You are on page 1of 8

Viewpoint

Cite This: ACS Catal. 2018, 8, 1520−1527 pubs.acs.org/acscatalysis

Direct Synthesis of H2O2 from H2 and O2 on Pd Catalysts: Current


Understanding, Outstanding Questions, and Research Needs
David W. Flaherty*
Department of Chemical and Biomolecular Engineering, University of Illinois, Urbana−Champaign, Urbana, Illinois 61801, United
States
1. MOTIVATION AND CHALLENGES FOR DIRECT 2. SELECTIVITY IN THE DIRECT SYNTHESIS REACTION
SYNTHESIS OF H2O2 The direct synthesis reaction involves reactions among
Catalytic reactions that utilize H2O2 as a terminal oxidant intermediates derived from H2 and O2 on transition metal
See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.

possess high selectivities for industrially relevant oxidations that nanoparticles in a liquid solvent at low temperatures (273−313
rival Cl2 mediated chemistry and reactions with organic K). A series of elementary steps appear to add H atoms
peroxides. H2O2 does not present, however, the same risks to sequentially to O2 to form H2O2 while avoiding the production
environmental and human health as those more commonly of H2O by irreversible O−O bond rupture in any intermediate
used oxidants.1 The direct synthesis of hydrogen peroxide which is followed by further reduction. Success requires
Downloaded via 86.19.122.92 on December 2, 2023 at 23:04:01 (UTC).

(reaction 1) over transition metal nanoparticles offers an kinetically trapping unstable surface intermediates such as
environmentally attractive method to enable a wide array of OO* and OOH* as well as H2O2 to prevent formation of
impactful selective oxidation reactions including epoxidations of thermodynamically preferred intermediates (O*, OH*) that
alkenes, oxidations of thiols, thiophenes, and sulfides, and lead to H2O. The primary challenge, therefore, is suppressing
conversion of alkanes to alcohols. O−O bond rupture on metal nanoparticles that must activate
H2 under conditions where O2 dissociation seems facile (∼300
H 2 + O2 → H 2O2 (1) K, 0.5−10 MPa H2, 0.5−10 MPa O2).
Data from nearly five decades of academic research show that
The appeal of the direct synthesis reaction stems from the H2O2 rates and selectivities on Pd depend strongly on a
potential of direct synthesis to create inexpensive H2O2 at number of factors. The fact that H2 and O2 pressures affect
midscale, geographically distributed facilities, because direct reaction rates4−6 is not unexpected. Figure 1 portrays several
synthesis facilities have much lower capital and operating costs
than comparably sized Riedl−Pfleiderer plants (the auto-
oxidation of anthraquinones, the incumbent technology).1,2
Direct synthesis has not emerged as a practical technology
due to several technological and scientific barriers. Process
designs reveal a few barriers, such as process safety. For
example, direct synthesis requires that both H2 and O2 gas are
present in a single reactor at pressures sufficiently high to
achieve acceptable rates. Yet these conditions require also the
use of significant amounts of diluent gases (e.g., N2, CO2) to
prevent the formation of explosive mixtures.1,3 The use of
diluted reactant streams decreases the H2O2 productivity per
unit volume of reactor for a given total pressure and also
introduces the need for a sizable recycle stream to recover the Figure 1. Direct synthesis of H2O2 on Pd proceeds in complex
diluent gas and minimize compression costs.3 Perhaps more mixtures of aqueous solvents that contain mineral acids, halide salts,
importantly, scientific challenges and questions remain that alcohols, and their reaction byproducts. The reasons why these
relate to the design of stable catalytic systems that convert H2 additives affect H2O2 formation are not well understood.
to H2O2 selectively. Specifically, how and why H2O2 formation
rates and selectivities on Pd nanoparticle catalysts depend on other attributes of the catalytic system that are, perhaps,
reactant pressures and the solvents (water, alcohols) and surprising to researchers not familiar with this chemistry. These
promoters (halide salts and acids) used are not understood. include the choice of solvent (e.g., water, methanol, or ethanol),
Further, the solvents and promoters used to achieve significant the presence of trace amounts of halide salts (e.g., ∼ 5 ppm
selectivities on Pd catalysts reduce their stability and useful NaBr), the addition of mineral acids (e.g., HCl, H2SO4), and
lifetime, which presents one of the greatest challenges for direct the identity of the diluent gas (e.g., N2 or CO2).1,4 These
synthesis. This Viewpoint introduces and describes current complex combinations of halide salts, acids, and alcohol
understanding of this deceptively complex chemistry as well as solvents are required to achieve even modest H2O2 selectivities
concepts important for developing selective, productive, and (>60%) on supported Pd nanoparticles (the most selective and
stable Pd catalysts. Outstanding questions related to this
promising catalytic reaction and needs for future investigations Received: November 30, 2017
are described. Published: January 24, 2018

