Robotic lepidoptery: Structural characterization of (mostly) unexpected palladium complexes obtained from high-throughput catalyst screening

Article abstract of DOI:10.1021/om900240a

In the course of a high-throughput search for optimal combinations of bidentate ligands with Pd(II) carboxylates to generate oxidation catalysts, we obtained and crystallographically characterized a number of crystalline products. While some combinations

Full text of DOI:10.1021/om900240a

Organometallics 2009, 28, 5017–5024 5017
DOI: 10.1021/om900240a
Robotic Lepidoptery: Structural Characterization of (mostly) Unexpected
Palladium Complexes Obtained from High-Throughput Catalyst
Screening
John E. Bercaw,*,† Michael W. Day,† Suzanne R. Golisz,† Nilay Hazari,† Lawrence
M. Henling,† Jay A. Labinger,*,† Susan J. Schofer,‡ and Scott Virgil†
†Arnold and Mabel Beckman Laboratories of Chemical Synthesis, California Institute of Technology,
Pasadena, California 91125, and ‡Symyx Technologies, Inc., 1263 East Arques Avenue, Sunnyvale,
California 94085
Received March 30, 2009
In the course of a high-throughput search for optimal combinations of bidentate ligands with Pd(II)
carboxylates to generate oxidation catalysts, we obtained and crystallographically characterized a
number of crystalline products. While some combinations afforded the anticipated (L-L)Pd(OC(O)R)2
structures (L-L = bipyridine, tmeda; R = CH3, CF3), many gave unusual oligometallic complexes
resulting from reactions such as C-H activation (L-L=sparteine), P-C bond cleavage (L-L=1,2-
bis(diphenylphosphino)ethane, and C-C bond formation between solvent (acetone) and ligand (L-L=
1,4-bis(2,6-diisopropylphenyl)-1,4-diaza-1,3-butadiene). These findings illustrate potential pitfalls of
screening procedures based on assuming uniform, in situ catalyst self-assembly.
Introduction
specimen.”3 In this sense, one potential role of an HTE robot
may be as an extremely effective butterfly net.
The role of automated high-throughput experimentation
(HTE) in homogeneous catalysis is steadily increasing.1,2
Efficient application of such methods virtually requires some
degree of catalyst self-assembly; having to presynthesize all
the candidates to be tested will almost always be far too time-
consuming. Most commonly members of a ligand library are
allowed to react in situ with a suitable metal complex, which
acts as the catalyst precursor. While this is often successful,
there is an obvious concern: in any given case, the targeted
species might not form, so that a negative result in catalytic
performance could reflect that failure rather than the actual
properties of the desired catalyst.
We have recently reported the palladium-catalyzed oxida-
tive aromatization of cycloolefins, using molecular oxygen as
the terminal oxidant: cyclohexene can be oxidized to ben-
zene, and 1,2-dihydronaphthalene to naphthalene, at room
temperature in the presence of palladium(II) trifluoroacetate
(tfa) and O2 (eq 1).4 Palladium complexes with diimine
ligands have also been found to oxidatively aromatize cyclo-
hexene to benzene, at elevated temperatures.5 For both of
these systems, C-H bond activation from an η2-olefinic
intermediate is proposed.
On the other hand, in HTE, reactions are often carried out
in multiwell plates, with the individual containers more or
less open to the environment, thus permitting gradual con-
centration by solvent evaporation; variations of relative and
absolute concentrations, solvent, and temperature may also
be included in the screening regime. These constitute excel-
lent conditions for the serendipitous generation of crystals-
of the intended product or something else-without much
(or any) additional effort. Long ago, Cotton summed up the
then-state-of-the-art in characterizing metal cluster struc-
tures: “Thus the student of cluster chemistry is in somewhat
the position of the collector of lepidoptera or meteorites,
skipping observantly over the countryside and exclaiming
with delight when fortunate enough to encounter a new
We wanted to extend these promising results to the oxida-
tion of more useful substrates such as linear alkenes and
alkanes, for which no examples of reaction were found under
the above conditions. While the ligand-free Pd(tfa)2 catalyst
system is more active than those of the diimine-ligated
complexes, it is also considerably less stable, decomposing
to Pd metal at temperatures much above ambient. Concei-
vably an appropriately ligated catalyst, even if inherently
less reactive, could be operated at sufficiently high tempera-
ture to overcome the differential and activate more diffi-
cult substrates. To explore this possibility, we initiated a
*Corresponding author. E-mail: bercaw@caltech.edu; jal@caltech.
edu.
