Publishing Group, CRC Press: New York, 2007; Chapter 10, pp 237-249.
(b) Severn, J. R.; Chadwick, J. C.; Duchateau, R.; Friedrichs, N. Chem. Rev.
€
ꢀ
Sons: New York, 1992. (c) Jakel, C.; Paciello, R. Chem. Rev. 2006, 106, 2912–
2005, 105, 4073–4147. (c) Coperet, C.; Chabanas, M.; Saint-Arroman, R. P.;
2942.
Basset, J.-M. Angew. Chem., Int. Ed. 2003, 42, 156–181.
r
2010 American Chemical Society
Published on Web 01/29/2010
pubs.acs.org/IC

2248 Inorganic Chemistry, Vol. 49, No. 5, 2010
Trovitch et al.
to be successful for the site isolation of olefin metathesis
catalysts,7 as well as several precious metal-based asymmetric
hydrogenation8 and transfer hydrogenation catalysts.9 The
second category includes cases where a reaction takes place
between the metal complex and the surface such that salt
metathesis or ligand protonation occurs, often resulting in
the formation of a metal-oxygen bond. Several advances
using this approach have been reported,7,10 with one notable
example being the preparation of a silica-supported tantalum
hydride species that is an active catalyst for alkane metath-
esis.11 This catalyst was prepared upon silanol protonation
of one or two neopentyl groups from Ta(CH2C(CH3)3)3-
(dCHC(CH3)3), followed by hydrogenation of the reaction
mixture. Although it was not the first report of alkane
metathesis,12 Basset’s system was the first to rely on a single
catalyst to effect this transformation, and turnover numbers
as high as 62 were realized for the conversion of butane into
other hydrocarbons.
This research sparked further interest in developing an
alkane metathesis process that could be used to upgrade low-
molecular-weight alkanes into a suitable transportation fuel
on a large scale, which could have a significant impact
when coupled with the Fischer-Tropsch process.13 A recent
approach to this problem, developed by the Goldman
and Brookhart groups,14 utilized homogeneous iridium
pincer dehydrogenation catalysts15 in tandem with either
a Schrock-type16 or a heterogeneous rhenium oxide17 ole-
fin metathesis catalyst to obtain a wide distribution
of linear alkanes from n-decane. Initially, the authors sought
to use Grubbs-type olefin metathesis catalysts;18 how-
ever, they determined that these complexes quickly deacti-
vated the iridium dehydrogenation catalysts.14 Additionally,
the authors reported that the rate of alkane metathesis had
decreased as a function of time when the molybdenum
alkylidene catalysts16 were employed. When Re2O7/Al2O3
was ultimately chosen as the olefin metathesis catalyst, it was
unclear as to whether the iridium-based dehydrogenation
catalyst had been adsorbed onto the alumina surface or
remained in solution.14
Surface-bound iridium complexes have been investigated
as catalysts for both alkene hydrogenation8b,c and alkane
dehydrogenation. Conversion of isopentane to isopentene
has been accomplished with several silica-supported iridium
complexes;19 however, these systems require temperatures
higher than 300 ꢀC to achieve reasonable conversion (>5%).
Site isolation of the highly active iridium pincer complexes
has also recently been accomplished by modification of the
para position of the ligand pyridyl ring, and their ability to
catalyze transfer dehydrogenation reactions has been inves-
tigated.20 Although these complexes are the most active
alkane dehydrogenation catalysts known to date, methods
of tethering them to a surface have thus far required compli-
cated ligand syntheses.20
In this contribution, our research team set out to prepare
immobilized, trivalent iridium complexes from easily synthe-
sized starting materials and to study their viability as alkane
dehydrogenation catalysts. With these goals in mind, we
decided to model our efforts after the well-studied immobi-
lization of Ir(acac)3,21 Ir(C2H4)2(acac),22 and Rh(allyl)3,23
because complexes of this type satisfy our requirement of
using a simple molecular synthon. Although the immobiliza-
tion reactions of Ir(acac)324 and Ir(C2H4)2(acac)25 have only
recently been reported to proceed through the protonation of
one acetylacetonate (acac) ligand, the site isolation of Rh-
(allyl)3 has been studied for decades and the outcome is
known tobehighly sensitive to both temperature and the type
of support used. The reaction of Rh(allyl)3 (in hexane at
ambient temperature) with a silica surface was first investi-
gated by the Schwartz group and was found to result in the
formation of a site-isolated bis(allyl)rhodium complex,
(tSiO)Rh(η3-allyl)2, along with 1 equiv of propene (eq 1).26
Foley and co-workers later suggested that (tSiO)Rh(η3-
allyl)2 was not the only product formed from this reaction
upon investigating its reactivity toward CO and H2. They
proposed that a fraction of the immobilized (tSiO)Rh(η3-
allyl)2 was further protonated by neighboring surface hydro-
xyl groups, resulting in the loss of 2 additional equiv of
propene and the formation of (tSiO)3Rh.27 A different
outcome altogether was proposed by the Iwasawa group,
whereupon the reaction of Rh(allyl)3 with TiO2 resulted in
ꢀ
(7) (a) Buchmeiser, M. R. Chem. Rev. 2009, 109, 303–321. (b) Coperet, C.;
Basset, J.-M. Adv. Synth. Catal. 2007, 349, 78–92.
(8) (a) Liu, G.; Liu, M.; Sun, Y.; Wang, J.; Sun, C.; Li, H. Tetrahedron:
Asymmetry 2009, 20, 240–246. (b) Liu, G.; Yao, M.; Wang, J.; Lu, X.; Liu, M.;
Zhang, F.; Li, H. Adv. Synth. Catal. 2008, 350, 1464–1468. (c) Sahoo, S.; Kumar,
P.; Lefebvre, F.; Halligudi, S. B. J. Catal. 2008, 254, 91–100.
(9) (a) Liu, G.; Yao, M.; Zhang, F.; Gao, Y.; Li, H. Chem. Commun. 2008,
347–349. (b) Jiang, D. M.; Gao, J. S.; Yang, Q. H.; Yang, J.; Li, C. Chem. Mater.
2006, 18, 6012–6018. (c) Jiang, D. M.; Yang, Q. H.; Yang, J.; Zhang, L.; Zhu,
G. R.; Su, W. G.; Li, C. Chem. Mater. 2005, 17, 6154–6160.
(10) (a) Fujdala, K. L.; Tilley, T. D. J. Catal. 2003, 216, 265–275. (b)
Lefebvre, F.; Thivolle-Cazat, J.; Dufaud, V.; Niccolai, G. P.; Basset, J.-M. Appl.
Catal., A 1999, 182, 1–8.
ꢀ
(11) Vidal, V.; Theolier, A.; Thivolle-Cazat, J.; Basset, J.-M. Science 1997,
276, 99–102.
(12) Burnett, R. L.; Hughes, T. R. J. Catal. 1973, 31, 55–64.
(13) Van der Laan, G. P.; Beenackers, A. A. C. M. Catal. Rev.;Sci. Eng.
1999, 41, 255–318.
€
(19) Bohmer, I. K.; Alt, H. G. J. Organomet. Chem. 2009, 694, 1001-1010
(14) Goldman, A. S.; Roy, A. H.; Huang, Z.; Ahuja, R.; Schinski, W.;
Brookhart, M. Science 2006, 312, 257–261.
(15) (a) Gupta, M.; Hagan, C.; Flesher, R. J.; Kaska, W. C.; Jensen, C. M.
and references cited therein.
(20) Huang, Z.; Brookhart, M.; Goldman, A. S.; Kundu, S.; Ray, A.;
Scott, S. L.; Vicente, B. C. Adv. Synth. Catal. 2009, 351, 188–206.
(21) For the first reported synthesis of Ir(acac)3, see: Dwyer, F. P.;
Sargeson, A. M. J. Am. Chem. Soc. 1953, 75, 984–985.
