Novel rhodium on carbon catalysts for the oxidation of benzyl alcohol to benzaldehyde: A study of the modification of metal/support interactions by acid pre-treatments

Article abstract of DOI:10.1016/j.apcata.2018.11.006

Rhodium nanoparticles or rhodium organometallic complexes are mainly used in catalysis for reduction or hydroformylation reactions. In this work instead, we explored the capabilities of Rh nanoparticles as an oxidation catalyst, applied to the oxidation o

Full text of DOI:10.1016/j.apcata.2018.11.006

Accepted Manuscript
Title: Novel rhodium on carbon catalysts for the oxidation of
benzyl alcohol to benzaldehyde: A study of the modification
of metal/support interactions by acid pre-treatments
Authors: Conor A. Wilde, Yulia Ryabenkova, Ian M. Firth,
Liam Pratt, James Railton, Mariela Bravo-Sanchez, Naoko
Sano, Peter J. Cumpson, Phil D. Coates, Xi Liu, Marco Conte
PII:
S0926-860X(18)30553-2
DOI:
Reference:
https://doi.org/10.1016/j.apcata.2018.11.006
APCATA 16875
To appear in:
Applied Catalysis A: General
Received date:
Revised date:
Accepted date:
5 September 2018
29 October 2018
8 November 2018
Please cite this article as: Wilde CA, Ryabenkova Y, Firth IM, Pratt L, Railton J, Bravo-
Sanchez M, Sano N, Cumpson PJ, Coates PD, Liu X, Conte M, Novel rhodium on
carbon catalysts for the oxidation of benzyl alcohol to benzaldehyde: A study of the
modification of metal/support interactions by acid pre-treatments, Applied Catalysis A,
General (2018), https://doi.org/10.1016/j.apcata.2018.11.006
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.

Novel rhodium on carbon catalysts for the oxidation of benzyl alcohol to benzaldehyde:
a study of the modification of metal/support interactions by acid pre-treatments
Conor A. Wilde,a Yulia Ryabenkova,b Ian M. Firth,a Liam Pratt,a James Railton,a Mariela
Bravo-Sanchez,c Naoko Sano,c,d Peter J. Cumpson,c Phil D. Coates,b Xi Liu,e and Marco
Conte a*
aDepartment of Chemistry, University of Sheffield, Sheffield, S3 7HF, UK
bSchool of Engineering, Polymer IRC, University of Bradford, Bradford, BD7 1DP, UK
cNEXUS, School of Mechanical & Systems Engineering, Newcastle University, Newcastle
upon Tyne, NE1 7RU, UK
dNara Women’s University, Nisi-machi, Kita Uoya, Nara, 630-8506, Japan
eSyncat@Beijing, Synfuels China Co. Ltd., Huairou, Beijing, 101407, P. R. China
*corresponding author E-mail: m.conte@sheffield.ac.uk
Graphical abstract
Highlights
 Study of the properties of novel Rh on carbon catalysts for alcohol oxidation under very
mild conditions and exploitation of this metal for oxidation reactions.
 A pre-treatment of the carbon support with HNO3 was reducing Rh particle size, whereas a
pre-treatment with HCl was either increasing particle size or had no effect.
 Carbon supports pre-treated with HNO3 led to enhanced catalytic activity.
1

 An upper diameter limit for the activity of Rh nanoparticles for benzyl alcohol oxidation
was estimated to be ca. 30 nm.
Abstract
Rhodium nanoparticles or rhodium organometallic complexes are mainly used in catalysis for
reduction or hydroformylation reactions. In this work instead, we explored the capabilities of
Rh nanoparticles as an oxidation catalyst, applied to the oxidation of benzyl alcohol to
o
benzaldehyde under very mild conditions (100 C, and atmospheric pressure) as a model
reaction. Here we report the preparation of novel Rh/C catalysts by using an impregnation
protocol, with particular emphasis on the pre-treatment of the carbon supports by using HNO3
and HCl, as well as the characterization of these materials by using an array of methods
involving TEM, XPS and XRPD. Our preparation method led to a wide Rh particle size
distribution ranging from 20 to 100 nm, and we estimate an upper limit diameter of Rh
nanoparticles for their activity towards benzyl alcohol oxidation to be ca. 30 nm.
Furthermore, a HNO3 pre-treatment of the activated carbon support was able to induce a
smaller and narrower particle size distribution of Rh nanoparticles, whereas a HCl pre-
treatment had no effect or sintered the Rh nanoparticles. We rationalise these results by
HNO3 as an acid able to create new nucleation sites for Rh on the carbon surface, with the
final effect of smaller nanoparticles, whereas for HCl the effect of sintering was most likely
due to site blocking of the nucleation sites over the carbon surface. The roles of acid centres
on the carbon surfaces for the oxidation reaction was also investigated, and the larger their
amounts the larger the amounts of by-products. However, by treatment with HNO3 we were
able to convert neutral or basic carbons into supports capable to enhance the catalytic activity
of Rh, and yet minimised detrimental effects on the selectivity of the oxidation to
benzaldehyde.
Keywords: Rhodium, nanoparticles, activated carbons, acid pre-treatment, alcohol oxidation.
Introduction
2

