Insights into Nitrate Reduction over Indium-Decorated Palladium Nanoparticle Catalysts

Article abstract of DOI:10.1021/acscatal.7b01371

Nitrate (NO₃^-) is an ubiquitous groundwater contaminant and is detrimental to human health. Bimetallic palladium-based catalysts have been found to be promising for treating nitrate (and nitrite, NO₂^-) contaminated waters. Those containing indium (In) are unusually active, but the mechanistic explanation for catalyst performance remains largely unproven. We report that In deposited on Pd nanoparticles (NPs) ( In-on-Pd NPs ) shows room-temperature nitrate catalytic reduction activity that varies with volcano-shape dependence on In surface coverage. The most active catalyst had an In surface coverage of 40%, with a pseudo-first order normalized rate constant of k_cat7.6 L g_{surface-metal}^-1 min^-1, whereas monometallic Pd NPs and In₂O₃ have nondetectible activity for nitrate reduction. X-ray absorption spectroscopy (XAS) results indicated that In is in oxidized form in the as-synthesized catalyst; it reduces to zerovalent metal in the presence of H₂ and reoxidizes following NO₃^- contact. Selectivity in excess of 95% to nontoxic N₂ was observed for all the catalysts. Density functional theory (DFT) simulations suggest that submonolayer coverage amounts of metallic In provide strong binding sites for nitrate adsorption and they lower the activation barrier for the nitrate-to-nitrite reduction step. This improved understanding of the In active site expands the prospects of improved denitrification using metal-on-metal catalysts.

Full text of DOI:10.1021/acscatal.7b01371

Subscriber access provided by READING UNIV
Article
Insights into Nitrate Reduction over Indium-
Decorated Palladium Nanoparticle Catalysts
Sujin Guo, Kimberly Heck, Sashank Kasiraju, Huifeng Qian,
Zhun Zhao, Lars C. Grabow, Jeff Miller, and Michael S. Wong
ACS Catal., Just Accepted Manuscript • DOI: 10.1021/acscatal.7b01371 • Publication Date (Web): 21 Nov 2017
Downloaded from http://pubs.acs.org on November 21, 2017
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
ACS Catalysis is published by the American Chemical Society. 1155 Sixteenth Street
N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 40
ACS Catalysis
1
2
3
4
5
6
7
8
9
Insights into Nitrate Reduction over Indiumꢀ
Decorated Palladium Nanoparticle Catalysts
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
,2†
†3
4
3
3
Sujin Guo , Kimberly Heck , Sashank Kasiraju , Huifeng Qian , Zhun Zhao , Lars C.
4
5
1,2,3,6,7
Grabow *, Jeffrey T. Miller and Michael S. Wong
*
1
2
Department of Civil & Environmental Engineering, Nanosystems Engineering Research Center
3
for NanotechnologyꢀEnabled Water Treatment, Department of Chemical and Biomolecular
4
Engineering, Rice University, 6100 S. Main Street, Houston, TX 77005, Department of
Chemical and Biomolecular Engineering, University of Houston, 4726 Calhoun Road, Houston,
5
TX 77204ꢀ4004, Department of Chemical Engineering, Purdue University, 480 Stadium Mall
6
7
Drive, West Lafayette, IN 47907, Department of Chemistry, Department of Materials Science
and NanoEngineering, Rice University, 6100 S. Main Street, Houston, TX 77005
*To Whom Correspondence Should Be Addressed Eꢀmail: mswong@rice.edu,
grabow@uh.edu
†
These authors contributed equally to the study
1
ACS Paragon Plus Environment

ACS Catalysis
Page 2 of 40
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Abstract
ꢀ
Nitrate (NO ) is an ubiquitous groundwater contaminant, and is detrimental to human health.
3
Bimetallic palladiumꢀbased catalysts have been found to be promising for treating nitrate (and
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ꢀ
nitrite, NO ) contaminated waters. Those containing indium (In) are unusually active, but the
2
mechanistic explanation for catalyst performance remains largely unproven. We report that In
deposited on Pd nanoparticles (NPs) ("InꢀonꢀPd NPs") shows roomꢀtemperature nitrate catalytic
reduction activity that varies with volcanoꢀshape dependence on In surface coverage. The most
active catalyst had an In surface coverage of 40%, with a pseudoꢀfirst order normalized rate
ꢀ
1
ꢀ1
constant of kcat ~ 7.6 L gsurfaceꢀmetal min , whereas monometallic Pd NPs and In O have
2
3
nondetectible activity for nitrate reduction. Xꢀray absorption spectroscopy (XAS) results
indicated that In is in oxidized form in the asꢀsynthesized catalyst; it reduces to zerovalent metal
ꢀ
in the presence of H and reꢀoxidizes following NO contact. Selectivity in excess of 95% to
2
3
nontoxic N was observed for all the catalysts. Density functional theory (DFT) simulations
2
suggest that submonolayer coverage amounts of metallic In provide strong binding sites for
nitrate adsorption and they lower the activation barrier for the nitrateꢀtoꢀnitrite reduction step.
This improved understanding of the In active site expands the prospects of improved
denitrification using metalꢀonꢀmetal catalysts.
Keywords: nitrate reduction; bimetallic catalyst; DFT; indium; palladium; coreꢀshell
2
ACS Paragon Plus Environment