© 2018 American Chemical Society 1520 DOI: 10.1021/acscatal.7b04107


ACS Catal. 2018, 8, 1520−1527
ACS Catalysis Viewpoint

frequently studied monometallic catalyst). Slight differences in intermediates. Perhaps surprisingly, the way in which O2
these attributes of the catalytic systems significantly change reduces to form H2O2 is still unclear.
H2O2 production rates and selectivities (several comparisons 3.1. H2O2 Formation in Water. Within purely aqueous
are provided in Table 2 of ref 1), and in some cases, these solvents (i.e., in the absence of alcohols), most studies measure
changes suggest that these species participate directly in the H2O2 formation in acidic solutions of either HCl or H2SO4 that
reaction. also contain NaBr or NaCl, because H2O2 will not form in
Efforts to increase H2O2 selectivities fall largely into two measurable quantities otherwise.5,6,25,26 Under these conditions,
categories; catalyst design and reaction engineering. One Lunsford reported that H2O2 formation increase in proportion
challenge is that the stability of O2* and OOH* surface to both H2 and O2 pressures at lower pressures and become
species that may be reduced to H2O2 are strongly correlated independent of reactant pressures at greater values (water
with those of reactive intermediates that ultimately form H2O formation rates were not reported).22 Complementary experi-
(O*, OH*) on extended metal surfaces.7−11 Therefore, a ments demonstrated rates did not differ between perhydro-
common principle of catalyst design for direct synthesis focuses genated and perdeuterated combinations of aqueous solvents,
on the formation of bimetallic catalysts (e.g., PdAu,12,13 PdSn,14 acids, and hydrogen.6 These data were interpreted as evidence
PdZn15) or the use of promoters and additives (e.g., Cl−, that the reaction rates were transport limited or that O2 binding
Br−)16−18 that may change or break scaling relationships among was rate determining,6 although these explanations seem
O-containing intermediates. Both bimetallics and promoted Pd inconsistent with the measured rate dependencies. The
catalysts give greater H2O2 selectivities perhaps by preventing addition of HCl or H2SO4 to aqueous solvents (or neat
the formation of binding configurations thermodynamically ethanol) increases rates significantly but the manner by which
preferred on uniform metal surfaces (η2-O2 on Pd(111)) in these proton donors promoted H2O2 formation was not clear in
favor of those that involve less electron exchange (η1-O2 on initial reports.6,22,27 Contemporaneous work by Stahl et al.
PdxAu(111)).19−21 Reaction engineering approaches attempt to illustrated that H2O2 forms upon reduction of bathocuproine
capitalize on differences between mechanisms and coreactants Pd−O2 complexes28,29 and inspired Lunsford to propose that a
for the kinetically relevant steps and transition states that form similar catalytic cycle may form H2O2 during direct syn-
H2O2 (e.g., OOH* + H* → H2O2* + *) and those that form thesis.6,30
H2O (e.g., OOH* + * → OH* + O*). Among other factors, Figure 2 shows steps that corrosively oxidize Pd to form
changes in H2 pressure, the identity of the solvent, and the soluble Pd-peroxo complexes that are protonated by HCl and
addition of mineral acids to the liquid phase affect H2O2
selectivities. Several plausible explanations have been proposed
for these effects within individual studies, but direct evidence to
support or disprove many hypotheses is difficult to obtain.
Aside from the potential technological impact of selective
and stable catalysts for the direct synthesis of H2O2, this
chemistry is an appealing probe reaction to develop greater
understanding (and quantitative descriptions) of the manner by
which distinct aspects of catalytic systems (e.g., solvent, surface
promoters, catalyst composition, etc.) dictate chemistry at the
liquid−solid interface. Many opportunities exist for exper- Figure 2. Proposed catalytic cycle for H2O2 formation by Pd within
imental and particularly computational investigations of this acidified water. O2 binds to soluble PdII complexes and reduces by
proton electron transfer as described by Lunsford et al.6
reaction, because, we understand much less about the direct
synthesis of H2O2 than many other reactions of small
molecules. Future studies need to embrace the complexity of subsequently reduced by H2 gas to reform metallic Pd and HCl.
this system and include the critical “additives” that permit H2O2 Importantly, this work hypothesized that proton−electron
to form in order to account for the complete set of species and transfer pathways were responsible for H2O2 formation on Pd
interactions required for this chemistry. Such work would help sites under nominally thermocatalytic conditions.6,30 Kinetic
to build a conceptual framework that describes this fascinating dependencies on reactant pressures similar to these early
chemistry. To build such a framework, we need first to studies (rH2O2 ∼ [H2]0−1[O2]0−1) were recently reported from
understand the mechanism for H2O2 formation. experiments in continuous flow microreactors that also used
water as a solvent (5−10 ppm NaBr, 0.05−0.5 M H2SO4).5,26,31
3. PROPOSED MECHANISMS FOR H2O2 FORMATION However, these data were interpreted as evidence for a two-site
surface reaction in which H*-atoms and O2* adsorb non-
ON PD
competitively and react on the surface to reduce O2 to H2O2.5
Detailed discussions of the mechanism for direct synthesis of Neither of these mechanisms contains quantitative descrip-
H2O2 date to seminal publications by Lunsford and co-workers tions that show protons or acids are coreactants or cocatalysts
who reported several key observations for H2O2 production for H2O2 formation, rather, the given interpretations state that
over Pd-SiO2 catalysts from reactions in semibatch reac- these species participate indirectly. Protons are frequently
tors.6,22−24 Analysis of the oxygen isotope labels in H2O2 described as modifying the electronic structure of Pd.32
produced from reactant mixtures of 18O2, 16O2, and H2 Similarly, halides are denoted as site blockers, despite the fact
demonstrated that O−O bonds cleave irreversibly and always that even trace quantities of H2O2 do not form in aqueous
lead to H2O formation.22 Consequently, achieving high solvents in their absence.32 These conclusions seem speculative
selectivities to H2O2 requires catalysts and active sites that and specific to aqueous systems. As described next, rate
bind O2* but lack e− of sufficient energy to populate the 2π* dependencies and kinetic behavior for H2O2 formation differ
orbitals and cleave the O−O bond in O2*, OOH*, or other significantly within alcohol solvents and also in aprotic solvents.
1521 DOI: 10.1021/acscatal.7b04107
ACS Catal. 2018, 8, 1520−1527
ACS Catalysis Viewpoint