Rhodium is primarily used in catalysis for reduction or hydroformylation reactions [1, 2].
Very important applications of this metal include the reduction of NOx to N2 and O2 in
catalytic converters [3], or the synthesis of acetic acid from methanol, CO and H2 [4]. In a
recent study though [5] we demonstrated that polymer incarcerated Rh nanoparticles can be
an efficient oxidation catalyst for the oxidation of aryl alcohols. In particular, a strong
catalytic activity was observed in the presence of a biphasic system containing water, in
which case Rh can be an oxidation catalyst competitive to more common catalytic systems
comprising metals like Au, Pd and Pt [6]. This background prompted us to further investigate
the potential of Rh as an oxidation catalyst for alcohol, and especially considering catalysts
capable to oxidise alcohols in the absence of a base and without the use of pressurised
systems [7]. Furthermore we were also interested to investigate the properties of Rh based
catalysts by preparing Rh based nanoparticles supported on activated carbon, using an
impregnation protocol [8]. This is a frequent choice in catalysis with a large number of
applications ranging from alcohol oxidation [9] to hydrocarbon oxidation [10], as well as
halogenation [11] and coupling reactions [12]. The choice of this preparation method is often
justified by being a relatively facile preparation protocol and its apparent statistical
robustness [13]; whereas the choice of activated carbon as supports is often justified by their
high surface area, which together with their structural features - like the presence of phenolic
or acid groups - can either reduce an impregnated metal to metal nanoparticles or stabilise
high oxidation states at the same time [14]. In addition, surface properties of activated carbon
may be altered by treatment with acids [15, 16], and we deliberately used this approach to
change the particle size distribution of Rh nanoparticles with the aim to gain structure-
activity correlations.
In this context, we were interested to explore the capability of Rh at a fundamental level as an
oxidation catalyst - that is by reverting its catalytic activity from a reducing catalyst [17, 18]
to an oxidizing species - by using activated carbon of different characteristics as supports for
the oxidation of benzyl alcohol to benzaldehyde in the presence of toluene as a solvent. The
selection of this reaction as a model system for our study is twofold: alcohol oxidation is one
of the most relevant oxidation reactions for the manufacture of fine chemicals [19, 20], and
for this class of reactions the oxidation of benzyl alcohol is among the most investigated and
understood [21, 22]. As such, this is an excellent benchmark for the investigation of the
capabilities of novel materials like those presented and discussed in this study.
In fact, we aim to identify structure-activity correlations for Rh to be used in catalysis as an
oxidation catalyst, and to identify particle size ranges and metal support interactions that will
3

be useful for catalyst development. To the best of our knowledge no information is present in
the literature up to which particle size limit Rh may be active as an oxidation catalyst, and
little is known [23] about the interactions that Rh centres may have with widely used
activated carbons as supports, or the effects that these interactions may have in the oxidation
process.
In view of these considerations, we foresee that this work may have broader applications than
those focused in this study, by expanding the uses of this metal to novel applications for
oxidation reactions and in turn developing the toolkit of nanoparticles available for the
synthesis of fine chemicals.
1. Experimental
2.1 Catalyst preparation
All the carbon-supported rhodium catalysts (Rh/C, 1 wt% of Rh content) were prepared using
an incipient wetness impregnation protocol [24], using water as a solvent for the dissolution
and impregnation of a RhCl3xH2O precursor (Sigma Aldrich, assay 38 wt%) into the carbon
matrices.
Three different kinds of activated carbon were used in this study: Darco-12, Norit GAC 1240
and Norit SA2 (all supplied by Sigma Aldrich and used without any further purification
unless otherwise specified). For each of these carbons the macroscopic impregnation volume
was preliminary determined by means of calibration curves reporting volume of absorbed
water needed to fill the carbon pores against the mass of carbon. These volumes were: 0.90
1
1
1
mL g , 0.89 mL g and 0.82 mL g for Darco-12, Norit GAC 1240 and Norit SA2
respectively. Solutions made of these volumes containing 53 mg of metal precursor were
added dropwise with stirring at room temperature to the activated carbons (1.98 g) in order to
obtain 2 g batches of catalysts with a final metal loading of 1 wt.%. Each of these products
1
were then dried in static air at 150 oC (temperature ramp 20 oC min ) for 16 h, and used as a
catalyst. Control samples in the absence of Rh but comprising acid pre-treated carbons were
prepared in an identical manner to the impregnated samples, but without initially adding the
RhCl3xH2O precursor.
The acid pre-treatment of the carbon supports was carried out preparing batches of 5 g of
carbon, which were impregnated with HNO3 (Fisher 68%, 15 M), or HCl (Fisher 32%, 12M),
by using incipient wetness impregnation volumes determined as described above. For carbons
pre-treated with HNO3, the stirring was continued at ambient temperature until NOx
4