(4) Bercaw, J. E.; Hazari, N.; Labinger, J. A. J. Org. Chem. 2008, 73,
8654–8657.
(5) Williams, T. J.; Caffyn, A. J. M.; Hazari, N.; Oblad, P. F.;
Labinger, J. A.; Bercaw, J. E. J. Am. Chem. Soc. 2008, 130, 2418–
2419. Bercaw, J. E.; Hazari, N.; Labinger, J. A.; Oblad, P. F. Angew. Chem.,
Int. Ed. 2008, 47, 9941–9943.
€
(1) Jakel, C.; Paciello, R. Chem. Rev. 2006, 106, 2912–2942.
(2) de Vries, J. G.; de Vries, A. H. M. Eur. J. Org. Chem. 2003, 799–
811, and references therein.
(3) Cotton, F. A. Q. Rev. Chem. Soc. 1966, 20, 389–401.
r
2009 American Chemical Society
Published on Web 08/20/2009
pubs.acs.org/Organometallics

5018 Organometallics, Vol. 28, No. 17, 2009
Table 1. Crystal and Refinement Data for Complexes 1, 2, 3, and 4
Bercaw et al.
1
2
3
4
empirical formula
fw
temperature (K)
C
546.70
100
30.4018(13)
13.6007(6)
18.8208(8)
90
90
90
14H8F6N2O4Pd C3H6O
3
C14H14N2O4Pd 5(H2O)
3
C10H16F6N2O4Pd
448.65
100
C38H62N4O8Pd3
1022.12
100
10.8368(6)
16.3596(9)
23.7139(12)
90
90
90
4204.1(4)
4
470.75
100
˚
a (A)
6.8790(3)
12.3760(5)
12.7548(6)
115.183(2)
96.790(2)
97.921(2)
954.15(7)
2
8.9413(5)
9.4538(5)
10.2243(5)
75.945(2)
74.203(2)
72.612(2)
781.24(7)
2
˚
b (A)
˚
c (A)
R (deg)
β (deg)
γ (deg)
volume (A )
Z
3
˚
7782.1(6)
16
cryst syst
space group
dcalc (Mg/m3)
θ range (deg)
)
abs correction
orthorhombic
Pnc2 (#30)
1.866
1.97 to 32.23
1.043
none
triclinic
P1 (#2)
1.639
1.80 to 39.31
1.020
semiempirical from equivalents
1.62
0.028, 0.042
triclinic
P1 (#2)
1.907
2.10 to 30.78
1.271
none
orthorhombic
P212121 (#19)
1.615
1.72 to 37.20
1.322
none
μ (mm-1
GOF
R1,a wR2b [I > 2σ(I)]
1.57
0.042, 0.072
P
2.19
0.036, 0.075
1.39
0.020, 0.033
P
P
P
a R1 =
Fo| - |Fc
/
|Fo|. b wR2 = [ [w(Fo2 - Fc2)2]/ [w(F2o)2]1/2
.
high-throughput study involving a variety of Pd(II) precur-
sors, chelating ligands, solvents, and reaction conditions.
While no effective metal-ligand combination was detected,
the study did generate a number of crystalline species. Some
of these displayed solid state structures quite different from
those expected for simple coordination of ligand to Pd
(which may be why catalytically active species were not
formed), often resulting from unexpected and unusual
reactivity. We report here on these findings.
considerable cross-contamination by volatile components.
Although we expect this problem could have been largely
eliminated by improving the experimental setup (which
employed a porous rubber mat held flush to the glass vials
with a metal plate affixed by screws), instead we simply
restricted screening to the less volatile 1,2-dihydronaphtha-
lene, for which no such transfer could be detected. Most
ligand-Pd combinations resulted in no oxidation activity
whatsoever; a couple of cases gave low activity, substantially
less than that found for the ligand-free catalysts. (We believe
incomplete complexation is probably responsible for these
low conversions; another possibility is that some ligated
palladium complexes are active, although inherently less
active than their ligand-free counterparts). Hence, this sur-
vey was abandoned.
A parallel set of substrate-free samples was generated, using
the robot, to investigate whether the desired catalysts assem-
ble properly. In many of these cases, solvent evaporation in air
directly provided X-ray quality crystals. All such reactions
were scaled up in the laboratory, to determine by NMR
spectroscopy whether the crystalline material was identical
to the main product generated in solution. Some of the
products were also examined by mass spectrometry.
