Amination of 1-hexanol on bimetallic AuPd/TiO2 catalysts

Article abstract of DOI:10.1039/c8gc02321b

AuPd/TiO₂ catalysts, synthesized using controlled surface reactions, are active for the gas-phase amination of 1-hexanol using ammonia. The bimetallic active sites for these catalysts have been characterized using CO chemisorption and XAS techniques, and the absence of monometallic Pd species in the AuPd catalysts was confirmed using UV-vis and STEM-EDS analysis. The bimetallic catalysts exhibit synergy between Au and Pd, as the rate of hexanol conversion increases from 8.7 μmol ks^-1 (μmol total Pd)^-1 over Pd/TiO₂ to up to 42 μmol ks^-1 (μmol total Pd)^-1 over AuPd/TiO₂ with a Pd/Au atomic ratio of 0.06. The rate of hexanol conversion is also enhanced with respect to Au content, with a 5-fold increase in the total Au-normalized rate from Au/TiO₂ to AuPd_0.67/TiO₂. As Pd is added to Au/TiO₂ in increasing quantities, the production rate of primary species (i.e., hexylamine and hexanenitrile) is preferentially increased. The rate of dihexylamine production increases to a lesser extent, while trihexylamine formation remains relatively constant across Pd loadings. Moreover, trihexylamine, which cannot be formed via the condensation of dihexylamine and hexanol, is shown to be produced via the secondary aldimine, N-hexylidene hexylamine. The AuPd bimetallic catalysts also exhibit reduced hydrogenolysis activity compared to monometallic Pd/TiO₂.

Full text of DOI:10.1039/c8gc02321b

View Article Online
View Journal
Green
Chemistry
Accepted Manuscript
This article can be cited before page numbers have been issued, to do this please use: M. Ball, T. S.
Wesley, K. R. Rivera-Dones, G. Huber and J. Dumesic, Green Chem., 2018, DOI: 10.1039/C8GC02321B.
Volume 18 Number
7
7
April 2016 Pages 1821–2242
This is an Accepted Manuscript, which has been through the
Royal Society of Chemistry peer review process and has been
accepted for publication.
Green
Chemistry
Cutting-edge research for a greener sustainable future
www.rsc.org/greenchem
Accepted Manuscripts are published online shortly after
acceptance, before technical editing, formatting and proof reading.
Using this free service, authors can make their results available
to the community, in citable form, before we publish the edited
article. We will replace this Accepted Manuscript with the edited
and formatted Advance Article as soon as it is available.
You can find more information about Accepted Manuscripts in the
author guidelines.
Please note that technical editing may introduce minor changes
to the text and/or graphics, which may alter content. The journal’s
standard Terms & Conditions and the ethical guidelines, outlined
in our author and reviewer resource centre, still apply. In no
event shall the Royal Society of Chemistry be held responsible
for any errors or omissions in this Accepted Manuscript or any
consequences arising from the use of any information it contains.
ISSN 1463-9262
CRITICAL REVIEW
G. Chatel et al.
Heterogeneous catalytic oxidation for lignin valorization into valuable
chemicals: what results? What limitations? What trends?
rsc.li/green-chem

Page 1 of 44
Green Chemistry
View Article Online
DOI: 10.1039/C8GC02321B
Amination of 1ꢀHexanol on Bimetallic AuPd/TiO Catalysts
2
1
†
1†
1
1
Madelyn R. Ball, Thejas S. Wesley, Keishla R. RiveraꢀDones, George W. Huber , and
1
*
James A. Dumesic
1
Department of Chemical and Biological Engineering, University of Wisconsin-Madison, 1415
Engineering Drive, Madison, WI 53706
†
These authors contributed equally to this work.
*
Corresponding Author: jdumesic@wisc.edu
Keywords: Bimetallic catalysts, gold, palladium, titania, amination, Fourier transform infrared
spectroscopy, scanning transmission electron microscopy, Xꢀray absorption spectroscopy

Green Chemistry
Page 2 of 44
View Article Online
DOI: 10.1039/C8GC02321B
Abstract
AuPd/TiO catalysts, synthesized using controlled surface reactions, are active for the gasꢀ
2
phase amination of 1ꢀhexanol using ammonia. The bimetallic active sites for these catalysts have
been characterized using CO chemisorption and XAS techniques, and the absence of
monometallic Pd species in the AuPd catalysts was confirmed using UVꢀvis and STEMꢀEDS
analysis. The bimetallic catalysts exhibit synergy between Au and Pd, as the rate of hexanol
ꢀ1
ꢀ1
ꢀ1
conversion increases from 8.7 µmol ks (µmol total Pd) over Pd/TiO to up to 42 µmol ks
2
ꢀ1
(
µmol total Pd) over AuPd/TiO with a Pd/Au atomic ratio of 0.06. The rate of hexanol
2
conversion is also enhanced with respect to Au content, with a 5ꢀfold increase in the total Auꢀ
normalized rate from Au/TiO to AuPd /TiO . As Pd is added to Au/TiO in increasing
2
0.67
2
2
quantities, the production rate of primary species (i.e., hexylamine and hexanenitrile) is
preferentially increased. The rate of dihexylamine production increases to a lesser extent, while
trihexylamine formation remains relatively constant across Pd loadings. Moreover,
trihexylamine, which cannot be formed via the condensation of dihexylamine and hexanol, is
shown to be produced via the secondary aldimine, Nꢀhexylidene hexylamine. The AuPd
bimetallic catalysts also exhibit reduced hydrogenolysis activity compared to monometallic
Pd/TiO2.
1

Page 3 of 44
Green Chemistry
View Article Online
DOI: 10.1039/C8GC02321B
Introduction
Amines are widely used in the chemical industry in the production of a range of useful
1
–3
chemicals, including pharmaceuticals, solvents, polymers, and dyes. Typical amine syntheses
are carried out using nitrile, carboxylic acid, halide, or carbonyl compounds as feedstocks, and
3
–5
current processes typically involve toxic species and create salt residues as a byproduct. Thus,
there exists motivation for development of more environmentally friendly processes for amine
synthesis. One potentially attractive process for amine synthesis is the direct amination of
alcohols to amines using ammonia. Alcohols, which are widely produced in environmentally
friendly biorefinery processes, can thus be converted to amines with water formed as the only
3
–5
byproduct. Thus, direct amination of alcohols offers a less toxic and more sustainable route for
production of amines. This work focuses on the gasꢀphase amination of an aliphatic alcohol (i.e.,
1
ꢀhexanol) with ammonia via “hydrogen borrowing” chemistry using AuPd/TiO catalysts.
2
“Hydrogen borrowing” generally refers to atomꢀefficient reactions that navigate through
dehydrogenated intermediates. These intermediates are subsequently reduced using hydrogen
temporarily “borrowed” by the catalyst from the reactants, such that no net hydrogen is
6
,7
consumed. In the case of alcohol amination, it is generally viewed that the alcohol is oxidized
to its corresponding ketone or aldehyde, which undergoes nucleophilic attack by the reactant
amine to form water (the only byproduct) and an imine. The imine, in turn, is hydrogenated to
the product amine. This chemistry has been widely implemented in industry, with most
commercial aliphatic amine syntheses relying on Coꢀ, Niꢀ, Cuꢀ, and Zrꢀbased catalysts and
8
–10
ammonia as the aminating agent.
Several studies have investigated heterogeneouslyꢀcatalyzed amination of aliphatic
7
alcohols using ammonia. Shimizu et al. demonstrated the use of Ni/Al O for the amination of
2
3
2