(22) For the first reported synthesis of Ir(C2H4)2(acac), see: Van Gaal,
H. L. M.; Van der Ent, A. Inorg. Chim. Acta 1973, 7, 653–659.
(23) For the first reported synthesis of Rh(allyl)3, see: Powell, J.; Shaw,
B. L. Chem. Commun. 1966, 323–325.
Chem. Commun. 1996, 2083–2084. (b) Gupta, M.; Hagan, C.; Kaska, W. C.;
€
Cramer, R. E.; Jensen, C. M. J. Am. Chem. Soc. 1997, 119, 840–841. (c) Gottker-
Schnetmann, I.; White, P.; Brookhart, M. J. Am. Chem. Soc. 2004, 126, 1804–
€
1811. (d) Gottker-Schnetmann, I.; Brookhart, M. J. Am. Chem. Soc. 2004, 126,
9330–9338. (e) Liu, F.; Pak, E. B.; Singh, B.; Jensen, C. M.; Goldman, A. S. J.
Am. Chem. Soc. 1999, 121, 4086–4087. (f) Zhu, K.; Achord, P. D.; Zhang, X.;
Krogh-Jespersen, K.; Goldman, A. S. J. Am. Chem. Soc. 2004, 126, 13044–
13053.
(16) (a) Schrock, R. R. Chem. Commun. 2005, 2773–2777. (b) Schrock,
R. R.; Murdzek, J. S.; Bazan, G. C.; Robbins, J.; DiMare, M.; O'Regan, M. J. Am.
Chem. Soc. 1990, 112, 3875–3886.
€
(24) Silvennoinen, R. J.; Jylha, O. J. T.; Lindblad, M.; Sainio, J. P.;
Puurunen, R. L.; Krause, A. O. I. Appl. Surf. Sci. 2007, 253, 4103–4111.
(25) (a) Uzun, A.; Bhirud, V. A.; Kletnieks, P. W.; Haw, J. F.; Gates, B. C.
J. Phys. Chem. C 2007, 111, 15064–15073. (b) Uzun, A.; Ortalan, V.; Browning,
N. D.; Gates, B. C. Chem. Commun. 2009, 31, 4657–4659.
(17) (a) Ivin, K. J.; Mol, J. C. Olefin Metathesis and Metathesis Polym-
erization; Academic Press: San Diego, 1997. (b) Pariya, C.; Jayaprakash, K. N.;
Sarkar, A. Coord. Chem. Rev. 1998, 168, 1–48.
(26) Ward, M. D.; Harris, T. V.; Schwartz, J. J. Chem. Soc., Chem.
Commun. 1980, 357–359.
(27) Foley, H. C.; DeCanio, S. J.; Tau, K. D.; Chao, K. J.; Onuferko,
J. H.; Dybowski, C.; Gates, B. C. J. Am. Chem. Soc. 1983, 105, 3074–3082.
(18) Grubbs, R. H. Tetrahedron 2004, 60, 7117–7140.

Article
Inorganic Chemistry, Vol. 49, No. 5, 2010 2249
the liberation of 2 equiv of propene and the sole formation of
(tTiO)2Rh(η3-allyl).28 These inconsistencies were later in-
vestigated in depth by Dufour et al., who determined that
the addition of Rh(allyl)3 to either silica, titania, or alumina
results in the rapid formation of the surface-bound bis-
(allyl)rhodium complex, (tMO)(tMOX)Rh(η3-allyl)2 (M =
Si, Ti, Al; X = H, Si), with the liberation of 1 equiv of
propene (eq 1).29 These authors proposed a two-site binding
to the surface, and when M = Si and X = H, they observed a
loss of additional propene due to silanol protonation of a
second allyl group at slightly elevated temperatures (80 ꢀC).
Figure 1. Comparison of the CP-MAS 13C NMR spectra of 1 (top) and
Ir(allyl)3 (bottom), in ppm.
observed over the course of 3 days at ambient tempera-
ture. At this time, the powder was collected by filtration,
washed several times with pentane, and dried in vacuo.
Analysis by a variety of spectroscopic methods revealed
the formation of the 5-coordinate surface-bound bis-
(allyl)iridium moiety, (dAlO)Ir(allyl)2 (1; eq 2). Although
the data collected on 1 discussed below favor this geo-
metry, we cannot definitively rule out the possibility that
the iridium center is 6-coordinate, having two surface
attachments in a fashion similar to that proposed for the
extensively investigated immobilized rhodium complexes,
(tMO)(tMOX)Rh(η3-allyl)2 (M = Si, Ti, Al; X = H,
Si; eq 1).29
In this paper, we describe our efforts to extend this
methodology to the immobilization of Ir(allyl)3 onto par-
tially dehydroxylated γ-alumina and to characterize the
resulting site-isolated iridium complex with a combination
of spectroscopic methods. Expanding upon our previous
studies regarding Lewis base addition to homoleptic me-
tal-allyl complexes,30,31 the reactivity of this supported
complex toward tertiary phosphorus ligands, CO, and 2,6-
dimethylphenyl isocyanide is discussed, along with degrada-
tion studies and preliminary attempts to catalyze alkane
dehydrogenation. We expected that the immobilization of
Ir(allyl)3, upon protonation of one or more allyl ligands,
would proceed easily based on the successful site isolation
of Rh(allyl)3,26-29 Ir(acac)3,24 and Ir(C2H4)2(acac).25 We
further expected that supported iridium complexes bearing
Lewis base ligands would exhibit better redox stability than
the site-isolated rhodium system32 because Ir(allyl)3(L)n com-
plexes have displayed greater stability than their second-row
congeners in solution.30b,33
Evaporation of the pentane filtrate in vacuo allowed re-
collection of up to 20% of the initial Ir(allyl)3, as judged
by H NMR spectroscopy upon integration against a
1
Results and Discussion
known amount of ferrocene. Similar percentages of Ir-
(allyl)3 recovery were achieved after 5 days under these
reaction conditions. Inductively coupled plasma (ICP)
analysis of 1 afforded iridium loadings between 1.2 and
1.6 wt %, depending on the batch of γ-alumina employed
(due to varying degrees of surface dehydroxylation).
Additionally, conducting the immobilization reaction in
a J. Young tube with minimal headspace allowed the
observation of a stoichiometric amount of propene by 1H
NMR spectroscopy, confirming the loss of one allyl
ligand from Ir(allyl)3 upon reaction with the surface
(Figure S1, Supporting Information).
In order to obtain valuable structural information
about the site-isolated complex, 1, cross-polarization
magic angle spinning (CP-MAS) NMR experiments were
conducted. The solid-state 13C NMR spectrum of pure
Ir(allyl)3 (Figure 1, bottom) was collected to allow for a
comparison with the immobilized material and was found
to be consistent with the solution data previously re-
ported for this complex.35 The CP-MAS 13C NMR
Preparation and Characterization of the Immobilized
Bis(allyl)iridium Moiety. Taking advantage of the pre-
viously reported high-yield synthesis of Ir(allyl)3,30b our
study commenced with the addition of this complex to γ-
alumina that had been partially dehydroxylated at either
500 or 1000 ꢀC under a helium flow, as previously
described.34 Upon stirring a pentane slurry of the reac-
tants, a change in color from white to light tan was
(28) (a) Iwasawa, Y.; Sato, H. Chem. Lett. 1985, 507–510. (b) Tada, M.;
Sasaki, T.; Iwasawa, Y. J. Phys. Chem. B 2004, 108, 2918–2930.
ꢀ
(29) Dufour, P.; Houtman, C.; Santini, C. C.; Nedez, C.; Basset, J. M.;
Hsu, L. Y.; Shore, S. G. J. Am. Chem. Soc. 1992, 114, 4248–4257.
(30) (a) John, K. D.; Salazar, K. V.; Scott, B. L.; Baker, R. T.;
Sattelberger, A. P. Chem. Commun. 2000, 581–582. (b) John, K. D.; Salazar,
K. V.; Scott, B. L.; Baker, R. T.; Sattelberger, A. P. Organometallics 2001, 20,
296–304.
(31) Trovitch, R. J.; John, K. D.; Martin, R. L.; Obrey, S. J.; Scott, B. L.;
Sattelberger, A. P.; Baker, R. T. Chem. Commun. 2009, 4206–4208.
(32) (a) Santini, C. C.; Scott, S. L.; Basset, J.-M. J. Mol. Catal. A: Chem.
1996, 107, 263–271. (b) Scott, S. L.; Dufour, P.; Santini, C. C.; Basset, J.-M.