Results and Discussion
HTE Methodology. The robot was used to generate solu-
tions of a wide range of combinations of catalyst precursor
(Pd(tfa)2 or Pd(OAc)2), ligand, and substrate (cyclohexene
or 1,2-dihydronaphthalene) in acetone (Pd(tfa)2 and Pd-
(OAc)2 have limited solubility in many other solvents,
and the control reaction described in eq 1 shows signi-
ficant solvent dependence, so the choice of solvent was not
varied in this work). The following N- and P-centered
bidentate ligands were expected to react according to eq 2:
2,20-bipyridine (bpy), N,N,N0,N0-tetramethylethylenedia-
mine (tmeda), (-)-sparteine, 1,2-bis(diphenylphosphino)-
ethane (dp), and 1,4-bis(2,6-diisopropylphenyl)-1,4-diaza-
1,3-butadiene (diimine). 2-Phenylpyridine was also included
in the study, anticipating that a chelate product would result
from C-H activation, as shown in eq 3.
Bipyridine Complexes. Both Pd(tfa)2 and Pd(OAc)2 react
with bpy according to eq 2, as confirmed by 1H NMR
spectroscopy. From the HTE screening, (bpy)Pd(tfa)2 (1)
was obtained as pale yellow needles (crystal and refinement
data are shown in Table 1). The inner coordination sphere of
1 (Figure 1) appears entirely unexceptional, with the bipyr-
idine lying in the Pd coordination plane and the two mono-
dentate trifluoroacetate groups bent out of the plane, both in
the same direction, very similar to the orientation found in an
earlier structure of a (bpy)Pd(carboxylate)2.6 The crystal
packing of 1 reveals intermolecular Pd-Pd and π-stacking
interactions between adjacent molecules (Figure 2), and the
separations fall within the expected range: Pd-Pd distances
˚
are between 3.0 and 3.2 A, and all spacings between the bpy
˚
rings are less than 3.8 A, with the closest contact being
˚
approximately 3.4 A. Each complex participates in a Pd-Pd
interaction with another complex directly above (or below)
When cyclohexene was used as substrate, control sam-
ples (without substrate) were found to contain substan-
tial amounts of both cyclohexene and benzene, indicating
(6) Canty, A. J.; Denney, M. C.; Skelton, B. W.; White, A. H.
Organometallics 2004, 23, 1122–1131.

Article
Organometallics, Vol. 28, No. 17, 2009 5019
Figure 1. Solid statestructure of 1. Hydrogens have been omitted
˚
for clarity. Selected bond distances of 1 (A): Pd1-N1 2.004(1),
Pd1-N2 1.993(1), Pd1-O1 2.016(1), Pd1-O3 2.007(1).
Figure 3. Comparison of the molecular conformation in the
two polymorphs of (bpy)Pd(tfa)2 with 1 as the solid bonds and 10
as the dashed bonds.
Figure 4. Solid statestructure of 2. Hydrogens have been omitted
˚
for clarity. Selected bond distances of 2 (A): Pd1-N1 1.998(1),
Pd1-N2 1.992(1), Pd1-O1 2.013(1), Pd1-O3 2.007(1).
A second polymorph of 1 (10) was also obtained. Although
these crystals were not of high enough quality to obtain
quantitative data, an overlay of the two molecules shows clear
differences (Figure 3). In particular, the trifluoroacetate
groups are no longer oriented in the same direction with
respect to the square plane as in 1. This allows the molecules
to align in the unit cell so that π-stacking is the predominant
interaction, with no short Pd-Pd contacts. For 1, it appears
that such an alignment, with only π-stacking (or only d8-d8
interactions), is prevented by steric repulsions between the
parallel-oriented tfa groups. Clearly there is only a small
energetic difference between 1 and 10, and the exact conditions
for crystallization influence which polymorph is obtained.
The yellow crystals of (bpy)Pd(OAc)2 (2) obtained through
HTE exhibit a structure very similar to that of 10 (Figure 4),
with the two acetate groups oriented above and below the
coordination plane (crystal and refinement data are shown
in Table 1). There is a slightly offset face to face π-stacking
interaction between the bpy rings of adjacent molecules
Figure 2. Unit cell diagram of 1 viewed along the c-axis.
and a π-stacking interaction between bpy rings with the
complex directly below (or above). These two interactions
combine to create a columnar structure comprised of alter-
nating d8-d8 and π-stacking interactions. While both
π-stacking between adjacent bpy rings7 and d8-d8 interac-
tions between metal centers8,9 are well established, to the best
of our knowledge, 1 is the first example of a structure that
contains both of these interactions in the same crystal.