These differences suggest that the solvent participates directly oxidation of H2 (i.e., H2 → 2H+ + 2e−) and shuttle via oxonium
within the catalytic cycle as either a coreactant or a cocatalyst. species during consecutive proton−electron transfer steps that
Moreover, H2O2 can form even without halide “site blockers.” reduce O2 to H2O2. This mechanism accounts for the need for
3.2. H2O2 Formation in Non-Aqueous Solvents. Direct H2 as a reductant (i.e., H2O2 did not form at detectable rates in
synthesis may become an economical method to produce H2O2 methanol without external H2) and provides a means to
for epoxidation or oxidation catalysis over transition metal maintain the solvent pH and the electronic charge on Pd
oxides.1 Water can significantly decrease rates in these systems, nanoparticles.4 Pd-based bimetallic nanoparticles also appear to
and therefore, H2O2 may be produced in alcohol solvents that directly involve alcohol molecules in the formation of H2O2.
are compatible with these chemistries.33 The significant Bimetallic AuPd nanoparticle catalysts exhibit similar depend-
differences in H2O2 formation rates and selectivities between encies of rates on H2 and O2 pressures and on the protic nature
aqueous and alcohol solvents are frequently attributed to of the solvent, even though H2O2 selectivities are nearly 6-fold
greater solubility of H2 in alcohols compared to water (vide greater on the most Au-rich nanoparticles.37 These differences
inf ra).34−36 A few specific observations suggest, however, that reflect changes in the electronic structure of the Pd active sites
the mechanism for H2O2 in methanol and ethanol may differ that increase activation enthalpies for O−O bond dissociation
fundamentally from the mechanism within water. and the formation of H2O, consistent with theoretical
Lunsford first reported that H2O2 formation rates increase in predictions for close-packed surfaces.19,38−40 Kinetic data and
proportion to H2 pressure and do not depend on O2 pressure solvent requirements for H2O2 formation on intermetallic β-
(rH2O2 ∼ [H2]1[O2]0) within ethanol (0.12 M H2SO4),23 but PdZn nanoparticles15 also match the mechanism shown in
the report did not propose a molecular mechanism to explain Figure 3.
these results. These results were quickly followed by reports for Despite these recent findings, the molecular mechanism
the effects of H2SO4, Cl−, and CH3COOH on H2O2 rates responsible for H2O2 formation is by no means resolved. The
within ethanol that were interpreted as evidence for high results reviewed here for both aqueous and nonaqueous
coverages of acetate (formed by ethanol oxidation) and Cl− on solvents (Sections 3.1and 3.2) are all consistent with proton−
Pd surfaces that block Pd ensembles needed to dissociate O−O electron transfer mechanisms. Based upon current findings,
bonds (vide inf ra).32 Again, halides and protons derived from such heterolytic reaction mechanisms seem most likely to be
mineral salts and acids were described as critical promoters for responsible for the direct synthesis of H2O2. However, the
H2O2 that modify the reactivity of the Pd surface but are not identity of the kinetically relevant step, the most abundant
involved directly. surface intermediate, and the details of how the protic solvent
Our group recently reported mechanistic analysis of steady (or the derived surface fragments) shuttle protons and
state H2O2 and H2O formation rates measured as functions of electrons to O2 may differ. Additional experimental and
reactant pressures along with complementary experiments computational investigations are needed to develop detailed
probing the influence of the solvent properties.4 In pure understanding of the steps that combine H atoms with O2 and
methanol (i.e., without water, NaBr or H2 SO 4 ), H 2 O 2 the ways in which halides, acids, and solvent molecules
formation rates increase with H2 pressure but do not depend participate. In situ spectroscopy and increasingly complex
on O2 pressure (rH2O2 ∼ [H2]1/2−1[O2]0; similar to reports in molecular simulations should be integrated into future studies
ethanol23).4 H2O formation rates depend more weakly on H2 to probe other ways in which solvent molecules may be
and are also independent of O2 (rH2O ∼ [H2]0−1/2[O2]0).4 involved.
These observations are inconsistent with competitive adsorp-
tion of H2 and O2 on a single active site. Independent 4. SOLVENTS HAVE MULTIPLE ROLES FOR H2O2
experiments demonstrated that H2O2 forms only in the FORMATION
presence of protic solvents (methanol, water) and was The solvent used for the direct synthesis of H2O2 has a
undetectable in aprotic solvents (acetonitrile, dimethyl significant effect on the rate of the reaction and the selectivity
sulfoxide, and propylene carbonate).4 Together these observa- to H2O2. Solvents such as water, methanol, and ethanol may
tions provide strong evidence that protic solvents cocatalyze influence H2O2 yields by facilitating transport of the reactants
H2O2 formation. to the active sites and by chemically binding to and modifying
Figure 3 depicts one method by which water or methanol exposed metal atoms. These effects are shown by comparisons
may participate directly in H2O2 formation by introducing low of catalytic rates performance to the solubility of O2 and H2 in
barrier pathways for O2* reduction. Protons form by heterolytic pure solvents and multiple component solutions.29,34,35 Several
reports suggest that the adsorption of organic molecules (e.g.,
CH3COO−) to catalyst surfaces increases H2O2 selectivities in
several reports.23,32,41 Solvent molecules seem likely to
influence H2O2 formation and decomposition pathways by
less frequently discussed methods as well. H2O2 only forms
within protic solvents,4,15,37 which strongly suggests that
solvents participate directly in proton or hydride transfer
processes. In addition, solvent molecules may stabilize critical
intermediates via intermolecular interactions such as hydrogen-
bonding.
Undoubtedly, the use of appropriate solvents to facilitate
Figure 3. Catalytically coupled heterolytic hydrogen oxidation and two mass transport is important for implementing direct synthesis
electron oxygen reduction reactions may form H2O2 on Pd chemistry, particularly in industrial reactors that would operate
nanoparticles. Adapted with permission from ref 4. Copyright 2016 near the limit of mass transfer rates. Yet, carefully designed
American Chemical Society. laboratory experiments can eliminate interphase and inter- and
1522 DOI: 10.1021/acscatal.7b04107
ACS Catal. 2018, 8, 1520−1527
ACS Catalysis Viewpoint