Page 3 of 40
ACS Catalysis
1
2
3
4
5
6
7
8
9
Introduction
ꢀ
Nitrate (NO ) is a contaminant detected globally in surface water and underground aquifers.
3
Nitrate pollution occurs due to the overuse of nitrogenꢀrich agriculture fertilizers, wastewater
1
–3
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
discharge, and contaminant leaching from landfills. This anion, in addition to its partially
ꢀ
reduced form, nitrite (NO ), can cause adverse health effects in humans including
2
methemoglobinemia (blue baby syndrome). When ingested, these compounds can be
4
,5
metabolized into Nꢀnitroso compounds, which may result in cancer and hypertension.
6
Additionally, nitrate is suspected to increase cocontamination of uranium. The United States
Environmental Protection Agency (US EPA) has mandated maximum contaminant levels (MCL)
ꢀ
ꢀ
7
of NO and NO of 10 mgꢀN/L (mgꢀnitrogen per liter of water) and 1.0 mgꢀN/L, respectively,
3
2
ꢀ
ꢀ
while the European Drinking Water Directive limits the concentration of NO and NO to 11.3
3
2
8
mgꢀN/L and 0.15 mgꢀN/L, respectively. The limits proposed by the World Health Organization
ꢀ
ꢀ 9
are 11.3 mgꢀN/L for NO and 0.91 mgꢀN/L for NO .
3
2
The most widely used technologies to treat nitrate/nitrite include ion exchange, reverse
2,10,11
osmosis and biological denitrification.
Ion exchange is used for largeꢀscale treatment, while
reverse osmosis is more suitable for smaller scale treatments. However, ion exchange and
reverse osmosis processes produce concentrated nitrate/nitriteꢀcontaining waste streams that
must undergo further treatment prior to disposal. Biological denitrification is sensitive to pH and
the presence of other salts at the contaminated site. Moreover, long startup times and concerns
1
2, 13
regarding pathogens and active biomass during intermittent operations limit its applicability.
The catalytic reduction of environmental contaminants has attracted attention as a promising
1
4–16
solution due to their high activity and selectivity to environmentally benign products.
For
1
6–20
20
example, monometallic and bimetallic Pd
and PdAu catalysts have been shown to be very
3
ACS Paragon Plus Environment

ACS Catalysis
Page 4 of 40
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
+
active and selective to the reduction of nitrite to dinitrogen (without the formation of NH /NH ).
3
4
2
1
Pd alone has very low reduction activity for nitrate, but its combination with certain secondary
2
2–27
metals greatly increases catalytic activity.
due to their high activity and high selectivity to dinitrogen gas over toxic ammonia.
Mechanistically, it is supposed that InPd catalysts contain metallic In that oxidizes in order to
Of these catalysts, InPd bimetallics are preferred
28–30
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ꢀ
ꢀ
reduce NO to NO . The nitrite then surface diffuses to neighboring Pd sites to be further
3
2
reduced to N (or NH ) through a series of surface reaction steps with hydrogen adatoms formed
2
3
from H chemisorption. These Hꢀadatoms are presumably responsible for regenerating the In
2
sites. Metal loading, atomic configuration, and metal nanoparticle size have been found to affect
31–34
the rate of nitrate reduction.
Since the first publication in 2000 by Vorlop and coworkers
28
though, the exact nature or structure of the In active sites remains unknown.
Our previous studies have focused on the synthesis of model PdꢀonꢀAu NP catalysts to better
3
5–37
understand hydrodehalogenation reactions
, nitrite reduction, glycerol oxidation, and
2
0,38–40
nitrophenol reduction.
In this work, we deposited In on the surface of monodisperse Pd
NPs, so as to better control the structure of the bimetallic InꢀonꢀPd NP catalysts. We prepared a
series of InꢀonꢀPd NPs with different surface coverages of In, and quantified their reactivity for
nitrate reduction. By using XPS and ex situ and in situ XAS characterization techniques, we
identified likely catalyst surface structures, and correlated them with observed catalyst activities
and selectivities. We performed DFT calculations using different InꢀonꢀPd slab models to
estimate the relevant adsorption and transition state energies, giving further insights about the
role and nature of the In surface structures.
4
ACS Paragon Plus Environment

Page 5 of 40
ACS Catalysis
1
2
3
4
5
6
7
8
9
Materials and methods
Materials
Palladium (II) acetate (Pd(CH COO) , 98%), 2ꢀethoxyethanol (C H O , 99%),
3
2
4
10
2
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
polyvinylpyrrolidone (PVP, M ~55,000), indium chloride tetrahydrate (InCl ·4H O, 97%),
w
3
2
sodium nitrate (NaNO , >99.0%), sodium nitrite (NaNO , >97.0%), 1 wt% Pd on Al O catalyst
3
2
2
3
(Pd/ Al O )and modified Griess’s reagent were purchased from SigmaꢀAldrich. Hydrogen gas
2 3
(99.999%), carbon dioxide gas (99.995%) and nitrogen gas (99.999%) were purchased from
Matheson. All experiments were conducted using deionized (DI) water (>18 Mꢁcm, Barnstead
NANOpure Diamond).
Catalyst preparation and characterization
To synthesize Pd NPs, 0.255 g PVP and 0.1332 g palladium (II) acetate were dissolved in 15 mL
2
ꢀethoxyethanol. Before heating, the solution was purged with nitrogen gas for 15 min to remove
dissolved oxygen. Using a microwave reactor (AntonꢀParr Monowave 300), the brownꢀcolored
solution was heated to 120 ºC and held at this temperature for 1 h. The solution was stirred at
6
00 rpm throughout the procedure. The color of the solution was black after heat treatment,
indicative of the formation of Pd NPs from the Pd (II) precursor. The InꢀonꢀPd bimetallic
catalysts with different In surface coverages (sc%) were prepared by adding the appropriate
amounts of In salt precursor to the Pd sol, and contacting with H gas (Table S1). The In surface
2
coverage was calculated as the percentage of In atoms needed to fill the 10th shell around a
magic cluster of 9 closed shells of Pd atoms (which approximates the 5.32 nm Pd NPs, Table
41
S2). Our model assumes that all the In precursor is fully reduced onto the Pd surface and that
the Pd particles are monodisperse. The various surface coverage percentages (sc%) of InꢀonꢀPd
NPs were obtained by mixing 0, 0.170, 0.425, 0.511, 0.681, 0.851, 1.191, 1.872 mL InCl ·4H O
3
2
5
ACS Paragon Plus Environment