Another noteworthy feature of the layered structure is the
presence of four independent molecules in the unit cell.
˚
(shortest contact 3.4 A) and no evidence for any Pd-Pd
interaction.7 One notable feature of 2 is a hydrogen-bonded
network of waters (not shown), which are also hydrogen
bonded to the acetate ligands. This water presumably origi-
nates from wet solvents that were used in the reaction.
Tetramethylethylenediamine and Palladium(II) Trifluoro-
acetate. The reaction of tmeda and Pd(tfa)2 in acetone
(7) Janiak, C. J. Chem. Soc., Dalton Trans. 2000, 3885–3896.
(8) Roundhill, D. M.; Gray, H. B.; Che, C. M. Acc. Chem. Res. 1989,
22, 55–61.
(9) Connick, W. B.; Marsh, R. E.; Schaefer, W. P.; Gray, H. B. Inorg.
Chem. 1997, 36, 913–922.
1
also followed eq 2 (confirmed by H NMR spectroscopy),
giving colorless crystals of (tmeda)Pd(tfa)2 (3). The struc-
ture (Figure 5) is similar to those of 1 and 2 but exhibits

5020 Organometallics, Vol. 28, No. 17, 2009
Bercaw et al.
Figure 7. Schematic representation of 4.
Figure 5. Solid state structure of 3. Hydrogens have been
˚
omitted for clarity. Selected bond distances of 3 (A): Pd1-N1
2.017(1), Pd1-N2 2.034(1), Pd1-O1 2.025(1), Pd1-O3 2.079(1)
or 2.095(1). (If only one value is given, the distances are the same
in both conformations.)
Figure 8. Solid state structure of 4. Hydrogens have been omitted
˚
for clarity. Selected bond distances of 4 (A): Pd1-N1 2.061(1),
Pd1-C7 2.010(1), Pd3-N3 2.069(1), Pd3-C30 1.997(1).
Figure 6. Overlay of the two conformations of 3.
disorder in both the tfa groups and the ethylene backbone
(crystal and refinement data are shown in Table 1). In the two
(uncoupled) conformations in this crystal, the carbonyl
oxygens of the tfa ligands are on either the same or opposite
sides of the coordination plane, and the two carbons of the
ethylene backbone of tmeda alternate between being up or
down with respect to the coordination plane (Figure 6).
Phenylpyridine Reactions. The reaction of phenylpyridine
with both Pd(OAc)2 and Pd(tfa)2 gave C-H activation
products as expected (eq 3). The yellow crystals of both pro-
ducts exhibit a “clamshell” dimer structure, as reported
previously for the acetate complex,10,11 in which the two
phenylpyridines are stacked on top of each other and the
bridging acetates are perpendicular to the palladium-ligand
coordination planes. The structure of the tfa complex,
which features a relatively short palladium-palladium inter-
action, will be fully discussed in a subsequent paper des-
cribing the electronic structure of d8-d8 bonding in Pd
complexes.12
Figure 9. Schematic representation of 5.
to palladium (eq 2),13-16 the gray crystals obtained from
mixing (-)-sparteine and Pd(OAc)2 in acetone instead con-
sist of (sparteinyl)Pd(μ-OC(CH3)O)2Pd(μ-OC(CH3)O)2Pd-
(sparteinyl) (4), the result of C-H activation at a bridgehead
position of sparteine (crystal and refinement data are shown
in Table 1). The resulting trimetallic complex contains three
palladium atoms (Figures 7 and 8), in an approximately linear
(-)-Sparteine and Palladium(II) Acetate. While there are
many examples of sparteine acting as an L2-type ligand
(13) Togni, A.; Rihs, G.; Pregosin, P. S.; Ammann, C. Helv. Chim.
Acta 1990, 73, 723–732.
(14) Trend, R. M.; Stoltz, B. M. J. Am. Chem. Soc. 2004, 126, 4482–
4483.
(15) Intini, F. P.; Pacifico, C.; Pellicani, R. Z.; Roca, V.; Natile, G.