intraparticle concentration gradients such that measured H2O2 quantities of H2O2.4,15,49 Further, alcohols may provide greater
and H2O formation rates reflect only the kinetics of the rates and selectivities than water by acting as coreagents for
catalytic reaction.42,43 Under these conditions, solvents still catalytic transfer hydrogenation of O2 by hydride-proton
change formation rates of H2O2 and H2O, which shows that the transfer steps akin to the Meerwein-Ponndorf-Verley mecha-
solvent influences this catalytic reaction in ways that are not nism for reduction of carbonyl groups.33 Fluid-phase solvent
related to transport processes or reactant solubility alone. (or adsorbed fragments) molecules can also influence catalysis
The potential ability to use solvent molecules or other by stabilizing kinetically relevant transition states via
species (e.g., halides, acids) to chemically modify the surface of intermolecular interactions such as hydrogen-bonding or by
nanoparticle catalysts or to stabilize transition states directly are solvating highly charged structures (e.g., those involved in
intriguing. Lunsford and co-workers suggested that ethanol and proton−electron transfer processes).50,51 Classical models for
acetic acid (added directly or produced in situ) may be these effects in catalysis are described in texts51,52 and can be
responsible for greater H2O2 selectivities in ethanol in quantified in the framework of thermodynamic activity
comparison to water.23,32 The formation of chelating acetate coefficients for solvent-species pairs, which can include
species32 may inhibit O−O bond rupture by disrupting transition states.
ensembles of Pd atoms required to bind η2-O2, η1-OOH, and
the transition states for their reduction or dissociation, similar 5. HALIDES AND ACIDS AFFECT H2O2 RATES AND
to proposals for bimetallic catalysts with improved H2O2 SELECTIVITIES
selectivities.38,44−46 Similarly, Nijhuis et al. demonstrated that The vast majority of publications that report the direct
H2O2 selectivities increase by a factor of two on Pd and AuPd synthesis of H2O2 in water or in alcohol solvents also include
nanoparticles with the addition of small amounts (≥2 vol %) of catalytically significant quantities of halide salts (e.g., NaBr) or
acetonitrile to an aqueous solvent (containing 0.05 M H2SO4, 9 inorganic acids (e.g., HCl, H2SO4). These additives are often
ppm NaBr), which was attributed to acetonitrile reducing the used with little discussion of their role in catalysis, which
coverage of O2 at Pd ensembles responsible for O−O bond suggests that their use is based on phenomenological
rupture.34 This concept could explain also one role of alcohol observations that they increase H2O2 selectivities. Pospelova
solvents (methanol), surface ligands (e.g., long-chain ammo- et al. first demonstrated that H2O2 formed in immeasurable
nium dihydrogen phosphates depicted in Figure 4),47,48 and quantities on pure supported Pd catalysts in water, but the
addition of millimolar quantities of acids and halides increased
H2O2 yields to nearly 60%.53 More recently, H2O2 selectivities
above 90% and nearing 100% were achieved through more
complex combinations of H2SO4, HCl, and Br− promoters.
Several publications have probed how these species influence
H2O2 formation and decomposition pathways, but few
investigations probe the intrinsic way in which these species
influence surface catalysis.
The additions of halides, and more specifically Cl− and Br−,
to solvents or to catalyst supports increase H2O2 selectivities
significantly. Adsorption and activation of O2, H2, and other
Figure 4. Long-chain alkyl ammonium phosphates coadsorbed on Pd surface intermediates are undoubtedly influenced by the
increase H2O2 selectivities and are predicted to increase barriers for presence of halide coadsorbates, and these species are
O−O bond rupture during the direct synthesis of H2O2. In general, frequently described as blocking specific sites responsible for
dense adlayers of solvent molecules or intentional ligands can modify cleaving O−O bonds.6,16−18,24,53 However, halides are likely to
binding modes of reactive intermediates and electron exchange with influence the density of states of greater numbers of metal
surfaces and can subsequently influence H2O2 selectivities. Repro- atoms in their vicinity and may confer greater selectivity for
duced with permission from ref 47. Copyright 2017 Wiley-VCH. H2O2 formation to multiple sites by limiting the extent of e−
back-donation to 2π* orbitals of O2 via through-surface
perhaps diluent CO2,13,14 which forms carbonic acid in situ.4 All interactions.54 This explanation is consistent with greater
of these species can form adlayers of chelating species that work functions for Pd(111)55 and weaker O*-atoms binding
reduce the number of accessible metal sites and hinder energies on Pt(111)56 and Au(111)57 in the presence of Cl*-
adsorption modes of reactive species such that O−O bonds adatoms. However, increasing concentrations of halides oxidize
are more difficult to cleave. In situ vibrational spectroscopy Pd to form soluble Pd2+-complexes (e.g., PdCl42−),49 which
would be useful to determine if the proposed carboxylate and leads to greater rates of metal dissolution from the support and
ligand adlayers exist during catalysis and to correlate the reduces the longevity (and utility) of the catalyst. More
coverage of these species to rates and activation energies for commonly, Br− is used to promote H2O2 formation on Pd,
H2O2 and H2O formation. These comparisons would directly because Br− leads to comparably high H2O2 selectivities yet
test how these adlayers influence the two reaction pathways. does not leach Pd from the catalyst as rapidly.27 The changes in
Apparent activation energies, selectivities, and rates for H2O2 selectivities, H2O2 formation rates, and Pd leaching rates do not
and H2O formation depend on the solvent likely because depend significantly on the precursor used to add Br− to the
solvent molecules participate directly in the catalytic cycle and solvent for the reaction58 but are sensitive to the concentration
assist by solvating reactive intermediates or transition states of acids used as copromoters.59
through intermolecular interactions. Methanol, water, and their The role of acids in H2O2 formation is more difficult to
protonated forms cocatalyze proton−electron transfer steps by understand, because their dissociation leads to two inter-
providing low barrier pathways to shuttle protons, which can mediates that may each modify surface reactivity. The anions
explain the need for protic solvents to form detectable derived from acids commonly used (e.g., HBr, HCl, and
1523 DOI: 10.1021/acscatal.7b04107
ACS Catal. 2018, 8, 1520−1527
ACS Catalysis Viewpoint