Green Chemistry
Page 4 of 44
View Article Online
DOI: 10.1039/C8GC02321B
various alcohols with NH , achieving more than 68% selectivity to the corresponding primary
3
3
amine. Recently, Dumon et al. studied nꢀoctanol amination using ammonia over Ni and Pd
4
catalysts. Density functional theory (DFT) calculations for the amination of methanol were
carried out on Ni and Pd surfaces, and they found adsorbed ammonia to be key in predicting
reactivity. The use of Ni catalysts for alcohol amination was further investigated by Tomer et al.,
employing a mixed metal oxide support or cyclodextrin complexes to increase conversion and
selectivity to the primary amine by improving the reducibility and dispersion of nickel oxide
2
,11
species.
The role of Pd in amination reactivity is less well understood, and thus we have
studied the impact of Pd structure on catalytic activity. Bimetallic catalysts offer an important
approach to tuning the reactivity of metal catalysts, having exhibited improved or tunable
reactivity, selectivity, or stability for a wide variety of reactions ranging from chemistries such as
1
2–15
16–19
20–25
hydrogenation,
hydrogenolysis,
and hydrogen production,
to more recent efforts
2
6–29
targeting the production of renewable chemicals and fuels via biomass upgrading
and
3
0–33
electrochemical processes.
The effects of bimetallic formation on amination reactions have been investigated in the
literature for selected cases. Recently, Takanashi et al. demonstrated synergy between rhodium
3
4,35
and indium for the amination of 1,2ꢀpropanediol.
The rate promotion was attributed to an
electronic perturbation of rhodium sites by indium that weakens the binding of ammonia.
However, they also observed reduced activity at high indium loadings, and they attributed this
behavior to an ensemble effect impeding the adsorption of propanediol. PtꢀSn/γꢀAl O has also
2
3
been shown to be effective for the conversion of alcohols to secondary and tertiary amines, with
3
6,37
Sn promoting both overall reactivity as well as selectivity to amines over imines.
Here, we use AuPd catalysts to modify the Pd structure with the goal of understanding
3

Page 5 of 44
Green Chemistry
View Article Online
DOI: 10.1039/C8GC02321B
bimetallic effects for amination chemistry. To develop robust design principles for optimizing
multimetallic catalysts, it is necessary to understand the implications of intermetallic bonding on
3
8,39
the properties of catalytic surfaces and, in turn, how these features dictate surface chemistry.
Although supported AuꢀPd catalysts have not yet to our knowledge been applied to the direct
amination of aliphatic alcohols using ammonia, they have been applied to similar reactions. Ono
and Ishida investigated the amination of phenols with ammonia using Pd/Al O ꢀbased catalysts,
2
3
and they observed that addition of Au to Pd improved catalyst stability, with a negligible effect
4
0
on initial massꢀspecific rates. More recently, Sankar, et al. employed titaniaꢀsupported AuꢀPd
and RuꢀPd alloy catalysts for the oneꢀpot synthesis of Nꢀbenzylideneaniline and Nꢀbenzylaniline
from nitrobenzene and benzyl alcohol via initial transfer hydrogenation to form aniline and
4
1
benzaldehyde intermediates. Both alloys were shown to be more active than monometallic Pd
for the conversion of the limiting reagent, nitrobenzene. The increased activity of AuꢀPd
bimetallics was attributed to an electronic effect whereby the rate of benzyl alcohol
dehydrogenation, the suggested rateꢀlimiting step, was increased. The high activity of RuꢀPd
catalysts was hypothesized to arise from a stabilization of Pd by the Ru in small nanoparticles.
It is generally accepted that the effects of bimetallic formation on catalysis may be
divided into two broad categories: electronic (or ligand) effects, and ensemble (or geometric)
3
9,42–47
effects.
Such electronic and ensemble effects have been thoroughly investigated for
4
8
bimetallic AuꢀPd systems. Lam, et al. observed 50ꢀfold higher turnover frequencies (on a Pdꢀ
basis) for AuPd than monometallic Pd for the reaction of molecular hydrogen and oxygen to
form water, and they attributed this result to weakened O binding to Pd when the latter is
4
9
coordinated to Au. However, Davis and Boudart observed negligible changes in ethylene and
cyclohexene hydrogenation turnover frequencies, reaction orders, and apparent activation
4

Green Chemistry
Page 6 of 44
View Article Online
DOI: 10.1039/C8GC02321B
5
0
energies upon the addition of Au to Pd/SiO catalysts. Due to the wellꢀestablished structureꢀ
2
1
9,51
insensitivity of those hydrogenation reactions,
convolution from ensemble effects was
unlikely, and thus the authors concluded that electronic effects were not important for these
5
2
systems. Conversely, for reactions such as acetylene cyclization to benzene and vinyl acetate
4
8,53,54
synthesis,
ensemble effects have been shown to dominate. Still other AuPdꢀcatalyzed
5
5,56
reactions are susceptible to both electronic and ensemble effects, such as CO oxidation
and
5
7–60
the direct synthesis of hydrogen peroxide.
Thus, the importance of electronic versus
ensemble effects is dependent on the surface chemistry. We aim to elucidate the relative
importance of electronic and ensemble effects for 1ꢀhexanol amination using AuPd catalysts.
Here, we report reactivity and selectivity trends for a series of AuPd/TiO catalysts for
2
the gasꢀphase amination of 1ꢀhexanol using ammonia. The catalysts were synthesized using
controlled surface reactions to achieve bimetallic formation, and catalysts were characterized
using CO chemisorption, Fourier transform infrared (FTIR) spectroscopy of adsorbed CO,
scanning transmission electron microscopy (STEM), energyꢀdispersive Xꢀray spectroscopy
(EDS), and Xꢀray absorption spectroscopy (XAS) techniques. We show that the addition of Pd to
Au catalysts initially promotes the formation of primary products, followed by a promotion of
secondary amine formation. Based on reaction kinetics studies and spectroscopic evidence, we
suggest that ensemble effects likely predominate over electronic effects.
Materials and Methods
Catalyst Synthesis
A monometallic, 1 wt% Au/TiO parent catalyst was synthesized using a deposition
2
6
1
precipitation method, as described by Li, et al. to achieve small Au particles. Briefly, TiO2
5

