Industrial Ziegler-type hydrogenation catalysts made from Co(neodecanoate)2 or Ni(2-ethylhexanoate)2 and AlEt3: evidence for nanoclusters and sub-nanocluster or larger Ziegler-nanocluster based catalysis.

Abstract

Ziegler-type hydrogenation catalysts are important for industrial processes, namely, the large-scale selective hydrogenation of styrenic block copolymers. Ziegler-type hydrogenation catalysts are composed of a group 8-10 transition metal precatalyst plus an alkylaluminum cocatalyst (and they are not the same as Ziegler-Natta polymerization catalysts). However, for ∼50 years two unsettled issues central to Ziegler-type hydrogenation catalysis are the nature of the metal species present after catalyst synthesis, and whether the species primarily responsible for catalytic hydrogenation activity are homogeneous (e.g., monometallic complexes) or heterogeneous (e.g., Ziegler nanoclusters defined as metal nanoclusters made from combination of Ziegler-type hydrogenation catalyst precursors). A critical review of the existing literature (Alley et al. J. Mol. Catal. A: Chem. 2010, 315, 1-27) and a recently published study using an Ir model system (Alley et al. Inorg. Chem. 2010, 49, 8131-8147) help to guide the present investigation of Ziegler-type hydrogenation catalysts made from the industrially favored precursors Co(neodecanoate)(2) or Ni(2-ethylhexanoate)(2), plus AlEt(3). The approach and methods used herein parallel those used in the study of the Ir model system. Specifically, a combination of Z-contrast scanning transmission electron microscopy (STEM), matrix assisted laser desorption ionization mass spectrometry (MALDI MS), and X-ray absorption fine structure (XAFS) spectroscopy are used to characterize the transition metal species both before and after hydrogenation. Kinetic studies including Hg(0) poisoning experiments are utilized to test which species are the most active catalysts. The main findings are that, both before and after catalytic cyclohexene hydrogenation, the species present comprise a broad distribution of metal cluster sizes from subnanometer to nanometer scale particles, with estimated mean cluster diameters of about 1 nm for both Co and Ni. The XAFS results also imply that the catalyst solutions are a mixture of the metal clusters described above, plus unreduced metal ions. The kinetics-based Hg(0) poisoning evidence suggests that Co and Ni Ziegler nanoclusters (i.e., M(≥4)) are the most active Ziegler-type hydrogenation catalysts in these industrial systems. Overall, the novelty and primary conclusions of this study are as follows: (i) this study examines Co- and Ni-based catalysts made from the actual industrial precursor materials, catalysts that are notoriously problematic regarding their characterization; (ii) the Z-contrast STEM results reported herein represent, to our knowledge, the best microscopic analysis of the industrial Co and Ni Ziegler-type hydrogenation catalysts; (iii) this study is the first explicit application of an established method, using multiple analytical methods and kinetics-based studies, for distinguishing homogeneous from heterogeneous catalysis in these Ziegler-type systems; and (iv) this study parallels the successful study of an Ir model Ziegler catalyst system, thereby benefiting from a comparison to those previously unavailable findings, although the greater M-M bond energy, and tendency to agglomerate, of Ir versus Ni or Co are important differences to be noted. Overall, the main result of this work is that it provides the leading hypothesis going forward to try to refute in future work, namely, that sub, M(≥4) to larger, M(n) Ziegler nanoclusters are the dominant, industrial, Co- and Ni- plus AlR(3) catalysts in Ziegler-type hydrogenation systems.

Full text

Page 1

Published:April 11, 2011
r2011 American Chemical Society
6279
dx.doi.org/10.1021/la200053f|Langmuir 2011, 27, 6279–6294
ARTICLE
pubs.acs.org/Langmuir
Industrial Ziegler-Type Hydrogenation Catalysts Made from
Co(neodecanoate)2or Ni(2-ethylhexanoate)2and AlEt3: Evidence for
Nanoclusters and Sub-Nanocluster or Larger Ziegler-Nanocluster
Based Catalysis
William M. Alley,†Isil K. Hamdemir,†Qi Wang,‡Anatoly I. Frenkel,‡Long Li,§Judith C. Yang,§
Laurent D. Menard,||Ralph G. Nuzzo,||Saim€ Ozkar,^Kuang-Hway Yih,#Kimberly A. Johnson,zand
Richard G. Finke*,†
†Department of Chemistry, Colorado State University, Fort Collins, Colorado 80523, United States
‡Department of Physics, Yeshiva University, New York, New York 10016, United States
§University of Pittsburgh, Pittsburgh, Pennsylvania 15261, United States
)
University of Illinois, Urbana, Illinois 61801, United States
^Department of Chemistry, Middle East Technical University, Ankara, Turkey 06531
#Hungkuang University, Taichung, Taiwan 433
zShell Oil Company, Houston, Texas 77082, United States
b
S Supporting Information
’INTRODUCTION
Ziegler-typehydrogenationcatalystsare,bydefinition,formed
from a non-zerovalent group 8?10 transition metal (M)
Received:
Revised:
January 5, 2011
March 11, 2011
ABSTRACT: Ziegler-type hydrogenation catalysts are important for industrial processes, namely, the
large-scale selective hydrogenation of styrenic block copolymers. Ziegler-type hydrogenation catalysts are
composedofagroup8?10transitionmetalprecatalystplusanalkylaluminumcocatalyst(andtheyarenot
thesameasZiegler?Nattapolymerizationcatalysts).However,for∼50yearstwounsettledissuescentralto
Ziegler-typehydrogenationcatalysisarethenatureofthemetalspeciespresentaftercatalystsynthesis,and
whether the species primarily responsible for catalytic hydrogenation activity are homogeneous (e.g.,
monometalliccomplexes)orheterogeneous(e.g.,Zieglernanoclustersdefinedasmetalnanoclustersmade
from combination of Ziegler-type hydrogenation catalyst precursors). A critical review of the existing
literature (Alley et al. J. Mol. Catal. A: Chem. 2010, 315, 1?27) and a recently published study using an Ir
model system (Alley et al. Inorg. Chem. 2010, 49, 8131?8147) help to guide the present investigation of
Ziegler-type hydrogenation catalysts made from the industrially favored precursors Co(neodecanoate)2or Ni(2-ethylhexanoate)2, plus
AlEt3. The approach and methods used herein parallel those used in the study of the Ir model system. Specifically, a combination of
Z-contrast scanning transmission electron microscopy (STEM), matrix assisted laser desorption ionization mass spectrometry (MALDI
MS),andX-rayabsorptionfinestructure(XAFS) spectroscopyareusedtocharacterizethetransitionmetalspeciesbothbeforeandafter
hydrogenation.KineticstudiesincludingHg(0)poisoningexperimentsareutilizedtotestwhichspeciesarethemostactivecatalysts.The
main findings are that, both before and after catalytic cyclohexene hydrogenation, the species present comprise a broad distribution of
metalclustersizesfromsubnanometertonanometerscaleparticles,withestimatedmeanclusterdiametersofabout1nmforbothCoand
Ni.TheXAFSresultsalsoimplythatthecatalystsolutionsareamixtureofthemetalclustersdescribedabove,plusunreducedmetalions.
The kinetics-based Hg(0) poisoning evidence suggests that Co and Ni Ziegler nanoclusters (i.e., Mg4) are the most active Ziegler-type
hydrogenationcatalystsintheseindustrialsystems.Overall,thenoveltyandprimaryconclusionsofthisstudyareasfollows:(i) thisstudy
examines Co- and Ni-based catalysts made from the actual industrial precursor materials, catalysts that are notoriously problematic
regarding their characterization; (ii) the Z-contrast STEM results reported herein represent, to our knowledge, the best microscopic
analysis of the industrial Co and Ni Ziegler-type hydrogenation catalysts; (iii) this study is the first explicit application of an established
method, using multiple analytical methods and kinetics-based studies, for distinguishing homogeneous from heterogeneous catalysis in
theseZiegler-typesystems;and(iv)thisstudyparallelsthesuccessfulstudyofanIrmodelZieglercatalystsystem,therebybenefitingfroma
comparisontothosepreviouslyunavailablefindings,althoughthegreaterM?Mbondenergy,andtendencytoagglomerate,ofIrversusNi
orCoareimportantdifferencestobenoted.Overall,themainresultofthisworkisthatitprovidestheleadinghypothesisgoingforwardto
tryto refute in future work, namely, that sub, Mg4to larger, MnZiegler nanoclusters are the dominant, industrial, Co- and Ni- plus AlR3
catalysts in Ziegler-type hydrogenation systems.

Page 2

6280
dx.doi.org/10.1021/la200053f |Langmuir 2011, 27, 6279–6294
Langmuir
ARTICLE
precatalyst such as Co(neodecanoate)2 or Ni(2-ethylhexan-
oate)2plus a trialkylaluminum cocatalyst such as triethylalumi-
num (AlEt3).Ziegler-type hydrogenation catalysts shouldnotbe
confused, however, with Ziegler?Natta or other common poly-
merization catalysts, which are not a subject of this study. The
relatively inexpensive Co- or Ni-based catalysts made from
Co(neodecanoate)2or Ni(2-ethylhexanoate)2, respectively, are
significant industrially, as they are used in the production of
∼1.7 ? 105metric tons of hydrogenated styrenic block copoly-
mers per year.1Several important fundamental questions about
Ziegler-type hydrogenation catalysts persist despite the use of
these catalysts for five decades.1?3One of the most important
remaining questions is the ∼50-year-old problem of whether the
true nature of Ziegler-type hydrogenation catalysis is homoge-
neous (e.g., single metal organometallic) versus heterogeneous
(e.g., nanoclusters).1,3?6
A recently published critical review of Ziegler-type hydroge-
nation catalysts includes an examination of the prior evidence
concerning their homogeneous versus heterogeneous nature.
That review reveals that the reasons for the longevity of this
problem, in this class of catalysts, include their sensitivity to
variables and conditions in their preparation and use, plus their
resistance to characterization by physical methods and isolation
for kinetic studies.2,3The literature review3led to the suggestion
that answering the homogeneous versus heterogeneous catalysis
question for Ziegler-type hydrogenation catalysts could be
facilitated through the use of a well characterized, third-row
transition-metalprecatalystincombinationwithamultipronged,
previously successful approach to solving the homogeneous
versus heterogeneous catalysis problem in a variety of other
catalystsystems.3,6?15Thecentralconceptsofthismultipronged
approach toward answering the homogeneous versus heteroge-
neous catalysis question are (i) identification of the potential
catalyst species using multiple complementary techniques, and
then(ii)kineticstudiestodeterminethecatalyticcompetencyof
those species.
SuchstudiesusingaZiegler-typehydrogenationcatalystmade
from the crystallographically characterized precatalyst, [(1,5-
COD)Ir(μ-O2C8H15)]2,plus
published.14Among the multiple analytical methods used were
Z-contrast scanning transmission electron microscopy (STEM),
matrix assisted laser desorption ionization mass spectrometry
(MALDI MS), and X-ray absorption fine structure (XAFS)
spectroscopy.14Since “catalysis is, by definition, a wholly kinetic
phenomenon”,16kinetic studies were performed as a necessary
component of addressing the homogeneous versus heteroge-
neous catalysis question.3,14Those studies revealed that the Ir
species present are different before versus after their use for
catalytic cyclohexene hydrogenation. Specifically, before hydro-
genation the catalyst solutions contain a wide range of Ir species
from mono-Ir complexes up to structurally disordered Ir∼100
Ziegler nanoclusters, with an estimated mean of 0.5?0.7 nm,
Ir∼4?15clusters,14whereas after hydrogenation, the Ir is in the
form of fcc Ir(0)∼40?150 Ziegler nanoclusters.14Moreover,
poisoning and other kinetic studies suggested that the fcc
Ir(0)∼40?150Ziegler nanoclusters are the kinetically dominant
catalysts.14
The goal of the present study is to repeat the analyses
performed on the Ir model Ziegler-type hydrogenation catalyst
system with Co- and Ni-based catalysts made from the authentic
Co(neodecanoate)2or Ni(2-ethylhexanoate)2precursor materi-
alsusedfor industrialpolymerhydrogenation.As such, this work
AlEt3
havebeenrecently
expands not only on our own previous study using the Ir model
system,14but also on the results of others—notably the valuable
studies by Schmidt and co-workers,17and B€ onnemann and co-
workers18—that suggest transition metal nanoclusters are pre-
sent in the Ziegler-type systems they studied.
Our main hypotheses for the present work are (i) that the
approach that proved useful with the homogeneous vs hetero-
geneous catalysis question in the Ir system14willbe applicable to
the industrial Co- and Ni-based systems, and (ii) that the results
will be similar in that the most active catalysts will be revealed to
consistofCoorNiZieglernanoclusters,evenifassmallasCo4or
Ni4, that is subnanometer clusters. Many of the same analytical
techniques are employed herein, namely, Z-contrast STEM,
MALDI MS, XAFS spectroscopy (through its two complemen-
tary modifications, X-ray absorption near edge structure, or
XANES, and extended XAFS, or EXAFS), and Hg(0) poisoning
kineticsstudies.Analogoustothepreviousstudyon theIrmodel
system,14the specific objectives entail (i) determining the
nuclearity of the Mnspecies present initially (M is Co or Ni),
(ii) establishing what Mnspecies are present directly after use of
the catalysts for cyclohexene hydrogenation, and (iii) using
Hg(0) poisoning as a kinetics-based test of the homogeneous
vs heterogeneous nature of the active catalyst.3The challenging,
yet crucial, issues of the form(s) taken and role(s) played by the
AlEt3component in Ziegler-type hydrogenation catalysts are
currently being investigated, and will be reported elsewhere in
due course.19
Before the use of catalyst solutions for cyclohexene hydro-
genation, the Z-contrast STEM and MALDI MS results which
follow reveal that Mnclusters with a wide range of sizes are
obtained from combining Co(neodecanoate)2 or Ni(2-
ethylhexanoate)2, and AlEt3, and the average cluster sizes are
between0.9and1.4nmindiameter.TheresultsoftheZ-contrast
STEM herein are, to the best of our knowledge, the best existing
microscopic analysis of industrial Co and Ni Ziegler-type hydro-
genationcatalysts.TheXANESspectroscopyresultssuggestthat
a combination of nanoclusters and unreduced metal ions exists,
withtheratioofthetwophasesdepending,asonemightexpect,3
on theAl/Mratio.EXAFSspectroscopicanalysisofbothCoand
Ni catalyst samples gives mean first-nearest-neighbor (1NN)
coordination number (N) values for both metals in the 3?4
range. The most plausible, self-consistent interpretation of the
evidence from multiple, complementary techniques is that the
transition metal contents of the catalyst solutions are a combina-
tion of disordered nanoclusters and unreduced, monometallic
species. In addition, Z-contrast STEM, MALDI MS, and XAFS
all show that the transition metal species in catalyst solutions
remain essentially unchanged by their use for cyclohexene
hydrogenation. Furthermore, Hg(0) poisoning studies suggest
that catalysis is heterogeneous (i.e., occurs via the observed
subnanometerornanoscaleM∼4?130clusters).Noteworthyhere
is that, since control experiments (vide infra and in the Support-
ingInformation)showthatAlEt3isrequiredtogenerateanactive
catalyst(thatXANESshowsisreducedfromCo(II)),specieslike
Co?Et that can β-hydrogen eliminate to ethylene plus Co?H,
andthusplausiblespeciessuch ashydridic Co4H4subnanometer
clusters, all become candidates for the true catalyst. Through the
useofanestablishedapproachtodistinguishhomogeneousfrom
heterogeneous catalysis,3,6?15and with the additional advantage
of now being able to compare the results to those from a parallel
study of an Ir model system, this study provides the best current
evidence suggesting catalysis by what appear to be Ziegler