H2SO4) may influence catalysis by their adsorption and indirectly. PdO nanoparticles supported on silica do not form
interaction with nanoparticle surfaces. For example, halide measurable amounts of H2O2 in semibatch reactors (5 kPa H2,
anions may coordinate to the surface of Pd nanoparticles and 75 kPa O2, 283 K, ethanol, 2 h).61 However, H2O2 forms
either inhibit reaction at Pd ensembles or electronically modify readily within the same system after sparging with 20 kPa H2
Pd sites by withdrawing charge (vide supra).54−56 In addition, (in N2, 1 h) and reintroducing the H2 and O2 coreactants.61
acids have been proposed to inhibit the decomposition of H2O2 Han et al. proposed that Pd-PdO sites located at the interface
by preventing its deprotonation to more reactive forms (e.g., with a TiO2 support were active sites for H2O2 formation based
OOH−) that bind strongly to metal surfaces and decompose.27 upon comparisons of measured H2O2 selectivities to ex situ
Thus, the addition of acids may shift the position of equilibria XAS and XPS assessments of Pd oxidation states.62 Our group
related to these deprotonated and more reactive forms. Finally, has observed that PdO nanoparticles exhibit significant
the changes in the concentrations of acids (and also halide induction times (∼0.5 h) prior to forming H2O2, whereas, Pd
salts) in a solvent will influence the ionic strength of the nanoparticles form H2O2 immediately (9 kPa H2, 5 kPa O2, 298
solution and affect the thermodynamic activity of charged K, methanol).63 Cumulatively, these observations indicate that
intermediates and transition states within the catalytic cycle.51 at least a fraction of the nanoparticles must exist as reduced Pd
Such changes may preferentially stabilize pathways for H2O2 in order to form H2O2, but it is not clear if the active sites that
formation over those for H2O. These potential effects have not bind and selectively reduce O2 to H2O2 are reduced, partially,
been deconvoluted in previous experimental investigations, but or fully oxidized Pd atoms.62
these points may be addressed with computational studies. A few publication have applied in situ X-ray absorption
spectroscopy (XAS) to probe the state of Pd during
6. THE STATE OF THE ACTIVE PD CATALYST IS catalysis.41,64 Results from Centomo et al. suggested that Pd
UNKNOWN nanoparticles on a resin support contained a significant fraction
Despite ongoing work to determine the mechanism(s) by of Pd0 (60%) in pure methanol, which increased slightly upon
which H2O2 forms, it is clear that molecular O2 binds to the adding the reactant stream (H2 and O2 in CO2).41 On the other
active site and reduces to form H2O2 without O−O bond hand, more recent results reported by Hensen and Schouten
rupture.22 Pristine Pd surfaces and nanoparticles readily indicated that Pd nanoparticles existed largely as PdH during
dissociate O2 at cryogenic temperatures, which suggests that catalysis in water.64 These differences likely reflect differences
active sites for direct synthesis differ significantly from these in reactant pressures and solvents used. Systematic comparisons
more reactive surfaces. Pd nanoparticles can form PdO and of H2O2 formation rates and selectivities to the state of the Pd
PdH phases when exposed to O2 and H2 respectively, which nanoparticles (determined by in situ XAS or Raman) across a
coexist at significant pressures during H2O2 formation. These wide-range of O2 and H2 pressures would be useful, because
observations and the effects of ligand and halide additives such data may shed some light on the state of Pd responsible
indicate that the active state(s) of Pd responsible for selective for selective H2O2 formation.
formation of H2O2 is not a clean, metallic surface. Rather, H2O2
likely forms either on surfaces of metallic Pd nanoparticles 7. RECOMMENDATIONS FOR FUTURE RESEARCH
covered by halides or reactive intermediates (e.g., O*, OH*), Many fundamental aspects of the direct synthesis of H2O2
surfaces of PdO or PdH nanoparticles, or on soluble Pd- remain uncertain, but the great potential of this chemistry to
complexes formed by dissolution of nanoparticles. enable selective and environmentally benign oxidations
Lunsford first showed that rates of H2O2 formation in provides considerable motivation to continue investigating
semibatch reactors did not correlate to the amount of Pd this complex reaction. The comparisons of the mechanistic
supported on SiO2 supports in the presence of HCl.49 The Pd studies highlight several questions that should be addressed in
nanoparticles dissolved into vividly colored Pd2+ complexes future studies, and which will need contributions from both
(e.g., PdCl42−, observed in many batch reactor experiments) theoretical and experimental methods to unravel the more
and redistributed as nanometer sized Pd colloids throughout intricate aspects of H2O2 formation. Specific points for further
the solvent and across the walls of the reactor (steps shown in investigation include: the mechanism for O2 reduction to H2O2
Figure 2).24,49 These Pd colloids were proposed to be the active and the potentially direct participation of solvent molecules in
catalysts responsible for the majority (>97%) of H2O 2 this process; the interaction between coadsorbates (i.e., halides,
produced, because, significant mass transfer limitations limited solvent molecules, and reactive intermediates) that influence
the productivity of the Pd-SiO2 catalyst particles (i.e., the reactions of dioxygen intermediates; and chemical or
effectiveness factors (η) ≪ 1).24,49 While this work suggests electronic characterization of the active sites that produce
that Pd colloids form H2O2, it is not clear if the experiments H2O2.
performed could rule out the possibility that soluble Pd- 7.1. Need for Computational Work with Complete
complexes are also active sites for H2O2 formation. We are not Models. Computational studies of H2O2 formation have the
aware of further studies that have provided evidence for or potential to provide atomistic insight that eludes even the most
against the idea that liquid-phase Pd2+ species contribute to carefully designed experiments. To date, density functional
measured H2O2 formation rates during direct synthesis. theory (DFT) calculations have been used to probe the
However, barriers for O2 dissociation on these and other elementary steps that determine selectivity and differences
mononuclear Pd complexes28,60 are greater than on pristine Pd between Pd, PdH and more selective Pd-based bimetallic
surfaces,19,21 and the populations of such complexes increase catalysts (e.g., PdAu).8,38−40,47,65 Investigations of AuPd
significantly with the addition of halide and acid promoters. bimetallic surfaces demonstrated that intrinsic activation
The surfaces of metal nanoparticles are most commonly energies for O−O bonds in O 2 * or OOH* surface
considered to expose the active sites for H2O2 formation. Yet intermediates increase systematically with the number of Au
knowledge of the chemical state and structure of these atoms within the close-packed (111) surfaces,19,38,39 whereas,
nanoparticles during direct synthesis is often obtained the activation energies for hydrogenation pathways (e.g., O2* +
1524 DOI: 10.1021/acscatal.7b04107
ACS Catal. 2018, 8, 1520−1527
ACS Catalysis Viewpoint