Inorg. Chem. 1996, 35, 869–875. (c) Scott, S. L.; Mills, A.; Chao, C.; Basset,
J.-M.; Millot, N.; Santini, C. C. J. Mol. Catal. A: Chem. 2003, 204-205, 457–463.
(33) Powell, J.; Shaw, B. L. J. Chem. Soc. A 1968, 583–596.
(34) He, M.-Y.; Xiong, G.; Toscano, P. J.; Burwell, R. L., Jr.; Marks, T. J.
J. Am. Chem. Soc. 1985, 107, 641–652.
(35) John, K. D.; Michalczyk, R.; Hernandez, G.; Green, J. C.; Martin,
R. L.; Baker, R. T.; Sattelberger, A. P. Organometallics 2002, 21, 5757–5766.

2250 Inorganic Chemistry, Vol. 49, No. 5, 2010
Trovitch et al.
Table 1. Simulated EXAFS Data for Ir(allyl)3 and the Supported Complexes 1
and 2
DWF
2
3
˚
˚
complex scatter CN ((20%) R ((0.02 A) (A ꢀ 10 ) E0 (eV)
Ir(allyl)3 Ir-C
6.8
3.9
3.4
2.15
2.13
2.02
5.9
5.9
5.9
2.0
2.5
0.7
1
2
Ir-C/O
Ir-C/O
back-bonding to the allyl groups.36 The ability of this
oxygen atom to participate in π-donation to iridium
bolsters the formulation of a 5-coordinate metal center
in 1 because this type of interaction is present in several
reported 5-coordinate iridium(III) complexes.37 The
shifts of the allyl resonances additionally argue against
the possibility that 1 is an ion pair with the formula
[Ir(allyl)2]þ[O-Ald]-. A cationic bis(allyl)iridium com-
plex of this type would be expected to have 13C NMR
chemical shifts at a lower field than those observed for
Ir(allyl)3.38
Extended X-ray Absorption Fine-Structure (EXAFS)
Characterization of Unsupported Tris(allyl)iridium Com-
plexes and 1. To further spectroscopically differentiate
Ir(allyl)3 and 1, EXAFS measurements were carried out
on both 1 and a series of unsupported allyliridium com-
plexes. To allow for comparison, XAFS data were first
collected on Ir(allyl)3 (Figure 2, solid lines). Identical
X-ray absorption near-edge structure (XANES) and k-
space EXAFS data were collected for this complex upon
dilution with either partially dehydroxylated γ-alumina
(1000 ꢀC) or boron nitride, providing evidence that a
solid-state reaction between Ir(allyl)3 and the γ-alumina
hydroxyl groups did not preclude data acquisition. Ad-
ditionally, exposure of these samples to the beam for
extended periods of time did not result in degradation of
the complex. Fitting of the EXAFS data of Ir(allyl)3 in R-
space (Table 1) revealed an Ir-C coordination number of
3
˚
6.8 at 2.15 A, rather than the expected value of 9 (3 per η -
allyl ligand).
Even though determination of the metal-ligand co-
ordination numbers by EXAFS data fitting is a well-
established technique, these values can be associated
with up to 20% error and bond lengths differing by less
39
˚
than 0.1 A are difficult to distinguish by this practice.
Figure 2. EXAFS data collected for Ir(allyl)3 (solid) and 1 (dashed). At
the top is a comparison of the XANES region, displayed alongside the
corresponding k-space (middle) and R-space (bottom; phase correction
not applied) plots.
Because the solid-state structure of several related η3-
allyl-ligated iridium complexes feature slightly shorter
Ir-Cmethine distances than Ir-Cmethylene distances (by
approximately 0.06 A),30b we expected that all nine Ir-C
˚
contacts would be accounted for by fitting the major peak
˚
at 2.15 A. However, the fitted coordination number of 6.8
spectrum of 1 (Figure 1, top) was collected over the course
of several days (100 000 scans) and was found to exhibit
two broad resonances centered at 70.6 ppm (CH) and 24.6
ppm (CH2), consistent with an ensemble of surface-
bound species that have two equivalent allyl ligands
bound to a 5-coordinate iridium center. The appearance
of only two resonances confirms that these allyl ligands
are bound in an η3 fashion because η1-allyl ligands exhibit
three significantly different carbon shifts at approxi-
mately 150, 100, and 15 ppm.30b On the other hand, we
cannot rule out a 6-coordinate species that undergoes
rapid, reversible dissociation of a neutral surface oxygen
donor in the solid state. The 13C NMR resonances
observed for 1 are shifted significantly upfield of the start-
ing material and are reflective of electron donation from
the surface oxygen atom with concomitant increased
is outside the 20% error range for a molecule with a
coordination number of 9. This is not the first time that a
lower than expected M-Callyl coordination number has
been extruded from the EXAFS data fitting of an allyl-
ligated organometallic complex. In fact, the fitting of
EXAFS data collected on Rh(allyl)3 was previously re-
ported by Iwasawa and co-workers, who determined a
(36) Ahn, H.; Nicholas, C. P.; Marks, T. J. Organometallics 2002, 21,
1788–1806.
(37) Auburn, M. J.; Holmes-Smith, R. D.; Stobart, S. R.; Bakshi, P. K.;
Cameron, T. S. Organometallics 1996, 15, 3032-3036 and references cited
therein.
(38) Wakefield, J. B.; Stryker, J. M. Organometallics 1990, 9, 2428–2430.
(39) Hsiao, Y.-W.; Tao, Y.; Shokes, J. E.; Scott, R. A.; Ryde, U. Phys.
Rev. B 2006, 74, 214101.

Article
Inorganic Chemistry, Vol. 49, No. 5, 2010 2251
Figure 3. Phosphine-ligated tris(allyl)iridium complexes investigated in
this study.
28b
˚
Figure 4. EXAFS data (R-space without phase correction, solid lines)
collected for A-D. Corresponding fits are shown with dashed lines.
Rh-C coordination number of 6.7 at 2.18 A,
values
nearly identical with those obtained here for Ir(allyl)3.
Before interpreting the EXAFS spectra of 1, we sought
to further understand this η3-allyl ligand coordination
number phenomenon by fitting the EXAFS data col-
lected on a series of phosphine-ligated tris(allyl)iridium
complexes, Ir(allyl)3(PR3)n. The addition of phosphine
ligands to Ir(allyl)3 is known to result in tuning of the
σ/π-allyl ligand manifold, yielding related tris(allyl) com-
plexes that have different combinations of η1- and η3-allyl
ligands.30 The R-space EXAFS data (solid lines) collected
on four such complexes, Ir(η3-allyl)2(η1-allyl)(PPh3) (A),
Ir(η3-allyl)(η1-allyl)2(κ2-Ph2PC6H4PPh2) (B), Ir(η3-allyl)-
(η1-allyl)2(PMe2Ph)2 (C), and Ir(η1-allyl)3(κ3-Ph2P(CH2)2-
P(Ph)(CH2)2PPh2) (D) (Figure 3), along with their respec-
tive fits (dashed lines), are illustrated in Figure 4. The Ir-C
and Ir-P shells were refined simultaneously to yield the fits
that are tabulated in Table 2.
the average experimental Ir-C values.30b Similar to com-
plex A, the fitted Ir-P distances obtained for B and D of
2.40 and 2.42 A, respectively, are significantly longer than
their average X-ray crystallographically determined dis-
˚
tances of 2.2783[6] and 2.324[1] A, respectively.
˚
It is worth mentioning here recent work to decon-
volute the EXAFS data collected for [(Ph2PCH2CH2-
PPh2)Pd(CH2CHCMe2)][O3SCF3]. The original EXAFS
data analysis by Tromp et al. was conducted such that the
Pd-C and Pd-P shells were refined separately, leading
the authors to report false minima and propose that the
allyl ligand in this complex adopts an η2-hapticity in
solution.40 In the follow-up comprehensive density func-
tional theory and EXAFS modeling study,41 Caulton and
co-workers explained that multiple minima could be
arrived at for this complex because the Pd-C and Pd-P
EXAFS oscillations are out-of-phase; therefore, small
changes in the Pd-C coordination number can be com-
pensated for by changes in the Pd-P coordination num-
ber and bond length. For the previously discussed
phosphine-ligated iridium complexes, we believe that a
similar interplay between the Ir-C coordination number
and the Ir-P contact distance could be the reason that we
observe longer than expected Ir-P distances when the
number of Ir-P contacts is fixed during the EXAFS
fitting. Unfortunately, this explanation does not reveal
why lower than expected π-allyl coordination numbers
have been obtained for Rh(allyl)328b and Ir(allyl)3.