production subsided, for approximately 20-30 min. Each carbon was dried in static air at 150
o
1
oC (temperature ramp 20 C min ) for 16 h [24, 25]. On these pre-treated carbons, incipient
wetness impregnation of Rh was carried out. In view of the catalyst preparation procedure
used, no filtration of the carbon or catalyst washing was carried out, unless otherwise
specified, and the metal loading should be considered as equal to the nominal amount of
metal impregnated into the support [26].
Using this procedure, nine different Rh/C catalysts were obtained, (three per each activated
carbon), and with each activated carbon comprising: a non pre-treated sample, a sample pre-
treated with HNO3 and a sample pre-treated with HCl). Nine different activated supports
were prepared in the same manner but without the addition of Rh, and used for control tests.
In this work we use the notation: raw activated carbon (RAC), for supports that have not been
treated with any acid, and denoting Darco-12, Norit GAC 1240 and Norit SA2 as RAC1,
RAC2 and RAC3 respectively. Activated carbons pre-treated with HNO3 or HCl are denoted
as NAC1, NAC2, NAC3 and HAC1, HAC2, HAC3 respectively. All rhodium containing
catalysts use this same notation, preceded by the prefix Rh/ to denote the presence of this
metal. For example, a Rh containing catalyst prepared using an activated carbon Darco-12
pre-treated with HNO3 is: Rh/NAC1.
1.2 Catalyst characterization
1.2.1 Transmission electron microscopy
Samples were prepared for transmission electron microscopy (TEM) analysis by dispersing
and sonicating the catalyst powders in high purity ethanol for ca. 10 min, and allowing a drop
of the suspension to dry on a lacey-carbon film supported on a 300-mesh Cu TEM grid.
Bright-field (BF) images were acquired using a Tecnai G20 TEM microscope operating at
200 keV and equipped with a LaB6 electron gun. The frequency count for the particle size
distribution was obtained from a set of 200 particles for each catalyst. Data analysis and
fitting of the particles size distributions by Gaussian and log-normal models was carried out
by using OriginPro 2017 software.
1.2.2 X-ray photoelectron spectroscopy
X-ray photoelectron spectroscopy (XPS) was performed with a Kratos Axis Nova
spectrometer using a monochromatised AlK X-ray source (225 W) with an analyser pass
energy of 160 eV for survey scans and 20 eV for high resolution scans. Three positions per
5

sample were analysed using charge neutralization. All XPS spectra were charge corrected by
setting the C1s C-C/H component to 284.8 eV [14].
1.2.3 X-ray powder diffraction
X-ray powder diffraction (XRPD) patterns were acquired using a Bruker D8 Advance
equipped with a LynxEye detector. The samples were sieved and deposited over an
amorphous silicon sample holder. The instrument was operating at 40 kV and 40 mA
selecting the CuK radiation as X-ray source. The samples were analysed in the range 10o to
80o 2 for a scan time of 71 min.
1.2.4 Acid-base properties of the carbons
Each carbon support RAC1, RAC2 and RAC3 (100 mg) was crushed with a mortar and
pestle, dispersed in deionised water (10 mL) and stirred at 700 rpm for 10 min. pH values of
the resultant solutions [27] were recorded upon stabilization using an Accumet AB150 pH-
meter equipped with a FB67978 ion selective electrode.
1.2.5 Inductive coupled plasma – mass spectrometry
Determination of Fe and Cu content in untreated carbon supports was carried out via ICP-MS
analysis. Each untreated carbon, RAC1, RAC2 and RAC3, (1 g) was dispersed in HCl (12 M,
10 mL) under stirring (48 h, room temperature). The resulting suspension was then filtered,
and known aliquots of the filtrated solution were analysed using an Agilent 7500CE ICP-MS
instrument which was calibrated up to 10 parts per billion (ppb) using standards prepared by
dilution from stock solutions containing 1000 parts per million (ppm) of Fe or Cu standards.
The concentrations of Fe and Cu in the samples were calculated against a calibration graph.
1.2.6 Inductive coupled plasma – optical emission spectroscopy
A full scan analysis for elemental composition (but C, H and O), was carried out on aqueous
extracts from activated carbons by using ICP-OES. Each untreated carbon, RAC1, RAC2 and
RAC3, (1 g) was dispersed in HCl (1 M, 5 mL) under stirring (24 h, room temperature). The
resulting suspension was then filtered, and the reaming carbon solid was washed 20 times
with 5 mL aliquots of deionised water. The resulting liquid filtrate was analysed using a
Spectro-Ciros-Vision ICP-OES instrument, which is calibrated for 70 elements. These were
are measured on at least two emission lines so any interelement interferences can be avoided.
6

1
The plasma conditions were: power 1400 W, coolant flow 12.0 L·min , auxiliary flow 1.0
1
1
L·min , and nebuliser flow 0.85 L·min .
1.3 Catalytic tests
The catalyst (200 mg, Rh 1 wt%), was dispersed in toluene (Fluka, 2 mL) and benzyl alcohol
(Sigma-Aldrich, 210 mg, 1.94 mmol) solution, in order to obtain a molar metal (Rh) to
substrate (benzyl alcohol) ratio, M:S, of 1:100 for each catalyst. For the tests using activated
carbon only, a mass of carbon equal to that present in the Rh/C catalysts minus the amount of
o
Rh was used. The reaction mixture was heated using a reflux condenser at 100 C for 24
hours with a magnetic stirrer operating at 700 rpm at atmospheric pressure in air.
2.4 Characterization of the reaction mixtures
2.4.1 Nuclear magnetic resonance
Analysis of the reaction mixture to determine product selectivity and conversion was
obtained via 1H NMR spectroscopy using a Bruker Avance IIIHD 400 spectrometer operating
at 400 MHz. NMR spectra were collected using CDCl3 as solvent. Chemical shifts were
reported in parts per million (ppm) from tetramethylsilane using the methyl group of toluene
1
resonance as the internal standard (toluene CH3, s, δ: 2.36 ppm) for H NMR [28]. All
conversions and selectivity values are expressed in mol %, and all conversion values are
based on NMR data (see Appendix A, Supplementary Data File, section S3), on the observed
products. In fact, GC/MS characterizations (vide infra), revealed only traces amounts of by-
products < 1% which are due to condensation reactions from isomers of methyl-
diphenylmethane. Due to their very low amount these could be neglected in our selectivity
calculations by normalizing all the selectivites to the presence of: benzaldehyde, benzyl ether
and methyl-diphenylmethane isomers. Functional groups for quantification of species in the
reaction mixture and comparison with standard compounds of benzyl alcohol, benzaldehyde,
dibenzyl ether, 4-methyl-diphenylmethane and benzoic acid are reported in detail in
supplementary data (Figs. S1-S4), as well as a description of the formulas used to calculate
conversion and selectivity (eqs. S1-S5).
2.4.2. Gas chromatography – mass spectrometry
Gas chromatography mass spectrometry analysis was used for the characterization of the
reaction mixtures and analysis of traces. GC/MS was carried out using a Perkin Elmer
7