H* → OOH* + H*) increase to a somewhat smaller Vibrational spectroscopy and kinetic measurements can be
extent.38,39 These trends reproduce the greater selectivities combined to determine if coverages of such coadsorbates
toward H2O2 on AuPd materials.12,13,49,66 Analysis of the correlate with activation enthalpies or selectivities for H2O2
densities of states upon Pd atoms within these bimetallic selectivities, the identity of these coadsorbates, and to inform
surfaces shows that the addition of Au leads to systematic shifts computational studies that probe adsorbate−adsorbate inter-
of the d-band to lower energies (presumably due to actions.
rehybridization of Pd atomic orbitals as a result of changes in In situ XAS or Raman can be used to determine the
strain)38,39 and a concurrent reduction in the extent of electron electronic and structural states of Pd nanoparticles (and
back-donation to 2π* orbitals of O2*. particularly their surfaces) under relevant reaction conditions.
Notably, most computational investigations into direct Pd nanoparticles may become oxidized by reaction with O2 and
synthesis of H2O2 utilize periodic boundary conditions for form regions of PdO62 or mixed Pd oxides14 that increase H2O2
two-dimensional unit cells (e.g., 3 × 3) and treat low coverages selectivities by possessing greater barriers for O−O dissocia-
of reactants (<0.25 monolayers) but do not incorporate solvent tion.71,72 Particularly, comparisons of how Pd−Pd coordination
molecules (explicitly or implicitly) or interactions between and mean oxidation states of Pd change between conditions
coadsorbates, such as halide promotors or other reactive that give low to undetectable H2O2 selectivities (e.g., modest
intermediates. As described above, H2O2 rates are immeasur- H2 pressures and aqueous solvents)6 to those that give high
able on most catalysts (and particularly on Pd) in the absence selectivities (e.g., high H2 pressures, alcohols solvents, and
of alcohol solvents or halide and acid promoters. Therefore, we halides or acid promoters)18 would provide valuable insight.
must carefully consider the meaning of predictions derived Such measurements would help identify subtle changes in the
from models that do not contain these species. These aspects electronic state of surface Pd atoms or more dramatic phase
are regularly treated in quantum chemical studies of the changes (e.g., formations of hydrides or oxides) that depend on
electrocatalytic oxygen reduction reaction (ORR) by two the concentrations of fluid phase species known to influence
electron (2e−) and four electron (4e−) pathways,44−46,67,68 H2O2 formation rates and selectivities.
which may have more similarities to direct synthesis of H2O2 We can still learn a significant amount about mechanistic
than initially expected.4,15,49 The computational expense to aspects of the direct synthesis reaction and the roles of the
implement models that include all the species crucial for H2O2 many additives with accessible experimental methods (e.g., rate
formation will be greater, but the understanding gained from measurements), but the measurements will need to strictly
these studies would be valuable. Such work will need to be reflect chemical kinetics (and not mass transfer)42 and the data
performed in close cooperation with experiments that provide analysis must be quantitative and model-based.4,5,31,37 More-
kinetic parameters (e.g., activation enthalpies) and measure over, researchers need to report fundamental and transferable
physical attributes of the catalyst surface (e.g., adsorbate kinetic quantities (e.g., turnover rates, activation enthalpies)
coverages, oxidation states) that can be used to develop and that provide direct comparison to model calculations.73
validate these models. Within such studies, it would be useful to probe how the
7.2. Areas for Further Experimental Study. Many apparent dependence of rates on H2 and O2 pressures change
previous investigations of the direct synthesis of H2O 2 with the identity of the solvents, such as aqueous and alcohol
compared H2O2 formation rates, yields and selectivities across solvents, and the presence of halides. Disparities between
different reaction conditions or between different forms of Pd observed rate dependencies in water,5,6,31 methanol,4,37 and
catalysts to learn a great deal. However, further progress will ethanol,23 suggest that certain aspects of the mechanism for
require application of experimental methods less commonly H 2 O 2 formation may change along with the solvent.
used in this topic (e.g., in situ spectroscopy) and more Comparisons between data and rate expressions need to be
quantitative analysis of intrinsically meaningful kinetic variables. built upon derivations from systems of elementary steps.
In situ vibrational spectroscopy could provide vital Measurements of both isotope scrambling and kinetic isotope
information regarding the identity and relative coverages of effects were useful in early mechanistic studies of H2O2
reactive species present on nanoparticles during catalysis. Such formation in water.6,22 Yet, isotope methods have been
measurements will, however, challenge researchers, because underutilized recently despite their ability to directly probe
liquid solvents and high gas pressures require combinations of the kinetic relevance and reversibility of specific steps and to
specialized spectroscopy methods (e.g., attenuated total implicate particular reactive species. For example, unpublished
reflectance, Raman, sum-frequency generation)69 and exper- data from our lab show that H2O2 formation rates differ
imental protocols (e.g., modulation excitation)70 to differentiate between H2 and D2 reactants but also between labeled alcohol
reactive surface intermediates from spectating surface species solvents (e.g., CH3OH, CH3OD, and CD3OH), which suggests
and the solvent. These experiments could directly test multiple that solvent molecules may participate in O2 reduction
hypotheses related to this chemistry. First, the proposal that directly.63
active sites are saturated with O2* and OOH* species during 7.3. Need for Benchmark Catalytic Measurements.
direct synthesis in alcohol solvents (suggested by rH2O2 ∼ Experimental studies should provide benchmark measurements
[O2]0)4,23,37 should be tested with vibrational spectroscopy, of H2O2 formation rates and selectivities at common conditions
because this suggestion conflicts with the facile dissociation of (and upon a standard catalyst).73 A survey of the literature on
O2 on pristine Pd nanoparticles and Pd(111).19−21 Use of direct synthesis shows the difficulty in making direct
infrared or Raman spectroscopy may detect vibrational features comparisons between publications and transferring knowledge
related to O2*-species (e.g., peroxides or superoxides) that from one study to the next. These difficulties are caused by the
would provide evidence for these intermediates. Second, Pd variability in methods for catalytic evaluation among different
catalysts that form H2O2 selectively may support dense adlayers groups and catalytically significant (yet quantitatively small)
of halides6,18 or solvent molecules (and their decomposition differences between composition and concentrations of the
products, such as acetate)32,34 that impede O−O bond rupture. additives and solvents used. Researchers in direct synthesis of
1525 DOI: 10.1021/acscatal.7b04107
ACS Catal. 2018, 8, 1520−1527
ACS Catalysis Viewpoint

H2O2 should agree upon the use of a standard catalyst and


solvent composition for benchmark measurements to be
■ ACKNOWLEDGMENTS
DWF acknowledges helpful discussions with Neil M. Wilson,
included in new investigations. Control experiments in our Pranjali Priyadarshini, Jason S. Adams, and Matthew Neurock.
own laboratory demonstrate that intraparticle diffusion of Financial support for this work was provided by a research
reactant gases and product H2O2 influence catalytic measure- grant from the National Science Foundation (CBET-
ments obtained on Pd nanoparticle catalysts with loadings 15531377) and the Energy Biosciences Institute funded by
greater than 0.05% wt on silica particles (0.25−0.5 mm in Royal Dutch Shell.