For complex A, an Ir-C coordination number of 7
would be expected if each η3-allyl ligand accounted for
three contacts, a number that is within 20% error of the
fitted value of 6.5. This value is also just outside the 20%
errorrangeexpected for a coordination numberof5, which
is based on the assumption that each η3-allyl ligand con-
tributes two Ir-C contacts. The determined Ir-C distance
˚
of 2.17 A is also in good agreement with the metrical
parameterspreviouslyreportedforthesolid-statestructure
˚
of this complex:30b Ir-Cσ-allyl = 2.167(4) A; Ir-Cπ-allyl
=
˚
2.145[5] and 2.217[5] A, for the methine and methylene
carbon atoms, respectively. Even though the fitting yielded
reasonable E0 values, it is important to note that the Ir-P
Upon immobilization, a decrease in the absorption
edge white line intensity of the iridium moiety was
observed (Figure 2, top), which is consistent with a
reduction in the coordination number. Similar differences
between the supported and unsupported complexes can
also be seen in the k-space (middle) and R-space (bottom)
plots, displayed in Figure 2. Importantly, the photon
energy at the absorption edge was slightly lower than that
˚
distance of 2.44 A is significantly longer than the experi-
˚
mentally determined value of 2.311(1) A.
Fitting the EXAFS data of the other phosphine com-
plexes (B-D) gave simulated Ir-C contactnumbers of 4.0,
4.6, and 3.5, respectively. For complexes B and C, the fitted
values are consistent with each η3-allyl ligand contributing
two Ir-C contacts and each η1-allyl ligand contributing
only the R-carbon contact. Similarly, the Ir-C coordina-
tion number of 3.5 obtained for D is within the 20% error
range expected for a value of 3 (for three η1-allyl ligands).
The solid-state structures of complexes B and D have
previously been reported, and in both cases, the fitted
(40) Tromp, M.; van Bokhoven, J. A.; van Haaren, R. J.; van Strijdonck,
G. P. F.; van der Eerden, A. M. J.; van Leeuwen, P. W. N. M.; Konings-
berger, D. C. J. Am. Chem. Soc. 2002, 124, 14814–14815.
(41) Clot, E.; Eisenstein, O.; Weng, T.-C.; Penner-Hahn, J.; Caulton,
K. G. J. Am. Chem. Soc. 2004, 126, 9079–9084.
˚
Ir-C distances (2.16 and 2.15 A) are slightly shorter than

2252 Inorganic Chemistry, Vol. 49, No. 5, 2010
Trovitch et al.
Table 2. EXAFS Fitting Results for A-D with Fixed Ir-P Coordination Numbersa
2
3
˚
˚
complex
scatter
expected CN
CN ((20%)
R ((0.02 A)
DWF (A ꢀ 10 )
E0 (eV)
A
Ir-C
Ir-P
Ir-C
Ir-P
Ir-C
Ir-P
Ir-C
Ir-P
5
1
4
2
4
2
3
3
6.5
1.0
4.0
2.0
4.6
2.0
3.5
3.0
2.17
2.44
2.16
2.40
2.17
2.41
2.15
2.42
5.9
2.4
5.9
2.4
5.9
2.4
5.9
2.4
2.6
5.6
1.8
1.1
1.8
1.3
1.7
-0.1
B
C
D
a Expected CN values were calculated considering that each η3-allyl ligand contributes two Ir-C contacts.
Figure 5. EXAFS data for 1 both before (solid line) and after (dashed
line) treatment with PPh3, shown in the k-space.
Figure 6. IR spectrum of complex 2.
observed for the parentcomplex (Figure 2, top), suggesting
that while the majority of iridium centers remain in the 3þ
oxidation state upon tethering to the surface, the reduction
of a small percentage of Ir3þ to Ir0 likely occurs. Because the
percentage of Ir0 formed is relatively small, a well-defined
Ir-Ir shell could not be identified in the k-space or R-space
plots of this sample. The formation of some Ir0 also
explains the lower than expected Ir-C/O coordination
analysisofthe surface byCP-MAS 31P NMR spectroscopy
revealed only a resonance at 5 ppm consistent with the
formation of [H-PPh3][O-Al].42 This resonance was also
observed upon the addition of Ir(allyl)3(PPh3) to γ-alumi-
na that was partially dehydroxylated at 1000 ꢀC. To verify
this lack of PPh3 affinity, EXAFS data were also collected
on 1 after treatment with excess PPh3 (Figure 5). Similar
results were obtained from the addition of Ir(allyl)3-
(PMe2Ph)2 to γ-alumina or treatment of 1 with 20 equiv
of the tied-back phosphite P(OCH2)3CCH3 for 4 days.
Because definitive IR spectral assignments could not be
made on 1 [Ir-X (X = C, O) or C-H bands not discretely
identified at high resolution (0.5 cm-1, 128 scans) because
of the complexity of the surface and low iridium content],
modification of the immobilized bis(allyl)iridium com-
plex with ligands that typically have strong vibrational
signatures such as CO and 2,6-dimethylphenyl isocyanide
was explored. In solution, Ir(allyl)3 is known to bind 3
equiv of 2,6-dimethylphenyl isocyanide to form Ir(σ-
allyl)3(CNAr)3 [Ar = 2,6-(CH3)2C6H4].30b Our investiga-
tion of this substrate commenced with the addition of 1
equiv of 2,6-dimethylphenyl isocyanide dissolved in pen-
tane to 1, allowing the formation of (dAlO)Ir(allyl)2-
(CNAr) (2) after 3 days at ambient temperature (eq 3).
Upon mixing of the reactants, the γ-alumina changed
from light tan to light orange. The IR spectrum of 2
featured a broad band at 2132 cm-1 (Figure 6), shifted 15
cm-1 from the free isocyanide stretch of 2117 cm-1. In
contrast to the previously explored solution chemistry,30b
2 was also formed when 20 or 100 equiv of isocyanide was
added under these reaction conditions, as judged by
EXAFS, CP-MAS 13C NMR, and IR spectroscopy.
Binding of more than 1 equiv in this case is likely
˚
number of 3.9 at an average bond distance of 2.13 A. It was
expected that upon immobilization a coordination number
of approximately 5 (or 6, assuming two Ir-O surface
contacts) would be reached, with each allyl ligand con-
tributing two contacts [based on what was observed for
Ir(allyl)3]. The Ir-C and Ir-O contacts could not be
˚
effectively separated in the data fitting (less than 0.1 A
apart),39 and all attempts to input a shorter or longer Ir-O
bond distance resulted in abnormal threshold energies.
Having already ruled out alternative origins for the low
Ir-C/O coordination number observed, such as the for-
mation of an electrostatically surface-bound, 14-electron
[Ir(allyl)2]þ complex or loss of an additional allyl ligand as
propene to form a bis(oxido) surface linkage as reported
for the Rh congener,28 formation of a small amount of Ir0
appears to be the likeliest source for this phenomenon.
Having established the identity of 1 through solid-state
NMR and EXAFS studies, the reactivity of this complex
toward Lewis bases was investigated in order to tune the
σ/π-allyl ligand manifold of the supported iridium com-
plex and definitively detect Ir0 formation.
Reactivity of (dAlO)Ir(allyl)2 toward Lewis Bases. In
contrast to the reactivity of Ir(allyl)3 and even some
reports on the supported rhodium congener, the addition
of tertiary phosphorus ligands to complex 1 did not
proceed readily. The addition of a large excess of PPh3
(15equiv, basedonthe iridium contentdeterminedbyICP)
to 1 at ambient temperature for 2 h and subsequent
(42) Hu, B.; Gay, I. D. J. Phys. Chem. B 2001, 105, 217–219.