Inorg. Chim. Acta 2008, 361, 1606–1615.
(16) Mueller, J. A.; Sigman, M. S. J. Am. Chem. Soc. 2003, 125, 7005–
7013.
(10) Thu, H.-Y.; Yu, W.-Y.; Che, C. M. J. Am. Chem. Soc. 2006, 128,
9048–9049.
€
(11) Dinc-er, M.; Ozdemir, N.; Gumar, M. E.; C-etinkaya, B. Acta
€
Crystallogr. Sect. E: Struct. Rep. Online 2008, E64, m381.
(12) Bercaw, J. E.; Durrell, A. C.; Gray, H. B.; Green, J. C.; Hazari,
N.; Labinger, J. A.; Winkler, J. R., manuscript in preparation.

Article
Organometallics, Vol. 28, No. 17, 2009 5021
˚
Figure 10. Solid state structure of 5. Hydrogens have been omitted for clarity. Selected bond distances of 5 (A): Pd1-P1 2.215(1), Pd1-P2
2.214(1), Pd1-O1 2.080(1), Pd1-O8 2.145(1), Pd2-P2 2.240(1), Pd2-C291.973(1),Pd2-O2 2.150(1), Pd2-O5 2.153(1), Pd3-P3 2.214(1),
Pd3-P4 2.216(1), Pd3-O3 2.084(1), Pd3-O6 2.149(1), Pd4-P4 2.227(1) Pd4-C55 2.004(1), Pd4-O4 2.169(1), Pd4-O7 2.128(1).
Table 2. Crystal and Refinement Data for Complexes 5 and 6
5
6
empirical formula
fw
temperature (K)
C
60H60O8P4Pd4
C310H428N20O30Pd10 8(C2H4O2) 2(H2O)
3
3
1458.56
100
12.8301(6)
20.9667(9)
21.6140(10)
90
100.018(3)
90
5725.6(4)
4
6495.17
100
˚
a (A)
12.6109(5)
28.7865(11)
29.2433(12)
113.477(2)
100.174(3)
93.227(3)
9489.3(7)
1
˚
b (A)
˚
c (A)
R (deg)
β (deg)
γ (deg)
volume (A )
Z
3
˚
cryst syst
space group
dcalc (Mg/m3)
θ range (deg)
)
abs correction
monoclinic
P21/c (#14)
1.692
1.91 to 30.44
1.402
semiempirical from equivalents
triclinic
P1 (#2)
1.137
1.66 to 22.50
0.523
semiempirical from equivalents
μ (mm-1
GOF
R1,a wR2b [I > 2σ(I)]
2.96
0.072, 0.126
2.32
0.079, 0.137
P
P
P
P
a R1 =
Fo| - |Fc
/
|Fo|. b wR2 = [ [w(Fo2 - Fc2)2]/ [w(F2o)2]1/2
.
˚
arrangement (—Pd1-Pd2-Pd3 = 176.03ꢀ; Pd-Pd≈ 2.9 A),
presumably the hydrogen generated by C-H activation is
released as acetic acid. It is probable that the solvent
(acetone) is partially responsible for promoting C-H bond
activation, as it has been previously reported that reaction
of sparteine with Pd(OAc)2 in dichloroethane results in the
formation of (sparteine)Pd(OAc)2.16 To the best of our
knowledge, this is the first example of a metal complex
containing a metalated sparteine ligand.
1,2-Bis(diphenylphosphino)ethane and Palladium(II) Ace-
tate. The colorless crystals obtained from the mixture of dp
and Pd(OAc)2 in acetone exhibit a tetrametallic structure,
[(Pd(μ-PPhCH2CH2PPh2)(μ-OC(CH3)O)PdPh)(μ-OC(CH3)-
O)]2 (5), resulting from C-P cleavage of the dp ligand
each exhibiting square-planar geometry. The central Pd is
˚
coordinated to four acetate groups (Pd-O ≈ 2.0 A); these
bridge in pairs to the two outer Pds, each of which is addi-
tionally coordinated by a (κ2-N,C)-sparteinyl group. The
sparteinyl ligand has undergone an inversion at one of the
nitrogenatoms (relative tofreesparteine), presumablytoassist
with coordination. Overall, the structure of the molecule is
S-shaped and seems to involve two d8-d8 interactions.12
When this reaction was performed on a large scale, the
presence of 4 as the major product was confirmed by HRMS
1
and H NMR spectroscopy. We are unable to propose a
detailed mechanism for formation of 4 at this time, although

5022 Organometallics, Vol. 28, No. 17, 2009
Bercaw et al.