ACS Catalysis
Page 6 of 40
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
solution (0.04 M) with 5 mL of the asꢀsynthesized Pd NPs, then adding DI water to the solution
to reach a volume of 6.872 mL (for 0, 10, 25, 30, 40, 50, 70, and 110 sc%; see Table S3 for
weight fractions). The solution was stirred for 1 min and then bubbled with H gas for 15 min at
2
ꢀ1
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
a flow rate of ~200 mL min .
For comparison, a supported PdIn catalyst was prepared with a Pd:In weight ratio of 95:5
(equivalent to InꢀonꢀPd NPs of 40 sc% In) through wet impregnation similar to that described by
4
2
Werth and coworkers. 350 mg of a commercially available 1 wt% Pd/Al O was weighed and
2
3
2
1
4 ꢂL of 0.2 M indium chloride solution was added. The resulting material was then air dried at
20 °C for 14 h and treated under H gas at 120 °C for 1 h.
2
Transmission electron microscopy (TEM) images were obtained using a JEOL 2010
transmission electron microscope operating at an accelerating voltage of 200 kV. The particle
size distribution was calculated by counting at least 200 particles with ImageJ. The Xꢀray
photoelectron spectroscopy (XPS) data was obtained by using a PHI Quantera System with
monochromatic Al K radiation (1486.7 eV). XPS samples were prepared by placing a vial of 10
R
mL of catalyst in a 50°C oven overnight to evaporate solvent to form a black plateꢀlike solid and
stored under atmospheric conditions at room temperature before analysis. pH measurements
were taken using a VWR sympHony SB20 meter with a standard pH electrode. In K (27.90 to
2
7.98 keV) edge or Pd K (24.32 to 24.40 keV) edge XAS measurements were carried out on the
insertion and bending magnet devices (beamline 10ꢀIDꢀB and 10ꢀBMꢀB) of the Materials
Research Collaborative Access Team (MRCAT) at the Advanced Photon Source at Argonne
2
2
National Laboratory. The Xꢀray beam was 0.5 × 0.5 mm at the ID beamline and 0.5 ×2.0 mm at
the BM beamline, and measurements were made in transmission mode with the ionization
chambers optimized for maximum current with linear response, and were obtained
6
ACS Paragon Plus Environment

Page 7 of 40
ACS Catalysis
1
2
3
4
5
6
7
8
9
simultaneously with In or Pd foil spectra for energy calibration. Samples of the InꢀonꢀPd NPs of
1
0 sc%, 40 sc%, and 110 sc% were prepared by adding 600 mg activated carbon to 20 mL of the
respective InꢀonꢀPd suspension. After stirring overnight, the carbon slurry was removed by
centrifugation, dried in an oven at 50°C overnight, and stored under ambient conditions. These
carbonꢀsupported NPs samples were also heatꢀtreated at 200 °C for 30 min under flowing 4% H2
gas. For the in situ experiments, solutions were prepared by bubbling DI water for 30 min to
saturate the water with the gas of choice (air, 4% H , or inert N ). The in situ XANES were
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
2
obtained by loading catalyst powder into a polymer, in situ solution cell designed for air sensitive
4
3
homogeneous and slurry catalysts and is described in detail in a previous publication. This cell
is equipped with inlet and outlet ports to flush the cell with the desired solution. Approximately 5
mL of the solution of interest (airꢀsaturated water, 4% H ꢀsaturated water, or 100 ppm nitrate
2
solution purged with N gas) was flushed through the cell (with ~3 mL volume), prior to the cell
2
being sealed for XANES measurements. At the Pd and In Kꢀedge spectra, a 1ꢀcm path length
was used for in situ treatments.
WINXAS 3.1 software was used to fit the XAS data. The EXAFS coordination parameters
2
were obtained by a leastꢀsquares fit in Rꢀspace of the k ꢀweighted Fourier transform data from
3
1
.0 to 12.0 Å. The first shell fit of the magnitude and imaginary parts were performed between
.6 to 2.9 Å for the Pd edge. At the In edge, the first shell fit was performed from 1.8 to 2.8 Å.
An average coordination number and bond distance were determined for PdꢀPd and PdꢀIn
scattering by fitting with the experimental phase shift and back scattering amplitude of PdꢀPd
scattering extracted from Pd foil XAS data (12 PdꢀPd at 2.75 Å). Such treatment was rationalized
based on the small difference in phase shift and amplitude between PdꢀPd and PdꢀIn scattering
7
ACS Paragon Plus Environment

ACS Catalysis
Page 8 of 40
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
due to the close atomic number of Pd and In. Fits were performed by refinement of the
2
coordination number (CN), bond distance (R), ꢃσ , and energy shift (E ).
0
Catalyst activity characterization
44
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Similar to our previous experiments, batch nitrate reduction experiments were conducted in a
screwꢀcap bottle (125 mL, Alltech) with PTFEꢀsealed threads and a PTFEꢀsilicone septum. The
initial water volume was controlled so that the final liquid reaction volume was 99.5 mL after
adding the catalyst NPs. The amount of InꢀonꢀPd NPs added was chosen such that the total In
ꢀ1
amount per reaction was 0.553 mg, the final catalyst charge concentration was 5.53 mgꢀIn L
(
Table S4). The solution was then bubbled simultaneously with H gas (120 mL/min, to serve as
2
reductant) and CO gas (120 mL/min, to buffer the solution to a pH value of 4ꢀ6) for 30 min to
2
displace dissolved oxygen and to fill the headspace with a hydrogen/CO atmosphere (1 atm).
2
All the catalytic reactions were conducted at room temperature (~23 °C). After hydrogen and
carbon dioxide bubbling, 0.5 mL of NaNO solution (10 mg/mL) was injected into the sealed
3
ꢀ
ꢀ
bottle to start the reaction (initial NO concentration was 50 mg NO per L). The reaction was
3
3
monitored periodically by withdrawing 2 mL aliquots from the reactor. The InꢀonꢀPd NPs
showed negligible UVꢀvis absorbance, so no separation of the particles from the reaction
medium was performed prior to UVꢀvis spectroscopy measurements.
The activity of the supported catalysts was tested similarly, except that the solid catalyst was
added directly to the reactor before the H /CO bubbling (see Table S5 for catalyst loading). As a
2
2
control, experiments using 3.51 mg (1.15 mL of 3.05 mg Pd/mL solution) of monometallic Pd
NPs (at the same Pd charge as that for 30 sc% InꢀonꢀPd NPs) were also conducted. To verify that
intimate contact between the In and Pd was required, we carried out a control experiment using a
combination of 10 mg of In O and 3.51 mg (1.15 mL of 3.05 mg Pd/mL solution) Pd NPs.
2
3
8
ACS Paragon Plus Environment