Page 3

6281
dx.doi.org/10.1021/la200053f |Langmuir 2011, 27, 6279–6294
Langmuir
ARTICLE
nanoclusters (i.e., Mg4) in Ziegler-type hydrogenation catalysts
made from the actual industrial Co and Ni precatalyst materials.
’EXPERIMENTAL SECTION
Materials and Instruments. Material sources used to prepare
catalyst solutions were kept consistent in order to obtain reproducible
results (vide infra). All materials were stored and handled under a N2
atmosphere in a Vacuum Atmospheres drybox, unless stated otherwise.
Drybox O2levels were continuously monitored via a Vacuum Atmo-
spheres O2-level indicator and maintained ate5 ppm.Gastight syringes
were used to carry out all solution measurements and additions done in
theFinkegroupdryboxatColoradoStateUniversity(CSU).Procedures
used to control the amount of H2O present were followed consistently
to ensure reproducibility (vide infra); glassware was rinsed with
nanopure water, dried overnight at 160 ?C, and cooled under a vacuum
orN2atmosphere.Cyclohexane(Sigma-Aldrich,99.5%,H2O<0.001%)
was kept over molecular sieves (Acros, 3 Å, activated by heating at
200 ?C for 6 h under vacuum) for g2 days prior to use with the Co
catalyst, but used as received with the Ni catalyst (vide infra). Cyclohex-
ene (Aldrich, 99%) was distilled over Na under argon. Precatalysts were
obtained from OM Group, Inc., (OMG) as solutions in mineral spirits,
Co(neodecanoate)2, 12% wt Co, and Ni(2-ethylhexanoate)2, 8% wt Ni
(productnames:12%Coten-cemand8%Nihex-cem).Theseindustrial
precatalyst sources of Co(neodecanoate)23aRCO2H3bH2O or Ni(2-
ethylhexanoate)23cRCO2H3dH2O (both from OMG) are neither rela-
tively pure nor well-characterized structurally compared to the Ir model
[(1,5-COD)Ir(2-ethylhexanoate)]2precatalyst,whichwascharacterized
via single crystal X-ray diffractometry and used as the pure crystalline
starting material for the preparation of catalyst solutions.1,14Hereafter,
we will refer to these as simply Co(neodecanoate)2 and Ni(2-
ethylhexanoate)2(i.e., a,b,c,d = 0). Both were used after diluting with
cyclohexane to 12.0 mM in [M] (molecular weights of Co-
(neodecanoate)2andNi(2-ethylhexanoate)2fordilutionswereassumed
to be the corresponding a,b,c,d = 0 values of 401.5 g/mol and 345.1
g/mol,respectively).AlEt3(StremChemicals, 93%) was used as asolution
in cyclohexane. Both Ar and H2gases were passed through moisture
(Scott Specialty Gases) and oxygen traps (Trigon Technologies) prior
to use. THAP (20,40,60-trihydroxyacetophenone, Aldrich, 98%), used in
the MALDI MS experiments as a matrix, was stored and used outside of
the drybox, and applied as an aqueous solution.
Catalyst Solution Preparation and Catalytic Cyclohexene
Hydrogenations. Previous investigation into both the existing
literature,3and the Ir model system14have made it clear that Ziegler-
type hydrogenation catalysts are sensitive to the conditions and proce-
dures used in their synthesis. We therefore carried out a variety of initial
control experiments—testing the effects of catalyst aging, the Al/M
ratio, the volume and concentration of catalyst solution prepared, the
amount of H2O present, temperature, concentration of AlEt3used, and
order and rate of precursor component combination—all with the goal
of ensuring that the characterization results obtained herein would be
both reproducible and representative of active Ziegler-type hydrogena-
tion catalysts. The results from these control experiments are briefly
summarized here and given in greater detail in the Supporting Informa-
tion for the interested reader. One of the important findings from these
control experiments is the presence of gas-to-solution mass transfer
limitation(MTL) effectsinourcurrent hydrogenationapparatus, which
limits the measurable hydrogenation uptake rate to the rate of H2gas
transfer into solution where the catalytic reaction takes place.20How-
ever, we have used catalyst preparation methods and conditions for this
studythat(i)resultincatalyticcyclohexenehydrogenationratesthatare
atleastasrapidaswecanobserveduetotheMTLeffectspresent,(ii)are
consistent with the most favorable methods and conditions described in
the majority of the literature,3and (iii) are similar to, or the same as,
those used for the model Ir Ziegler-type hydrogenation catalyst made
from [(1,5-COD)Ir(μ-O2C8H15)]2and AlEt3.1,14In short, the MTL
kinetics present for these exceptionally active, industrial Ziegler-type
hydrogenation catalysts did not preclude our determination of condi-
tions and procedures for catalyst synthesis necessary to give results that
are both reproducible and representative of active Ziegler-type hydro-
genation catalysts standardized to the MTL limit of our apparatus.
Once established, the procedures for preparing and using catalyst
solutions(referredtohereafter asthestandardconditions)werefollowed
consistently for repeat experiments unless specified otherwise. Control
experiments demonstrate that the presence of (deliberately added)
water during catalyst synthesis negatively affects the cyclohexene
hydrogenation activity of the resulting catalysts. Therefore, all glassware
was carefully dried as was the cyclohexane solvent for use with the Co-
basedcatalyst(cyclohexanedryingwasnotbeneficialfortheNicatalysts;
see the Supporting Information). The catalyst solutions were made
under a N2atmosphere by combination of a 36.0 mM cyclohexane
solution of AlEt3 with a 12.0 mM Co(neodecanoate)2 or Ni(2-
ethylhexanoate)2precatalyst stock solution. The ratios Al/Co = 3 and
Al/Ni = 2 were used for the standard conditions on the basis of control
experiments testing catalysts prepared with a range of Al/M values.
ControlexperimentswereperformedwithanAl/Mratioofzeroforboth
Co and Ni, and it was found that no H2gas uptake occurred without
added AlEt3, which shows the importance of the alkylaluminum
cocatalyst in making active Ziegler-type hydrogenation catalysts.
Synthesis of catalyst solutions in batches up to 20 mL, as opposed to
the2.5mLofcatalystsolutionpreparedforuseinasinglehydrogenation
run,hadnoobservableeffectoncatalystactivity.Likewise,batchcatalyst
preparation at 7.2 mM in [M] had no observable effect on catalyst
activityincomparisontothe1.44mMin[M]catalystsolutionsprepared
for use in a single hydrogenation run (diluted after preparation to
1.2 mM in [M] with the addition of 0.5 mL of cylcohexene). Therefore,
it was possible to prepare catalyst solutions either individually or
batchwise as necessary, and at concentrations necessary for the subse-
quent type of analysis. Catalyst synthesis carried out with solutions
heated to 60 ?C resulted in catalyst solutions with lower cyclohexene
hydrogenation activity (Supporting Information); hence,catalyst synth-
esis at the ambient drybox temperature of ∼25 ?C was established as a
standard condition. For the sake of consistency, and unless noted
otherwise, catalyst solutions were prepared by adding the AlEt3solution
to either the Co(neodecanoate)2or Ni(2-ethylhexanoate)2solution
dropwise but rapidly (at a rate g1 drop every 5 s), and with 1000 (
200 rpm stirring (measured with a Monarch Instruments Pocket-
Tachometer 100). As an example of batch catalyst preparation, 20 mL
of catalyst solution was prepared by first adding 16.8 mL of cyclohexane
to a 20 mL glass vial containing a new 5/8 ? 5/16 in Teflon-coated
magnetic stir bar. Next, 1.6 mL of a 12.0 mM cyclohexane solution of
Ni(2-ethylhexanoate)2was added. Stirring was started, followed by
addition of 1.6 mL of a 36.0 mM AlEt3solution. Stirring in the drybox
was continued for 30 min, after which aliquots of the catalyst solution
were taken for analysis or transferred to a new 22 ? 175 mm Pyrex
borosilicate culture tube containing a new 5/8 ? 5/16 in Teflon-coated
magnetic stir bar for kinetic studies via use in cyclohexene hydrogena-
tion. Since, as noted above, volume and concentration had no effect on
hydrogenation, catalyst solutions were also prepared directly in the
culture tubes for individual hydrogenation runs by, for example, first
adding 1.9 mL of cyclohexane to a culture tube followed by 0.3 mL of a
cyclohexane solution of Co(neodecanoate)2, 12.0 mM in [Co]. Stirring
was started and then 0.3 mL of the 36.0 mM AlEt3 solution in
cyclohexanewasadded.Cyclohexene,0.5mL,wasaddedlast.Ingeneral,
theproceduresusedinthisstudywereverysimilarto,andinanumberof
cases the same as, those used previously for the Ir model system.14
After combination of the precursor components, cyclohexene was
added to catalyst solutions used for catalytic hydrogenation runs.