Page 7 of 44
Green Chemistry
View Article Online
DOI: 10.1039/C8GC02321B
(
Degussa P25) was added to deionized (to a resistivity of 18.2 Mꢀ cm) water and stirred for 30
minutes. The pH of this slurry was adjusted to pH 10 using 0.1 M sodium hydroxide (NaOH,
7%, Sigma Aldrich) and then stirred for 90 minutes. A 0.2 M solution of gold chloride hydrate
HAuCl , 99.995%, Sigma Aldrich) containing the desired quantity of gold was added to 15 mL
9
(
4
of 0.01 M NaOH solution and let sit for 30 minutes. The Au solution and 0.1 M NaOH were then
added to the TiO slurry simultaneously such that the pH remained constant at 10. The slurry was
2
stirred for 2 hours, then centrifuged and washed with water. The Au/TiO catalyst was dried in a
2
vacuum desiccator overnight, and then reduced under flowing hydrogen (Industrial Grade,
Airgas) at 623 K for 6 hours (with a heating rate of 1.5 K min) and finally passivated with
flowing 1 mol% O in He (Airgas).
2
Bimetallic AuPd/TiO catalysts were synthesized using controlled surface reactions, a
2
2
4,62–65
method developed in previous work for other multimetallic systems.
A portion of the
Au/TiO parent catalyst was reduced at 623 K under hydrogen flow for 6 hours (heating at 1.5 K
2
ꢀ1
min ) then transferred using Schlenk techniques to a glove box with an inert argon (Ultraꢀhigh
purity, Airgas) atmosphere. A solution of cyclopentadienyl palladium allyl (Cp(Pd)allyl) (98%,
Strem Chemicals) in nꢀpentane (≥99%, Sigma Aldrich) was then added to a Schlenk tube
containing the Au catalyst and the slurry was stirred for 1 h. The remaining pentane was
evaporated using standard Schlenk techniques. Without exposure to air, the catalyst was then
ꢀ1
reduced under flowing hydrogen at 673 K for 6 hours (with a heating rate of 1.5 K min ) and
passivated with flowing 1% O in He. Bimetallic catalysts are designated as AuPd /TiO , where
2
x
2
x is the atomic Pd/Au ratio as measured by inductively coupled plasmaꢀoptical emission
spectroscopy. The AuPd0.06/TiO and AuPd /TiO catalysts were synthesized in one cycle of
2
0.23
2
controlled surface reactions, while the AuPd0.55/TiO and AuPd /TiO catalysts were
2
0.67
2
6

Green Chemistry
Page 8 of 44
View Article Online
DOI: 10.1039/C8GC02321B
synthesized in two cycles. When multiple cycles were used, catalysts were returned to the glove
box after reduction and the synthesis repeated. Equal amounts of Cp(Pd)allyl were used in each
cycle to maintain a low concentration of precursor during synthesis, which, as discussed below,
promotes selective bimetallic formation (as opposed to Pd deposition on the support).
A control experiment was carried out where palladium was introduced to the titania
support alone. TiO was treated at 623 K under hydrogen flow for 6 hours (with a heating rate of
2
ꢀ1
1
.5 K min ) then transferred to an inert atmosphere glovebox. Following the same procedure as
described above, Cp(Pd)allyl was introduced in a nꢀpentane solution and mixed with the TiO2,
then the remaining pentane was evaporated, and the final material was treated in hydrogen.
A monometallic 0.5 wt% Pd/TiO catalyst was synthesized using incipient wetness
2
impregnation to produce a material with similar sized metal particles as the Au/TiO catalyst. An
2
aqueous solution of palladium nitrate hydrate (Pd(NO ) , 40% Pd basis, Sigma Aldrich) was
3
2
added dropwise to TiO and continuously mixed until the incipient wetness point was reached.
2
The catalyst was then dried in air overnight at 383 K, reduced under flowing hydrogen at 533 K
ꢀ1
for 6 hours (with heating rate of 1 K min ), and finally passivated with flowing 1% O in He.
2
Characterization
UVꢀVisible Spectroscopy
UVꢀvisible absorption spectra of cyclopentadienyl Pd allyl in pentane were collected
using a Thermo Scientific Evolution 300 UVꢀvisible spectrometer. Spectra were collected over
wavelengths ranging from 190 to 700 nm using a 1 cm pathꢀlength quartz cuvette to contain the
samples.
CO Chemisorption
7

Page 9 of 44
Green Chemistry
View Article Online
DOI: 10.1039/C8GC02321B
The Pd dispersion was measured by CO (CP Grade, Airgas) chemisorption using a
custom volumetric and gasꢀhandling apparatus. CO was purified by passing over copper turnings
at 503 K to remove metal carbonyls and 4 Å molecular sieves to remove water prior to use.
Catalyst samples were reduced in flowing hydrogen (Industrial Grade, Airgas) at 533 K for 1 h
ꢀ1
(
with a heating rate of 2 K min ) prior to analysis. After reduction, the samples were evacuated
ꢀ5
to 10 Torr and cooled to room temperature to collect the isotherms. CO adsorption was carried
out at 293 K and the number of surface Pd atoms was taken to be equal to the irreversible CO
uptake.
Fourier Transform Infrared Spectroscopy
Infrared spectra of CO adsorbed on catalyst samples were collected using a Nicolet 6700
spectrometer connected to a custom vacuum manifold. Catalyst samples were pressed into selfꢀ
6
6,67
supporting pellets and loaded into a transmission cell described elsewhere.
Samples were
reduced in flowing hydrogen (Industrial Grade, Airgas) at 573 K then cooled to room
ꢀ5
temperature and evacuated to 10 Torr. After a background scan was collected, the cell was
cooled to 223 K using flowing liquid nitrogen. Then, 300 Torr of 1 mol% CO in He (Airgas) was
introduced to the sample and the sample was further cooled to 103 K. Spectra of adsorbed CO
ꢀ1
were collected in the temperature range of 103 to 293 K with a resolution of 4 cm .
Inductively Coupled PlasmaꢀOptical Emission Spectroscopy
The total metal content of catalysts was determined using a Varian VistaꢀMPX CCD
Simultaneous inductively coupled plasmaꢀoptical emission spectrometer (ICPꢀOES).
Calibrations for ICP analysis were obtained by analyzing dilutions of commercial Au and Pd ICP
ꢀ1
standards (Fluka, 1000 mg L ).
Scanning Transmission Electron Microscopy – Energy Dispersive Xꢀray Spectroscopy
8

Green Chemistry
Page 10 of 44
View Article Online
DOI: 10.1039/C8GC02321B
Scanning transmission electron microscopy (STEM) imaging was performed using a FEI
Titan STEM with Cs aberration correction operated at 200 kV in highꢀangle annular dark field
(HAADF) mode. Energy dispersive Xꢀray spectroscopy (EDS) data was collected using the same
microscope with an EDAX SiLi detector. Particle composition distributions of each sample were
determined by EDS analysis of at least 30 particles. Samples were prepared by dropping the
passivated catalyst, suspended in ethanol, on a holey carbon coated Cu grid. The samples were
plasma cleaned before loading in the microscope.
Xꢀray Absorption Spectroscopy
Ex-situ Xꢀray absorption spectroscopy (XAS) measurements were taken at beamline 10ꢀ
ID at the Advanced Photon Source at Argonne National Laboratory. Measurements were taken at
the PdꢀK edge (24350 eV) and the AuꢀL edge (11919 eV). Samples were pressed into wafers
3
inside a stainlessꢀsteel sample holder, which was placed inside a quartz tube with Kapton
windows on each end. The tube was flushed with helium prior to analysis and spectra were
collected in transmission mode. For the catalyst with the lowest Pd loading (Pd/Au = 0.06), a
poor signalꢀtoꢀnoise ratio at the Pd edge in transmission mode prohibited quantitative analysis of
the Pd species. Spectra of Au and Pd foils were collected with each sample to calibrate the
energy. Fitting of spectra was performed using the Demeter software package over an R range of
ꢀ
1
2
1
0
.4ꢀ3.4 Å and a kꢀrange of 3ꢀ10 Å . The amplitude reduction factor was determined to be S =
0
2
.82 for Au and S = 0.86 for Pd by fitting the spectra of the reference foils.
0
Reactivity Studies
Reactivity studies for the gasꢀphase amination of 1ꢀhexanol using ammonia were
conducted in a 0.7ꢀinch OD fixed bed, borosilicate glass flow reactor operated in a downꢀflow
configuration. In each run, the catalyst (typically between 5 and 25 mg) was diluted in 1.5 g of
9