Turbomass GC-MS, equipped with a Phenomenex Zebron ZB-5MS column 30 m  0.25 mm,
1
0.25 m film thickness. The carrier gas was helium at 1 mL min , the injection volume was
1.0 L, using an injector temperature of 250 oC, and a temperature programme from 60 oC to
1
260 oC using a ramp at 10 oC min and then hold for 10 minutes.
2.5. Computational studies
Quantum chemical calculations were performed using Gaussian 09, version D.01, [29]
running on a 32 processor (Viglen) 64-bit computational Linux cluster. Mixed basis sets
coupled with density functional theory (DFT) methods and the effective core potential were
used. In detail, mixed basis sets SDD (Stuttgart-Dresden triple ζ incorporating effective core
potentials) for the transition metal Rh [30], and 6-311G(d,p) (Pople triple split valence
Gaussian basis functions with added polarization) for all other non-transition-metal atoms
have been employed [31], using the B3LYP functional of DFT [32]. The nature of stationary
states (minimum structures) was confirmed by frequency calculations. Stabilization energies
of the complexes were obtained by subtraction from energetic values of isolated species, and
considering zero point energy (ZPE) corrections [33] for all of the molecular structures.
2. Results and discussion
3.1 Catalytic activity of Rh/C catalysts prepared on untreated activated carbon
supports
Rhodium supported nanoparticles (Rh 1 wt%) were prepared over activated carbons by using
three among the most common activated carbons used in catalysis, namely: Darco-12 [34],
Norit GAC 1240 [35] and Norit SA2 [36]. These are used for their large surface area,
acid/base properties and granulometry and they are here abbreviated for simplicity as RAC1,
RAC2 and RAC3 respectively. These carbons were initially used untreated as supports for Rh
nanoparticles. The resultant Rh/C catalysts, as well as the carbon supports only as control
tests, were tested for the oxidation of benzyl alcohol to benzaldehyde as a model reaction
(Table 1), and under very mild conditions (T = 100 oC, P = atmospheric pressure of air). This,
with the aim to determine the catalytic capabilities of Rh species for oxidation reactions, by
using a facile impregnation protocol.
All the Rh/RAC catalysts are able to oxidise benzyl alcohol to benzyl aldehyde (Scheme 1(a),
and Figs. S5-S7), although to a different extent. The most active catalyst, in terms of
8

conversion, is Rh/RAC1 (52%) with the trend: Rh/RAC1 > Rh/RAC2 > Rh/RAC3. The
selectivity to benzaldehyde follows an opposite trend instead: Rh/RAC1 < Rh/RAC2 <
Rh/RAC3, with a selectivity up to 100% for the latter catalyst. In contrast, when Rh/RAC1 is
used, products like dibenzyl ether and methyl-diphenylmethane are detected (Fig. S5).
Whereas the presence of dibenzyl ether may be related to condensation of two molecules of
benzyl alcohol with de-hydration (Scheme 1(b)) [37], the presence of methyl-
diphenylmethane as a by-product (Scheme 1(c)) is rather surprising, as this compound is
usually obtained via benzylation of toluene with benzyl alcohol over strongly acidic zeolites
[38], and not by using activated carbons. Nevertheless, the presence of this by-product was
confirmed via GC-MS and comparison with standards (Figs. S8 and S9), with the
identification of two isomers (2- and 4-methyl-diphenylmethane).
These four compounds: benzyl alcohol, benzaldehyde, dibenzyl ether and methyl-
diphenylmethane are also the major components of all of our reaction mixtures with a carbon
mass balance > 95% in all cases. This was obtained by calibration of the signal of toluene
used as internal standard (see supplementary data, eqs. S6-S11), thus confirming these are the
only products of our reaction.
In this context, it does worth to highlight that we did not detect any formation of benzoic acid
in reaction mixtures using catalysts prepared using the untreated supports, nor in reaction
mixtures from catalytic tests using the catalysts prepared via acid pre-treatment. As benzoic
acid is obtained from benzaldehyde via a free-radical pathway involving the formation of
acyl radicals which evolve to peracids [39], and in turn decompose to acids, we speculate that
the carbon supports may quench acyl radicals, thus inhibiting this reaction pathway.
Furthermore, in order to explain the product distribution described in scheme 1, it should be
noted that although activated carbons are not as acidic as zeolites, they may still be acidic or
have different acid-base properties. In particular, we confirmed [27] RAC1 is an acidic
carbon (pH water extract = 4.7), whereas RAC2 and RAC3 are basic carbons (pH water
extract = 9.9, and 10.5 respectively). This suggests that the presence of dibenzyl ether when
Rh/RAC1 and Rh/RAC2 are used, is a consequence of the acid-base properties of the
activated carbon supports rather than an effect induced by Rh nanoparticles. As confirmed by
control tests by using the activated carbons only in the absence of any Rh (Table 1 bottom).
In fact, Rh/RAC1 and RAC1 as well as Rh/RAC2 and RAC2 both led to dehydration
9