Article
Inorganic Chemistry, Vol. 49, No. 5, 2010 2253
Figure 7. CP-MAS 13C NMR spectrum of 2, in ppm.
Figure 8. EXAFS data collected for complex 2 shown in R-space (solid
line; without phase correction) along with the data fit (dashed line).
prevented by the steric protection imparted by the sur-
face.
The EXAFS data collected on complex 2 were also
reflective of 2,6-dimethylphenyl isocyanide ligation. The
R-space data after Fourier transformation (not phase-cor-
rected; solid line) are displayed in Figure 8, along with the
fit, which is shown as a dashed line. The data fitting yielded
˚
an Ir-C/O coordination number of 3.4 at 2.02 A (Table 1).
This short average Ir-C/O bond distance reflects the bind-
ing of isocyanide to 1 because the Ir-CNAr distances in
Ir(allyl)3(CNAr)3 were determined by X-ray crystallogra-
phy to be an average of 1.979[5] A.30b The Ir-C/O coordi-
˚
nation number determined in this case is once again lower
than expected because it was prepared directly from 1.
The reaction of 1 with CO resulted in a different out-
come than that observed upon 2,6-dimethylphenyl iso-
cyanide addition. When an evacuated bomb containing 1
was charged with an atmosphere of CO and the gas was
removed after 5 min at ambient temperature, the dicar-
bonyl complex (dAlO)Ir(allyl)2(CO)2 (3; eq 4) was iden-
tified spectroscopically. The IR spectrum of this complex
features symmetric and asymmetric CO stretching vibra-
tions at 2068 and 1992 cm-1, respectively (Figure 9, top).
Upon labeling with 13CO, these bands shift accordingly to
2020 and 1944 cm-1, respectively. Unfortunately, this
method of preparation is not a valid way to isolate this
complex because only a small amount of 1 is converted to
3 under these conditions, as judged by CP-MAS 13C
NMR (29 200 scans needed to observe a weak carbonyl
resonance for 3-13CO) and XAFS (sample nearly identi-
cal to the starting material).
In the IR spectrum of 2, a second much smaller iso-
cyanide stretch was observed at 1989 cm-1, regardless of
the number of substrate equivalents added. To determine
whether or not this lower-frequency stretch was due to
isocyanide binding to Ir0, a sample of 1 was prereduced
under a constant flow of 6% H2/94% Ar at 170 ꢀC for 4 h,
in an attempt to increase the ratio of Ir0 to 2 present in the
sample (see the decomposition studies in the following
section). The subsequent addition of 20 equiv of 2,6-di-
methylphenyl isocyanide to this sample (4 days at ambi-
ent temperature) significantly increased the ratio of the
IR stretch at 1989 cm-1 relative to the stretch at 2132
cm-1 (Figure S2, Supporting Information), suggesting
that the lower-frequency stretch is, in fact, due to iso-
cyanide binding to Ir0. Although this technique is not
quantitative because each Ir0 cluster may bind 1 equiv or
more of isocyanide, the IR spectrum of 2 confirms that a
small amount of Ir0 is present in 1.
The CP-MAS 13C NMR spectrum of 2 was particularly
valuable in determining the hapticity of the allyl ligands in
this complex. In this spectrum, displayed in Figure 7, two
distinct allyl ligand methylene resonances can be ob-
served at 30.7 and 24.5 ppm, providing evidence that
there are two inequivalent η3-allyl methylene environ-
ments in 2 that are likely differentiated by the surface. The
methine resonances associated with these ligands appear
as an unresolved broad peak centered at 70.9 ppm. In
addition, resonances for the bound isocyanide ligand
were observed at 135.0 (aryl), 126.8 (aryl), and 18.2
ppm (methyl).
When 1 was exposed to an atmosphere of CO for more
than 5 min, the alumina acquired a blue/green hue and the
presence of a second carbonyl-containing species was
detected by IR and CP-MAS 13C NMR spectroscopies.
As judged by IR spectroscopy, this newly formed moiety

2254 Inorganic Chemistry, Vol. 49, No. 5, 2010
Trovitch et al.
Figure 10. Mixture of 3-13CO and 13CO adsorbed onto Ir0 as observed
by CP-MAS 13C NMR spectroscopy, in ppm.
Although relatively intense spinning side bands from both
Ir-13CO resonances complicated the remainder of the
NMR spectrum, resonances consistent with the presence
of one η1-allyl ligand (101.3 and 13.4 ppm) and one η3-allyl
ligand (70.2 and 24.5 ppm) were identified (see Figure S3 of
the Supporting Information).
Exposure of 1 to an atmosphere of CO for periods
of time longer than 1 week did not result in further
conversion to Ir0-CO. Additionally, placing the mixture
of complexes under vacuum for 3 h did not increase the
ratio of Ir0-CO to 3 observed. The formation of Ir0-CO
also does not result from exposure of 3 to air or moisture
because monitoring the IR spectrum of the mixture (or of
3 itself) in open air did not result in a change of the
product ratio after 10 min. One approach in particular,
the addition of Ir(allyl)3(CO)330b to partially dehydroxy-
lated γ-alumina, did improve the ratio of Ir0-CO to 3,
although both products were still observed by IR spec-
troscopy. Within 2 h of mixing of the reagents, the slurry
turned dark purple. After 4 days at ambient temperature,
the reaction mixture was worked up and an olive-green
solid was collected. As determined by IR spectroscopy,
the change in the surface color from tan to purple, blue, or
green appears to reflect the aggregation of iridium clus-
ters under ambient conditions.
Figure 9. Solid-state IR spectrum of 3 after 5 min under an atmosphere
ofCO (top), shown along with spectra collectedfollowing exposure of1 to
an atmosphere of CO for 1 week (middle) and upon exposure of 1 to an
atmosphere of 13CO for 1 week (bottom). Ir-CO and Ir-13CO represent
the product of carbon monoxide addition to an Ir0 cluster.
Alkane Dehydrogenation Experiments and Decomposi-
tion Studies. After the isolation and characterization of 1,
catalytic experiments were conducted to determine the
viability of the supported bis(allyl)iridium complex to
catalyze a series of alkane dehydrogenation transforma-
tions. Initially, turnover numbers greater than 100 h-1
were achieved when 1 was employed as the catalyst for the
conversion of cyclohexane to benzene and of heptane to
toluene at 200 and 270 ꢀC, respectively.44 However,
contains only one CO band at 2035 cm-1 (Figure 9,
middle), which is consistent with that reported previously
for ligation of CO to an Ir0 cluster.43 This band also
shifted appropriately to 1988 cm-1 when 13CO was em-
ployed (Figure 9, bottom). Preparing a mixture of complex
3-13CO with the newly formed Ir-13CO species greatly
facilitated the collection of CP-MAS 13C NMR data. The
equivalent, or nearly equivalent, carbonyl ligands in 3-13CO
gave rise to a broad singlet centered at 161.0 ppm (Figure 10).
Because this shift is similar to the one reported for the
equivalent CO ligands in Ir(allyl)3(CO)2,30b these complexes
have a similar electronic environment about the metal
center, which also argues against the likelihood that 3 (or
1) is cationic in nature. The 13C NMR shift of the second
carbonyl-containing species was also detected at 180.6 ppm.
(44) In a typical dehydrogenation reaction, 30-50 mg of 1 was loaded
into a 100 mL Schlenk tube containing 0.1 mL of an alkane substrate. The
reaction vessel was then cooled to -78 ꢀC, evacuated on a Schlenk line,
sealed, and heated to either 200 or 270 ꢀC for approximately 15 h. Upon
completion of the reaction, the products were extracted into approximately
1 mL of CD2Cl2 and analyzed by 1H NMR spectroscopy. In one trial, a
turnover number of 121 for the dehydrogenation of cyclohexane to benzene
was achieved at 200 ꢀC (41% yield) when 0.050 g of 1 was employed as the
catalyst. Using the same quantity of 1, a turnover number of 156 was
observed for the conversion of heptane to toluene at 270 ꢀC (72% yield).
(43) Zeinalipour-Yazdi, C. D.; Cooksy, A. L.; Efstathiou, A. M. Surf. Sci.
2008, 602, 1858–1862.