Page 11 of 44
Green Chemistry
View Article Online
DOI: 10.1039/C8GC02321B
silica chips (Silicon Dioxide, fused (granular), 4ꢀ20 mesh, 99.9% trace metals, SigmaꢀAldrich),
which had been crushed and sieved to between 106 and 212 µm to mitigate channeling artifacts.
Prior to each reaction, catalysts were reduced under flowing H (Industrial Grade, Airgas) at 533
2
ꢀ1
K for 1 h with a heating rate of 2 K min . All reactions were conducted at 503 K and 1
atmosphere total pressure with a gas composition of 0.07 kPa hexanol, 2.5 kPa NH , 51 kPa H ,
3
2
and balance He. The high H pressure was required to achieve sufficiently high hexanol
2
conversion. The reactor temperature was read using a Kꢀtype thermocouple seated in a well
(which extended into the catalyst bed) in the wall of the reactor. Temperature control was
achieved using a resistively heated furnace connected to a variable autotransformer supplied by a
selfꢀtuning PID process controller (Series 16A, Love Controls). To prepare the gas mixture, a
3
ꢀ1
flow of 20 cm (STP) min hydrogen was sparged through a 1ꢀhexanol (≥99%, SigmaꢀAldrich)
3
ꢀ1
saturator maintained at 303 K using a heated circulating bath. A flow of 20 cm (STP) min of 5
mol% NH in He (Airgas) was mixed with the hexanolꢀsaturated hydrogen stream in a tee union
3
(heated to avoid hexanol condensation) directly upstream of the reactor. To trap all products, the
reactor effluent was fed to a borosilicate glass trap filled with 1ꢀbutanol (99.9% HPLC Grade,
SigmaꢀAldrich) and immersed in an iceꢀwater bath. Using a simplified variant of the method of
6
8
Jou and Freeman, the butanol was purified by refluxing with NaBH (>98%, SigmaꢀAldrich)
4
(
1.3 g NaBH per L of butanol) at 323 K under bubbling Ar, followed by distillation; the middle
4
fraction was taken for use. To quantify product formation, the reactor was periodically
approximately every 2 h) isolated, during which time the butanolꢀcontaining product trap was
(
replaced with another equivalent one. The butanol solution from the justꢀremoved trap was then
injected on a gas chromatograph (Shimadzu GCꢀ2010) equipped with a flameꢀionization detector
(hereafter referred to as GCꢀFID) and a Zebron ZBꢀ50 column (30 m x 0.25 mm OD,
10

Green Chemistry
Page 12 of 44
View Article Online
DOI: 10.1039/C8GC02321B
Phenomenex). All product identification was substantiated by comparison to analytical standards
and mass spectrometry using a Shimadzu GCMSꢀQP2010 S. Only data reflecting steadyꢀstate
catalytic reactivity are reported. All reactions were carried out at fractional hexanol conversions
ranging from 16ꢀ25%; however, due to the high sensitivity of hexanol conversion to fractionꢀofꢀ
aꢀdegree uncertainty in the saturator temperature, we compute total hexanol conversion rates
based on the appropriate linear combination of production rates of all products observable by
GCꢀFID and GCꢀMS techniques. Additionally, due to the complexity of the reaction network, we
report reactivity in terms of siteꢀtime yields (STY), which are the rates at which products are
formed per defined catalyst site (where different methods are used to count sites, as discussed
below). After each reaction, the reactor was cleaned by soaking in nitric acid to be used again.
Results and Discussion
Selective Deposition of Pd
The loadings of the Au, Pd, and AuPd catalysts are shown in Table 1. Up to a Pd/Au ratio
of about 0.5, the Pd/Au molar ratio as determined by ICP is in agreement with the ratio targeted
during synthesis, indicating that most of the Pd introduced during the synthesis procedure was
taken up by the Au/TiO parent catalyst. This conclusion is corroborated by UVꢀvis spectra of
2
the synthesis solutions, shown in Figure 1 and in the Supplemental Information (Figure S1). The
large decrease in peaks at 260 and 320 nm after mixing the Cp(Pd)allyl solution with the reduced
Au/TiO parent catalyst indicates that essentially all of the Cp(Pd)allyl has been taken up by the
2
Au/TiO (Figure 1a). Furthermore, the spectra in Figure 1b indicate that the Cp(Pd)allyl was not
2
taken up by the TiO support during a control experiment, as shown by no change in the intensity
2
of peaks at 260 and 320 nm. Because the control experiment was performed with 1.5 mg
11

Page 13 of 44
Green Chemistry
View Article Online
DOI: 10.1039/C8GC02321B
Cp(Pd)allyl/g pentane, whereas all catalyst syntheses were carried out with precursor
concentrations of at most 1.0 mg Cp(Pd)allyl/g pentane, we conclude that most of the uptake of
Cp(Pd)allyl onto Au/TiO during synthesis is due to PdꢀAu, rather than PdꢀTiO , interactions.
2
2
When the concentration of precursor was increased to 3.6 mg Cp(Pd)allyl/g pentane for a control
experiment, the precursor was taken up entirely by the TiO support, indicating the importance of
2
maintaining low concentrations during synthesis to achieve selective deposition of Pd onto Au.
Table 1. Compositions of Au/TiO , Pd/TiO , and AuPd/TiO catalysts.
2
2
2
Targeted Atomic Ratio
Pd/Au
ICP Atomic Ratio
Pd/Au
ICP Au
(wt%)
ICP Pd
(wt%)
Catalyst
Au/TiO2
N/A
0.05
0.2
0.5
1
N/A
0.06
0.23
0.55
0.67
N/A
1.09%
1.06%
1.25%
1.09%
1.03%
N/A
N/A
AuPd0.06/TiO2
AuPd0.23/TiO2
0.03%
0.16%
0.31%
0.37%
0.47%
a
AuPd0.55/TiO2
AuPd0.67/TiO2
a
Pd/TiO2
N/A
a
Synthesized in two cycles with equal amounts of Pd deposited during each cycle
The catalysts were also analyzed by STEMꢀEDS to determine particleꢀsize distributions
and qualitative compositional distributions of the individual AuPd nanoparticles. The
composition distributions are shown in Supplemental Figure S2. The average composition
calculated from this EDS analysis indicates a lower Pd content than that determined by ICP. This
discrepancy can be attributed either to damage to the small nanoparticles by the electron beam
during analysis or to large Pd particles not present in the areas observed by STEM. The
likelihood of large monometallic Pd particles will be discussed further below with respect to the
chemisorption and FTIR results. It is important to note, however, that on the sample areas
analyzed by STEMꢀEDS, monometallic Pd particles were not observed and few monometallic
Au particles were observed, indicating formation of AuPd bimetallic nanoparticles.
12