Page 4

6282
dx.doi.org/10.1021/la200053f |Langmuir 2011, 27, 6279–6294
Langmuir
ARTICLE
Control experiments show that aging prepared catalyst solutions
resulted in decreased catalyst activity (Supporting Information), so
catalysts were used for hydrogenation or otherwise analyzed as soon
as possible after preparation. The procedure and apparatus used for
catalytic cyclohexene hydrogenation have been described in detail
elsewhere.21Briefly, the culture tube containing the catalyst solution
was placed in a Fisher-Porter (F?P) bottle, sealed, and transferred out
of the drybox. The F?P bottle was placed in a temperature regulating
bath, stirring was begun, and the F?P bottle was connected to a
pressurized H2line using Swagelock quick-connects. The F?P bottle
was purged 15 times (1 purge/15 s) before setting the pressure to 40
psig. Data collection was then started at 4 min after the first purge. H2
pressure data as a function of time were collected using an Omega PX
624?100 GSV pressure transducer, which was connected to a PC
running LabView 7.0 by an Omega D1131 analog-to-digital converter.
Data were subsequently handled using MS Excel and Origin 7. Standard
conditionsforhydrogenationrunsareasfollows:solvent=cyclohexane,
[M]=1.2mM,initial[cyclohexene]=1.65M,temp=22.0?C,initialH2
pressure = 40psig, and stirring rate =1000 (10rpm. Themainpoint is
that,inbothcatalystsynthesisandsubsequenthydrogenations,variables
with the potential to influence the resulting catalytic activity have been
tested and optimized (to the MTL limit), thereby allowing the devel-
opment of standard conditions for the preparation and use of the highly
active Ziegler-type hydrogenation catalysts used herein. This in turn
ensures that the subsequent analytical results should be both reprodu-
cible and representative of active Ziegler-type hydrogenation catalysts.
Z-Contrast STEM. Catalyst samples were prepared according to
standardconditionsasdescribed;theywerethencollectedforZ-contrast
microscopy both before and after use in cyclohexene hydrogenation.
Sample solutions were double-sealed airtight, and shipped to the Uni-
versity of Pittsburgh for imaging (2?3 days between preparation and
analysis). Preparation of samples on TEM grids was carried out in a
glovebag filled with dry N2at >1 atm, and located in the TEM room.
Sample solutions were diluted with cyclohexane to twice their original
volume,and2?3dropsweredispersedontoaTEMgridwithanultrathin
carbonfilmonaholeycarbonsupport(TedPella,Inc.).Theseweredried
atroomtemperatureunderN2forg10minbeforebeingtransferredinto
the TEM instrument. Transfer was done quickly to reduce possible
oxidation of the sample. Samples were first treated with a high-intensity
electron beam (electron beam shower) for ∼15 min each time in the
TEM column (with vacuum better than 3 ? 10?6Torr). Images were
acquired using a field-emission JEM 2010 (scanning) transmission
electron microscope operated at 200 kV. The high-angle scattering
electrons were collected with a JEOL ADF detector at a camera length
of 8 cm, with a 0.2 nm (nominal) diameter probe. High-angle annular
dark-field (HAADF) images were collected at 2 M (million) magnifica-
tion, andwere 1024 ? 1024 pixels in dimension.Cluster diameters were
measured manually at the full width at half-maximum (fwhm) of the
intensity profile across g600 clusters from images at the same levels of
magnification and contrast using Gatan Digital Micrograph.
Controlexperimentswereperformedtodeterminewhetherthemetal
clusters observed were artifacts of the microscopy itself. Co-
(neodecanoate)2, without added AlEt3, was deposited on a TEM grid
(ultrathin carbon film supported by a lacey carbon film on a 400 Mesh
copper grid, TedPella), andimaged following themethods notedabove
(i.e., including the electron beam shower). No Co clusters could be
observed, suggesting that neither the sample preparation procedures,
nor the Z-contrast STEM conditions, are responsible for creating the
observed clusters in catalyst samples. No Co clusters were observed
whenthissamecontrolexperimentwascarriedoutusinghighresolution
(HR)TEM. (The fact that Co in Co(neodecanoate)2could not be
observedinZ-contrastSTEMimageswithoutCoclusterformationhasa
bearing on the interpretation of the EXAFS results, vide infra; specifi-
cally, it leaves open the possibility that monometallic, unreduced metal
ions are present.) Additionally, Co(neodecanoate)2, without added
AlEt3, was deposited on special TEM grids with 25-nm-thick SiO2
windows (Dune Sciences).22However, for this sample on the special
SiO2grids, imaging using bright field TEM, Z-contrast STEM, and
HRTEM all revealed the presence of nanometer-scale clusters, osten-
sibly the result of Co cluster formation under the TEM beam. These
control experiments suggest that the clusters observed using Z-contrast
STEM imaging of catalyst samples deposited on ultrathin carbon grids,
and resultant cluster size histograms, are not artifacts resulting from the
required sample handling or microscopy itself. Images from control
experiments and additional microscopy are provided in the Supporting
Information for the interested reader.
MALDI MS. Catalyst samples were prepared for analysis by MALDI
MSinamanneralmostidenticaltothatdescribedpreviouslyusingtheIr
model system.14A0.5 μL, 100mMaqueous NaIionizingagentsolution
was hand-spotted on a steel MS sample plate and air-dried, which was
followed by 1 μL of 20,40,60-trihydroxyacetophenone (THAP) over the
samespotandthenalsoair-dried.Theplatewasthentransferredintothe
drybox where sample solutions (1 μL, [M] = 1.44 mM) were applied
ontothespotofdepositedionizingagentandmatrix.Theplatewasthen
covered withitsplasticcapping plateandplaced intoadesiccator, which
wassealedandremovedfromthedrybox.Theplatewastransferredinair
(exposure of ∼30 s) from the desiccator to the vacuum of the MALDI
MS instrument, and MALDI MS spectra were taken immediately
thereafter. Mass spectra were obtained at CSU on a Bruker Ultraflex
TOF-TOF instrument in linear mode, with acceleration voltage at 25
kV, and in positive ion mode. A nitrogen laser (λ = 337 nm) with a 3 ns
pulsewidthwasfocusedovera1-mm-diameterspot.Datawerecollected
with the highest laser power possible, for a higher S/N, but which still
maximized resolution and avoided sample fragmentation. Calibration
was done using Bradykinin, Angiotensin_I, Angiotensin_II, Substan-
ce_P, Bombesin, Renin_Substrate, ACTH_clip and Somatostatin
(purchased as a mixture of all these peptides from Bruker-Daltonics).
XAFS. Procedures forXAFS spectroscopy herein are similar to those
used previouslyfortheanalysisoftheIrmodelsystem.14Solutionsamples
of Co(neodecanoate)2, Ni(2-ethylhexanoate)2, and catalysts made from
these plus AlEt3were prepared at Colorado State University, in 6.0 mL
batchesat7.2mMconcentrationin[M].Aliquotsofcatalystsampleswere
used for cyclohexene hydrogenation in order to obtain both pre- and
posthydrogenation catalyst samples. All samples were then sealed airtight,
and transported to the National Synchrotron Light Source (NSLS) at
Brookhaven National Laboratory (BNL), Upton, NY (2 days transit). At
the NSLS, catalyst samples were handled and stored in an N2atmosphere
glovebox maintained at e10 ppm O2. Catalyst samples were loaded, via
glass pipet, into a custom-designed, airtight, ∼1.5 mL capacity, solution
sample cell composed of a stainless steelframe made to press Kaptonfilm
windowsontoaTeflonblock.ThreadedportsintheTeflonblockallowfor
sample loading, whichwere then sealedusing Teflon screws. Airtightseals
in the threaded ports and windows were ensured by using Kalrez o-rings.
XAFS experiments were performed at room temperature on beamline
X18borX11a,whicharesourcedbybendingmagnets,andemploySi(111)
double-crystal monochromators. Samples were loaded into an airtight
sample cell, then mounted and positioned at 45? in the beam path. Three
30-cm-long ion chambers filled with suitable gas mixtures were employed
to record in transmission mode the incident, transmitted, and reference
beam. A Lytle detector was used to measure fluorescence data simulta-
neously with transmission, but the fluorescence spectra were deemed of
inferior quality to the transmission spectra and, therefore, not used in the
analysis. Co or Ni foils were used both for absorption edge calibration of
theCo(7709eV)andNi(8333eV)KedgespriortoXAFSscans.Coand
Ni foils were also used to obtain reference spectra simultaneously in
transmission mode for all sample scans. Six to eight scans were typically
performed for each sample; during data processing multiple scans of a
single sample were merged (averaged).

Page 5

6283
dx.doi.org/10.1021/la200053f |Langmuir 2011, 27, 6279–6294
Langmuir
ARTICLE
DataprocessingwasaccomplishedusingIFEFFIT.23Forbackground
removal, thresholdenergyvalues(E0)forbothCoandNiwereassigned
values corresponding to the inflection point in the normalized absorp-
tionedges.AHanningwindowfunctionwasusedtoselectdatarangesin
k-spacewithsufficientsignal-to-noiseratioforFouriertransforms(FTs)
(Supporting Information). The passive electron reduction factors (S02)
for Co and Ni were acquired from fitting the Co and Ni foil standards,
respectively (Supporting Information). Parameters including the co-
ordination numbers (N), bond lengths (R), and their disorders (σ2)
were varied in the fitting of catalyst sample spectra, as well as the
correction to the photoelectron energy origin (ΔE0). Details of fitting
EXAFS spectra are given in the Supporting Information.
Hg(0) Poisoning. Catalyst solutions for use in Hg(0) poisoning
experiments were first prepared in the drybox according to the standard
conditions as described with [M] concentration of 1.2 mM (M is Co or
Ni), an Al/Co ratio of 3.0, or an Al/Ni ratio of 2.0, and initial
cyclohexene concentrations of 1.65 M. In one version of the Hg(0)
poisoning experiment, a standard conditions hydrogenation was
stopped after about half the cyclohexene had been consumed by filling
andpurgingtheF?PbottlefivetimeswithArgaspressurizedto40psig.
The F?P bottle was then transferred back into the drybox where the
Hg(0)wasaddedandallowedtomixforthespecifiedtime(24hforCo,
1.5 h for Ni). The F?P bottle was then reconnected to the hydrogena-
tion line, refilled with H2gas using the standard procedure and data
acquisition was restarted. Time and pressure values collected after
Hg(0) addition were corrected to fit with the data collected before
Hg(0) addition (see Figure 10). In another version of the Hg(0)
poisoning experiment, Hg(0) was added to the catalyst solutions before
cyclohexene hydrogenation catalysis was started and allowed to mix for
thespecifiedtime.ThebottlecontainingthecatalystsolutionandHg(0)
was then transferred to the pressurized H2to collect pressure data using
normal procedures.
TheresultsofHg(0)poisoningcontrolexperiments areshowninthe
Supporting Information for the interested reader. In the case of the Co
catalyst, eight control experiments carried out independently by two
differentresearchers(WMAandKHY)using∼1700equivofHg(0)per
Co and 24 h of mixing show that the degree of poisoning of the Co
catalyst when Hg(0) is added prior to the start of cyclohexene hydro-
genationisirreproducible.FortheNicatalyst,controlexperimentsusing
various quantities of Hg(0) added to prepared catalyst solutions,
followed by various mixing times before their use in hydrogenation,
show that a procedure using g300 equiv of Hg(0) per Ni and g1.5 h of
stirring (at 1000 rpm in a sealed FP bottle in the drybox) is adequate to
thoroughly contact the Hg(0) with all of the Ni catalyst in solution; this
procedure was then strictly followed and proved reproducible. Other
control experiments show that both the Co and Ni catalyst solutions
retain catalytic activity when subjected to the handling procedures
required for Hg(0) addition, but in the absence of Hg(0). Restated,
those additional controls show that it is the Hg(0) itself, and not the
procedures,thatpoisonthecatalysis(seetheSupportingInformationfor
details).
’RESULTS AND DISCUSSION
InitialObservations, Plus anOverviewoftheKeyPre- and
Posthydrogenation Characterization Results. As noted in a
review of the literature of the homogeneous versus heteroge-
neous catalysis problem,6initial observations of the catalyst
solutions alone make industrial Ziegler-type catalysts candidates
for study regarding the homogeneous vs heterogeneous catalysis
question. Specifically, dark brown or black solutions are fre-
quently observed in literature catalyst systems now known to
involveheterogeneous(e.g.,nanoparticle)catalysis,makingsuch
anobservation,byitself, suggestive ofheterogeneouscatalysis.6In
the present study, there are several noteworthy observations from
thesynthesisoftheindustrialCo-andNi-basedcatalysts,especially
in comparison with the observations from the Ir model system.14
Forexample,additionoftheclearandcolorlesssolutionofAlEt3to
the clear, deep-blue Co(neodecanoate)2solution results in an
immediate change to a dark brown, almost black solution. Likewise,
addition of the AlEt3solution to the clear, light-green solution of
Ni(2-ethylhexanoate)2causes an immediate change to a dark
brown solution (but one that is a lighter shade of brown than the
Co/AlEt3catalyst solution). Unlike with the [(1,5-COD)Ir(μ-
O2C8H15)]2plus AlEt3catalyst system, which is a much lighter,
yellow-brownsolutionafteradditionofAlEt3butdarkensduringa
cyclohexene hydrogenation run, and will occasionally precipitate
a dark-brown powder a few days after the completion of a
hydrogenation run,14these industrial Co- or Ni-based catalysts
do not exhibit observable color change or insoluble particle
formation during, or post, hydrogenation. Using the Ir model
catalyst,itwasfoundthatH2uptakebeginsinitiallyataslowerrate,
then accelerates to achieve its maximum rate after the start of
hydrogenation (i.e., the initial rate is not the maximum rate).14
Furthermore, this increase in cyclohexene hydrogenation rate
during the hydrogenation using the model Ir catalyst is accom-
panied by the observation of, on average, Ir∼4?15clusters pre-
hydrogenation, but fcc Ir(0)∼40?150clusters posthydrogenation.
In contrast, with the industrial Co- or Ni-based catalysts, H2
uptake begins immediately at the apparent H2gas-to-solution
MTL rate (∼80 ( 20 psig/h at 1000 rpm stirring; Supporting
Information) or at ∼30% of the apparent H2gas-to-solution
MTL rate, respectively (Figure 1). This implies that, in the
Figure 1. Top: general steps for the synthesis of Co- or Ni-based
Ziegler-type hydrogenation catalyst solutions; M(O2CR)2is either of
the authentic industrial precatalysts, Co(neodecanoate)2 or Ni(2-
ethylhexanoate)2. Catalyst solutions were made by combining a cyclo-
hexane solution of one of the precatalysts, 12.0 mM in [M], with a
36.0 mM cyclohexane solution of AlEt3. Example catalytic cyclohexene
hydrogenation curves using standard conditions of solvent = cyclohex-
ane,[M]=1.2mM,initial[cyclohexene]=1.65 M,temp=22.0 ?C,and
stirring rate = 1000 ( 10 rpm are shown. The apparent MTL value,
depicted here as ablack line, is∼80 (20 psig/h in our apparatus and at
these conditions (e.g., at the 1000 ( 10 rpm stirring rate).