Page 9 of 40
ACS Catalysis
1
2
3
4
5
6
7
8
9
Unlike the NPꢀonly experiments, the supported Pd catalysts and In O powder had significant
2
3
absorbance in the visible light regime, and the powder was separated from the reaction medium
with syringe filters (0.45ꢀꢂm PES filter media) before nitrate, nitrite and ammonium
concentrations were determined.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
45
Nitrite concentrations were analyzed using the Griess test. A stock solution of the Griess
reagent was prepared by dissolving 0.1g of Nꢀ(1ꢀnaphthyl)ethylꢀenediamine dihydrochloride, 1.0
g of sulfanilamide and 2.94 mL H PO in 100 mL DI water, such that the final concentrations
3
4
were 0.1 wt% Nꢀ(1ꢀnaphthyl)ethylꢀenediamine dihydrochloride, 1 wt% sulfanilamide, and 5%
46
H PO . In a typical colorimetric assay (concentration range from 0.02 to 2.5 ppm), the Griess
3
4
reagent solution (0.2 mL), a nitriteꢀcontaining solution (0.2 mL), and water (1.6 mL) were mixed
together and allowed to react at room temperature for 10 min, after which no further color
change was observed. The sulfanilamide reacts with the nitrite anion to form a diazonium salt,
which further reacts with the amine to form an azo dye (magenta in color). The absorbance at
ꢀ
5
40 nm is measured via UVꢀvis spectroscopy, and the NO concentration is determined in the 0
2
to 2.0 ppm range using a standard calibration curve. Nitrate and ammonium measurements were
made using a nitrate ion selective electrode (ColeꢀParmer, lower detection limit 0.1 ppm) and an
ammonium ion selective electrode (ColeꢀParmer, lower detection limit 0.01 ppm).
ꢀ1
The observed reaction rate constant kmeas (with unit of min ) was calculated by assuming
pseudo firstꢀorder dependence on nitrate concentration (H gas was in excess):
2
ꢀ1
where CNO3− is the concentration of nitrate (with units of mg L ) and t is reaction time (with
units of min).
9
ACS Paragon Plus Environment

ACS Catalysis
Page 10 of 40
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
20,44
Using methods described in our previous work
and in the Supporting Information, mass
transfer effects were quantified by varying the mass loading of catalyst at a constant stir rate. We
used 40 sc% InꢀonꢀPd NPs as the test case, such that the metal content varied (1.75 mg, 3.51 mg,
and 5.26 mg Pd NPs, and 277, 553, and 830 ꢂg In, respectively) while keeping the total reaction
volume constant at 99.5 mL.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
The measured rate constants were corrected for gasꢀliquid mass transfer to obtain the corrected
rate constant kcorr values. The surface metal normalized reaction rate constant k (with units of L
cat
ꢀ1
ꢀ1
g surface metal min ) was then calculated from k
(via k = k /C , where C is the total
cat corr cat cat
corr
ꢀ
concentration of amount of surface In and Pd in the reaction medium; with units of gsurface metal
L
1
).
The initial molar concentration of nitrate and the concentration of nitrate after time t are C and
0
C. Prior studies indicate that only trace amounts of intermediate products NO and N O are
2
2
8,29,42
formed in the bulk during nitrate reduction.
the following way:
Thus we calculate the selectivity to N (S ) in
2 N2
,
where the selectivity to ammonium and selectivity to nitrite (SNH4 and SNO2 respectively) are
,
and
respectively.
10
ACS Paragon Plus Environment

Page 11 of 40
ACS Catalysis
1
2
3
4
5
6
7
8
9
Computational methodology
4
7
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
All periodic density functional theory (DFT) calculations were performed using the Vienna abꢀ
48–51
initio simulation (VASP) package.
Electron exchange and correlation were described by the
5
2,53
revised PerdewꢀBurkeꢀErnzerhof (GGAꢀRPBE) functional.
The projector augmented wave
5
4,55
(PAW)
method was used to describe the core and valence electrons with a kinetic energy
cutꢀoff of 400 eV for surfaces and 540 eV for bulk materials. Partial occupancies for the
5
6
wavefunctions were calculated using Gaussian smearing with a Fermi temperature (k T) of 0.1
b
eV. The total energies were extrapolated to k T = 0 eV at the end of ionic convergence. Bader
b
57–60
charge analysis
was performed to assess the extent of oxidation of the indium atoms.
A flat Pd(111) surface was constructed using a (4×4) unit cell, and a (1×3) unit cell was
used to describe the stepped Pd(211) surface. For both structures we used an equivalent of four
atomic layers, where the top two layers were fully relaxed and the bottom two layers were fixed
at their bulk positions corresponding to the optimized lattice constant value of 3.99 Å. A vacuum
of 20 Å was added along the z axis normal to the surface along with the corresponding dipole
6
1
correction. For the bulk and surface calculations a (13×13×13) and (4×4×1) Monkhorstꢀpack kꢀ
62
point mesh were used, respectively. The bulk structure of Indium was also optimized using the
6
3,64
RPBE functional, with the initial crystal structure obtained from ICSD.
The final optimized
lattice constants were a = 3.34 Å, b = 3.35 Å, c = 5.10 Å. The monometallic In(001) surface was
constructed using a (2×2) unit cell, with a (4×4×1) Monkhorstꢀpack kꢀpoint mesh and other
calculation settings were kept consistent with the palladium system.
11
ACS Paragon Plus Environment

ACS Catalysis
Page 12 of 40
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
The bimetallic InꢀonꢀPd models were created by successively replacing Pd(111) surface
atoms with In, as shown in Figure 1. The In sc% was then approximated by the number of
indium atoms in the surface divided by the total number of surface atoms, e.g., a substitution of 4
Pd atoms with In in a (4×4) unit cell yields 4/16 = 25 sc% In. When the In surface coverage
reached 50 sc%, the larger In atoms could no longer be accommodated in the Pd lattice without
causing significant surface buckling and reconstruction. Thus, we increased the Pd(111) unit cell
to (6×6) and added a layer of 25 In atoms to reduce the surface strain when modeling a full
monolayer of In on Pd(111) (100 sc%). Here a (3x3x1) Monkhorstꢀpack kꢀpoint mesh was used.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 1. Successive In atom (red) substitution in a Pd(111) 4×4 unit cell to form compact In
ensembles. The progressive substitution sequence is indicated by the numerals 1ꢀ7 and a
decreasing red color intensity. Substitution of all seven sites results in a nominal In surface
content of 7/16 = 43.8 sc%.
ꢀ
To avoid problems of using periodic DFT methods to describe the charged NO species
3
as reference, we used neutral HNO as gas phase reference instead, and obtained the energy for
3
the nitrate ion according to the thermodynamic cycle shown in Scheme 1. This reference neglects
the solvation energy, ꢁG , of nitrate ions in aqueous environment. Since the value of ꢁG is
sol
sol
12
ACS Paragon Plus Environment