products, whereas Rh/RAC3 and RAC3, the most basic of the carbons, led to the formation of
benzaldehyde only.
To further corroborate this experimental evidence, and support our hypothesis for the
formation of dibenzyl ether and methyl-diphenylmethane methyl-diphenylmethane isomers,
we added acids: HCl, HNO3 and H2SO4 into an alcohol/toluene mixture (Table S1). The
principle of these tests was the following: HCl to act as an acid catalyst only, HNO3 as an
acid but capable of oxidation processes, and H2SO4 as an acid with strong dehydration
capabilities. HCl led to a small conversion (< 10%) and a mixture of dibenzyl ether and
benzyl methyl benzene, HNO3 led to a conversion of ca. 10% and benzaldehyde only as
product. H2SO4 led to a conversion > 90% and was selective to methyl-diphenylmethane
(Table S1). This trend broadly fits the properties of these acids. In view of these tests, we
conclude that Brønsted acid sites on RAC1, and in minor part on RAC2, are those responsible
for the observed by-products and product distribution. It does worth noting though, that even
if HNO3 is capable to carry out the oxidation of benzyl alcohol to benzaldehyde, it cannot be
practically used as a reagent as it would eventually corrode the walls of a catalytic (non-
glass) reactor.
Furthermore as the formation of methyl-diphenylmethane via benzylation is also a
consequence of a dehydration reaction but with toluene as a reagent, and not just as a mere
solvent (Scheme S2), we tested the Rh/RAC catalysts and solvent-free benzyl alcohol as a
substrate (in this case by using a M:S ratio of 1:1000, Tables S2-S4) No methyl-
diphenylmethane was detected, thus confirming this by-product is a consequence of a
reaction between the solvent and the substrate and the acid properties of the carbon support
(RAC1).
The solvent-free tests are important also for an additional reason. Alcohols may
disproportionate to lead to the formation a carbonyl containing compound and an alkane [40].
In our case this reaction would be: Ph-CH2-OH + Ph-CH2-OH  Ph-CHO + Ph-CH3 + H2O.
However, we did not detect any formation of toluene during our tests, most probably as this
reaction is promoted in basic media [41] or biphasic systems [5].
In summary, the catalytic activity of Rh/C catalysts (Table 1) should be considered as the
result of two distinct contributions: (a) a contribution form Rh species oxidising benzyl
alcohol to benzaldehyde, and (b) a contribution from the activated carbon supports (mainly
10

RAC1) responsible for the formation of by-products: benzyl ether and methyl-
diphenylmethane.
On the other hand, the activated carbons alone, especially RAC1, were also capable of some
catalytic activity towards the formation of benzaldehyde in the absence of Rh. In contrast the
activity of RAC3 is the same, within experimental error, to the background activity of the
blank reaction (ca 5%). To explain these carbon behaviours, and verify the origin of this
minor oxidation process, we carried out additional control tests. Activated carbons are known
to contain Cu of Fe, which can oxidise an alcohol via reaction pathways involving peroxides
in solution [42-44]. As a consequence, we quantified the amounts of these metals in our
carbons via extraction and ICP-MS methods. The metal content for Cu and Fe was ranging
from 0.01 to 0.1 wt% (Table S5). In order to rule out possible interference to the catalytic
activity from the presence of these metals, the activated carbons were then washed by using
diluted HCl (see supplementary data for experimental details) to remove metal contaminants
natively present within the carbon [11] .
However, upon testing the activated carbons after these treatments (Table S6) both the
conversion and the selectivity were virtually identical (i.e. within our experimental error) to
those of un-washed carbons (Table 1). This leads us to conclude that although traces of Cu
and Fe are present in our samples, these do not contribute to the activity or our catalysts, and
the oxidation activity that we observe from the carbons may originate from reaction pathways
involving the formation of peroxides in solution in the absence of metals [45]. Activity
which, however, is very close to that of blank tests. As a consequence, we conclude the
activated carbons can be used for our reaction without any further purification steps.
Nevertheless, in order to gather data for a more comprehensive comparison on the effect that
the different activated carbons can induce to our reaction or to the Rh nanoparticles, we
carried out a full elemental scan on aqueous extracts (see experimental section 2.2.6) of the
activated carbons and analysed these samples via ICP-OES (Tables S7-S10). Up to 26
elements were identified, with Fe being indeed among the most abundant species, together
with Al, Ca and Mg. Unlike RAC2 and RAC3 though, we observed the presence of
significant amount of S for RAC1 (2.5 mg of S per gram of activated carbon, accounting for
ca. 30 at% of the total impurities). Whereas the presence of S is not uncommon in activate
carbon via thiols groups [46] these could act as nucleation centres for Rh. A factor to be
considered for future catalysts development, for metals supported nanoparticles not restricted
to Rh, but extendable also to metals like Pd or Au which show a greater affinity for S rather
than O.
11