Green Chemistry
Page 14 of 44
View Article Online
DOI: 10.1039/C8GC02321B
(a)
(b)
Figure 1. UVꢀvis spectra of Cp(Pd)allyl in pentane before (―) and after (―) contact with (a) reduced
Au/TiO catalyst during the synthesis of AuPd /TiO and (b) TiO as a control. Precursor concentrations
2
0.23
2
2
were 1.0 and 1.5 mg Cp(Pd)allyl/g pentane for the synthesis (a) and control (b), respectively.
Bimetallic Catalyst Structure
Surface site densities, dispersions, and STEMꢀbased particle sizes of the Au/TiO2,
Pd/TiO , and AuPd/TiO catalysts are shown in Table 2. The Au fractional dispersion, D, was
2
2
estimated from the Au nanoparticle sampleꢀaverage diameter in nm, d, assuming hemispherical
69
particles using D = 1.16/d as described by Bergeret and Gallezot. Surface Pd was quantified
using CO chemisorption assuming that the number of exposed Pd atoms was equal to the
irreversible CO uptake. This stoichiometry was chosen because it has been previously reported
that Pd:CO chemisorption stoichiometries are approximately 1:1 for goldꢀrich supported AuPd
7
0
catalysts. Furthermore, this stoichiometry reflects predominantly linear CO adsorption atop
surface Pd atoms, which is supported by our FTIR results (see below). On monometallic Pd
13

Page 15 of 44
Green Chemistry
View Article Online
DOI: 10.1039/C8GC02321B
7
0
catalysts, however, Pd:CO stoichiometries are known to be higher (up to Pd:CO = 2 or higher)
4
8,55
due to the presence of bridgeꢀbonded CO.
For consistency with the bimetallic AuPd catalysts
here however, we have also elected to use Pd:CO = 1:1 for Pd/TiO because this assumption
2
provides a lower bound on the monometallic Pd site density, and thus an upper bound on the
monometallic Pd siteꢀtime yields and a conservative estimate for the promotional effect of
bimetallic formation.
Table 2. Catalyst site densities, dispersions, and particle sizes.
c
Site Density Dispersion Average Particle Size
Catalyst
ꢀ1
(
ꢁmol g )
(%)
(nm)
a
a
Au/TiO2
34
64
1.8 ± 1.4
3.4 ± 1.7
2.9 ± 1.5
4.2 ± 5.1
3.6 ± 2.4
1.1 ± 0.4
b
b
AuPd0.06/TiO2
AuPd0.23/TiO2
AuPd0.55/TiO2
AuPd0.67/TiO2
Pd/TiO2
0.14
4.9
b
b
1.0
6.6
b
b
2.0
7.0
b
b
2.5
7.3
b
b
7.3
15
a
Au surface site density and dispersion, estimated from the average particle size using D = 1.16/d (Ref
7
2) as determined by STEM.
Pd surface site density and dispersion, measured by CO chemisorption, using a Pd:CO stoichiometry of
b
1
:1.
c
Measured by STEM.
Under the definition of a surface site as determined by CO uptake, it is clear from Table 2
that the Pd in the bimetallic catalysts has a lower dispersion than the Pd in the monometallic Pd
catalyst, and that the Pd dispersion in the bimetallic catalysts modestly decreases as Pd is
increasingly diluted in Au. Because the bimetallic catalysts were reduced at a higher temperature
(673 K) during synthesis than the monometallic Pd catalyst (533 K) to ensure complete
decomposition of the cyclopentadienyl ligands, it was necessary to determine whether the
14

Green Chemistry
Page 16 of 44
View Article Online
DOI: 10.1039/C8GC02321B
reduction in Pd dispersion upon bimetallic formation was caused by strong metalꢀsupport
interactions (SMSI) between the Au or Pd and titania.
Strong metalꢀsupport interactions are reported to cause the migration of metal oxide
7
1
species onto metal particles, potentially blocking metal sites. Indeed, the chemisorptive
capacity of titaniaꢀsupported catalysts decreases upon reduction at temperatures as low as 573
7
2
71
K; however, these effects may be reversed by calcining the catalyst at 673 K. To determine
the impact of SMSI on the AuPd/TiO catalysts, we first calcined a portion of the AuPd0.55/TiO2
2
catalyst at 673 K to remove any TiO species from the surface of the metal nanoparticles, and we
x
subsequently reduced the sample at 533 K, the reduction temperature of the monometallic
Pd/TiO . If SMSI were responsible for the lower Pd dispersion in the bimetallics, then we would
2
expect the irreversible CO uptake on this new sample to exceed the value reported in Table 2;
ꢀ1
however, calcination had the opposite effect: CO uptake decreased to 1.1 ꢁmol g , likely due to
sintering (Supplemental Information Table S1). To circumvent possible effects of sintering, in a
second experiment the monometallic Pd/TiO catalyst was reduced at 673 K—the reduction
2
temperature used during bimetallic synthesis—prior to CO chemisorption, to determine if the
higherꢀtemperature reduction would induce SMSI effects and thus decrease CO uptake on the
ꢀ1
monometallic catalyst. The CO uptake on Pd/TiO after reduction at 673 K was 5.7 ꢁmol g ,
2
2
2% lower than the value reported in Table 2, which could be due to mild SMSI and/or particle
sintering. This reduced CO uptake is consistent with the uptake reduction reported by Bracey and
7
2
Burch in a similar comparison. However, this effect is not sufficient to explain the factorꢀofꢀ
two discrepancy in dispersion between the monometallic Pd and bimetallic AuPd catalysts. We
also note that work by Gubó et al. determined that a Au rich shell formed in AuPd particles with
7
3
a high Au concentration which was less susceptible to SMSI than Pd rich particles. Therefore,
15

Page 17 of 44
Green Chemistry
View Article Online
DOI: 10.1039/C8GC02321B
while modest SMSI effects may occur in our bimetallic catalysts, they are not the primary cause
of the low Pd dispersion in the bimetallics. The low dispersion of the bimetallic catalysts could
be attributed to the presence of some large monometallic Pd particles, which may partially
explain the difference between the EDS and ICP compositions, as discussed above. We note,
however, that the dispersion of the monometallic Pd/TiO catalyst (which has particles less than
2
2
nm in diameter) is higher than that of the bimetallic catalysts. Thus, while some large Pd
particles may be present on the bimetallic catalysts, we do not expect this behavior to be the
main explanation for the apparent low dispersion. We instead hypothesize that the unaccountedꢀ
for Pd is subsurface within the AuPd nanoparticles.
The preferential segregation of Au to the surface of AuPd nanoparticles is wellꢀ
established in the literature, and can be the result of Au having a lower partial molar surface free
4
8,74,75
energy than Pd.
In general, however, the surface concentration of bimetallic AuꢀPd
nanoparticles is dependent on the synthesis method and reaction conditions (e.g., temperature,
5
5,75
gas composition) used.
been observed to be surfaceꢀenriched in Pd.
surface of bimetallic nanoparticles in the presence of CO has been well studied.
For example, there are conditions under which Pdꢀrich particles have
5
3,76
Most notably, the segregation of Pd to the
7
7–79
However,
because of the low Pd content in the catalysts studied here, we expect that most Pd is sufficiently
subsurface as to encounter prohibitively high barriers for diffusion to the nanoparticle surface,
even in the presence of CO. Furthermore, work by Seemala, et al. found that the metal structure
8
0
of bimetallic catalysts was influenced by the TiO support. In CuNi bimetallic particles, the
2
TiO support was found to induce segregation such that Cu was preferentially found on the
2
particle surface while Ni was found at the interface with the support. This type of bimetallic
segregation, induced by the differential interactions of reducible TiO with the two metals, could
2
16