Figure 11. Solid state structure of 6. Hydrogens have been omitted for clarity. Selected bond distances of 6 are listed in the Supporting
Information.
(Figure 9). Two representations of 5 are shown in Figure 10
(crystal and refinement data are shown in Table 2). The four
Pd(II) centers are each square planar and are arranged in
pairs. One member of each pair (Pd2, Pd4) is bonded to a
phenyl group, and the other (Pd1, Pd3) to the intact Ph2P end
of a dp. The other (phosphide) end of the dp, along with one
acetate, bridge the two members of each pair; and then the
pairs are joined together by two additional bridging acetates
(Pd1 to Pd4, Pd2 to Pd3). The overall arrangement is
an approximately planar macrocycle containing the four
Pd atoms (torsion angle=0.43ꢀ), the two bridging phosphide
P atoms, and the two interpair bridging acetates, with the
other (intrapair) bridging acetates and the phosphine P
atoms above and below the plane.
Repeated large-scale reactions of dp and Pd(OAc)2 in
acetone gave only (dp)Pd(OAc)2 (5m), the expected mono-
meric product according to eq 2. This complex has been
Figure 12. Schematic representation of 6, showing key bond
lengths (values averaged over the five independent units).
previously reported, along with the crystal structure (which
includes a coordinated CH2Cl2 molecule).17 The structural
parameters of 5m were identical to those in the literature,
with the replacement of CH2Cl2 by half a molecule of acetone
in the unit cell (see Supporting Information for crystal and
refinement data). We were unable to synthesize 5 on a large
scale, and thus designate it as a minor product in the reaction
of dp and Pd(OAc)2.
in acetone yielded small red crystals, for which X-ray crystal-
lographic studies revealed the remarkable cyclic decameric
structure shown in Figure 11 (crystal and refinement data
in Table 2). The connections between units involve -OC-
(CH3)X- groups that are O-bound to one Pd and X-attached
to a backbone carbon of a diimine ligand on the neighboring
Pd. At first we thought that X was either O or CH2, corre-
sponding respectively to bridging acetate or the enolate of
acetone, either of which would give overall charge neutrality.
However, closer inspection of the crystallographic data shows
1,4-Bis(2,6-diisopropylphenyl)-1,4-diaza-1,3-butadiene and
Palladium(II) Acetate. The reaction of diimine with Pd(OAc)2
(17) Bianchini, C.; Lee, H. M.; Meli, A.; Oberhauser, W.; Peruzzini,
M.; Vizza, F. Organometallics 2002, 21, 16–33.

Article
Organometallics, Vol. 28, No. 17, 2009 5023
Figure 13. Diagram of the crystal packing of 6.
clearly that X is in fact a dCH- group. The structure refines
much more poorly with X = O than X = C; electron density
peaks corresponding to a single H can be located at the
expected positions adjacent to each C; and the relative bond
lengths are consistent only with this formulation, as is the 1H
NMR spectrum, which includes signals (all of intensity corre-
sponding to one proton) at δ 4.05 (s), 3.15 (d), and 4.83 (d) for
the dCH- and the twononequivalent backbone CH2 protons,
respectively. (A parent ion for the intact decamer could not be
detected by mass spectrometry, perhaps not surprisingly.) The
molecular structure is shown schematically in Figure 12, along
with the key bond distances.
Stoichiometrically, the formation of 6 corresponds to net
double deprotonation of acetone, with the addition of the
carbon end to one of the backbone imine carbons and
migration of the hydrogen originally on that carbon to the
adjacent carbon; this has taken place 10 times per macro-
cycle, in which each Pd is square-planar, coordinated by one
imine N, one amidic N, one normal (monodentate) anionic
acetate, and an O from what has become a substituted
acetonyl. Since three of those are anionic ligands, each Pd
is anionic; hence there must be one additional proton per Pd
that was not crystallographically detected. Most probably
those are hydrogen-bonded to the dangling oxygens on the
periphery of the macrocycles; alternatively, they may be
bonded to solvent and/or water molecules inside the chan-
nels that are formed by the stacking of decamers in the crystal
packing (Figure 13). There is no covalent bonding between
the channels. We were unable to obtain the high-angle
diffraction data that would be needed to determine whether
any solvent or other molecules are present in the channels,
and thermogravimetric analysis was inconclusive, but it
seems highly unlikely that they are empty considering the
intimate involvement of the solvent.