diameter). Inadvertently, we have learned that exposure of
stainless steel reactors to halide salts or mineral acid promoters REFERENCES
irreversibly corrupts all catalytic measurements obtained within (1) Wilson, N. M.; Bregante, D. T.; Priyadarshini, P. In Catalysis;
the same reactor for a period of months, presumably due to Spivey, J., Han, Y.-F., Eds.; Royal Society of Chemistry: 2017; Vol. 29,
corrosion of metal from the reactor walls. These same pp 122−212.
phenomena may affect also catalytic data from other groups. (2) Goor, G.; Glenneberg, J.; Jacobi, S. Hydrogen Peroxide; Wiley-
Within our laboratory, we now measure H2O2 formation VCH: Weinheim, Germany, 2007; Vol.18, pp 393−427.
rates (0.02 (mol H2O2) (mol PdS s)−1), selectivities (20%), and (3) García-Serna, J.; Moreno, T.; Biasi, P.; Cocero, M. J.; Mikkola, J.-
activation enthalpies (−8 ± 3 kJ mol−1) on Pd-SiO2 catalysts P.; Salmi, T. O. Green Chem. 2014, 16, 2320−2343.
(<1 nm Pd nanoparticles; 0.05% wt Pd on Davisil 646, 60 kPa (4) Wilson, N. M.; Flaherty, D. W. J. Am. Chem. Soc. 2016, 138, 574−
586.
H2, 60 kPa O2, 277 K) within pure methanol to validate new
(5) Paunovic, V.; Ordomsky, V.; D’Angelo, M. F. N.; Schouten, J. C.;
reactors, benchmark catalysts, and train researchers (see Nijhuis, T. A. J. Catal. 2014, 309, 325−332.
citation for details of catalyst synthesis and testing).4,37 High (6) Chinta, S.; Lunsford, J. H. J. Catal. 2004, 225, 249−255.
purity methanol, without halides or acids promoters, appears to (7) Rankin, R. B.; Greeley, J. ACS Catal. 2012, 2, 2664−2672.
be the most suitable solvent for these benchmark measure- (8) Ford, D. C.; Nilekar, A. U.; Xu, Y.; Mavrikakis, M. Surf. Sci. 2010,
ments, because it does not introduce possible contamination 604, 1565−1575.
and H2O2−methanol solutions are industrially relevant for (9) Greeley, J.; Stephens, I. E. L.; Bondarenko, A. S.; Johansson, T.
applications such as alkene epoxidation (e.g., in the hydrogen P.; Hansen, H. A.; Jaramillo, T. F.; Rossmeisl, J.; Chorkendorff, I.;
peroxide−propylene oxide process).24 Norskov, J. K. Nat. Chem. 2009, 1, 552−556.
(10) Norskov, J. K.; Rossmeisl, J.; Logadottir, A.; Lindqvist, L.;
Kitchin, J. R.; Bligaard, T.; Jonsson, H. J. Phys. Chem. B 2004, 108,
8. SUMMARY 17886−17892.
Successful implementation of the direct synthesis of H2O2 (and (11) Viswanathan, V.; Hansen, H. A.; Rossmeisl, J.; Norskov, J. K.
ACS Catal. 2012, 2, 1654−1660.
the resultant inexpensive H2O2) would decrease the environ- (12) Edwards, J. K.; Hutchings, G. J. Angew. Chem., Int. Ed. 2008, 47,
mental impact of industrial oxidation reactions and could 9192−9198.
enable a number of ground-breaking processes (e.g., selective (13) Edwards, J. K.; Solsona, B.; Ntainjua N, E.; Carley, A. F.; Herzig,
oxidations of alkanes to alcohols). The greatest barriers that A. A.; Kiely, C. J.; Hutchings, G. Science 2009, 323, 1037−1041.
stand in the way of implementation stem from outstanding (14) Freakley, S. J.; He, Q.; Harrhy, J. H.; Lu, L.; Crole, D. A.;
scientific challenges. Namely, the combinations of solvents and Morgan, D. J.; Ntainjua, E. N.; Edwards, J. K.; Carley, A. F.; Borisevich,
promoters that provide high selectivities to H2O2 on Pd A. Y.; Kiely, C. J.; Hutchings, G. J. Science 2016, 351, 965−968.
catalysts (and Pd-based bimetallic catalysts) also decrease the (15) Wilson, N. M.; Schroder, J.; Bregante, D. T.; Kunz, S.; Flaherty,
D. W. to be submitted.
stability of those catalysts by increasing rates of Pd dissolution
(16) Ntainjua N., E.; Piccinini, M.; Pritchard, J. C.; Edwards, J. K.;
and leaching. Current understanding of this presumably simple Carley, A. F.; Moulijn, J. A.; Hutchings, G. J. ChemSusChem 2009, 2,
reaction between small molecules (H2 and O2) on nanoparticle 575−580.
surfaces is surprisingly limited. Consequently, the pathway to (17) Gallina, G.; Garcia-Serna, J.; Salmi, T. O.; Canu, P.; Biasi, P. Ind.
develop new combinations of solvents and promoters that may Eng. Chem. Res. 2017, 56, 13367−13378.
provide both selective and stable catalysts is uncertain. Further (18) Liu, Q.; Lunsford, J. H. J. Catal. 2006, 239, 237−243.
progress in understanding this deceptively intricate chemistry (19) Yu, W.-Y.; Zhang, L.; Mullen, G. M.; Henkelman, G.; Mullins,
will require rigorous studies that combine kinetic, spectro- C. B. J. Phys. Chem. C 2015, 119, 11754−11762.
scopic, and computational methods and which develop (20) Imbihl, R.; Demuth, J. E. Surf. Sci. 1986, 173, 395−410.
(21) Nolan, P. D.; Lutz, B. R.; Tanaka, P. L.; Mullins, C. B. Surf. Sci.
quantitative descriptions of how the many components of the 1998, 419, L107−L113.
system influence intrinsic kinetic parameters and reaction (22) Dissanayake, D. P.; Lunsford, J. H. J. Catal. 2003, 214, 113−120.
mechanisms. To do so, researchers will need to embrace the (23) Liu, Q.; Lunsford, J. H. Appl. Catal., A 2006, 314, 94−100.
complexity of this reaction and consider that Pd nanoparticles (24) Bassler, P.; Weidenbach, M.; Goebbel, H. Chem. Eng. Trans.
are only one part of the complete catalytic system that produces 2010, 21, 571−576.
H2O2 selectively. (25) Voloshin, Y.; Lawal, A. Appl. Catal., A 2009, 353, 9−16.


(26) Voloshin, Y.; Lawal, A. Chem. Eng. Sci. 2010, 65, 1028−1036.
(27) Choudhary, V. R.; Samanta, C. J. Catal. 2006, 238, 28−38.
AUTHOR INFORMATION (28) Stahl, S. S.; Thorman, J. L.; Nelson, R. C.; Kozee, M. A. J. Am.
Corresponding Author Chem. Soc. 2001, 123, 7188−7189.
(29) Steinhoff, B. A.; Fix, S. R.; Stahl, S. S. J. Am. Chem. Soc. 2002,
*Phone: (217) 244-2816. E-mail: dwflhrty@illinois.edu.
124, 766−767.
ORCID (30) Lunsford, J. H. J. Catal. 2003, 216, 455−460.
David W. Flaherty: 0000-0002-0567-8481 (31) Voloshin, Y.; Halder, R.; Lawal, A. Catal. Today 2007, 125, 40−
47.
Notes (32) Han, Y.-F.; Lunsford, J. H. J. Catal. 2005, 230, 313−316.
The author declares no competing financial interest. (33) Gilkey, M. J.; Xu, B. ACS Catal. 2016, 6, 1420−1436.