Green Chemistry
Page 18 of 44
View Article Online
DOI: 10.1039/C8GC02321B
explain the low amount of surface Pd (which is more oxophilic than Au) in the AuPd particles.
Other modeling work by Xu, et al. showed that the structure of AuPd particles was dependent on
8
1
the redox properties of the support. They determined that for particles on a TiO support under
2
reducing conditions, Pd preferred to be at the interface of the metal particle and oxide support or
in the core of the particle. Thus, the subsurface Pd observed in the AuPd/TiO catalysts used in
2
this study can be explained based on the relative stabilities of Au and Pd in various particle
configurations. Because of the Pd being primarily subꢀsurface or at the support interface in the
bimetallic catalysts, the Pd which remains on the particle surface has a different structure than
the structure found in monometallic Pd catalysts, as indicated by the FTIR results discussed
below.
(a)
(b)
Figure 2. FTIR spectra of CO adsorbed on catalysts at (a) 283 K and (b) 153 K. The spectra were
collected after introducing 300 Torr of 1 mol% CO in He.
17

Page 19 of 44
Green Chemistry
View Article Online
DOI: 10.1039/C8GC02321B
Results from CO FTIR spectroscopy provide insight into the structure of the surface Pd
species on the bimetallic catalysts. Figure 2 shows spectra taken at (a) 283 K and (b) 153 K after
exposing the reduced catalysts to 300 Torr of 1 mol% CO in He. At low temperature, peaks
corresponding to CO on Pd are not visible in the bimetallic catalysts because the Pd loadings are
low and because the extinction coefficient of CO bound to Pd is much lower than that of CO
8
2,83
ꢀ1
bound to Au.
The peak observed at 2103 cm for all catalysts at 153 K can be assigned to the
7
9
stretching of CO bound linearly to the top of a Au atom (Figure 2b). When the temperature is
ꢀ1
increased to 283 K, this peak shifts to 2112 cm and decreases in intensity as CO desorbs from
7
9,84,85
Au (Figure 2a). This blueꢀshift with decreasing CO coverage has been previously observed
and has been primarily attributed to the fact that CO chemisorption induces coverage dependent
8
5,86
changes to the electronic band structure of Au.
This chemical effect of changing CO
coverage was further shown to dominate over dipolar coupling between coꢀadsorbed CO
8
6,87
88
molecules,
the latter of which increases CO vibrational frequency with increasing coverage.
The changes in this COꢀAu peak are not likely attributable to interaction with Pd,
however, because the Pd/Au ratio is low such that we do not expect Pd to greatly influence bulk
Au electronically. Au atoms that are locally coordinated with Pd are likely to be influenced by
Pd to some extent; however, this electronic interaction is expected to affect a low fraction of Au
atoms and therefore not be represented in the bulk FTIR measurement for Au.
For the monometallic Pd catalyst at room temperature (Figure 2a), we observe a peak at
ꢀ1
53,79,89
ꢀ1
2
090 cm corresponding to linear CO on Pd.
A large, broad peak at 1924 cm with a
,
ꢀ1
shoulder at 1986 cm is also observed, corresponding to CO on threeꢀfold Pd sites and bridge
7
9,90–92
ꢀ1
sites, respectively.
The linear CO peak at 2090 cm for monometallic Pd/TiO is red
2
ꢀ1
shifted to 2076 cm for the AuPd bimetallic catalysts. Likewise, as Pd is increasingly diluted
18

Green Chemistry
Page 20 of 44
View Article Online
DOI: 10.1039/C8GC02321B
ꢀ1
with Au, the bridging CO peak at 1986 cm in the monometallic Pd also shifts to lower
wavenumber. This shift is consistent with the findings of Kugler and Boudart, who studied CO
4
5
adsorption on AuPd/SiO catalysts. Using isotopically labelled CO to distinguish electronic
2
alloying effects from coꢀadsorbate interactions, they concluded that the shift in linearꢀbound CO
predominantly arises from electronic AuꢀPd interactions, while the shift in bridgeꢀbonded CO is
primarily due to coꢀadsorbate interactions. Increasing Pd content is expected to lead to larger
ensembles of contiguous Pd atoms and thus to more extensive CO interactions and dipolar
5
5,88
coupling, which increases the frequency of observed vibrational modes.
Electronic effects may be rationalized by a combined transfer of s and p electron density
from Pd to Au and a concomitant, partially chargeꢀcompensating d electron transfer in the
3
8,48,75,93
reverse direction.
Because metalꢀadsorbate interactions are mediated primarily by the
9
4
metal dꢀstates, as opposed to the spꢀstates, the increased dꢀelectron density in Pd atoms from
Au coordination leads to increased occupancy of molecular orbitals exhibiting antibonding
9
5,96
character with respect to carbon and oxygen.
The CꢀO bond is thus weakened, which we
observe in the FTIR spectra as a redꢀshifted CO vibration. This observation is wellꢀestablished in
4
5,48,92
the literature.
Thus, the electronic effects induced by AuPd bimetallic formation lead to
changes in the adsorption of species onto the catalyst surface.
The relative intensity of the peaks for each of the CO binding modes on Pd provides
information about the surface structure of the AuPd catalysts. The monometallic Pd catalyst
ꢀ1
shows a large, broad peak between 1800ꢀ2000 cm corresponding to the adsorption of CO on
bridge and 3ꢀfold Pd sites. For all the bimetallic catalysts, this peak corresponding to multifold
CO adsorption has a much smaller area, and linear CO adsorption dominates. At low Pd loadings
(Pd/Au = 0.06 and 0.23) multifold adsorption of CO is negligible, while at higher loadings
19

Page 21 of 44
Green Chemistry
View Article Online
DOI: 10.1039/C8GC02321B
(Pd/Au = 0.55 and 0.67) a multifold CO peak is observed. The binding energy of CO on bridge
and 3ꢀfold Pd sites is stronger than the binding energy of CO on top Pd sites and thus CO
9
7,98
adsorption is preferred on the multifold sites when available.
The low area of the multifold
adsorption peak then indicates that most of the surface Pd on the bimetallic catalysts is present in
isolated species surrounded by Au. The high ratio of the isolated Pd peak area to multifold Pd
peak area on the bimetallic catalysts indicates a different geometric surface structure than on the
monometallic Pd catalyst.
It has been reported in the literature that the interaction between Pd surface atoms is low,
and as a result, Pd atoms are not likely to form clusters when no surface adsorbates are
7
7,78
present.
energy of CO on bridge Pd sites rather than top Pd sites. However, the catalysts with Pd/Au =
.06 and 0.23 do not show a multifold CO adsorption peak, which indicates that if the Pd
When CO is present, Pd clusters are likely to form as a result of the stronger binding
7
8
0
concentration is sufficiently low, nearly all of the Pd can be isolated from other Pd atoms. This
isolation may be at least partially explained by the Coulombic PdꢀPd repulsion induced by net
4
8
charge transfer from Pd to Au.
Xꢀray absorption spectroscopy on the reducedꢀandꢀpassivated catalysts also provides
evidence for interactions between Pd and Au in the bimetallic catalysts. Au and Pd speciation in
the reducedꢀandꢀpassivated catalysts was analyzed by Xꢀray absorption near edge spectroscopy
(XANES) using a linear combination analysis, and the results are shown in Table 3. We note that
some of these composition values are smaller than the corresponding error; however, this
uncertainty does not affect the key conclusions from these experiments. In all of the Auꢀ
containing catalysts, Au is present as primarily metallic Au. For the Pd containing catalysts,
however, Pd speciation shifts in favor of metallic Pd as the Pd/Au ratio decreases, ranging from
20