Page 6

6284
dx.doi.org/10.1021/la200053f |Langmuir 2011, 27, 6279–6294
Langmuir
ARTICLE
industrial Co- or Ni-based catalysts, very active catalyst species
are present initially after the addition of AlEt3, or possibly are
formed essentially immediately upon the introduction of H2gas.
In short, the initial observations from catalyst preparation alone
are consistent with the presence of Co and Ni Ziegler nanoclus-
ters in catalyst solutions both initially and throughout the
hydrogenation process.
These initial observations of the dark colors of the catalyst
solutions explain why the specific objectives herein necessarily
entail (i) determining the nuclearity of the Mnspecies present
initially, and (ii) establishing what Mnspecies are present directly
after use of the catalysts for cyclohexene hydrogenation. These
are the necessary first steps in probing the homogeneous versus
heterogeneous nature of the most active catalyst in these
industrial systems.
Asummaryoftheresultsobtainedfromtheanalysisofcatalyst
samples pre- and posthydrogenation by Z-contrast STEM and
MALDI MS is given in Table 1 alongside the results from the Ir
model system for comparison.14The key findings for both the
Co- and Ni-based catalysts are the following: (i) Z-contrast
STEM and MALDI MS reveal nanometer-scale clusters for both
CoandNisamples,bothbeforeandafterhydrogenation;and(ii)
the XAFS data indicate that unreduced metal ions are present in
solution,dependingontheAl/Mratio,withthenanometer-scale
Conor Ninclusters present. In addition, the XAFS shows those
Conand Ninclusters possess disordered atomic structures. In
short, disordered transition metal Ziegler nanoclusters appear to
bethepredominantclustersformedbytheindustrialCo-andNi-
based precatalysts upon addition of AlEt3, both before and after
hydrogenation, yet monometallic (homogeneous) species ap-
pear to be present as well. In addition, the ability to directly
comparetheresults obtained herein totheresults fromtheprior,
analogous study of the model Ir system,14is a valuable, unique
feature of the present study.
Nuclearity of MnSpecies before Hydrogenation: Z-Con-
trast STEM. Samples of the Co(neodecanoate)2 plus AlEt3
catalyst, with an Al/Co ratio of 3.0, before use for cyclohexene
hydrogenation were imaged using Z-contrast STEM. Measure-
ment of 604 clusters shows a range of Co cluster sizes from
0.6 to 3.3 nm in diameter, with a mode and median of 1.3 nm
clusters, and a mean Co cluster diameter of 1.4 ( 0.4 nm.
These cluster diameters correspond to cluster nuclearities with a
range from Co∼10to Co∼1700, a mode and median of Co∼100,
and a mean of Co∼130.24?26Figure 2 shows an example image
and the histogram.
SamplesoftheNi(2-ethylhexanoate)2plusAlEt3catalyst,with
an Al/Ni ratio of 2.0, before use for cyclohexene hydrogenation
werealsoimagedusingZ-contrastSTEM.Anexampleimageand
the histogram are shown in Figure 3. Measurement of 650
clusters in Z-contrast STEM images reveals a range of Ni cluster
sizes from 0.4 to 3.5 nm in diameter. The mode, median, and
mean Ni cluster diameters are 1.1 nm, 1.2 nm, and 1.3 ( 0.5 nm,
respectively. These diameters correspond to cluster nuclearites
ranging from Ni∼3to Ni∼2050, the mode, median, and mean
being Ni∼60, Ni∼80, and Ni∼100, respectively.24?26
For both Co and Ni samples, Z-contrast STEM shows the
presence of metal clusters with a broad distribution of sizes
ranging from subnanometer to several nanometers in diameter.
Cluster diameter measurements were made using the full width
at half-maximum (fwhm) of line intensity profiles across indivi-
dualclusters.TheseZ-contrastmicroscopyresultsbythemselves
should not be considered absolutely definitive, however, due to
the possibility that the observed clusters could be artifacts of the
microscopy itself, especially given that lighter (first-row) transi-
tion metal clusters and precursors are known to be less stable in
TEM electron beams than their heavier (third-row) analogs—a
key reason we began our studies with our now-published third-
rowmetal,Ir-modelsystem.14,27?29Morespecifically,NiZiegler-
type hydrogenation catalysts have been observed to be sensitive
to electron microscopy sample treatment processes, specifically,
drying of the Ni catalyst solution on TEM grids.2However, the
possibility of artifactual results is mitigated herein by the use of
scanning TEM,30which diminishes the potential for beam-
induced sample damage via a small electron probe, low beam
current, and minimal beam exposure time.31The images herein
were watched during image acquisition for signs of the influence
of the TEM beam on the catalyst sample, and no changes in
cluster size or shape were observed. In addition, control experi-
ments (described in the Experimental Section, images shown in
the Supporting Information) suggest that the clusters observed
using Z-contrast STEM, and measured to construct the cluster
size histograms, are not artifacts. To summarize, Z-contrast
microscopy shows that Co and Ni catalyst samples, before
hydrogenation, each contain a wide range of Mnclusters, 1.4 (
0.4 nm, Co∼130, and 1.3 ( 0.5 nm, Ni∼100, being the mean
clustersizeandnuclearityineachcase,respectively.Totheextent
Table 1. Summary of Results from Investigation of Metal Cluster Sizes Using Z-Contrast STEM and MALDI MS for Industrial
Ziegler-TypeHydrogenationCatalystsMadefromCo(neodecanoate)2orNi(2-ethylhexanoate)2plusAlEt3(Al/Cois3.0,Al/Niis
2.0),andforComparisonanIrZiegler-TypeHydrogenationCatalystmadefrom[(1,5-COD)Ir(μ-O2C8H15)]2plusAlEt3(Al/Iris
2.0), Both before and after Use for Cyclohexene Hydrogenation
precatalysispostcatalysis
analytical methodrange (nm)averagea(nm)averageaMnnuclearityrange (nm)averagea(nm)averageaMnnuclearity
Co Z-contrast STEM
MALDI MS
Z-contrast STEM
MALDI MS
Z-contrast STEM
MALDI MS
0.6?3.3
0.8?1.8
0.4?3.5
0.8?1.7
0.2?1.4
0.5?1.1
1.4
1.2
1.3
0.9
0.5
0.7
Co∼130
Co∼80
Ni∼100
Ni∼34
Ir∼4
Ir∼15
0.5?2.5
0.8?1.8
0.6?4.0
0.8?1.6
0.4?1.9
0.6?1.4
1.4
1.1
1.4
0.9
1.0
0.8
Co∼130
Co∼60
Ni∼130
Ni∼34
Ir∼40
Ir∼20
Ni
Irb
aThe average values are calculated mean cluster diameters from Z-contrast STEM, and estimated mean nuclearities from MALDI MS. Explanations for
how these values were determined, and how the cluster diameter-nuclearity conversion is performed, are given below.bResults from a previously
published study,14provided here for comparison.

Page 7

6285
dx.doi.org/10.1021/la200053f |Langmuir 2011, 27, 6279–6294
Langmuir
ARTICLE
of our knowledge, the results of the Z-contrast STEM herein are
the best existing microscopic analysis of industrial Co and Ni
Ziegler-type hydrogenation catalysts.
Nuclearity of MnSpecies before Hydrogenation: MALDI
MS. Samples of the Co(neodecanoate)2plus AlEt3catalyst, with
an Al/Co ratio of 3.0, were also analyzed using MALDI MS
before their use in cyclohexene hydrogenation. A broad peak is
observed with a maximum intensity at ∼4500 m/z (figures are
shown in the Supporting Information). With the assumptions
that the ions forming the broad peaks are composed of only Co
atoms,32?34and that the ionic charge is þ1,14,32,34,35the max-
imum intensity of the MALDI MS peak at ∼4500 m/z corre-
sponds to Co∼80 clusters. This, in turn, corresponds to a
diameter approaching ∼1.2 nm (used as an estimate of the
averageCoclustersreportedinTable1).Furthermore,thebroad
MALDI MS peak also indicates a wide size dispersity of the Co
clusters present, similar to the wide size dispersity of the Co
clusters observed using Z-contrast STEM. The fwhm of the
broad, asymmetrically shaped MALDI MS peak is from
∼2000?9000 m/z, and tails off toward higher m/z values. The
peakreaches one-fourth maximumintensityat∼12000m/zand
one-eighthmaximumintensityat∼16000m/z;thesem/zvalues
correspond to approximately Co∼30?150, Co∼200, and Co∼270
clusters, respectively, which in turn correspond to approximately
0.9?1.5, 1.6, and 1.8 nm Co clusters, respectively.
SamplesoftheNi(2-ethylhexanoate)2plusAlEt3catalyst,with
anAl/Niratioof2.0,werealsoanalyzedusingMALDIMSbefore
theiruseincyclohexenehydrogenation.Abroadpeakisobserved
withamaximumintensityatm/zof2000.However,thepresence
Figure 2. Example Z-contrast STEM image of the Co(neodecanoate)2
plus AlEt3catalyst, with an Al/Co ratio of 3.0, and before its use for
cyclohexene hydrogenation. The histogram from measuring 604 Co
clusters reveals an overall range of Co clusters observed from 0.6 to
3.3nmindiameter,whichcorrespondtoCo∼10toCo∼1700clusters.The
Co clusters measured have a mode and median of 1.3 nm, and a mean
diameter of 1.4 ( 0.4, corresponding to Co∼100and Co∼130clusters,
respectively.24?26
Figure
ethylhexanoate)2plus AlEt3catalyst with an Al/Ni ratio of 2.0, and
before use for cyclohexene hydrogenation. The histogram made from
measurement of 650 Ni clusters shows Ni cluster sizes ranging from 0.4
to 3.5 nm in diameter, which correspond to Ni∼3to Ni∼2050clusters.
The Ni clusters measured have a mode of 1.1 nm, a median of 1.2 nm,
and a mean diameter of 1.3 ( 0.5 nm, corresponding to Ni∼60, Ni∼80,
and Ni∼100, respectively.24?26
3. ExampleZ-contrastSTEMimage oftheNi(2-