Page 13 of 40
ACS Catalysis
1
2
3
4
5
6
7
8
9
constant when different catalysts are compared, we chose to lump it as a first order
approximation into the estimation of the equilibrium constant for nitrate adsorption. Upon
adsorption on the metal surface we further assume that any charges on adsorbates are quenched
and delocalized throughout the metal, such that surface energetics are expected to be within
typical DFT errors. Despite our best efforts to make reasonable assumptions for the adsorption
and reaction of the charged nitrate species in aqueous solution, this approach is subject to
uncertainties as a result of neglecting the explicit treatment of the negative charge and solvation
effects. The solvent does not only stabilize the nitrate ion in solution, but also alters the
energetics of adsorbed (possibly charged) species and empty surface sites. Reliable
quantification methods for these effects are currently unavailable, but we estimate that the
catalystꢀadsorbate interactions across our model systems outweigh the significance of solvation
effects for similar surface intermediates and transition states. The gas phase energy values for H,
H , H O, HNO , and N were obtained from Gamma point only calculations by placing a single
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
2
3
2
molecule in a cubic box with edges of 10 Å.
ꢀ
Scheme 1. Thermodynamic cycle to estimate the reference DFT energy for the NO ion using
3
6
5
reference values for the ionization energy of H (I.E.) and the heat of reaction for HNO (g)
deprotonation (ꢁHr).
3
6
6
13
ACS Paragon Plus Environment

ACS Catalysis
Page 14 of 40
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Transition state geometries were found using the climbing image nudged elastic band
6
7–72
73–76
(NEB)
and dimer methods.
A single imaginary mode in the subsequent vibrational
analysis using the harmonic oscillator approximation and a Cartesian displacement of 0.01 Å
verifies these geometries as true saddle points.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
All rate constants in the lumped kinetic model are estimated from free energies obtained
from first principles and corrected for the zeroꢀpoint energy, enthalpy and entropy contributions
at 300 K. For adsorbed species, we obtained these corrections from a frequency analysis in the
harmonic oscillator approximation. For gas phase free energies, we adopted enthalpy and entropy
corrections calculated using the Shomate equation and gas phase partial pressures totaling 1 bar.
Results and discussion
Representative TEM images for Pd NPs and 110% sc InꢀonꢀPd NPs are shown in Figure 2. The
bare Pd NPs were monodisperse, with a mean diameter of 5.0±0.4 nm. The addition of In slightly
increased the particle size. The sample with the most In, 110% sc InꢀonꢀPd NPs (e.g., more than
one calculated monolayer of In on the Pd NPs) had a mean diameter of 5.7±0.6 nm.
14
ACS Paragon Plus Environment

Page 15 of 40
ACS Catalysis
1
2
3
4
5
6
7
8
9
(
a)
(b)
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
50 nm
50 nm
80
80
Pd NPs
average size: 5 ± 0.4 nm
110 sc% In-on-Pd NPs
average size: 5.7 ± 0.6 nm
7
6
0
0
70
6
5
4
3
2
1
0
0
0
0
0
0
0
50
4
3
0
0
20
1
0
0
0
1
2
3
4
5
6
7
8
9
10
0
1
2
3
4
5
6
7
8
9
Diameter (nm)
Diameter (nm)
Figure 2. Transmission electron microscopy (TEM) images and corresponding particle size
distributions of (a) Pd NPs and (b) InꢀonꢀPd NPs (with 110 sc% In). Each bar represents the total
count percentage of NPs with a measured diameter ± 0.25 nm.
The XPS spectra of the Pd NPs and InꢀonꢀPd NPs with 10ꢀ110 sc% are shown in Figure 3.
Figure 3a shows the spectra between 440ꢀ450 eV, the region for the In 3d5/2 peak (previous
literature reports the In O peak at 445 eV and the peak for In(0) is reported to be around 444
2
3
7
7
eV). As can be seen, the peak binding energy blueshifted with added In, from 444.2 eV (for the
1
0 sc% InꢀonꢀPd NPs) to 444.9 eV (for the 110 sc% PdꢀonꢀIn NPs), indicating that the Pd had an
15
ACS Paragon Plus Environment

ACS Catalysis
Page 16 of 40
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
electronic effect on the In; the samples with lower loadings had more of a reduced character. On
the other hand, the electronic state of Pd was unaffected by the In (Figure 3b). The peak at 335.5
eV matches that reported for Pd 3d5/2 for Pd(0), and the peak at 340.5 eV can be attributed to the
Pd 3d3/2 for Pd(0). The shoulder for the latter peak could be due to the presence of oxidized Pd
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(the Pd3/2 of PdO is 342 eV) which possibly formed during sample preparation and storage at
atmospheric conditions. Unlike the In peak, the Pd peaks did not shift with In loading.
Figure 3. XPS spectra of InꢀonꢀPd NPs with different surface coverages: (a) In 3d5/2 energy
range and (b) Pd 3d energy range.
To further verify the oxidation state of In and Pd in the InꢀonꢀPd NPs and also infer more
information about the catalysts, we used Xꢀray absorption spectroscopy (XAS). We have
previously used this technique to extensively characterize the surface structure of PdꢀonꢀAu NP
20,36,39
catalysts for various reactions.
Figure 4a shows the XAS results of the Pd edge for Pd NPs
and InꢀonꢀPd NPs of different surface coverages and with different pretreatments (asꢀsynthesized
and reduced in H , Table S6).
2
16
ACS Paragon Plus Environment

Page 17 of 40
ACS Catalysis
1
2
3
4
5
6
7
8
9
1.5
1.5
0
1
sc% (Pd NPs)
0 sc%
40 sc%
10 sc%
10 sc%
40 sc%
110 sc%
In O
(a)
(b)
1
2
3
1
0
0
.0
.5
.0
1.0
0.5
0.0
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
4.32
24.34
24.36
24.38
24.40
27.90
27.92
27.94
27.96
27.98
Photon Energy (keV)
Photon Energy (keV)
(
c)
(d)
1
.2
.8
0
0.4
.0
0
Air
H
2
2
7.98
nd
2
Air
27.96
nd
2
H
rd
27.94
2
27.92
3
Air
2
7.90
Figure 4. XANES spectra of the asꢀsynthesized samples at the (a) Pd K and (b) In K edges. (c)
XANES spectra (at In K edge) of InꢀonꢀPd NPs with different In sc% after reduction with H at
200 °C. (d) in situ XANES (In K edge) of the 40 sc% InꢀonꢀPd NPs under cycles of airꢀ and 4%
H ꢀsaturated water at room temperature.
2
2
In agreement with the XPS data, the spectra showed that regardless of pretreatment and
composition, all samples had zeroꢀvalent Pd. We note that this is in contrast to our previous
37,39,40
results,
which showed that monometallic Pd NPs were partially oxidized after handling in
air. We attribute this to the presence of PVP, as other researchers have also observed the absence
7
8,79
of oxidized metal in metal NPs prepared using PVP.
Significantly, the PdꢀPd bond distance
was close to 2.75 Å (Table S6) for all the catalysts, and is identical to bulk Pd, suggesting that
the In did not alloy into the Pd NPs and remained on the NP surface. In contrast, fully alloyed
17
ACS Paragon Plus Environment