Green Chemistry
Page 22 of 44
View Article Online
DOI: 10.1039/C8GC02321B
0
0
Pd :PdO ≅ 1:1 in the monometallic Pd/TiO catalyst to Pd :PdO ≅ 9:1 for AuPd /TiO . (The
2
0.23
2
low Pd loading in AuPd0.06/TiO precluded this analysis for that catalyst.) The PdO species is
2
characteristic of surface Pd in the passivating layer, which suggests that Pd increasingly favors
migrating subsurface as the Pd loading decreases. This behavior would corroborate the low
dispersions measured by CO chemisorption.
Table 3. Fractional composition of Au and Pd oxidation states in Pd/TiO , AuPd/TiO , and Au/TiO
2
2
2
catalysts, as determined by XANES analysis.
Au L ꢀEdge
Pd KꢀEdge
3
Catalyst
Au/TiO2
Au O3
Au
PdO
Pd
2
0.04
0.00
0.03
0.03
0.04
N/A
0.96
1.00
0.97
0.97
0.96
N/A
N/A
a
N/A
a
AuPd0.06/TiO2
AuPd0.23/TiO2
AuPd0.55/TiO2
AuPd0.67/TiO2
Pd/TiO2
0.10
0.27
0.28
0.54
0.90
0.73
0.72
0.46
a
The Pd loading in the AuPd0.06/TiO catalyst was too low for reliable Pd speciation.
Errors for the linear combination fit are ± 2%
2
b
The EXAFS fits at both the Au and Pd edges are shown in Error! Reference source not
found.. For all the bimetallic catalysts, the PdꢀAu coordination number is above 8 and the PdꢀPd
coordination number is less than 1.6. These values indicate that Pd is well diluted in Au, in
agreement with the results from the CO FTIR spectra. This result also suggests that
monometallic Pd particles make up a negligible portion of the bimetallic catalysts. From the Au
L3ꢀedge EXAFS analysis, we observe an increase in the AuꢀPd coordination number as the Pd
content of the catalysts increases. The monometallic Pd/TiO catalyst has a PdꢀPd coordination
2
number of approximately 5, which is the coordination expected for a small particle and is in
9
9,100
agreement with the average particle size calculated from STEM images.
21

Page 23 of 44
Green Chemistry
View Article Online
DOI: 10.1039/C8GC02321B
Table 4. Exꢀsitu EXAFS fits of the Pd Kꢀedge and Au L ꢀedge for Pd/TiO , AuPd/TiO , and Au/TiO
3
2
2
2
catalysts.
a
b
2
Catalyst
Edge
Scatter Path
Pd − Pd
CN
R (Å)
ꢂσ
E (eV)
R factor
0.006
0
5.4
2.3
2.73
1.92
Pd/TiO2
Pd
0.008
0.008
ꢀ1.2
Pd − O
Pd − Pd
Pd − Au
Pd − O
1.6
11.0
0.4
2.86
2.80
1.97
Pd
13.3
0.015
AuPd0.67/TiO2
Au − Au
Au − Pd
Au − O
Pd − Pd
Pd − Au
Pd − O
9.1
3.1
0.2
1.0
8.6
1.0
10.0
2.2
0.6
1.0
10.8
0.6
9.9
0.7
0.5
10.0
0.6
0.4
8.8
2.85
2.77
1.97
2.75
2.82
2.01
2.85
2.80
1.98
2.63
2.81
2.04
2.84
2.82
1.90
2.82
2.85
1.91
2.85
Au
0.008
0.008
0.008
0.008
0.008
8.5
13.4
7.5
0.039
0.002
0.051
0.057
0.061
Pd
AuPd0.55/TiO2
Au − Au
Au − Pd
Au − O
Pd − Pd
Pd − Au
Pd − O
Au
Pd
12.8
5.1
AuPd0.23/TiO2
Au − Au
Au − Pd
Au − O
Au − Au
Au − Pd
Au − O
Au − Au
Au
c
AuPd0.06/TiO2
Au/TiO2
Au
Au
0.008
0.008
4.0
5.4
0.035
0.015
22

Green Chemistry
Page 24 of 44
View Article Online
DOI: 10.1039/C8GC02321B
Au − O
0.2
2.06
a
b
c
Coordination number, estimated uncertainty ± 20%
Distance between the absorber and backꢀscatterer, estimated uncertainty ± 0.02 Å
Data too noisy to obtain quality fits for the Pd edge
Reactivity Trends
Scheme 1 shows a representation of the reaction network for the hydrogen borrowingꢀ
6
,7
mediated amination of 1ꢀhexanol using ammonia.
We were able to identify hexane,
hexylamine, hexanenitrile, Nꢀhexylidene hexylamine (the secondary aldimine), dihexylamine,
and trihexylamine as products by GCꢀFID and GCꢀMS. The aldehyde (hexanal) and primary
aldimine (hexylimine) intermediates were not observed by GCꢀFID or GCꢀMS techniques. We
attribute their absence to their high reactivity and the large molar excess of ammonia
(
NH /hexanol = 36) used: under these conditions, any hexanal formed would react with ammonia
3
on the catalyst surface to form the 1° aldimine since aldehydeꢀaldimine interconversion is
typically equilibrated. The reactive 1° aldimine would then be converted to more stable products.
Gaussianꢀbased thermochemical calculations of the twoꢀcarbon analogs show that both
hexanenitrile and hexylamine are favored at equilibrium (see Supplemental Information Table
S2).
23

Page 25 of 44
Green Chemistry
View Article Online
DOI: 10.1039/C8GC02321B
Scheme 1. Reaction network for the amination of 1ꢀhexanol using ammonia. Boxed species were
identifiable (by GCꢀFID and GCꢀMS) and quantified in this work.
Hexanol conversion rates based on product formation at 16ꢀ25% hexanol conversion are
shown versus timeꢀonꢀstream for the various catalysts in Supplementary Figure S3. From the
timeꢀonꢀstream data, it is apparent that most of the catalysts are stable over the time periods
studied, typically about 8 hours. We compare catalysts using measurements after about four
hours onꢀstream, after which point catalyst deactivation was minimal. Although ammonia and
hydrogen bind more strongly to Pd than hexanol and could poison those sites, the stability of the
catalyst over at least 8 hours indicates that sufficient vacant sites are available at steady state for
hexanol binding.
Comparisons of the hexanol conversion rates and siteꢀtime yields (based on product
formation) over the various catalysts are shown in Table 5. The rate of hexanol conversion
ꢀ1
ꢀ1
ꢀ1
increases from 8.7 ꢁmol ks (µmol total Pd) on Pd/TiO to between 12 and 42 ꢁmol ks (µmol
2
ꢀ1
total Pd) on the bimetallic catalysts. The highest rate (per total Pd) is observed on the
AuPd0.06/TiO catalyst, which exhibits a rate nearly 5 times higher than that over Pd/TiO .
2
2
24