Page 8

6286
dx.doi.org/10.1021/la200053f |Langmuir 2011, 27, 6279–6294
Langmuir
ARTICLE
ofNiatomsinspeciesbelow1500m/zisruledoutbytheabsenceof
characteristicNiisotopepeakdistributionsinthatregion.Inacontrol
experiment, the MALDI MS of a blank sample containing only the
matrix, trihydroxyacetophenone (THAP), and ionizing agent, NaI,
containspeaksinthe0?1500m/zrange(SupportingInformation).
Therefore, the 0?1500 m/z range was excluded from the mass
spectrum region used to calculate number of transition metal atoms
(M) in the Mnclusters and corresponding diameters, for both Co
andNicatalystsamples;them/zvaluesof1500?16000forCoand
1500?13500 for Ni were used to calculate the cluster diameter
ranges reported in Table 1. Using the same assumptions employed
fortheCosystemabove,aswellaspreviouslyintheliterature,14,32?35
themaximumintensityofthebroadpeakatm/zof∼2000indicates
Ni∼34clusters,correspondingto∼0.9-nm-diameterNinanoclusters
(usedasanestimate ofthe average NiclustersreportedinTable1).
Much like the MALDI MS peak of the Co catalyst (and of the Ir
model system14), the broad, asymmetrically shaped peak of the Ni
catalyst also tails off toward higher m/z values reaching ∼6000 m/z
at half-maximum intensity, ∼9000 m/z at one-fourth maximum
intensity,and∼13500m/zatone-eighthmaximumintensity,which
correspond to approximately Ni∼100, Ni∼150, and Ni∼230, respec-
tively. These nuclearities correspond, in turn, to approximately 1.3,
1.5, and 1.7 nm Ni nanoclusters, respectively.
Somewhatasanaside,thisstudy,andthepreviousoneoftheIr
modelsystem,14areinteresting,ifnotuniquetestsofthevalueof
MALDI MS as an analytical method for measuring the size and
size distribution of transition metal nanoclusters since they
obtain MALDI MS data on systems where Z-contrast STEM
(andXAFS,videinfra)dataareavailableforcomparison.Overall,
the MALDI MS-estimated nanocluster sizes and size distribu-
tionsforbothCoandNiprehydrogenationcatalystsaregenerally
consistent with those determined using Z-contrast STEM in
showing cluster sizes in the ranges 0.8?1.8 nm for Co and
0.8?1.7 nm for Ni are present.
NuclearityofMnSpeciesbeforeHydrogenation:XAFS(i.e.,
XANES plus EXAFS) Spectroscopy. The XANES spectra of
both Co and Ni catalysts are compared to those of the corre-
sponding metal foils and catalyst precursors in Figure 4. In each
case, the XANES spectra of the catalyst solution becomeless like
that of the precursor solution and more like that of the metal foil
with higher Al/M ratios. This suggests that, in terms of compo-
siteaverageformaloxidationstate,theCoorNimetalsincatalyst
solutions become progressively less like their M(II) precatalysts
and progressively more similar to M(0), as the Al/M ratios
increase from 1.0 to 3.0. These results imply that unreduced
metal ions are likely present in catalyst solutions in amounts that
decrease with additional AlEt3. Given the Mn nanoclusters
observed using both Z-contrast STEM and MALDI MS, these
results suggest that catalyst solutions contain a combination of
Mnclusters with a wide range of diameters and unreduced metal
ions, with the proportion of M atoms in the cluster versus ion
phasesdependingontheAl/Mratiousedincatalystpreparation.
The potential of EXAFS spectroscopy for the characterization
of Ziegler-type hydrogenation catalysts, especially the indust-
riallyfavoredCoandNicatalysts,wasmadeapparenttousbythe
valuable prior studies of Goulon and co-workers.36Specifically,
thoseauthorsfoundNi?Nifirstnearestneighborsindicatingthe
presence of Ni metal clusters.36However, additional study
proved worthwhile using modern EXAFS analysis methods that
use ab initio theory for the quantitative modeling and analysis of
experimental EXAFS spectra,37especially when considered
alongside results of complementary Z-contrast STEM and
MALDI MS techniques used herein, the Hg(0) poisoning
studies,andthenow-possible comparison totheresultsobtained
from the Ir model system.14
First, EXAFS data were collected separately for Co and Ni
foils, and cyclohexane solutions of the Co(neodecanoate)2and
Ni(2-ethylhexanoate)2precatalysts, without added AlEt3, for use
as reference samples (see Supporting Information for the full
results, including fits to the data). Solution samples of the
catalysts prepared by addition of AlEt3, but before their use in
cyclohexene hydrogenation, were then analyzed by EXAFS.
Spectra were collected for catalyst samples with Al/M ratios of
0.5,1.0,1.5,2.0,2.5,3.0,and5.0.However,theEXAFSspectraof
many of these samples were of sufficiently poor quality to make
fitting and interpretation unreliable. The highest quality spectra
were obtained for the Al/M =1.0 and 3.0 samples; therefore, the
spectra and fitting results of the Al/M = 1.0 and 3.0 samples are
shown here, but the spectra and fitting results from samples
prepared at other Al/M ratios are shown in the Supporting
Information.38For both Co and Ni catalysts, sample spectra
show peaks that correspond to the first nearest neighbor (1NN)
M?O peak in the precatalyst spectra, and to the first nearest
neighbor(1NN)M?MpeakintheMfoilspectra,Figure5.This
is analogous to the catalyst spectra of the Ir model catalyst
Figure4. (a)XANESspectraofCofoil(black)theCo(neodecanoate)2
catalyst precursor without added AlEt3(blue), and Co(neodecanoate)2
plus AlEt3catalysts with Al/Co ratios of 1.0 (red) and 3.0 (green). (b)
XANES spectra of Ni foil (black), the Ni(2-ethylhexanoate)2catalyst
precursor withoutaddedAlEt3(green), andNi(2-ethylhexanoate)2plus
AlEt3catalysts with Al/Ni ratios of 1.0 (pink) and 3.0 (blue). In each
case, with additional AlEt3, the XANES spectra of the catalyst solution
becomes less like the precursor solution and more like the metal foil.

Page 9

6287
dx.doi.org/10.1021/la200053f |Langmuir 2011, 27, 6279–6294
Langmuir
ARTICLE
system;14hence,thefittingstrategyusedhereinfortheCo-orNi-
based catalysts is analogous to the one employed to fit the
EXAFSspectraoftheIrmodelcatalystsamples.14TheCoandNi
catalystspectrawerefitusingcompositemodelscreatedfromthe
1NN M?O path of the precatalyst and the 1NN M?M path of
the bulk metal. Examples of fitting results are shown in Figure 6,
and given in Tables 2 and 3.
The main results from EXAFS are as follows: (i) peaks in the
3?6ÅrangeintheR-spaceEXAFSspectra(indicativeofordered
metallic structures and evident in the Co and Ni foil reference
spectra,Figure5) are absent for both CoandNi catalyst samples.
The lack of the large distance peaks observed here suggests that
Co and Ni catalyst samples are (a) composed of metal species
such as subnanometer metal clusters too small to have contribu-
tions in that interatomic distance range, (b) are composed of
larger metal nanoclusters with a high degree of atomic disorder,
or (c) are some combination of the two. The next main result
from the EXAFS data is that (ii) spectra are fit reasonably well
using a composite model generally analogous to the one em-
ployed for the Ir model system.14Significantly, and unlike in the Ir
Figure 5. (a) Fourier transform magnitudes of the k2-weighted EXAFS
spectra of Co metal foil (black), the Co(neodecanoate)2precatalyst
without added AlEt3(blue), and a sample of the Co(neodecanoate)2
plus AlEt3 catalyst with an Al/Co ratio of 1.0 before its use for
hydrogenation (red). (b) Fourier transform magnitudes of the k2-
weighted EXAFS spectra of Ni foil (black), the Ni(2-ethylhexanoate)2
precatalyst without added AlEt3(green), and a sample of the Ni(2-
ethylhexanoate)2plus AlEt3catalyst with an Al/Ni ratio of 1.0 before its
use for hydrogenation (pink). Upon addition of AlEt3, the Co and Ni
catalystsamplesstillshowapeakcorrespondingtothe1NN,M?Opeak
of the Co(neodecanoate)2 and Ni(2-ethylhexanoate)2 precatalysts,
respectively, but also display a peak corresponding to the 1NN, M?M
peakfromthespectrumofthebulkmetal.Also,andsignificantly,catalyst
samples lack peaks in the 3?6 Å range characteristic of ordered, metallic
structure. Spectra for Co and Ni foils are shown at one-fourth intensity
scale for the purpose of comparison.
Figure 6. Data and fits for (a) Co(neodecanoate)2plus AlEt3catalyst
and(b)Ni(2-ethylhexanoate)2plusAlEt3catalyst,withanAl/Mratioof
1.0 in each case. The highest quality spectra were obtained for the
Al/M = 1.0 and 3.0 samples; the experimental spectra and fits to the
Al/M =1.0 data are shownhere asexamples—spectra andfittingresults
fromsamplespreparedatotherAl/MratiosareshownintheSupporting
Information.
Table 2. Fitting Results from EXAFS Spectroscopic Analysis
of Co Reference Samples and Co(neodecanoate)2plus AlEt3
Catalyst Samples before Hydrogenation
Sample Al/CoCo foilCo(O2CR)2aCo catalyst 1.0 Co catalyst 3.0
NCo?Co
NCo?O
RCo?Co(Å)b
RCo?O(Å)b
σ2Co?Co(Å2)c
σ2Co?O(Å2)c
aCo(O2CR)2is the catalyst precursor Co(neodecanoate)2without
added AlEt3. The full analysis of Co(neodecanoate)2is given in the
Supporting Information.bR stands for the interatomic distance corre-
sponding to the single scattering paths.
square variation in R due to both static and dynamic disorder (also
known as the EXAFS Debye?Waller factor), and values shown
are ?103.dFor Co foil, this parameter was defined as the value shown
(i.e., not varied in the fit).
12d
3(2
3.5(0.9
2.51( 0.02
1.95(0.02
15( 6
7( 3
3.9(0.4
3(2
2.432(0.009
1.86(0.02
12(1
20(7
4.7(0.4
2.492(0.002
1.959(0.005
6.7(0.3
4.6(0.7
cσ2represents the mean

Page 10

6288
dx.doi.org/10.1021/la200053f |Langmuir 2011, 27, 6279–6294
Langmuir
ARTICLE
modelsystemhowever,thecatalystsampleswithanAl/Mratioof3.0
did not require incorporating a backscattering contribution from
M?Al into the model, although a reasonable possibility here is
thatthequality(or,thesignal-to-noise)oftheCoandNiEXAFS
data is insufficient to distinguish a M?Al feature. Furthermore,
thespectrathemselves(Figure6)lackthefeatureobservedinthe
spectra of the Ir model system that “grew in” with successively
greater Al/M ratios. From fitting the data, (iii) the 1NN M?M
coordination numbers observed for Co and Ni samples are, like
those observed in the Ir model system studied previously,14
roughly in the 3?4 range, and could point toward the predomi-
nance of, on average, subnanometer, M∼4?6, metal clusters in
catalyst solutions before hydrogenation.39Alternatively, low
1NN M?M coordination numbers could signify a large degree
of structural disorder in relatively large metal nanoclusters.14,40
The σ2M?Mvalues of the catalyst samples are approximately
twice the experimentally determinedbulk metal values (Tables 2
and 3), which is also suggestive of disordered nanoclusters.
Another possibility is that the metal species in catalyst solutions
exist as some combination of disordered clusters and unreduced
metal ions.
An additional main result from EXAFS is that (iv) the closest
M?M distances, given by 1NN RM?Mvalues, overlap within
experimental error with the corresponding bulk metals for both
CoandNisampleswithAl/Mratiosof1.0,butforAl/M=3.0are
shorter than the bulk metal M?M distances. M?M distances in
nanometer scale metal particles with a bulk-like atomic structure
are expected to be shorter on average than the corresponding
bulkM?MdistancesduetoM?Mbondcontractionrequiredto
counteract (i.e., decrease) the high surface free energy of the
smallmetalclusters.40a?d,41Therefore,theimplicationisthatthe
Co or Ni catalyst materials are becoming structurally more like
nanoscale metal particles with increasing amounts of AlEt3, but
nottothepointthatthe1NNNM?Mvaluesincreasesignificantly
or long-range metallic order becomes apparent in the 3?6 Å
range in the R-space EXAFS spectra (which is also consistent
with the changes in the XANES spectra given above).
Interpretation of the EXAFS results from the Co and Ni
samples must be carried out in light of the Z-contrast STEM,
MALDI MS, and XANES results. For example, the 1NN NM?M
values from EXAFS of roughly 3?4 seem, at first take, to imply,
on average, M∼4?6clusters analogous to the Ir results. However,
theZ-contrast STEMrevealsmeanCoorNiclusterdiametersof
1.4 or 1.3 nm, respectively, that is, M∼130to M∼100clusters.
Therefore, the most plausible explanation for the results from
combining the Z-contrast STEM, MALDI MS, and XAFS (i.e.,
XANES and EXAFS) spectroscopy appears to be that a combina-
tionofnanoclusters(which arestructurallydisorderedresulting in
the absence of peaks at larger distances in the R-space EXAFS
spectra and distorted 1NN NM?Mvalues from fits of the EXAFS
spectra42) and unreduced metal ions are present, with these two
phases of M species both contributing to the mean NM?M
value.40i,43The possibility of monometallic, unreduced metal
ions being present is supported by the control experiments for
Z-contrast STEM in which no Co was observable when only
Co(neodecanoate)2, without AlEt3, was on the sample grid. In
other words, the metal-containing species in Co and Ni catalyst
solutions appear to consist of disordered metal clusters with a
Table 3. Fitting Results from EXAFS Spectroscopic Analysis
of Ni Reference Samples and Ni(2-ethylhexanoate)2plus
AlEt3Catalyst Samples before Hydrogenation
Sample Al/NiNi foilNi(O2CR)2aNi catalyst 1.0 Ni catalyst 3.0
NNi?Ni
NNi?O
RNi?Ni(Å)b
RNi?O(Å)b
σ2Ni?Ni(Å2)c
σ2Ni?O(Å2)c
aNi(O2CR)2is the catalyst precursor Ni(2-ethylhexanoate)2without
added AlEt3. The full analysis of Ni(2-ethylhexanoate)2is given in the
Supporting Information.bR stands for the interatomic distance corre-
sponding to the single scattering paths.cσ2represents the mean square
variationinRduetobothstaticanddynamicdisorder(alsoknownasthe
EXAFSDebye?Wallerfactor),andvaluesshownare?103.dForNifoil,
thisparameter was definedasthevalueshown(i.e.,notvariedinthefit).
12d
3(1
2.8(0.5
2.51( 0.02
2.00(0.02
13( 4
8(3
4.4(0.3
1.2(0.3
2.447(0.006
1.85(0.01
12.4(0.8
14(5
5.8(0.3
2.490(0.003
2.035(0.005
6.9(0.5
7.4(0.7
Figure 7. Example Z-contrast STEM image of a Co(neodecanoate)2
plus AlEt3catalyst sample, Al/Co = 3.0, after its use in hydrogenation.
The histogram shows the results from measuring the diameters of 614
Coclustersinsuchimages;measuredclusterdiametersrangefrom0.5to
2.5 nm, which correspond to Co cluster nuclearities from Co∼6to
Co∼740. The mode, median, and mean diameters of Co clusters are 1.3,
1.4, and 1.4 ( 0.3 nm, corresponding to Co∼100or Co∼130accordingly.