3.2 Expected effects on metal-support interaction by activated carbon pre-treatment
with HNO3
By comparing the data reported in table 1, it is evident that Rh displays some appreciable
activity, however we speculated how to increase the conversion of these catalysts but still
preserving good selectivity for benzaldehyde. An obvious parameter to be considered in this
sense is Rh particle size. In fact, control tests by using RhCl3xH2O did not show any
appreciable catalytic activity, conversion < 5%, and consistent with blank values (Table 1),
and therefore not statistically showing any activity. With our preparation method it is the
carbon itself to be a reducing agent [24] for Rh3+ centres to Rh0 species, or to a combination
of Rh3+/Rh0 species. On the other hand, in order to trigger the nucleation and formation of
nanoparticles, Rh3+ species from a precursor in solution need to anchor to the carbon support
in the first instance [47]. This circumstance prompted us to consider a pre-treatment of our
carbon matrices before the deposition of Rh. This with the aim to: (i) induce a change in Rh
particle size distribution, (ii) estimate a Rh particle size upper limit for the activity of this
metal for our model reaction, and (iii) gather an estimate on the efficiency of an impregnation
protocol in terms of uniformity of final Rh size without any post-reduction treatment. The
latter to form the basis of a future work involving the identification of the active sites for this
reaction, which at present is beyond the purpose of the current manuscript.
Activated carbons are complex matrices containing phenol, ethers, esters and carboxylic
acids on their surfaces [48]. Among these functional groups, carboxylic acids are excellent
functionalities for the coordination of metallic cationic species [49]. In view of this, the
principle to change particle size distribution, and in turn change or enhance the activity of our
catalysts is the following: as HNO3 is an oxidizing agent, we would oxidise either hydroxyl
groups to aldehydes or acids [50], or methyl and methylene groups to acid (eq. 1) and
carbonyl (eq. 2) groups respectively [51]:
R-CH3 + 3HNO3  RCOOH + 3HNO2 + H2O
(eq. 1)
R-CH2-R + 2HNO3  R-CO-R + 2HNO2 + H2O (eq. 2)
A pre-treatment of the activated carbon with this acid should then increase the amount of
carboxylic acids on the carbon surface, and in turn the amount of nucleation centres available
12

for the initial coordination of Rh3+ species. A similar process has been observed for Cu over
activated carbon materials via the general scheme (eq. 3) [51]:
X-OH + Mn+  X-O-M(n-1)+ + H+ (eq. 3)
Where X is any organic branch, including carbonyl groups, of the carbon framework, most
often a phenol or a carboxylic acid. It should also be noted that any residual HNO3 in our
o
catalysts is eliminated by virtue of our calcination process in air at 150 C, which is used
promote the reduction reaction (eq. 4) [25]:
4HNO3  4NO2 + 2H2O + O2
(eq.4)
and in turn oxidation of the activated carbon. As the total amount of Rh precursor is a
constant across all our tests, a higher amount of nucleation centres should translate to smaller
Rh particles over our catalysts, and possibly a higher activity (Scheme 2). For this procedure
to be effective though, we would not have to increase the amount of acid centres so much to
lead to the formation of undesired by-products by Brønsted acid centres.
In order to further corroborate this working hypothesis, we carried out some DFT studies
aimed to mimic the interaction of a RhCl3 precursor with different functional groups on the
activated carbon surface. The principle of this approach is the more stable the interaction of
RhCl3 with a carbon functional group, the higher the likelihood to act as a nucleation centre,
and in turn initiate the nucleation process, or formation of nanoparticles on that group.
To this purpose, we used a simplified, and yet useful, model by evaluating the interactions of
RhCl3 with: benzene, p-xylene and naphthalene to mimic the interaction of a Rh centre with a
carbon basal plane (Figs. S10(A-C)). As well as propane, propanol and propionic acid (Figs.
S10(D-F)), p-xylene, p-toluic alcohol and p-toluic acid (Figs. S10(G-I)) to investigate the
evolution on adsorption capabilities, and in turn nucleation, of a -CH3 end to a R-OH and R-
COOH functional groups.
Although basal planes can have a strong stabilization of the RhCl3 precursor, from just 29
1
1
kJmol for a single benzene ring, to 336 kJmol if naphthalene is considered (Table S11),
these centres would not allow for the dissociation of Cl and in turn initiate the nucleation
process.
13