Article
Inorganic Chemistry, Vol. 49, No. 5, 2010 2255
Figure 12. Percent conversion of cyclohexane dehydrogenation as a
function of the temperature for 1 (dashed line) and preformed Ir0 (solid
line).
Figure 11. TPR of 1 under a flow of cyclohexane-saturated helium, as
observed through in situ EXAFS measurements.
because high temperatures were required to effect these
transformations and a lack of consistency in the turnover
frequency was observed, it was hypothesized that the
iridium nanoparticles rather than a well-defined immo-
bilized iridium may be responsible for the observed
catalysis.
In order to determine the true origin of this catalytic
activity, a series of in situ XAFS experiments were con-
ducted along with continuous-flow reactor studies. Be-
cause H2 would inevitably be generated during catalytic
alkane dehydrogenation, the stability of 1 under a flow of
H2 was investigated with ambient-temperature in situ
XAFS measurements. However, only a slight reduction
in the Ir-C coordination number was observed after
approximately 15 min under a constant flow of 4% H2/
96% He with no observable metallic iridium. To more
accurately mimic the conditions of the catalytic alkane
dehydrogenation reactions, in situ temperature-pro-
grammed-reduction (TPR) experiments were conducted
under a flow of cyclohexane gas in helium. This experiment
revealed a clear decrease in the Ir-C coordination number
and an increase in the Ir-Ir coordination number as a
function of the temperature for 1 (Figure 11). The Ir-C
coordination number decreased from approximately 6 to 3
upon heating from ambient temperature to 215 ꢀC, with a
further reduction in the Ir-C coordination number ob-
served upon heating to 309 ꢀC. Importantly, the formation
of Ir-Ir contacts can be observed at 215 ꢀC, providing
evidence of additional Ir0 nanoparticle formation under
conditions that are milder than those employed in the
catalytic alkane dehydrogenation reactions.45
In continuous-flow reactor studies, independently pre-
pared batches of 1 were monitored for cyclohexane
dehydrogenation activity in the temperature ranges of
either 180-220 ꢀC (Figure 12, dashed line) or 200-240 ꢀC.
In the lower-temperature trial, conversions of greater
than 5% were never achieved. In the second trial, iden-
tical conversions were observed while heating from 200
to 220 ꢀC, with conversions greater than 5% realized at
240 ꢀC. In a third trial, a sample of 1 (same iridium
loading as that in trial 1) was treated at 370 ꢀC for 1 h with
Figure 13. R-space EXAFS plot (not phase-corrected) of1 after heating
to 240 ꢀC under a cyclohexane/argon flow (solid line). Fitted data are
presented as a dashed line.
Table 3. EXAFS Fitting Results of 1 upon Completion of the Continuous-Flow
Reactor Study
DWF
2
3
˚
˚
complex scatter CN ((20%) R ((0.02 A) (A ꢀ 10 ) E0 (eV)
1 (240 ꢀC) Ir-C
Ir-Ir
3.1
2.4
2.12
2.77
5.9
2.0
5.4
2.8
a flow of 5% H2 in argon, conditions under which 1 would
be fully reduced to Ir0. The activity observed for the
prereduced sample (Figure 12, solid line) was significantly
higher (∼15% conversion at 220 ꢀC) than the activity
observed using unreduced 1 under the same reaction
conditions, providing a second line of evidence that the
alkane dehydrogenation reactions were catalyzed predo-
minantly by Ir0 nanoparticles rather than a well-defined
monometallic species.
Upon completion of the second trial of the reactor
conversion study (heating from 200 to 240 ꢀC), the
iridium catalyst 1 was collected without exposure to air
and an XAFS measurement was taken. After Fourier
transformation, a clearly defined Ir-Ir shoulder was
observed to the right of the Ir-C/O peak (Figure 13).
EXAFS data analysis revealed the formation of 2.4 Ir-Ir
˚
contacts at 2.77 A during the course of the catalytic
reaction (Table 3). This measurement provides direct
evidence that additional iridium nanoparticles are, in
fact, formed under the alkane dehydrogenation condi-
tions and that they are the most active species.
(45) Because the catalytic trials were conducted in a closed system, a
buildup of H2 pressure in the reaction vessel would result in more reducing
conditions than those in the in situ XAFS experiments, which were
conducted under a constant flow of cyclohexane gas in helium.

2256 Inorganic Chemistry, Vol. 49, No. 5, 2010
Trovitch et al.
Discussion and Comparison to Rh(allyl)3 Immobilization
Studies. Because the reactivity of (tSiO)(tSiOX)Rh(η3-
allyl)2 (X = H, Si) toward Lewis bases such as phosphines
and CO has been thoroughly investigated,32 comparisons
between the immobilized bis(allyl)rhodium and iridium
systems warrant discussion. The reactivity observed for
(tSiO)(tSiOX)Rh(η3-allyl)2 was found to vary greatly
with the degree of support dehydroxylation. For example,
when excess CO was added to the rhodium complex
supported on silica that was partially dehydroxylated
at 550 ꢀC, reductive elimination of 1,5-hexadiene was
observed along with a small amount of 1,6-heptadien-4-
one and a species proposed to be the surface-bound
rhodium dimer, [(tSiO)Rh(CO)2]2.32a However, when
the silica was partially dehydroxylated at 200 ꢀC, the
binding of 3 equiv of CO and liberation of 1 equiv of
propene were observed, affording a species proposed to
be (tSiO)2Rh(CO)3(η1-allyl).32a Both this product and
its proposed phosphine-ligated analogue, (tSiO)2Rh-
(PMe3)3(η1-allyl), exhibited poor redox stability because
either acyl (following CO insertion) or allyl group transfer
to the surface was observed, respectively.32a
during the immobilization reaction. TPR of the supported
bis(η3-allyl)iridium complex revealed that its integrity was
slowly compromised at temperatures above 150 ꢀC, render-
ing it inadequate for high-temperature catalytic studies. On
the other hand, because this complex proved more resistant
to redox changes than its well-studied rhodium congener, the
potential remains that the covalent immobilization of
judiciously selected Ir3þ complexes to a surface may lead to
the development of well-defined alkane dehydrogenation
catalysts.
Experimental Section
General Considerations. All synthetic reactions were per-
formed in an MBraun glovebox under an atmosphere of purified
nitrogen. Aldrich or Acros anhydrous solvents were sparged
˚
with argon and stored in the glovebox over activated 4 A
molecular sieves and sodium before use. Benzene-d6 and methy-
lene chloride-d2 were purchased from Cambridge Isotope La-
˚
boratories and dried over 4 A molecular sieves prior to use.
Triphenylphosphine, dimethylphenylphosphine, and ferrocene
were used as received from Acros, while cyclohexane, heptane,
and boron nitride were purchased from Aldrich and dried before
use. All of the gases used in this study were obtained from Airgas
with the exception of 13CO, which was purchased from Cam-
bridge Isotope Laboratories. Several of the allyliridium com-
plexes studied in this manuscript, including Ir(allyl)3, Ir(η3-
allyl)2(η1-allyl)(PPh3), Ir(η3-allyl)(η1-allyl)2(κ2-Ph2PC6H4PPh2),
Ir(η1-allyl)3(κ3-Ph2P(CH2)2P(Ph)(CH2)2PPh2), and Ir(allyl)3(CO)3,
were prepared according to literature procedures.30b The tied-back
phosphite, P(OCH2)3CCH3, was also prepared according to a
The reactivity studies on the rhodium system highlight
the importance of achieving a good degree of surface
homogeneity prior to metal complex immobilization. For
the immobilization of Ir(allyl)3 discussed in this manu-
script, the relative inertness of this complex, when com-
pared to that of Rh(allyl)3, prevented the immobilization
reaction from taking place between Ir(allyl)3 and silica. In
1
literature procedure.46 Solution H NMR spectra were recorded
turn, we chose highly Bronsted acidic, partially dehy-
e
at room temperature on a Bruker AVANCE 400 MHz spectro-
meter. All 1H and 13C NMR chemical shifts are reported relative to
SiMe4 using 1H (residual) and 13C chemical shifts of the solvent as
secondary standards. 31P NMR data are reported relative to H3PO4.