Page 11

6289
dx.doi.org/10.1021/la200053f |Langmuir 2011, 27, 6279–6294
Langmuir
ARTICLE
broaddistributionofsizes,themeandiametersofwhicharegiven
by Z-contrast STEM and MALDI MS, plus some monometallic
complexespresentasunreducedmetalionicspecies.Noteworthy
here is that it is the industrial Co and Ni catalysts, and Al/M
ratios and conditions, that have been examined, to which these
conclusions refer.
Nuclearity of MnSpecies after Hydrogenation: Z-contrast
STEM. The Co(neodecanoate)2plus AlEt3catalyst, with an Al/
Co ratio of 3.0, and after its use for cyclohexene hydrogenation
was imaged using Z-contrast STEM. Measurement of 614
clusters shows a range of Co cluster sizes 0.5?2.5 nm in
diameter. The mode, median, and mean Co cluster diameters
are 1.3, 1.4, and 1.4 ( 0.3 nm, corresponding to Co∼100and
Co∼130, accordingly. Figure 7 shows an example image and the
histogram.
The Ni(2-ethylhexanoate)2plus AlEt3catalyst, with an Al/Ni
ratio of 2.0, after its use for cyclohexene hydrogenation was also
imaged using Z-contrast STEM. Measurement of 650 clusters in
Z-contrast STEM images reveals a range of Ni cluster sizes
0.6?4.0 nm in diameter. The mode and median Ni cluster
diameter is 1.4 nm and the mean is 1.4 ( 0.4 nm. These
diameters correspond to Ni∼130. An example image and the
histogram are shown in Figure 8.
Z-contrast STEM shows that using these Co and Ni Ziegler-
type hydrogenation catalysts for cyclohexene hydrogenation
does not induce a change in the sizes of the metal cluster species
present in either Co or Ni catalyst samples, at least under the
conditions used herein. Although this differs from the distinct
increaseinmetalclustersizeandchangeinstructureexhibitedby
the Ir model system,14it is consistent with the lack of changes in
catalyst solution color, no observation of precipitates in post-
hydrogenationsolutions(unliketheIrmodelsystem14).Inshort,
catalytic cyclohexene hydrogenation induces essentially no
changes in size or size distribution of the Co or Ni clusters
observed by Z-contrast STEM.
Nuclearity of MnSpecies after Hydrogenation: MALDI MS.
Samples of the Co(neodecanoate)2plus AlEt3catalyst, with an
Al/Co ratio of 3.0, were analyzed using MALDI MS after their
use in cyclohexene hydrogenation (figures are shown in the
Supporting Information). MALDI MS of the Co catalyst results
inabroadpeakwithmaximumintensityat∼3500m/z(reported
astheaverageCoclusterinTable1)andashoulderat∼6000m/
z.Usingthesamenecessaryassumptionsasbefore,thatthebroad
peaks are composed of only þ1 charged ions,14,32?35the peak at
∼3500 m/z indicates Co∼60clusters, corresponding to a dia-
meter of ∼1.1 nm. The peak of the posthydrogenation Co
catalyst tails off toward higher m/z values; fwhm of the peak is
from ∼1500?9500 m/z, the peak reaches one-fourth maximum
intensity at ∼12000 m/z, and one-eighth maximum intensity at
∼17000 m/z (1500?17000 is used to report the range of Co
clusters in Table 1), which correspond to 0.8?1.5 nm,
Co∼25?160; 1.6 nm, Co∼200; and 1.8 nm, Co∼290 clusters,
respectively—essentially the same as the prehydrogenation
results.
The Ni(2-ethylhexanoate)2plus AlEt3catalyst, with an Al/Ni
ratioof 2.0, was also analyzed using MALDI MS after it had been
used for cyclohexene hydrogenation, giving a broad peak with a
maximum intensity at ∼2000 m/z, which again indicates Ni∼34
clusters, corresponding to ∼0.9-nm-diameter Ni nanoclusters
(reported as the average cluster size in Table 1). (As in the
catalyst sample before hydrogenation, the presence of Ni atoms
in species below 1500 m/z is ruled out by the absence of
characteristic Ni isotope peak distributions in that region.) The
broad, asymmetrically shaped MALDI MS peak of the catalyst
sample after hydrogenation also tails off toward higher m/z
values, but is not completely identical to the peak of the sample
before hydrogenation; the posthydrogenation peak displays two
slight shoulders at ∼3000 and ∼6000 m/z. Nevertheless, the
broadpeakinthesample afterhydrogenation reaches ∼6500m/
zathalf-maximumintensity,∼8500m/zatone-fourthmaximum
intensity, and ∼11000 m/z at one-eighth maximum intensity
(1500?11000 m/z is used to report the range of Ni clusters in
Table 1), which correspond to 1.3 nm, Ni∼110; 1.5 nm, Ni∼145;
and 1.6 nm, Ni∼190, respectively. These Ni cluster size and
nuclearity values are very similar to those from the prehydro-
genation sample. In short, the MALDI MS-determined sizes and
size distributions of both Co and Ni clusters in posthydrogenation
Figure 8. Example Z-contrast STEMimage ofaNi(2-ethylhexanoate)2
plus AlEt3catalyst sample, Al/Ni = 2.0, after its use in hydrogenation.
The corresponding histogram shows the results from measuring the
diameters of 650 Ni clusters in such images, and reveals a range of Ni
clusters with diameters from 0.6 to 4.0 nm, corresponding to Ni∼10to
Ni∼3060.Themodeandmediandiametersare1.4nm,andthemeanisNi
1.4 ( 0.4 nm, corresponding to mean Ni∼130clusters.

Page 12

6290
dx.doi.org/10.1021/la200053f |Langmuir 2011, 27, 6279–6294
Langmuir
ARTICLE
samples (i) agree closely with the analysis of posthydrogenation
catalyst samples using Z-contrast STEM, are consistent with the
Z-contrastSTEM,and(ii)indicatenosignificantchangeinthesizes
of the metal clusters present upon their use for the catalytic
hydrogenation of cyclohexene.
Nuclearity of MnSpecies after Hydrogenation: XAFS (i.e.,
XANES and EXAFS) Spectroscopy. Solution samples of both
Co(neodecanoate)2plus AlEt3and Ni(2-ethylhexanoate)2plus
AlEt3catalysts, with Al/M ratios of 1.0, were analyzed using
XAFS after their use in hydrogenation reactions. The XANES
spectraoftheCoandNicatalystsolutionsposthydrogenationare
nearly the same as their prehydrogenation counterparts. XANES
spectra collected after hydrogenation are shown and compared
to the prehydrogenation spectra in the Supporting Information
for the interested reader. For both Co and Ni catalysts, the
EXAFS spectra after hydrogenation also appear very similar to
the sample spectra before hydrogenation. The spectra are fit
using the same models employed for fitting the catalyst samples
before hydrogenation. The results are shown in Figure 9 and
summarized in Table 4. Complete fit information and additional
spectra are in the Supporting Information.
The most plausible interpretation of the EXAFS spectra and
fitting results is essentially the same for the catalyst samples after
hydrogenationasforthesamplesbeforehydrogenation.Thelack
ofpeaksinthe3?6Årangecorrespondingtothepeaksobserved
in this range for Co and Ni bulk metals implies that no Co or Ni
specieswithorderedmetallic structureson thatscalearepresent,
and1NNsinglescatteringNM?Mvaluesof∼3wereobtainedfor
both Co and Ni catalysts. Additionally, the RM?Mvalues from
bothCoandNisamplesposthydrogenationarethesameastheir
Figure 9. Data and fits of (a) the Co(neodecanoate)2plus AlEt3
catalyst, Al/Co ratio of 1.0; and (b) the Ni(2-ethylhexanoate)2plus
AlEt3 catalyst, Al/Ni ratio of 1.0; both after use of the catalytic
hydrogenation of cyclohexene. The R-ranges used for the fits of these
samples are 1.0?2.8 Å and 1.0?2.6 Å for Co and Ni, respectively.
Figure 10. Poisoning experiments using the Co(neodecanoate)2or
Ni(2-ethylhexanoate)2plusAlEt3catalysts,withanAl/Coratioof3.0or
an Al/Ni ratio of 2.0, are shown next to standard example cyclohexene
hydrogenation runs for comparison (black curves). (a) For the Co
catalyst, an example poisoning experiment shows immediate and
complete poisoning of catalysis by Hg(0) addition partway through a
cyclohexene hydrogenation run (red curve). This result was reproduci-
ble(3trials).(b)FortheNicatalyst,immediateandcompletepoisoning
of catalysis is observed upon addition of Hg(0) both partway through a
catalytic run (red), and after preparation of the catalyst but before
hydrogenationisbegun(blue).Theseresultssuggestthatcatalysisinthe
industrial Co and Ni Ziegler-type hydrogenation systems is heteroge-
neous (i.e., via the observed Mnnanoclusters, n g 4).
Table 4. Summary of Fit Results for Posthydrogenation Co
and Ni Catalyst Spectra
SampleCoNi
NM?M
NM?O
RM?M(Å)a
RM?O(Å)a
σ2M?M(Å2)b
σ2M?O(Å2)b
aR stands for the interatomic distance corresponding to the single
scattering paths.bσ2represents the mean square variation in R due to
both static and dynamic disorder (the EXAFS Debye?Waller factor),
and values shown are ?103.
3(2
3(1
3(1
2.7(0.4
2.52 (0.01
2.02 (0.01
13(3
7(2
2.48(0.02
1.96(0.02
15( 7
7(4