By considering the series propane, propanol and propionic acid to assess the evolution in
adsorption strength from a –CH3 end to a –COOH end. The acid can form a complex with
1
RhCl3 with a stabilization energy of stabilise the RhCl3 precursor of 68 kJmol against a
1
stabilization energy of just 10 kJmol for an alkyl –CH3 end. Thus showing the
appropriateness of our approach and the need to functionalise the activated carbon (Table
S11). From our simplified model it would also appear an alcoholic function could stabilise a
1
RhCl3 precursor more than the acid, with a stabilization energy of 106 kJmol . It does
worth nothing though, that in all our complexes, the H atom of the COOH group is facing in
the opposite direction with respect to the RhCl3 complex. In fact, if complexes were built
with the H atom of the COOH facing the RhCl3 complex, unstable species comprising
imaginary frequencies were obtained, and implying the attack of a Cl ligand by H to form
HCl, and dissociate the complex. This same trend was observed if p-toluic acid was used for
our models, and similar stabilization energies (Table S11) were also observed. This
phenomenon on the other hand, is precisely the start of the nucleation process. At present
stage, and with our simplified model, we can corroborate our hypothesis that acid functional
groups are indeed nucleation centres for Rh nanoparticles. It follows that an oxidation of the
carbon support would lead indeed to a larger amount of nucleation sites, either these be al
alcohol or an acid, and ultimately smaller nanoparticles.
3.3 Catalytic activity of Rh/C catalysts with supports pre-treated with HNO3
The effect of a HNO3 pre-treatment of the activated carbon to the catalytic activity, was
systematically evaluated for both: Rh/AC catalysts and activated carbons only as control tests
(Table 2).
It is possible to observe (Table 2 and Figs. S11-S13) that a HNO3 pre-treatment was capable
to increase the activity of all untreated Rh/RACs carbons systematically, with relative
conversion enhancements ranging from +40% for Rh/NAC1 to nearly +90% for Rh/NAC3, if
compared to Rh/RAC1 and Rh/RAC3 (Table 1). However, this also led also to an increase in
by-products (especially MDM) for Rh/NAC1 and Rh/NAC2. In contrast, the conversion of
Rh/NAC3 was nearly doubled, to a significant 37%, but still preserving a 100% selectivity to
benzaldehyde. As such, we consider this kind of pre-treatment a powerful tool to modify, and
possibly enhance, the catalytic activity of Rh species, which could potentially find a wider
applicability to other metals or supports. We ascribed this enhancement in catalytic activity to
a narrower particle size distribution and smaller Rh particle size as compared to the original
Rh/RACs catalysts (see section 3.4).
14

Furthermore, in analogy to the tests and data showed in section 3.1, the pre-treated carbons
without any deposited rhodium were also tested (Table 2 bottom). It is evident that the HNO3
treatment increases the background activity of the activated carbon matrices too, with the
formation of non-negligible amounts on dibenzyl ether and methyl-diphenylmethane for
NAC1 and NAC2. On the other hand, as we previously demonstrated (see section 3.1) these
by-products originate from Brønsted acidity from the activated carbons, these data also
indirectly prove that by using HNO3 we are indeed converting carbon functional groups to
carboxylic functional groups.
3.4 Catalytic activity of Rh/C catalysts with supports pre-treated with HCl
In view of the effects induced by HNO3, we pre-treated the activated carbons with HCl before
the deposition with Rh. As HCl does not have any oxidising nature, this acid should not affect
the activity of the carbon matrices, i.e. without inducing any changes in carbon functional
groups. On the other hand, the presence of Cl ions could still affect the final Rh particle size,
in this case Rh should behave like Au or Pt, as for these metals experimental evidence shows
a sintering effect of Cl ions towards the formation of Au and Pt nanoparticles [52]. In
analogy to the experimental approach used in sections 3.1 and 3.2, Rh was impregnated into
activated carbon matrices pre-treated with HCl, and pre-treated activated carbons in the
absence of Rh were also tested as control test (Table 3).
Interestingly the activity of these catalysts, both in terms of conversion or selectivity, is the
same or lower than the untreated catalysts (Table 1 and Figs. S14-S16). Thus proving HCl
does not significantly affect the surface of the activated carbons, but it may increase the Rh
particle size or have a small poisoning effect. Regarding the effect on the particle size of
impregnated metals, studies on the adsorption or binding of Cu centres on carbon surfaces
[51] showed that HCl may hinder the availability of nucleation sites (eq. 6) and especially if
Cl ions are present in large excess.
XOH + HCl  XCl + H2O
(eq. 6)
In this hypothesis Cl ions that would coordinate on the surface of the carbon in place of the
metals. If this model is correct, and by virtue of our approach (Scheme 2) in our case the
metal nanoparticles may be bigger not because the Cl ions are directly affecting the sintering
of metallic species, but because diminishing the number of functional groups available as
nucleation centres, and only in turn an increase in the particle size. Incidentally, this effect
15

could be operating also for Au and Pt and explain the sintering effect of Cl for these metals
in a different perspective [52].
3.5 Characterization of the catalysts and structure-activity correlations
In order to identify and confirm structure-activity correlations to our product distribution,
both with respect to Rh particles size and changes in activated carbon structure (especially
changes of functional groups), the catalysts were characterised by using TEM, XPS and
XRPD.
3.5.1 Transmission electron microscopy, Rh particle size and distribution
Changes in particle size are crucial in many catalytic applications [53] and the principle of
the use of HNO3 was precisely to induce these changes. In order to experimentally verify this
hypothesis, particle size and distribution of Rh nanoparticles were evaluated systematically
via TEM for all of the nine Rh/ACs catalysts.
It should also be underlined that TEM analysis of these samples is not trivial due to the
extreme dispersion of the active metal component, which presents a 1 wt% metal loading
1
1
1
over materials with surface areas of ca. 600 m2g [54], 1060 m2g [55] and 660 [56] m2g
for RAC1 (Darco-12), RAC2 (Norit GAC 1240) and RAC3 (Norit SA2) respectively.
Nevertheless, we were able to collect images comprising sets of 200 particles per catalyst and
therefore appropriate for statistical purposes and a meaningful data analysis of our results
[57]. Representative TEM images for each catalyst, together with their particle size
distribution are shown in Fig. 1(A-I).
TEM analysis of these samples lead to a number of interesting features: (i) a very large
average particle size range (mostly from 20 to 100 nm, with a small population of particles
even in the range of 200 nm), (ii) the lack of uniform size distributions, and with large
standard deviations from 10 to 50 nm, and (iii) only a few distributions appear to have a
Gaussian profile (for example Rh/NAC1, Rh/RAC3 and Rh/HAC3), whereas most of the
others appear to be highly skewed. It is worth to consider that impregnation protocols are
known to induce both large particle sizes and non-uniform distributions [58]. Nevertheless,
when applied to Rh, this method appears to lead to a much larger distribution and non-
uniform particles if compared to metals like Au or Pd. For these metals particle sizes in the
range of 2 to 20 nm can be obtained [11,15], and by using similar activated carbons as well as
16