IR spectra of the supported iridium complexes were recorded on a
Thermo Nicolet Nexus 470 FT-IR spectrometer running OMNIC
software. ICP mass spectrometry (ICP-MS) experiments were
conducted on a Varian MPX Simultaneous ICP-AES instrument,
following EPA SW 846 Method 6010. Elemental analyses were
performed at Atlantic Microlab, Inc., in Norcross, GA.
droxylated γ-alumina as the support. Unfortunately, this
support bears microenvironments that are much more
reactive than similar sites on a silica surface, a factor that
undoubtedly contributed to small amounts of Ir0 forma-
tion during the immobilization of Ir(allyl)3.
The observation that allyliridium complexes are more
stable toward redox events than their rhodium counter-
parts in solution30b,33 also carried over to the site-isolated
reactivities of these complexes. For example, when 1 was
subjected to an atmosphere of CO over the course of a
week, no evidence for CO insertion was observed by IR
and the resulting dicarbonyl 3 was remarkably stable,
even upon exposure to air. Although this inherent robust-
ness could prove invaluable for the discovery of well-
defined complexes that can function as catalysts at high
temperatures, better methods of tethering these allylir-
idium complexes to a surface will need to be developed.
Future studies in this regard will focus on the pretreat-
ment of alumina-based surfaces in an attempt to saturate
the highly reactive alumina microenvironments that ap-
pear to contribute to Ir0 formation. The addition of
iridium complexes bearing chelating ligands to highly
acidic surfaces may also prove to be valuable in circum-
venting metallic iridium formation.
Solid-State NMR Measurements. Solid-state CP-MAS NMR
experiments were conducted on a Bruker AVANCE system with
the resonance frequency of 13C at approximately 100.6 MHz.
Experiments were typically performed on 50-75 mg of solid
material packed under argon in airtight 4 mm rotors. The π/2
pulse for 1H was 4.00 μs, and the Hartmann-Hahn match
condition was optimized for a 4 mm rotor at a spinning speed
of 10 000 Hz. The contact time for magnetization transfer was 5
ms, and the relaxation delay between scans was 3 s. Approxi-
mately 4000 acquisitions were necessary for the Ir(allyl)3 sample,
100 000 for 1, 140 000 for 2, and 66 000 for 3-13CO. Additionally,
CP-MAS 31P NMR spectroscopy was utilized as a probe for
Lewis base reactivity toward 1.
Support Activation. γ-Alumina (Strem Chemical, 97%) was
activated under flowing gas using a standard tube furnace prior
to employment in the immobilization reactions. Approximately
2.5 g of γ-alumina was placed in a quartz tube, a flow of helium
was established, and the solid was heated at either 500 or 1000 ꢀC
for 20 min to yield partially dehydroxylated γ-alumina.34 Im-
mediately following activation, the alumina was quickly trans-
ferred to the glovebox and stored in a sealed container under
dinitrogen.
Concluding Remarks
A series of γ-alumina-supported iridium complexes were
characterized by a suite of techniques including XAFS, IR,
and solid-state NMR spectroscopy. Although a highly acidic
support was required for the covalent attachment of Ir(allyl)3
to the surface, this characteristic also contributed to the
undesired deposition of small amounts of zerovalent iridium
(46) Wadsworth, W. S., Jr.; Emmons, W. D. J. Am. Chem. Soc. 1962, 84,
610–617.

Article
Preparation of (dAlO)Ir(allyl)2 (1). In the glovebox, a 20 mL
Inorganic Chemistry, Vol. 49, No. 5, 2010 2257
XANES spectra of Ir(allyl)2(O-)/Al2O3 and fully reduced Ir0/
Al2O3. The EXAFS coordination parameters were obtained by
a least-squares fit in the q and r spaces of the isolated nearest-
neighbor, k2-weighted Fourier transform data. The quality of
the fits was equally good with both k1- and k3-weightings.
Preparation of Ir(η3-allyl)(η1-allyl)2(PMe2Ph)2 (C). In the
glovebox, a 20 mL scintillation vial was charged with 0.100 g
(0.317 mmol) of Ir(allyl)3 and approximately 3 mL of toluene.
While stirring, a second solution of 0.087 g (90 μL, 0.633 mmol)
of dimethylphenylphosphine in approximately 3 mL of toluene
was added. The solution became yellow over the course of 2.5 h
at ambient temperature. At this point, the solvent was removed
in vacuo to yield a yellow oil. This oil was redissolved in pentane,
and the solvent was evacuated several times until 0.114 g (61%)
of a yellow solid identified as C was collected. Anal. Calcd for
scintillation vial was charged with 0.570 g of γ-alumina
(partially dehydroxylated at either 500 or 1000 ꢀC, as described
above), 0.013 g (0.041 mmol) of freshly sublimed Ir(allyl)3, and
approximately 5 mL of pentane. The slurry was set to stir at
ambient temperature, and a color change from white to light tan
was observed over the course of 24 h. After 3 days, the reaction
mixture was filtered and the isolated tan solid was washed five
times with approximately 10 mL of pentane to remove any
residual Ir(allyl)3. Upon drying in vacuo, 0.505 g (∼88% yield)
of 1 was collected. Removal of pentane from the filtrate allowed
re-collection of approximately 20% of the original Ir(allyl)3, as
1
judged by H NMR spectroscopy upon integration against a
known quantity of ferrocene. Using a modified procedure, a
solution of 0.003 g (0.01 mmol) of Ir(allyl)3 and 0.002 g (mmol)
of ferrocene in 1.5 mL of toluene-d8 was added to a J. Young
NMR tube that was preloaded with 0.600 g of partially dehy-
droxylated γ-alumina (1000 ꢀC). After 5 h, the evolved propene
was quantified by 1H NMR spectroscopy, upon integration
against the added ferrocene. The iridium loading was deter-
mined to be between 1.2 and 1.6 wt % by ICP-MS, depending on
the batch of partially dehydroxylated γ-alumina employed. CP-
MAS 13C{1H} NMR: δ 70.6 (CH2CHCH2), 24.6 (CH2CHCH2).
EXAFS Measurements. In situ H2 and cyclohexane TPR
EXAFS studies of 1 at the Ir L3 edge were performed at the
Materials Research Collaborative Access Team (MRCAT) and
CMC beamlines at the Advanced Photon Source, Argonne
National Laboratory. The setup used was similar to that out-
lined by Castagnola et al.47 In an argon-filled Vacuum Atmo-
spheres glovebox with a large-capacity recirculator (0.25 ppm of
O2 and 0.5 ppm of H2O), approximately 25 mg of each sample
was loaded as a self-supporting wafer without binder in the
channels (i.d. = 4 mm) of a stainless steel multisample holder.
The sample holder was then placed in the center of a quartz tube,
which was equipped with gas and thermocouple ports and
Kapton windows. The amount of sample used was optimized
for the Ir L3 edge, considering absorption by the aluminum in
the support. With the gas ports closed, the setup was transferred
into the experiment hutch. The quartz tube was placed in a
clamshell furnace, and both were mounted on a positioning
platform. Helium flow at 200 mL/min was established to remove
the residual argon. The beam was set to scan the sample cell
holder by manipulating the position of the platform. Once the
sample positions were fine-tuned, the reactant gas (4% H2 in
helium or cyclohexane-saturated helium) was flowed through
the samples (50 mL/min) and a temperature ramp of 2 ꢀC/min
was initiated for the furnace. The EXAFS spectra were recorded
in transmission mode, and a platinum foil spectrum was mea-
sured simultaneously with each sample spectrum for energy
calibration purposes. Spectra for each sample were collected
from 11 000 to 12 258 eV, with a step size of 0.45 eV and an
acquisition time of 0.03 s/step. The change in the sample
temperature from the absorption edge through the end of the
scan was approximately 2.8 ꢀC, while each sample was probed
approximately every 15 ꢀC. On the basis of the observed
temperature dependence of the transitions, the local structure
remains essentially the same throughout a single spectrum from
the absorption edge to the end of the scan.