1526 DOI: 10.1021/acscatal.7b04107


ACS Catal. 2018, 8, 1520−1527
ACS Catalysis Viewpoint

(34) Paunovic, V.; Ordomsky, V. V.; Sushkevich, V. L.; Schouten, J. (70) Urakawa, A.; Burgi, T.; Baiker, A. Chem. Eng. Sci. 2008, 63,
C.; Nijhuis, T. A. ChemCatChem 2015, 7, 1161−1176. 4902−4909.
(35) Burch, R.; Ellis, P. R. Appl. Catal., B 2003, 42, 203−211. (71) Weaver, J. F. Chem. Rev. 2013, 113, 4164−4215.
(36) Crole, D. A.; Freakley, S. J.; Edwards, J. K.; Hutchings, G. J. Proc. (72) Hinojosa, J. A., Jr.; Kan, H. H.; Weaver, J. F. J. Phys. Chem. C
R. Soc. London, Ser. A 2016, 472, 20160156. 2008, 112, 8324−8331.
(37) Wilson, N. M.; Priyadarshini, P.; Kunz, S.; Flaherty, D. W. J. (73) Bligaard, T.; Bullock, R. M.; Campbell, C. T.; Chen, J.-G.; Gates,
Catal. 2018, 357, 163−175. B. C.; Gorte, R. J.; Jones, C. W.; Jones, W. D.; Kitchin, J. R.; Scott, S.
(38) Ham, H. C.; Hwang, G. S.; Han, J.; Nam, S. W.; Lim, T. H. J. L. ACS Catal. 2016, 6, 2590−2602.
Phys. Chem. C 2009, 113, 12943−12945.
(39) Ham, H. C.; Stephens, J. A.; Hwang, G. S.; Han, J.; Nam, S. W.;
Lim, T. H. Catal. Today 2011, 165, 138−144.
(40) Todorovic, R.; Meyer, R. J. Catal. Today 2011, 160, 242−248.
(41) Centomo, P.; Meneghini, C.; Sterchele, S.; Trapananti, A.;
Aquilanti, G.; Zecca, M. Catal. Today 2015, 248, 138−141.
(42) Madon, R. J.; Boudart, M. Ind. Eng. Chem. Fundam. 1982, 21,
438−447.
(43) Weisz, P. B.; Prater, C. D. Adv. Catal. 1954, 6, 143−195.
(44) Siahrostami, S.; Verdaguer-Casadevall, A.; Karamad, M.; Deiana,
D.; Malacrida, P.; Wickman, B.; Escudero-Escribano, M.; Paoli, E. A.;
Frydendal, R.; Hansen, T. W.; Chorkendorff, I.; Stephens, I. E. L.;
Rossmeisl, J. Nat. Mater. 2013, 12, 1137−1143.
(45) Stephens, I. E. L.; Bondarenko, A. S.; Grønjberg, U.; Rossmeisl,
J.; Chorkendorff, I. Energy Environ. Sci. 2012, 5, 6744−6762.
(46) Verdaguer-Casadevall, A.; Deiana, D.; Karamad, M.;
Siahrostami, S.; Malacrida, P.; Hansen, T. W.; Rossmeisl, J.;
Chorkendorff, I.; Stephens, I. E. L. Nano Lett. 2014, 14, 1603−1608.
(47) Lari, G. M.; Puertolas, B.; Shahrokhi, M.; Lopez, N.; Perez-
Ramirez, J. Angew. Chem., Int. Ed. 2017, 56, 1775−1779.
(48) Schoenbaum, C. A.; Schwartz, D. K.; Medlin, J. W. Acc. Chem.
Res. 2014, 47, 1438−1445.
(49) Dissanayake, D. P.; Lunsford, J. H. J. Catal. 2002, 206, 173−176.
(50) Gounder, R. Catal. Sci. Technol. 2014, 4, 2877−2886.
(51) Anslyn, E. V.; Dougherty, D. A. Modern Physical Organic
Chemistry; University Science Books: Sausalito, CA, 2006.
(52) Boudart, M.; Djega-Mariadassou, G. Kinetics of Heterogeneous
Catalytic Reactions; Princeton University Press: Princeton, NJ, 1984.
(53) Pospelova, T. A.; Kobozev, N. I.; Eremin, E. N. Russ. J. Phys.
Chem. 1961, 35, 143−147.
(54) Marshall, S. T.; Medlin, J. W. Surf. Sci. Rep. 2011, 66, 173−184.
(55) Erley, W. Surf. Sci. 1980, 94, 281−292.
(56) Hohenegger, M.; Bechtold, E.; Schennach, R. Surf. Sci. 1998,
412−413, 184−191.
(57) Gao, W.; Zhou, L.; Pinnaduwage, D. S.; Friend, C. M. J. Phys.
Chem. C 2007, 111, 9005−9007.
(58) Samanta, C.; Choudhary, V. R. Appl. Catal., A 2007, 330, 23−
32.
(59) Gallina, G.; García-Serna, J.; Salmi, T. O.; Canu, P.; Biasi, P. Ind.
Eng. Chem. Res. 2017, 56, 13367−13378.
(60) Popp, B. V.; Wendlandt, J. E.; Landis, C. R.; Stahl, S. S. Angew.
Chem., Int. Ed. 2007, 46, 601−604.
(61) Liu, Q.; Gath, K. K.; Bauer, J. C.; Schaak, R. E.; Lunsford, J. H.
Catal. Lett. 2009, 132, 342−348.
(62) Ouyang, L.; Tian, P.-F.; Da, G.-J.; Xu, X.-C.; Ao, C.; Rui Si, T.-
Y.; Xu, J.; Han, Y.-F. J. Catal. 2015, 321, 70−80.
(63) Adams, J. S.; Flaherty, D. W. Unpublished.
(64) D’Angelo, M. F. N.; Kanungo, S.; van Haandel, L.; Ordomiskiy,
V.; Hensen, E. J. M.; Schouten, J. C. In 25th North American Catalysis
Society Meeting, Denver, CO, 2017.
(65) Plauck, A.; Strangland, E. E.; Dumesic, J. A.; Mavrikakis, M.
Proc. Natl. Acad. Sci. U. S. A. 2016, 113, E1973−E1982.
(66) Edwards, J. K.; Thomas, A.; Carley, A. F.; Herzing, A. A.; Kiely,
C. J.; Hutchings, G. J. Green Chem. 2008, 10, 388−394.
(67) Keith, J. A.; Jacob, T. Angew. Chem., Int. Ed. 2010, 49, 9521−
9525.
(68) Keith, J. A.; Jacob, T. In Theory and Experiment in
Electrocatalysis; Balbuena, P., Subramanian, V. R., Eds; Springer-
Verlag: New York, 2010; Vol. 50, pp 89−132.
(69) Niemanstverdriet, J. W. Spectroscopy in Catalysis: An
Introduction, 3rd ed.; Wiley-VCH: Weinheim, 2007.

1527 DOI: 10.1021/acscatal.7b04107


ACS Catal. 2018, 8, 1520−1527

You might also like