Green Chemistry
Page 26 of 44
View Article Online
DOI: 10.1039/C8GC02321B
Table 5. Rate of hexanol conversion and siteꢀtime yields (STY) for hexanol amination. The hexanol
conversion rate is calculated as the sum of product formation rates, weighted by the number of moles of
hexanol needed to form one mole of product. Conditions: 503 K, 1 atm (0.07 kPa hexanol, 2.5 kPa NH3,
5
1 kPa H , and balance He), 16ꢀ25% hexanol conversion.
2
Rate /
Rate /
Rate /
STY /
STY /
ꢀ1
ꢀ1
ꢀ1
ꢀ1
ꢀ1
Catalyst
µmol ks (µmol µmol ks (µmol µmol ks (µmol
ks
ks
ꢀ1
ꢀ1
ꢀ1
a
b
total Pd)
total Au)
3.0
total metal)
[Pd sites]
[total surface metal]
Au/TiO2
N/A
42
3.0
2.1
4.3
7.3
4.7
8.7
N/A
1200
340
340
300
77
5.6
9
AuPd0.06/TiO2
AuPd0.23/TiO2
AuPd0.55/TiO2
AuPd0.67/TiO2
2.2
22
5.3
13
45
45
12
21
11
12
7.9
Pd/TiO2
8.7
N/A
a
STY for AuPd and Pd catalysts calculated from surface Pd sites determined by CO chemisorption.
Total surface metal site density to calculate STY was calculated using the average particle size as determined by
b
STEM. For the Au and AuPd catalysts dispersion was calculated using D=1.16/d and for the Pd catalyst, D=1.11/d
was used. (Ref. 72)
25

Page 27 of 44
Green Chemistry
View Article Online
DOI: 10.1039/C8GC02321B
To determine the reactivity of Au in the conversion of hexanol over the bimetallic
catalysts, we also calculate rates with respect to the Au metal content. The introduction of Pd to
ꢀ1
the Au parent increases the rate of hexanol conversion slightly from 3.0 ꢁmol ks (µmol total
ꢀ1
ꢀ1
ꢀ1
Au) on Au/TiO to 5.3 ꢁmol ks (µmol Au) on AuPd /TiO . A further rate enhancement to
2
0.23
2
ꢀ1
ꢀ1
1
1 ꢁmol ks (µmol Au) is observed as the Pd loading increases to Pd/Au = 0.55. The rate
ꢀ1
ꢀ1
decreases slightly to 7.9 µmol ks (µmol Au) as the Pd/Au ratio is further increased to 0.67 and
this trend is attributed to the higher fraction of contiguous Pd ensembles, discussed in more detail
below. Thus, independently of whether the rate is normalized by total Au or Pd content, we
observe an increase in activity for bimetallic catalysts over monometallic catalysts. Since the Au
parent catalyst is identical for all the bimetallic catalysts, the differences in reactivity as the Pd
loading changes must be attributed to changes in either the intrinsic activity of Pd or to changes
in Au activity because of interaction with Pd. We note that when the rate is normalized by the
total metal content, rather than surface metal, the highest rate is observed for the Pd/TiO2
catalyst. This result can be attributed to the higher dispersion for this catalyst – a higher fraction
of the active species is accessible.
The above results suggest that AuPd bimetallic formation creates new catalytic sites that
are more active for hexanol amination than either monometallic Au or Pd sites. We note that
bimetallic moieties exhibiting favorable electronic interactions and/or ensemble effects comprise
the active sites regardless of whether, mechanistically, Au promotes Pd catalysis, Pd promotes
Au catalysis, or both metals directly participate in the chemistry.
The coordination numbers as determined by EXAFS indicate that Pd is highly
coordinated to Au, while the Au coordination to Pd is much lower. This stems from the low
Pd/Au ratio, and thus each surface Pd likely exists in a bimetallic ensemble, while only a fraction
26

Green Chemistry
Page 28 of 44
View Article Online
DOI: 10.1039/C8GC02321B
of the surface Au is part of a bimetallic site. Thus, we expect that the number of surface
bimetallic active sites will be proportional to the number of surface Pd sites. To calculate siteꢀ
time yields (for Pdꢀcontaining catalysts), we may normalize the rates by the number of Pd
surface species as measured by CO chemisorption (Table 5). Based on Pd sites, the hexanol
conversion STY over AuPd0.06/TiO exceeds that of Pd/TiO by two ordersꢀofꢀmagnitude, from
2
2
ꢀ1
ꢀ1
7
7 ks to 1200 ks . At intermediate Pd/Au ratios ranging from 0.23 to 0.67, the STY promotion
is more modest.
Siteꢀtime yields can also be calculated using site densities of total surface metal, as
determined from the STEM average particle size, shown in Figure 3. For all Au containing
catalysts, the total number of surface metal sites can be approximated using sites calculated by
the catalyst particle size determined from STEM using D = 1.16/d. Similarly, the total number of
surface metal sites for the Pd/TiO catalyst was calculated from the Pd/TiO particle size using
2
2
6
9
D=1.11/d. By this normalization, the hexanol conversion STY increases as the Pd loading in
ꢀ1
the bimetallic catalysts increases. The highest STY of 45 ks is observed over the AuPd0.55/TiO2
and AuPd0.67/TiO catalysts, with totalꢀsurface STYs that are 5.6 and 3.8 times higher than that
2
of Au/TiO and Pd/TiO , respectively. We note that the standard deviation for the average
2
2
particle size is higher on the AuPd catalysts than the monometallic Au catalyst, and thus the error
in the calculated STY is also increased for the bimetallic catalysts.
27

Page 29 of 44
Green Chemistry
View Article Online
DOI: 10.1039/C8GC02321B
Figure 3. Totalꢀsurface siteꢀtime yield (STY) of hexanol conversion as a function of Pd and Au mol
fraction. The STY values were calculated using total surface metal as estimated by the average particle
size of each catalyst determined by STEM. Errors bars represent the STY calculated using sites from the
average particle size ± standard deviation. Conditions: 503 K, 1 atm (0.07 kPa hexanol, 2.5 kPa NH , 51
3
kPa H , and balance He), 16ꢀ25% hexanol conversion.
2
Based on chemisorption and FTIR results, we know that the number of accessible Pd
sites, which we assume to be bimetallic sites, increases with increasing Pd loading. Thus, the
increase in STY when a higher fraction of the total surface sites are bimetallic sites, rather than
Au sites, corroborates our conclusion that the bimetallic AuPd sites are responsible for enhanced
activity. This result is in agreement with previously reported AuꢀPd synergy for the AuPd/TiO2ꢀ
4
1
catalyzed hydrogenꢀborrowing amination of alcohols using nitrobenzene.
Figure 4a shows the siteꢀtime yields for the formation of each product observed for the
28