ACS Catalysis
Page 18 of 40
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
80
PdIn nanoparticles have EXAFS observable PdꢀIn bond distances of 2.81 Å, which were not
observed here.
The amount of oxidized In in the samples can be estimated by dividing the InꢀO
coordination number determined from EXAFS analysis of the In edge by six, since In
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3+
coordinated with six O atoms is consistent with fully oxidized In . As prepared, the In was fully
oxidized ("In(OH) ") for all the catalysts, and no InꢀIn scattering pairs were detected (Table S6).
3
Exposure of the catalysts to 4% H gas at 200 °C for 30 min was sufficient to partially reduce the
2
In, leading to InꢀIn scattering pairs with nonzero coordination numbers. InꢀonꢀPd NPs with the
1
0 sc%, 40 sc%, and 110 sc% contained 65%, 60%, and 45% In as zerovalent metal, respectively,
after reduction treatment (Figure 4c, Table S2). This result suggested the presence of oxidized In
that can be reduced under reaction conditions.
We gained additional insights about the state of In metal in the InꢀonꢀPd NPs, after
analysis of the calculated free energies of In atoms in different positions on Pd terraces and steps
(
Figure 5). An In adatom on the Pd(111) surface is highly mobile and can diffuse with an
activation barrier of less than ~0.1 eV to a Pd (211) step, where it is ~0.7 eV more stable (Fig.
aꢀb). Its stability further increases by ~0.2 eV if it replaces a higher coordinated Pd atom in the
111) surface (Fig. 5c). If located in a fully coordinated subsurface layer position though, In is
.05 eV less stable (Fig. 5d). 2D In ensembles containing 7 atoms or less are energetically more
5
(
0
stable compared to a single In adatom, with an energy penalty that increases with ensemble size
(Figures 5eꢀj). The 2D ensembles may grow to form a monolayer of In atoms as illustrated in
Figures 5k, or further agglomerate to form 3D In clusters as depicted in Figure 5l. We did not
attempt to model Pdꢀsupported In clusters; instead we chose to approximate them as pure In(001).
18
ACS Paragon Plus Environment

Page 19 of 40
ACS Catalysis
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 5. Stability of In atoms on Pd relative to the most stable configuration: (a) In adatom on
Pd(111) terrace; (b) In adatom at Pd(211) step; (c) In atom substitution in the top layer; (d) In
st
atom substitution in 1 sublayer of Pd(111); 2D In ensembles within the Pd(111) top layer
containing (e) 2, (f) 3, (g) 4, (h) 5; (i) 6 , and (j) 7 In atoms; (k) In monolayer (ML) on Pd(111)
(equivalent to 100 sc%); and (l) a schematic 3D In ensemble.
The combined experimental and computational characterization of the InꢀonꢀPd NPs are
largely consistent with metallic In substituting Pd on surface terrace sites. At low In sc%, In
atoms are wellꢀdispersed and remain metallic. The theoretical maximum In surface concentration
on Pd(111) with isolated In atoms is 33 sc%, suggesting that the InꢀonꢀPd NPs contain 2D and
3
D ensembles at higher In content. The Inꢀrich materials are expected to show more bulk In
metal character and are more likely to oxidize.
Nitrate reduction catalytic behavior
Figure 6 shows a representative plot of the time evolution of nitrate, nitrite, ammonium and
dinitrogen concentrations on the 40 sc% InꢀonꢀPd catalyst sample. The decrease in nitrate
concentration fit well to a pseudo firstꢀorder model, and the observed reaction rate constant, kmeas
,
19
ACS Paragon Plus Environment

ACS Catalysis
Page 20 of 40
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
was obtained from the slope of the natural log of nitrate concentration versus reaction time using
linear least squares fitting.
14
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
N2
10
8
-
NO3
6
4
2
+
NH4
-
NO2
0
0
20
40
60
80
100
Time (min)
ꢀ
ꢀ
+
Figure 6. Concentrationꢀtime curves of NO , NO , NH and N . Reaction conditions: 40 sc%
3
2
4
2
InꢀonꢀPd NPs with 0.553 mg/L In in reactor, 600 rpm stirring rate, 1 atm pressure, 120 mL/min
H , 120 mL/min CO , pH = 5. The initial nitrate/indium molar ratio was 166/1.
2
2
Control experiments using In O , Pd NPs, and 1 wt% Pd/Al O all showed that the
2
3
2
3
monometallic catalysts are inactive for nitrate reduction (Table 1). Thus catalytic activity was
quantified by normalizing the mass transfer corrected kmeas to total surface metal content (kcat),
not to surface Pd or surface In content.
Table 1. Rate constants for InꢀonꢀPd NPs, Pd /Al O , InꢀPd/Al O and In O
2
3
2
3
2
3
Sample name
k
kcorr
(min )
kcat
meas
ꢀ
1
ꢀ1
ꢀ1
ꢀ1 a
(
min )
(L gsurface metal min )
0 sc% InꢀonꢀPd NPs
inactive
inactive
0.035
0.034
0.072
0.087
inactive
0.86±0.30
1.94±0.32
4.97±0.18
7.57±0.65
1
2
0 sc% InꢀonꢀPd NPs
5 sc% InꢀonꢀPd NPs
0.032±0.01
0.031±0.0056
0.059±0.002
0.069±0.015
3
0 sc% InꢀonꢀPd NPs
40 sc% InꢀonꢀPd NPs
20
ACS Paragon Plus Environment