Page 13

6291
dx.doi.org/10.1021/la200053f |Langmuir 2011, 27, 6279–6294
Langmuir
ARTICLE
prehydrogenation counterparts within experimental error, and
are very close to the experimental bulk metal values (within
e0.03 Å). Recall from the discussion of the prehydrogenation
XAFS results that bulk metal-like RM?Mvalues are in contrast to
the larger RM?Mvalues expected for subnanometer Mnclusters
ligated byLewis acid species (i.e., AlEt3and its derivates).Lastly,
the σ2M?Mvalues of the catalyst samples are again roughly twice
the experimentally determined bulk metal values. Considered in
light of the posthydrogenation Z-contrast and MALDI MS
results, which reveal a predominance of nanometer scale clusters
as part of wide size distributions, the self-consistent interpreta-
tionofallmeasurements(madealreadyfortheprehydrogenation
samples) is that a combination of disordered nanoclusters and
unreduced,monometallicspeciesarepresentincatalystsolutions
posthydrogenation. In short, both the XANES and EXAFS
spectra confirm that use of catalyst solutions for cyclohexene
hydrogenation has a negligible effect on the oxidation state and
form of the transition metal catalyst material.
KineticsStudies:Hg(0)CatalystPoisoning.Theobservation
of Mnclusters before and after catalysis does not necessitate that
these species are the active hydrogenation catalysts—kinetic
studies are required to determine the most active catalyst(s)
from sample solutions. Catalyst poisoning by Hg(0) is a
useful, kinetics-based test for distinguishing homogeneous from
heterogeneous Ziegler-type hydrogenation catalysis, as has been
shown previously.14Hence, Hg(0) poisoning experiments
were utilized to test whether the observed catalytic activity of
the industrial Ziegler-type hydrogenation catalysts made from
Co(neodecanoate)2 or Ni(2-ethylhexanoate)2 and AlEt3 is
“homogeneous” (e.g., via single metal organometallic) or “het-
erogeneous” (e.g., via small M4 or larger nanoclusters)
(Figure 10). One benefit of using Hg(0) poisoning in this case
is that the results should be largely unaffected by MTL kinetics
(vide supra, and in the Supporting Information).
Two different versions of the Hg(0) poisoning experiment
were performed in order to emulate the study of the Ir system.14
In one version, the Hg(0) is added after about half the cyclohex-
enehasbeenconsumed.Intheotherversion,theHg(0)isadded
aftertheinitialcatalystsynthesis,priortothestartofcyclohexene
hydrogenation (i.e., before the catalyst solution is placed under
pressurized H2). When Hg(0) is added to the Co catalyst
solution after about half the cyclohexene has been consumed,
the Hg(0) reproducibly poisons the catalysis immediately and
completely(Figure10a).However,whentheHg(0)isaddedafter
the initial Co catalyst synthesis, eight attempts using ∼1700
equiv of Hg(0) per Co and 24 h of mixing, conducted separately
by two different researchers, show that the extent of poisoning
observed is irreproducible (Supporting Information).
In the case of the nickel catalyst, when Hg(0) is added to the
Ni catalyst solution after about half the cyclohexene has been
consumed, the Hg(0) reproducibly poisons the catalysis imme-
diately and completely (Figure 10b). Unlike the observation with
the Co catalyst, Hg(0) addition to the Ni catalyst prior to the
start of cyclohexene hydrogenation also poisons catalysis im-
mediately, completely, and reproducibly (Figure 10b).
An explanation for the irreproducibility observed when at-
tempting to poison the Co catalyst prior to the start of cyclohex-
ene hydrogenation is not immediately apparent. One possibility
is that the active catalyst is changing rapidly at the start of
cyclohexene hydrogenation from one that is not poisoned by
Hg(0) (i.e., a homogeneous catalyst) to one that is poisoned by
Hg(0) (i.e., a heterogeneous catalyst). However, this is merely
speculation; elucidation of the cause of the irreproducible
poisoning would require additional investigation and an im-
proved understanding of the fundamental interaction between
Hg(0) and the Co speciespresent. It is known thatone potential
difficulty with Hg(0) poisoning experiments is that it may be
difficulttothoroughlycontacttheHg(0)withallofthecatalystin
solution due to the insolubility of Hg(0).44However, for the Ni
system, control experiments allowed the determination that a
procedure using g300 equiv of Hg(0) per Ni and g1.5 h of
1000 rpm stirring is adequate to thoroughly contact the Hg(0)
with all of the Ni catalyst in solution. Nevertheless, and with the
possible exceptions implied by the irreproducible results
using the Co system, the Hg(0) poisoning results ultimately
suggestthatcatalysisintheindustrialZiegler-typehydrogenation
systems, made from Co(neodecanoate)2 or Ni(2-ethylhexa-
noate)2precatalysts plus AlEt3are “heterogeneous”, that is,
catalysis occurs via the observed sub, M∼4?6to larger M∼130
nanoclusters.
Conclusions and Needed Future Studies. Catalysts made
from either of the industrial precursors Co(neodecanoate)2or
Ni(2-ethylhexanoate)2, plus AlEt3, were analyzed by Z-contrast
STEM, MALDI MS, XAFS (i.e., XANES and EXAFS), and
Hg(0) poisoning studies, producing the following, key observa-
tions: (i) Co and Ni Ziegler-type hydrogenation catalyst solu-
tions turn dark brown upon the initial combination of the
Co(neodecanoate)2or Ni(2-ethylhexanoate)2precatalyst solu-
tions with the AlEt3solution, and not during hydrogenation
catalysis as observed with the Ir model system; and (ii) hydro-
genation proceeds immediately with the start of data acquisition
at, or very near, the maximum observable rate. (iii) Z-contrast
STEM reveals, for the prehydrogenation Co sample, a
0.6?3.3 nm range of particle diameters with a mean of 1.4 (
0.4nm,whichcorrespondstoCo∼130.Fortheprehydrogenation
Ni sample, Z-contrast STEM reveals a 0.4?3.5 nm range of
particle diameters with a mean of 1.3 ( 0.5 nm, which corre-
sponds to Ni∼100. (iv) MALDI MS is used to estimate, for the
prehydrogenation Co sample, a 0.8?1.8 nm range of particle
diametersandanaverageof1.2nm,whichcorrespondstoCo∼80.
For the prehydrogenation Ni sample, MALDI MS is used to
estimatea0.8?1.7nmrangeofparticlediametersandanaverage
of0.9nm,whichcorrespondstoNi∼34.(v)XANESspectrashow
that the Co or Ni metals in prehydrogenation catalyst solutions
become progressively less like their M(II) precatalysts, in terms
of composite average formal oxidation state, and progressively
more like the M(0) metal foils as the Al/M ratios increase from
1.0 to 3.0, implying that unreduced metal ions are present in
catalystsolutions inamounts thatdecreasewithadditionalAlEt3.
(vi) EXAFS spectroscopic analysis of prehydrogenation samples
reveals a lack of the R-space peaks in the 3?6 Å range indicating
thelackoforderedmetallicstructures. Fittingthespectraofboth
metals using composite models analogous to that used for the Ir
model system14gives mean 1NN M?M coordination numbers
in the 3?4 range. Fitting the EXAFS spectra also gives 1NN
RM?Mvalues that overlap, within experimental error, with the
corresponding bulk metals for both Co and Ni samples with Al/
M ratios of 1.0, but 1NN RM?Mvalues that are shorter than the
bulkmetalM?MdistancesforbothCoandNiAl/M=3.0samples,
consistent with greater M?M bonding in these Al/M = 3.0
samples.FittingtheEXAFSspectraalsorevealσ2M?Mvaluesthat
are approximately twice the experimentally determined bulk
metal values, indicative of disordered metal clusters. In addition,
(vii) the Z-contrast STEM, MALDI MS, and XAFS results all

Page 14

6292
dx.doi.org/10.1021/la200053f |Langmuir 2011, 27, 6279–6294
Langmuir
ARTICLE
show that cyclohexene hydrogenation does not significantly
change the speciation of the catalyst solutions. Finally, (viii)
Hg(0) poisons both Co and Ni catalysts immediately and
completely, when Hg(0) is added in the middle of a hydrogena-
tion run. When Hg(0) is added prior to the start of hydrogena-
tion, the degree of poisoning of the Co catalyst is irreproducible,
but the Ni catalyst is poisoned immediately, completely, and
reproducibly.
The self-consistent interpretation of all results from the
complementary techniques used herein is that the transition
metal components of catalysts made from either of the industrial
precursors Co(neodecanoate)2or Ni(2-ethylhexanoate)2, plus
AlEt3, consist of a combination of Mnclusters with abroad range
of sizes and a large degree of structural disorder, plus unreduced,
monometallic species, the distribution between the two phases
dependingontheAl/Mratio.Furthermore,theHg(0)poisoning
results suggest that Ziegler nanoclusters are the most active
catalysts in the industrial Ziegler-type hydrogenation catalyst
systems (i.e., that the catalysis is heterogeneous, and if one
includesMg4withinthedefinitionofheterogeneous).Thiswork
expands on the results of others—notably the important studies
by Schmidt and co-workers,17and B€ onnemann and co-
workers18—which show transition metal nanoclusters are pre-
sent in the Co, Pd, Ni, and Pt Ziegler-type systems they studied.
The combined results present the best evidence to date that the
“Ziegler nanocluster hypothesis” is the correct answer to the
∼50-year-old problem of what is the true nature of the industrial
Ni-,andpresumablyalsoCo-basedcatalysts.Restated,thenotion
that industrial Ziegler-type hydrogenation catalysis proceeds via
Ziegler nanoclusters is the leading hypothesis going forward to
try to disprove.
Much remains to be done, however. Operando spectroscopy
studies of the formation of, and catalysis by, both the Ni and Co
industrial catalyst systems remain to be accomplished.45A full
kinetic study and rate law determination under non-MTL
conditions also remain to be done, and it promises to be
challenging due to the high rates of these superior catalysts. In
addition, the differences regarding the backscattering contribu-
tion from M?Al between the EXAFS spectra of the Ir model
system (which show the presence of Al)14and those of the
industrial Co and Ni-based catalysts studied herein (which do
not show the presence of Al) are surprising and remain to be
explored—is this simply an artifact of the signal-to-noise of the
current EXAFS data, or could a M4H4type catalyst for M = Co
and Ni explain this discrepancy, forexample?Another important
difference between the Ir and Co, Ni catalysts is that catalyst
aging slows the rates for the Co, Ni catalysts, opposite what is
seen for Ir, so that future studies characterizing the aged Co and
Ni catalysts is another, important future objective. Furthermore,
specificdeterminationoftheform(s)takenandrole(s)playedby
the AlEt3component, both in the initial synthesis of the catalyst
and during catalytic cyclohexene hydrogenation, remain to be
fully understood.19
Despite the work remaining to be done, this investigation of
the homogeneous versus heterogeneous nature of Ziegler-type
hydrogenation catalysts is significant for at least four reasons:
(i)thisstudyexaminesCo-andNi-basedcatalystsmadefromthe
actual industrial precursor materials, which make catalysts that
arenotoriouslyproblematic tocharacterize;2,3(ii)theZ-contrast
STEM results reported herein represent, to our knowledge,3the
bestmicroscopicanalysisoftheindustrialCoandNiZiegler-type
hydrogenation catalysts; (iii) this study is the first explicit
application of an established method, using multiple analytical
methodsandkinetics-basedstudies,fordistinguishinghomogeneous
from heterogeneous catalysis;3,6?15and (iv) this study parallels the
successful study of an Ir model Ziegler catalyst system, thereby
benefiting from a comparison to those previously unavailable
findings,14although the greater M?M bond energy, and tendency
toagglomerate,ofIrversusCoorNiareimportantdifferencestobe
noted.46Overall, the main result of this work is that it provides the
leading hypothesis going forward to try to refute in future work:
namely, that sub, Mg4to larger, MnZiegler nanoclusters are the
dominant, industrial, Co- and Ni- plus AlR3catalysts in Ziegler-type
hydrogenation systems.
’ASSOCIATED CONTENT
b
S
SupportingInformation.
results of control experiments for cyclohexene hydrogenations
used to help establish standard conditions for catalyst prepara-
tion and use; additional TEM images; figures showing the
MALDI MS results; EXAFS spectra with fits; Hg(0) poisoning
control experiments; and a full list of the authors of ref 18d. This
materialisavailablefreeofchargeviatheInternetathttp://pubs.
acs.org.
Experimentalinformationand
’AUTHOR INFORMATION
Corresponding Author
*E-mail: rfinke@lamar.colostate.edu.
’ACKNOWLEDGMENT
Work was supported by NSF Grant CHE-0611588 at Color-
ado State University. MALDI MS were obtained with the expert
assistance of Phil Ryan (now deceased) of the Macromolecular
ResourcesLabatCSU.AIF,JCYandRGNacknowledgesupport
by DOE BES Grant DE-FG02-03ER15476. We thank Nebojsa
Marinkovic and Syed Khalid at the NSLS, the use of which was
supported by the U.S. Department of Energy, Office of Science,
Office of Basic Energy Sciences, under Contract No. DE-AC02-
98CH10886.BeamlineX18BattheNSLSissupportedinpartby
the Synchrotron Catalysis Consortium, U.S. Department of
Energy Grant No DE-FG02-05ER15688. We thank JoAn Hud-
son of Clemson University for the additional TEM imaging.
’REFERENCES
(1) Alley, W. M.; Girard, C. W.;€ Ozkar, S.; Finke, R. G. Inorg. Chem.
2009, 48, 1114–1121.
(2) Johnson, K. A. Polym. Prepr. 2000, 41, 1525–1526.
(3) Alley,W.M.;Hamdemir,I.K.;Johnson,K.A.;Finke,R.G.J.Mol.
Catal. A: Chem. 2010, 315, 1–27.
(4) Collman, J. P.; Hegedus, L. S.; Norton, J. R.; Finke, R. G.
Principles and Applications of Organotransition Metal Chemistry; Univer-
sity Science Books: Mill Valley, CA, 1987.
(5) Schwartz, J. Acc. Chem. Res. 1985, 18, 302–308.
(6) Widegren, J. A.; Finke, R. G. J. Mol. Catal. A: Chem. 2003,
198, 317–341.
(7) Lin, Y.; Finke, R. G. Inorg. Chem. 1994, 33, 4891–4910.
(8) Aiken,J.D.,III;Lin,Y.;Finke,R.G.J.Mol.Catal.A:Chem.1996,
114, 29–51.
(9) Widegren, J. A.; Bennett, M. A.; Finke, R. G. J. Am. Chem. Soc.
2003, 125, 10301–10310.
(10) Hagen,C.M.;Widegren,J.A.;Maitlis,P.M.;Finke,R.G.J.Am.
Chem. Soc. 2005, 127, 4423–4432.