Figure 14. Schematic representation of 6-dn showing the sites of
H/D scrambling.
It appears that 6 is not merely a minor component of the
reaction mixture, as it could be synthesized in reasonable
yield (36%) on a large scale. We would expect that a
compound of the form (diimine)Pd(OAc)2 might be an
1
intermediate, and H NMR changes during early stages of
the reaction between diimine and Pd(OAc)2 in acetone-d6
suggest that is the case. We were subsequently able to isolate
(impure) samples of (diimine)Pd(OAc)2 from the reaction of
the diimine ligand and Pd(OAc)2 in CHCl3, and dissolution
of isolated (diimine)Pd(OAc)2 in acetone-d6 led to the for-
mation of 6-dn, as confirmed by X-ray crystallography and
1H NMR spectroscopy. Detailed comparison of the 1H
NMR spectrum with that for the all-protio analogue (see
the Supporting Information) shows that the acetone-derived
methyl group is entirely deuterated, but remaining deuterons
on the deprotonated end have completely exchanged with the
backbone protons of the ligand (Figure 14), as revealed by
signals corresponding to both CH2 and CHD in the methy-
lene position and reduction of intensity for all three of the
signals cited above. At present we do not have a mechanistic
proposal to account for the formation of 6 and the H/D
scrambling process.

5024 Organometallics, Vol. 28, No. 17, 2009
Bercaw et al.
Conclusions
Bruker KAPPA APEX II instrument. Structures were deter-
mined using direct methods or, in some cases, Patterson maps
with standard Fourier techniques using the Bruker AXS
software package.
While the above findings remind us of the potential pitfalls
of screening protocols that rely on catalyst self-assembly,
they also illustrate their capability for generating intriguing
structures. More than half of the ligand-metal combina-
tions tested afforded products resulting from unexpected
chemistry-C-H bond activation, C-P bond cleavage, ad-
dition of solvent to ligand-that led to further assembly into
oligomeric structures with intricate and unusual intercon-
nections, which (while gladdening the collector’s heart) are
less likely to function as effective catalysts. Because the
efficiency of HTE screening for homogeneous catalysts relies
on some assumed uniformity of catalyst self-assembly, it is
important to carry out some structural verifications of
(pre)catalysts so generated, for at least representative li-
gand/metal complex/solvent combinations. Researchers
making use of these powerful methodologies should stay
alert to both the risks and the opportunities.
General Procedure for Large-Scale Reactions. In order to
confirm whether the crystals formed in the HTE screening were
the major products, reactions between metal precursors and
ligands were performed on a larger scale in the laboratory. The
conditions utilized in the HTE screening were replicated. In the
case of reactions between bpy and Pd(tfa)2, bpy and Pd(OAc)2,
tmeda and Pd(tfa)2, and dp and Pd(OAc)2, 1H NMR spectros-
copy was used to monitor the reactions and indicated the clean
formation of compounds 1, 2, 3, and 5m, respectively. The
chemical shifts for 1,19 2,19 3,14 and 5m17 were consistent with
those previously reported in the literature.
(sparteinyl)Pd(μ-OC(CH3)O)2Pd(μ-OC(CH3)O)2Pd(sparteinyl)
(4). A solution of (-)-sparteine (104 mg, 0.44 mmol) in ace-
tone (4.4 mL) was added to a solution of Pd(OAc)2 (100 mg,
0.44 mmol) in acetone (4.4 mL). The solution was allowed to
evaporate in air to give gray crystals (37 mg, 24% yield), which
were filtered from the supernatant before analysis. See SI for 1H
NMR spectrum. HRMS (FABþ): obsd Mþ 1021.160, calcd for
C38H62N4O8Pd3, 1021.160.
Experimental Section
[(diimine)(Pd)(OAc)(OC(CH3)CH)]10 (6). A solution of dii-
mine (167 mg, 0.44 mmol) in acetone (4.4 mL) was added to a
solution of Pd(OAc)2 (100 mg, 0.44 mmol) in acetone (4.4 mL).