Page 21 of 40
ACS Catalysis
1
2
3
4
5
6
7
8
9
5
7
0 sc% InꢀonꢀPd NPs
0 sc% InꢀonꢀPd NPs
0.057±0.013
0.069
0.015
7.26±0.49
2±0.53
0.014±0.003
.0086±0.0005
inactive
110 sc% InꢀonꢀPd NPs
0
0.0088
inactive
inactive
inactive
1.6±0.10
inactive
inactive
inactive
In O powder
2
3
Pd NPs
Pd/Al O (1 wt% Pd)
inactive
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
3
inactive
Pd NPs and In O
physical mixture
2
3
inactive
inactive
–
inactive
4.6
In/Pd/Al O (0.13 wt%
2
3
0
.075
In, 1 wt% Pd)
a
Calculated from kcorr and magic cluster model
Figure 7 shows that the activity of InꢀonꢀPd NPs (as assessed by k ) demonstrating the
cat
volcanoꢀshape dependence on In surface coverage. The rate constants increased from 0 sc% to
4
0 sc% and decreased from 40 sc% to 110 sc%. Maximum nitrate reduction was measured at 40
ꢀ
1
ꢀ1
ꢀ1
sc% with kcat = 7.57 L g
min . The In/Pd/Al O supported catalyst contained the
2 3
surface metal
same Pd and In mass ratio as the 40 sc% NP material but showed 60% less activity, attributable
to the deposition of inactive indium on the Al O support surface.
2
3
1
0
8
6
4
2
0
0
20
40
60
80
100
InPd/Al O
2 3
Calculated In surface coverage (sc%)
Figure 7. Experimentally determined nitrate reduction rate constants for InꢀonꢀPd NPs as a
function of In surface coverage and for InPd/Al O (1 wt% Pd, 0.13 wt% In).
2
3
21
ACS Paragon Plus Environment

ACS Catalysis
Page 22 of 40
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
The catalytic nitrate reduction reaction presumably involves a bifunctional effect in
which the InꢀonꢀPd NPs carry two active site types that participate in different parts of the
surface reaction pathway. The In site has the primary role of adsorbing and reducing nitrate from
solution, and the Pd site has the roles of generating hydrogen adatoms; reducing the surface
nitrite species; and reducing the oxidized In via hydrogen spillover (Scheme 2). The activity of
the InꢀonꢀPd NPs increased with indium amounts at surface coverages below 40 sc%, where
isolated In atoms and small In ensembles prevail. These sites adsorb and convert more nitrate,
which leads to a higher reaction rate. However, deposition of too much In is counterproductive,
as large In ensembles may block Pd active sites necessary for dissociating hydrogen used in the
further reduction of nitrite and regeneration of metallic In sites.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
c
Pd0
NO3-
NO2-
N2
a
b
H*
In3+
d
In0
Pd0
H2
2H*
H+
H*
Scheme 2. Schematic illustration of the nitrate reduction mechanism using InꢀonꢀPd NPs. H*
represents hydrogen atoms adsorbed on the Pd surface. (a) Reduction of nitrate to nitrite and
0
3+
0
oxidation of In to In , (b) regeneration of In by H*, (c) further reduction of nitrite to dinitrogen
0
0
over Pd , and (d) dissociation of H into H* over Pd .
2
Further support for the synergistic effect of the bimetallic InꢀonꢀPd NPs for nitrate
reduction was provided by DFT simulations on a Pd(111) surface with ensembles of varying
22
ACS Paragon Plus Environment

Page 23 of 40
ACS Catalysis
1
2
3
4
5
6
7
8
9
number of In atoms and monometallic Pd and In surfaces. Since the Pd ability to reduce nitrite is
1
5,17,19,20,81–83
well established
, we focused only on the following elementary reactions (nitrate
adsorption and oxygen abstraction) and assume all other steps to be fast (* represents the active
site):
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 8 shows the transition state geometries of the two reaction steps on selected
geometries and their corresponding free energy diagrams. A complete description of all initial
states (IS), transition states (TS), final states (FS), and the reaction energetics is provided in
Figure S3. While an isolated In atom (Figure 8b) did not promote nitrate binding or activation
compared to monometallic Pd(111) (Figure 8a), a small 3ꢀInꢀatom ensemble (18.8 sc%) did
result in stronger nitrate binding and higher oxygen abstraction activity. Notably, the In trimer
stabilized either the IS (Figure 8c) or FS (Figure 8d). The In promotion effect was more
pronounced (i.e., a lower activation barrier) when the dissociated O atom was located in the 3ꢀ
fold hollow site in the FS (Figure 8d).
Adding a fourth In atom into this ensemble (25 sc%) resulted in a substantial
improvement, because the IS, TS, and FS were all stabilized simultaneously by In atoms
throughout the reaction path. Increasing the ensemble size to 6 In atoms (37.5 sc%, Figure 8f)
lowered the activation barrier further, but this was associated with a weaker binding of the final
state. The weaker binding of the final state was consistent with increased compressive lateral
th
strain caused by the larger In atoms. In fact, substituting a 7 In atom into the Pd(111) (4×4) unit
cell resulted in sufficient lattice strain to cause surface buckling (43.8 sc%, Figure 8g).
23
ACS Paragon Plus Environment

ACS Catalysis
Page 24 of 40
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 8. Nitrate reduction pathways. (a)ꢀ(h) side and top view of the TS geometry on selected
Pd(111)/In ensembles. (a) 0 sc% (b) 6.3 sc%, one In atom, (c) 18.8 sc% ,In trimer (IS favored),
(d) 18.8 sc%, In timer (FS favored), (e) 25.0 sc% , In 4ꢀmer, (f) 37.5 sc%, In 6ꢀmer, (g) 43.8 sc%,
In 7ꢀmer), and (h) 100 sc%, In monolayer. (i) Free energy diagram for nitrate adsorption and
24
ACS Paragon Plus Environment