Page 15

6293
dx.doi.org/10.1021/la200053f |Langmuir 2011, 27, 6279–6294
Langmuir
ARTICLE
(11) Finney, E. E.; Finke, R. G. Inorg. Chim. Acta 2006, 359, 2879–
2887.
(12) Jaska, C. A.; Manners, I. J. Am. Chem. Soc. 2004, 126, 1334–
1335.
(13) Jaska, C. A.; Manners, I. J. Am. Chem. Soc. 2004, 126, 9776–
9785.
(14) Alley, W. M.; Hamdemir, I. K.; Wang, Q.; Frenkel, A.; Li, L.;
Yang,J.C.;Menard,L.D.;Nuzzo,R.G.;€ Ozkar,S.;Johnson,K.A.;Finke,
R. G. Inorg. Chem. 2010, 49, 8131–3147.
(15) Zahmakiran, M.;€ Ozkar, S. Inorg. Chem. 2009, 48, 8955–8964.
(16) Halpern, J. Inorg. Chim. Acta 1981, 50, 11–19.
(17) (a)Shmidt,F.K.;Nindakova,L.O.;Shainyan,B.A.;Saraev,V.V.;
Chipanina, N. N.; Umanetz, V. A. J. Mol. Catal. A: Chem. 2005,
235, 161–172. (b) Belykh, L. B.; Titova, Yu. Yu.; Umanets, V. A.; Shmidt,
F. K. Russian Journal of Applied Chemistry 2006, 79, 1271–1277.
(c)Nindakova,L.O.;Shmidt,F.K.;Saraev,V.V.;Shainyan,B.A.;Chipanina,
N. N.; Umanets, V. A.; Belonogova, L. N.; Toryashinova, D.-S. D. Kinetics
andCatalysis.2006, 47, 54–63.(d)Belykh,L. B.;Goremyka,T.V.;Skripov,
N.I.;Umanets,V.A.;Shmidt,F.K.KineticsandCatalysis.2006,47,367–374.
(e)Schmidt,F.K.;Ratovskii,G.V.;Dimitrieva,T.V.;Ivleva,I.N.;Borodko,
Y. G. J. Organomet. Chem. 1983, 256, 309–329.
(18) Studies by B€ onnemann and co-workers do not focus on
hydrogenation catalysis using Ziegler-type systems, but do demonstrate
the synthesis and identification of nanoclusters from Ni or Pt Ziegler-
type systems.(a) B€ onnemann, H.; Brijoux, W.; Brinkmann, R.;
Endruschat,U.;Hofstadt,W.;Angermund,K.RevueRoumainedeChimie
1999, 44, 1003–1010. (b) B€ onnemann, H.; Wald€ ofner, N.; Haubold,
H.-G.; Vad, T. Chem. Mater. 2002, 14, 1115–1120. (c) Angermund, K.;
B€ uhl, M.; Dinjus, E.; Endruschat, U.; Gassner, F.; Haubold, H.-G.;
Hormes,J.;K€ ohl,G.;Mauschick,F.T.;Modrow,H.;M€ ortel,R.;Mynott,
R.; Tesche, B.; Vad, T.; Wald€ ofner, N.; B€ onnemann, H. Angew. Chem.,
Int.Ed.2002,41,4041–4044.(d) Angermund, K.;etal. J.Phys.Chem.B
2003, 107, 7507–7515. (e) Haubold, H.-G.; Vad, T.; Wald€ ofner, N.;
B€ onnemann, H. J. Appl. Crystallogr. 2003, 36, 617–620. (f) Wen, F.;
B€ onnemann, H.; Mynott, R. J.; Spliethoff, B.; Weidenthaler, C.; Palina,
N.; Zinoveva, S.; Modrow, H. Appl. Organomet. Chem. 2005,
19, 827–829.
(19) Hamdemir, I. K.; € Ozkar, S.; Johnson, K. A.; Finke, R. G.,
manuscript in preparation.
(20) Aiken, J. D., III; Finke, R. G. J. Am. Chem. Soc. 1998,
120, 9545–9554.
(21) (a) Lin, Y.; Finke, R. G. Inorg. Chem. 1994, 33, 4891–4910. (b)
Watzky, M. A.; Finke, R. G. J. Am. Chem. Soc. 1997, 119, 10382–10400.
and(c) Widegren, J. A.; Aiken, J. D., III;€ Ozkar, S.; Finke, R. G. Chem.
Mater. 2001, 13, 312–324.
(22) Kearns, G. J.; Foster, E. W.; Hutchison, J. E. Anal. Chem. 2006,
78, 298–303.
(23) (a) Newville, M. J. Synchrotron Rad. 2001, 8, 322–324. (b)
Ravel, B.; Newville, M. J. Synchrotron Rad. 2005, 12, 537–541.
(24) Lin, Y.; Finke, R. G. J. Am. Chem. Soc. 1994, 116, 8335–8353.
(25) CRC Handbook of Chemistry and Physics; 77th ed., Lide, D. R.,
Frederikse, H. P. R., Eds.; CRC Press: Boca Raton, 1996.
(26) Thenumberofatomsinatransitionmetalnanocluster(n)may
be estimated making the necessary approximation that the nanoclusters
have the same close-packed atomic structure as the bulk metal (either
face centered cubic (fcc) or hexagonal close packed (hcp)), and using
the following formula: n = (N0F(4/3)π(D/2)3)/W.24According to this
approach, N0is 6.022 ? 1023mol?1, F is the room temperature density
of the pure bulk metal, D is the measured cluster diameter, and W is the
atomic weight of the transition metal. For Ni, F is 8.90 g/cm3, and W is
58.69 g/mol.25For Co, F is 8.86 g/cm3, and W is 58.93 g/mol.25For
example, a Ni cluster diameter of 1.3 nm corresponds to Ni∼100.
(27) StarkeyOtt,L.;Cline,M.L.;Deetlefs,M.;Seddon,K.R.;Finke,
R. G. J. Am. Chem. Soc. 2005, 127, 5758–5759.
(28) Hagen,C.M.;Vieille-Petit,L.;Laurenczy,G.;S€ uss-Fink;Finke,
R. G. Organometallics 2005, 24, 1819–1831.
(29) Williams, D. B.; Carter, C. B. Transmission Electron Microscopy;
Plenum Press: New York, 1996.
(30) Catalyst samples, both before and after hydrogenation, were
also analyzed using bright field and high resolutuion (HR)TEM as
control experiments. Images and a short discussion of those results can
be found in the Supporting Information.
(31) Pyrz, W. D.; Buttrey, D. J. Langmuir 2008, 24, 11350–11360.
(32) Whetten, R. L.; Khoury, J. T.; Alvarez, M. M.; Murthy, S.;
Vezmar, I.; Wang, Z. L.; Stephens, P. W.; Cleveland, C. L.; Luedtke,
W. D.; Landman, U. Adv. Mater. 1996, 8, 428.
(33) Khitrov, G. A.; Strouse, G. F. J. Am. Chem. Soc. 2003,
125, 10465.
(34) Kuzuya, T.; Tai, Y.; Yamamuro, S.; Sumiyama, K. Chem. Phys.
Lett. 2005, 407, 460–463.
(35) Maya, L.; Chen, C. H.; Stevenson, K. A.; Kenik, E. A.; Allman,
S. L.; Thundat, T. G. J. Nanoparticle Res. 2002, 4, 417.
(36) Goulon, J.; Georges, E.; Goulon-Ginet, C.; Chauvin, Y.;
Commereuc, D.; Dexpert, H.; Freund, E. Chem. Phys. 1984, 83, 357–366.
(b) Esselin, C.; Bauer-Grosse, E.; Goulon, J.; Williams, C.; Chauvin, Y.;
Commereuc, D.; Freund, E. J. Phys. Colloq. 1986, 47, C8–243?C8?248.
(37) Frenkel, A. I.; Hills, C. W.; Nuzzo, R. G. J. Phys. Chem. B 2001,
105, 12689–12703.
(38) SincemuchoftheotherdatapresentedhereinisfortheAl/Ni=
2.0 catalyst, we carefully considered if it was important to have higher
quality data for the Al/Ni = 2.0 sample in order to support the
conclusions in the present study. We reasoned that such data are not
necessary, primarily on the basis of the following four reasons. First, the
XANES spectra of these samples show a smooth progression of the
samples from the Al/Ni = 1.0 to Al/Ni = 3.0 ratios. Second, there is
considerable similarity between the Al/Ni = 2.0 and 3.0 spectra when
plotted asχ(k),anddespitetheadditionalnoiseapparentinthe Al/Ni=
2.0 sample (relevant spectra comparisons are given in the Supporting
Information). Third, these first two observations, together with the
similarity between the fitting results from catalyst samples with Al/Ni
ratios of 1.0 and 3.0, demonstrate that reliable fitting results for the
Al/Ni=2.0sampleshouldhavevaluesbetweenthoseobtainedfromthe
Al/Ni = 1.0 and 3.0 samples. And fourth, the interpretation of the XAFS
data herein, that 1NN M?M coordination numbers are not representa-
tive of cluster nuclearity, means that a direct comparison between the
results from XAFS and cluster size measurements using other methods
such as Z-contrast STEM is not straightforward; this is the case even if
higher quality data for the Al/Ni = 2.0 sample were in hand. In short,
obtaininghigherqualitydatafortheAl/Ni=2.0sampleisnotexpectedto
improve or change any of the conclusions reached in the present study.
(39) One possible interpretation is that M∼4?6 clusters remain
tightly associated (i.e., (M∼4?6)n). For some precedent for this, see:
(a) Fulton, J. L.; Linehan, J. C.; Autrey, T.; Balasubramanian, M.; Chen,
Y.; Szymczak, N. K. J. Am. Chem. Soc. 2007, 129, 11936–11949. (b)
Harada,M;Asakura,K.;Toshima,N.J.Phys.Chem.1994,98,2653–2662.
(40) (a) Ankudinov, A. L.; Rehr, J. J.; Low, J. J.; Bare, S. R. J. Chem.
Phys. 2002, 116, 1911–1919. (b) Petkov, V.; Ohta, T.; Hou, Y.; Ren, Y.
J.Phys.Chem.C2007,111,714–720.(c)Petkov,V.;Bedford,N.;Knecht,
M.R.;Weir,M.G.;Crooks,R.M.;Tang,W.;Henkelman,G.;Frenkel,A.
J. Phys. Chem. C 2008, 112, 8907–8911. (d) Sun, Y.; Zhuang, L.; Lu, J.;
Hong,X.;Liu,P.J.Am.Chem.Soc.2007,129,15465–15467.(e)Vila,F.;
Rehr, J. J.; Kas, J.; Nuzzo, R. G.; Frenkel, A. I. Phys. Rev. B 2008,
78, 121404–1?121404?4. (f) Sanchez, S. I.; Menard, L. D.; Bram, A.;
Kang, J. H.; Small, M. W.; Nuzzo, R. G.; Frenkel, A. I. J. Am. Chem. Soc.
2009, 131, 7040–7054. (g) Gilbert, B.; Huang, F.; Zhang, H.; Waychu-
nas,G.A.;Banfield,J.F.Science2004,305,651–654.(h)Billinge,S.J.L.;
Levin, I. Science 2007, 316, 561–565. (i) Sun, Y.; Frenkel, A. I.; Isseroff,
R.; Shonbrun, C.; Forman, M.; Shin, K.; Koga, T.; White, H.; Zhang, L.;
Zhu, Y.; Rafailovich, M. H.; Sokolov, J. C. Langmuir 2006, 22, 807–816.
(j) Menard, L. D.; Xu, H.; Gao, S.-P.; Twesten, R. D.; Harper, A. S.;
Song, Y.; Wang, G.; Douglas, A. D.; Yang, J. C.; Frenkel, A. I.; Murray,
R. W.; Nuzzo, R. G. J. Phys. Chem. B 2006, 110, 14564–14573.
(41) Finke, R. G.;€ Ozkar, S. Coord. Chem. Rev. 2004, 248, 135–146.
(42) Yevick, A.; Frenkel, A. I. Phys. Rev. B 2010, 81, 115451.
(43) In addition, the explanation of the low XAFS NM?Mvalues
involving disordered structure alone requires a degree of NM?M