The solution was allowed to evaporate in air to give red-orange
crystals (94.5 mg, 36% yield), which were filtered from the super-
natant before analysis. 1H NMR (500 MHz, CDCl3) δ 0.62 (d, J =
6, 3H, iPr), 0.82 (m, 6H, iPr), 0.90 (d, J = 6, 3H, iPr), 1.00 (s, 3H,
iPr*), 1.13 (d, J = 6, 3H, iPr), 1.19 (d, J = 6, 3H, iPr), 1.22 (d, J =
6, 3H, iPr), 1.99 (s, 3H, CH3), 2.13 (s, 3H, CH3), 2.90 (m, 1H, CH),
3.05 (m, 1H, CH), 3.15 (d, J = 13, 1H, CH2), 3.61 (m, 1H, CH),
4.05 (s, 1H, CH), 4.45 (m, 1H, CH), 4.83 (d, J = 13, 1H, CH2), 6.60
(d, J = 7, 1H, Ar), 6.88 (m, 2H, Ar), 7.03 (t, J = 7, 1H, Ar), 7.11
(t, J = 7, 2H, Ar). The resonance with an asterisk corresponds to
an unknown peak that has appeared in all spectra of all samples
of 6. This unknown is coincident with an isopropyl resonance.
General Considerations. All manipulations were performed in
air. Acetone, palladium(II) acetate, palladium(II) trifluoroace-
tate, (-)-sparteine, 2-phenylpyridine, and tmeda were purchased
from Aldrich and used as received. 2,20-Bipyridine was purchased
from Acros Organics and used as received. 1,2-Bis(diphenyl-
phosphino)ethane was purchased from Pressure Chemical
Company and used as received. 1,4-Bis(2,6-diisopropylphenyl)-
1,4-diaza-1,3-butadiene was prepared by literature methods.18
Acetone-d6 and chloroform-d3 were purchased from Cambridge
Isotopes and used as received. 1H NMR spectra were recorded on
a Varian INOVA 500 MHz instrument using the VNMRJ soft-
ware program, version 2.2d, at room temperature. Proton che-
mical shifts were reported using the residual solvent signal as the
internal standard. High-resolution mass spectra (HRMS) were
obtained from the California Institute of Technology Mass
Spectrometry Facility.
High-Throughput Equipment. Experiments were conducted
using a Symyx Technologies Core Module robotic system
(Santa Clara, CA). The core module was operated with library
designs and operating protocols developed using Library Studio
software version 7.1.9.50 and Automation Studio version
1.1.1.8. Stock solutions of metal complexes and ligands were
prepared at 0.1 M in acetone. The metal solutions were added to
the wells followed by an equimolar amount of the ligand
solutions. Acetone was ultimately added to each of the wells
to keep a constant total volume of 500 μL. Each combination of
metal and ligand was performed in quadruplicate, with a con-
centration gradient for each set of four wells such that the least
concentrated contained 5 mmol each of metal and ligand and the
most concentrated contained 15 mmol each of metal and ligand.
The wells were then left open to the air on the benchtop until
crystals formed. Crystal formation was complete in less than
24 h for some complexes.
Acknowledgment. The authors acknowledge bp for
funding through the MC2 program and thank John C.
McKeen of Caltech for TGA and Drs. Thomas J. Crevier
and Jeffrey C. Yoder of Symyx for suggestions relating to
the robotics experiments. The Bruker KAPPA APEXII
X-ray diffractometer was purchased via an NSF CRIF:
MU award to the California Institute of Technology,
CHE-0639094.
Supporting Information Available: Selected 1H NMR spectra,
crystal and refinement data for 5m, selected bond distances of 6,
and all CIFs. This material is available free of charge via the
Internet at http://pubs.acs.org. CCDC 708866 (1), 713343 (2),
717612 (3), 711734 (4), 717605 (5), 723085 (5m), and 724780 (6)
contain the supplementary crystallographic data for this paper.
These data can be obtained free of charge from The Cambridge
Crystallographic Data Centre via www.ccdc.cam.ac.uk/
data_request/cif.
X-ray Crystallography. The crystals were mounted on a
glass fiber with Paratone-N oil. Data were collected on a
(19) Milani, B.; Alessio, E.; Mestroni, G.; Sommazzi, A.; Garbassi,
F.; Zangrando, E.; Bresciani-Pahor, N.; Randaccio, L. J. Chem. Soc.,
Dalton Trans. 1994, 1903–1911.
(18) Kliegman, J. M.; Barnes, R. K. J. Org. Chem. 1970, 35, 3140–
3143.