Page 25 of 40
ACS Catalysis
1
2
3
4
5
6
7
8
9
oxygen abstraction on the same selected Pd(111)/In ensembles. The nitrate adsorption energy
(
) and the transition state free energy (
for the In 6ꢀmer are marked as example.
In an attempt to lower the compressive strain in the limiting case of 1 ML In on Pd(111),
we used an enlarged (6×6) Pd(111) unit cell and added a ML with only 25 In atoms. The average
InꢀIn distance in the relaxed ML structure became 3.1 Å, but there was still some compressive
strain compared to the InꢀIn bulk distance of 3.35 Å. While this In monolayer model had the
strongest stabilization of the dissociated final state, the calculated dissociation barrier was greater
than in the 4ꢀ or 6ꢀatom ensembles. These results demonstrate that the ensemble size and its
atomic arrangement greatly influence the synergistic effect of InꢀonꢀPd NPs and that an optimal
In sc% can be expected.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
We can generally conclude from the free energy diagram (Figure 8i) corroborates the
notion that In promotes nitrate adsorption, and that In ensembles of 4ꢀ6 atoms have the lowest
activation barrier for NꢀO bond cleavage. Yet, the IS, TS, and FS of nitrate reduction do not
follow a consistent trend with increasing ensemble size and multiple overlaps in the free energy
diagram exist. A direct interpretation of this free energy diagram in terms of nitrate reduction
activity is not straightforward because a lower NꢀO scission barrier may be compensated by
weaker NO * binding, i.e., lower NO * coverage, and vice versa. This motivated us to develop a
3
3
qualitative kinetic model to predict ensemble effects on the nitrate reduction rate.
Kinetic modeling
We used a simple lumped kinetic model to understand the promotional effect of In on Pd for
nitrate reduction leading to the remarkable activity of InꢀonꢀPd catalysts. The kinetic model was
developed to qualitatively explain activity changes with In content, and quantitative agreement
was not expected. In this model, the adsorption of a nitrate ion is quasiꢀequilibrated, followed by
25
ACS Paragon Plus Environment

ACS Catalysis
Page 26 of 40
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
an irreversible NO * reduction to NO * and O*. Since Pd is known to rapidly reduce NO * to N
2
3
2
2
or NH , we assumed that the remaining elementary reaction steps are fast and not rateꢀlimiting.
3
Thus, this model helps us understand how each of these catalyst models perform for the initial
NO * to NO * reduction step. The simplified reaction mechanism is written as:
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
2
(1)
(2)
(3)
(4)
We note that this lumped mechanism fails for active sites that easily oxidized, such that oxygen
ꢀ
removal becomes rateꢀlimiting (step 4). Similarly, the catalyst must be a good NO reduction
2
catalyst for step 3 to be fast. Under the stated assumptions, we derived an analytical expression
ꢀ
for the NO reduction rate. Using the quasiꢀequilibrium approximation for step 1 we obtain
3
.
The forward rate of the irreversible step 2 is then
in which K is the equilibrium constant for reversible NO adsorption calculated from ꢁG =
1
3
1
+
ꢁ
Gads ꢀ ꢁG , k is the forward reaction rate constant for NO * reduction,
is the nitrate
sol
2
3
concentration, and
respectively.
and
are the surface coverages of NO * and vacant sites ‘*’,
3
Finally, we assumed that all active site models have a similar fraction of empty sites,
ꢀ
because the adsorption of NO is highly endothermic as shown in Figure 8, hydrogen binds
3
weakly as we discuss later in Figure 10, and NO * disappears quickly on Pd sites. This leaves O*
2
as the most important surface intermediate to consider, but as previously stated, the simplified
26
ACS Paragon Plus Environment

Page 27 of 40
ACS Catalysis
1
2
3
4
5
6
7
8
9
model assumes that O* removal is not rateꢀlimiting. If the fraction of vacant sites is similar, then
the
term cancels out when we define the final expression for the NO reduction rate r’ relative
3
to Pd(111)
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
.
At this point we note that the constant contribution of ꢁGsol to K on different catalyst models
1
also cancels out and we consider only ꢁGads for the estimation of K . This relative rate of
1
disappearance of nitrate species is shown for all investigated In ensemble models in Figure 9a.
The 6ꢀatom In ensemble, corresponding to 37.5 sc%, outperforms all other tested ensembles and
coincides with the experimentally determined activity peak between 40ꢀ50 sc% (Figure 7). From
the semiꢀlog scale plot in Figure 9b one can more clearly see that In ensembles of 4 or more
atoms (>25 sc%) show a drastically improved nitrate reduction rate (6 orders of magnitude or
more) that extends to 100 sc%. It can also be shown that the effective, lumped rate constant
implies that the apparent activation energy is equal to the transition state energy G =
TS
ꢁ
Gads + E , where ꢁG describes the free energy of nitrate adsorption and E is the activation
a ads a
barrier for nitrate reduction. Thus, the high activity for the 6ꢀatom In ensemble can be attributed
to its low transition state energy G in Figure 8i.
TS
27
ACS Paragon Plus Environment

ACS Catalysis
Page 28 of 40
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 9. Nitrate reduction rate r’ relative to pure Pd(111) as function of In sc% in (a) linear
scale and (b) semiꢀlog scale.
Overall, the theoretical results support the first part of our hypothesis, namely that In
atoms are necessary to improve nitrate adsorption and lower the transition state energy for nitrate
reduction. As discussed in the previous section, the key contribution of In to stabilize the nitrate
abstraction transition state stems from its affinity towards oxygen. This may ultimately prove
detrimental, however, when the In ensembles become so large that they resemble bulk In and
form an oxide phase, rendering our assumption of rapid oxygen removal invalid.
28
ACS Paragon Plus Environment

Page 29 of 40
ACS Catalysis
1
2
3
4
5
6
7
8
9
Removing oxygen from In ensembles
Oxygen removal as water according to reaction (4) requires the activation of dihydrogen and
formation of OꢀH bonds. While Pd(111) terraces readily activate H and bind the resulting H
2
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
atoms favorably with ꢁE = ꢀ0.91 eV for dissociative adsorption, In(001) terraces are hardly
capable of activating H . The activation barrier for H dissociation on In(001) is E = 1.58 eV
2
2
a
and the step is endothermic by ꢁE = 1.33 eV. Similar results are obtained for 1 ML In on Pd(111)
Ea = 1.51 eV, ꢁE = 1.20 eV). Therefore, a plausible mechanism for O* removal from In sites
(
would invoke the bifunctional nature of the InꢀonꢀPd NPs, where Pd can supply H* atoms to In
ensembles to keep them reduced.
Figure 10. Potential energy contour map for H* binding on a static 7ꢀatom In ensemble with O*
in a In 3ꢀfold site (In ꢀO). The dashed and solid circles outline the position of metal atoms and
3
adsorbed oxygen, respectively.
29
ACS Paragon Plus Environment