1
C25H37IrP2: C, 50.74; H, 6.30. Found: C, 50.61; H, 6.17. H
NMR (benzene-d6): δ 7.18 (m, 2H, phenyl), 7.07-6.96 (m, 8H,
phenyl), 5.68 (overlapping m, 2H, CH2CHdCH2), 4.89 (dd,
JHH = 10.0 and 2.5 Hz, 1H, CH2CHdCH2), 4.68 (m, 3H,
CH2CHdCH2), 3.85 (quintet, JHH = 9.0 Hz, 1H, CH2CHCH2),
2.69 (m, 2H, CH2CHCH2), 2.23 (m, 2H, CH2CHCH2), 1.85 (m,
2H, CH2CHdCH2), 1.35 (d, JHP = 8.5 Hz, 6H, PPh(CH3)2),
1.29 (d, JHP = 8.5 Hz, 6H, PPh(CH3)2). 13C{1H} NMR
(benzene-d6): δ 148.9 (t, 4JCP = 3.0 Hz, CH2CHdCH2), 147.6
(t, 4JCP = 3.0 Hz, CH2CHdCH2), 141.1 (dd, 2JCP = 43.0 Hz,
3
4JCP = 3.0 Hz, 1-phenyl), 130.1 (d, JCP = 9.0 Hz, 2-phenyl),
129.0 (s, phenyl), 128.3 (s, phenyl), 110.4 (s, CH2CHCH2), 105.2
(s, CH2CHdCH2), 105.1 (s, CH2CHdCH2), 48.8 (dd, 3JCP-trans
=
33.0 Hz, 3JCP-cis = 2.5 Hz, CH2CHCH2), 14.8 (d, 2JCP = 33.5 Hz,
PPh(CH3)2), 14.6 (d, 2JCP = 33.5 Hz, PPh(CH3)2), 0.8 (t, 3JCP
=
4.5 Hz, CH2CHdCH2), -1.2 (t, 3JCP = 5.0 Hz, CH2CHdCH2).
31P{1H} NMR (benzene-d6): δ 39.5 (s).
Preparation of (dAlO)Ir(allyl)2(CNAr) [2; Ar = 2,6-(CH3)2-
C6H4]. In the glovebox, a 20 mL scintillation vial was charged
with 0.200 g of 1 and 0.034 g (0.258 mmol, 20 equiv) of 2,6-
dimethylphenyl isocyanide. Upon the addition of approximately
5 mL of pentane, the solid immediately changed from light tan to
yellowish-orange. The resulting slurry was allowed to stir for 3
days at ambient temperature before the light-orange solid was
collectedviafiltration. This solid waswashed twice with 10 mL of
toluene and then twice with 10 mL of pentane. Upon drying,
0.183 g (∼91%) of 2 was obtained. The iridium loading in 2 was
determined to be 1.6% by ICP-MS. CP-MAS 13C{1H} NMR: δ
135.0 (aryl), 126.8 (aryl), 70.9 (π-allyl), 30.7 (π-allyl), 24.5 (π-
allyl), 18.7 (CH3). IR (solid state): νCN 2132 cm-1
.
Observation of (dAlO)Ir(allyl)2(CO)2 (3). In the glovebox, a
thick-walled glass vessel was charged with 0.160 g of 1 and
sealed. The bomb was evacuated on a Schlenk line and then
charged with 1 atm of CO. After 5 min, the CO atmosphere was
evacuated, the vessel was brought back into the glovebox, and
0.106 g of CO-modified 1 was collected. Upon spectroscopic
investigation, this material was found to contain small amounts
of 3. IR (solid state): νCO 2068, 1992 cm-1. A similar procedure
was used to prepare the 13CO-labeled version of this complex at
early reaction times (3-13CO). IR (solid state): ν13CO 2020, 1944
cm-1. Allowing the reaction between 1 and 13CO to remain at
ambient temperature for 24 h rather than 5 min allowed greater
conversion to 3-13CO, which could then be characterized by CP-
MAS 13C NMR spectroscopy. CP-MAS 13C{1H} NMR: δ 161.0
(s, 13CO), 101.3 (σ-allyl), 70.2 (π-allyl), 24.5 (π-allyl), 13.4 (σ-
allyl). A small amount of this complex was also identified by IR
spectroscopy upon mixing of a toluene solution containing
0.007 g (0.017 mmol) of Ir(allyl)3(CO)3 with 0.400 g of partially
dehydroxylated γ-alumina for 4 days. Within 2 h of mixing the
reactants, the slurry turned dark purple and turned olive green
over the course of the reaction because of the formation of CO-
ligated Ir0 clusters, which were identified as the major product
XAFS Data Analysis. Because iridium is next to platinum in
the periodic table, phase shifts and backscattering amplitudes of
Pt-Pt and Pt-N scattering, which were obtained from a
platinum foil and Pt(NH3)4Cl2, respectively, were utilized for
Ir-Ir and Ir-C contributions, respectively. Standard proce-
dures based on WINXAS 3.1 software were used to fit the XAS
data. XANES spectra of the supported iridium catalyst under
various conditions were fitted as linear combinations of the
(47) Castagnola, N. B.; Kropf, A. J.; Marshall, C. L. Appl. Catal., A 2005,
290, 110–122.
by IR spectroscopy. IR (solid state): νCO 2035 cm-1
.

2258 Inorganic Chemistry, Vol. 49, No. 5, 2010
Trovitch et al.
Cyclohexane Dehydrogenation Reactor Studies. In an argon-
filled Vacuum Atmospheres glovebox with a large capacity
recirculator (0.25 ppm of O2 and 0.5 ppm of H2O), a predeter-
mined amount of 1 was loaded into a 1/2-in. quartz tube reactor
equipped with an Omega K-type thermocouple and two on/off
valves. With argon purging, the reactor assembly was attached to
a catalyst testing unit and the catalyst bed was positioned at the
center of a clamshell furnace. The gas flow rates were controlled
by Brooks Instrument 5850E mass flow controllers. Cyclohexane
was fed by flowing reaction gas (argon) through a cyclohexane
bubbler kept at room temperature. The effluent gas from the
reactor was analyzed with an online Hewlett-Packard 5890 series
II Plus gas chromatograph equipped with a HayeSep D column,
Acknowledgment. We thank the Office of Basic Energy
Sciences, Division of Chemical Sciences, Geosciences, and
Biosciences, for financial support through the Catalysis Science
Program. Use of the Advanced Photon Source was also
supported by the Office of Basic Energy Sciences of the U.S.
Department of Energy under Contract DE-AC02-06CH11357.
MRCAT (Sector 10) operations are supported by the Depart-
ment of Energy and the MRCAT member institutions.
The collection of XAFS data was also carried out at the XOR
Beamlines (Sector 9), which are supported in part by the Office
of Basic Energy Sciences of the U.S. Department of Energy and
by the National Science Foundation Division of Materials
Research. We thank Drs. Jeremy Kropf and Trudy Bolin
for assistance with the collection of XAFS measurements and
Brandy Duran for conducting ICP-MS experiments. We are
also grateful to Drs. Worajit Setthapun and Weiling Deng
for helpful discussions involving the continuous flow reactor
studies. LANL is operated by Los Alamos National Security,
LLC, for the National Nuclear Security Administration of the
U.S. Department of Energy under contract DE-AC52-
06NA25396.
˚
an Alltech-washed 5 A molecular sieve column, and a thermal
conductivity detector, which had been calibrated for hydrogen,
cyclohexane, cyclohexene, cyclohexadiene, and benzene. Organic
products were trapped in a chilled receiver and later were
subjected to gas chromatography-mass spectrometry analysis.
Preparation of Supported Zerovalent Iridium. In an argon-
filled Vacuum Atmospheres glovebox with a large-capacity
recirculator (0.25 ppm of O2 and 0.5 ppm of H2O), 158 mg of
1 was loaded into a 1/2-in. quartz tube reactor equipped with an
Omega thermocouple and on/off valves. With argon purging,
the reactor assembly was attached to the catalyst testing unit.
The catalyst was treated with 5% H2 in argon at 370 ꢀC for 1 h,
under which the supported Ir3þ was reduced to Ir0.
Supporting Information Available: Stoichiometric propene
determination and IR, NMR, and EXAFS spectra. This ma-
terial is available free of charge via the Internet at http://
pubs.acs.org.