low metal loading like those we have used. This experimental evidence may be due to
intrinsic properties of Rh when this protocol is applied to this metal. Surface energy
calculations by embedded atom methods [59] show that Rh has a higher surface energy
compared to Au and Pd. If these models are correct, Rh may minimize this energy by leading
to nanoparticles with a lower surface-to-bulk ratio i.e. larger nanoparticles. It should be noted
though that it is possible to synthesise Rh nanoparticles as small as 3 nm, but by using
incarceration methods [5].
Furthermore, in order to account for the skewing of the particle size distributions, these
distributions were fitted by using a standard Gaussian model [60] (eq. S12 and Fig. S17(A-
I)), and a log-normal distribution model [61] (eq. S13 and Fig. S18(A-I)). The latter
distribution is not uncommon to fit nanoparticles [62], although it is more often found in
nanoparticle distributions obtained from colloids or embedded into polymeric matrices [63].
A summary of these fittings is reported in table 4.
Although for the catalysts Rh/RAC1, Rh/NAC1, Rh/NAC2 and Rh/HAC1 the differences
between these two fitting models is small and their averages are similar (and effectively
statistically identical given the large dispersion of our data), the catalysts: Rh/RAC3,
Rh/NAC1, and Rh/HAC3 are better described by a Gaussian distribution, whereas the marked
skewed particle size distribution for Rh/RAC2, Rh/NAC3 and Rh/HAC2 is better described
by a log-normal distribution.
By using this data treatment and by selecting the best fitting model for each catalyst, we can
estimate the Rh/RAC catalysts have an average particle size of: 30 nm, 29 nm and 69 nm for
Rh/RAC1, Rh/RAC2 and Rh/RAC3 respectively. With the catalyst with the largest particles
being also the least active (Table 1).
In contrast, when the carbon support is pre-treated with HNO3 a smaller average particle size,
and in general also a narrower distribution, is obtained with diameters of: 23 nm, 31 nm and
32 nm for Rh/NAC1, Rh/NAC2 and Rh/NAC3 respectively. And as such corroborating our
hypothesis of a smaller particle size by a HNO3 pre-treatment, and in turn catalysts inherently
more active for the oxidation reaction.
For the effect of HCl towards the formation of Rh nanoparticles, this appears more difficult to
predict or interpret. Diameters of 20 nm, 36 nm and 106 nm were obtained for Rh/HAC1,
Rh/HAC2 and Rh/HAC3 respectively. However, Rh/HAC1 presents smaller particles than
17

the untreated sample Rh/RAC1, but a similar range if compared to Rh/RAC2, and a much
larger diameter if compared to Rh/RAC3.
Taking these trends as a whole, it appears that a pre-treatment with HNO3 lead to smaller Rh
particles or a narrower particle size distribution compared to samples obtained from untreated
carbons. Whereas a pre-treatment of the support by using HCl does not lead to any obvious
trend. Given these results though and the different functionalities over our activated carbons,
these effects may also be a function of the concentrations and amounts of the acids used to
carry out these treatments [64]. A parameter, the evaluation of the concertation effect, which
although beyond the proof of concept of the present manuscript, could be considered for
future catalyst development.
3.5.2 Rh particle size range and activity
It is obvious that these particle size ranges are very wide, and we think it is unrealistic to
consider such large particles (like those from Rh/HAC3 in the range of 100 nm) to be active
towards the oxidation reaction. In other terms, a factor or parameter is the effect that a pre-
treatment has on the Rh particle size, but another factor or parameter is the effect of changes
in Rh particle size has on the catalytic activity. As a consequence, and in order to provide
some estimate beyond which limit these nanoparticles are not active anymore, we grouped all
the data in one diagram, and reported also the overall range of activity of the activated carbon
supports (Fig. 2). Noteworthy, for particles above ca. 30-40 nm no changes or correlations
with the catalytic conversion are observed anymore, suggesting the particle size would not be
an active parameter anymore.
Therefore, we consider this value as an approximate upper limit diameter for the activity of
Rhodium for this oxidation of benzyl alcohol, or conversely to be active for this reaction
requires to be smaller than roughly ca. 30 nm. We deem this conclusion, although being a
broad estimate, to be relevant for future catalyst development based on this metal.
3.5.3 X-ray photoelectron spectroscopy of Rh/ACs catalysts
XPS is often used to characterize carbon matrices [65] to investigate changes in surface
functional groups like the presence of single or double C-C bonds, as well as carbonyl,
alcohol or ether groups [66], by analysing C1s and O1s signals, and in the present work XPS
has been used precisely for this purpose. On the other hand, one of the major uses of XPS is
also to investigate atomic composition and oxidation state of active metal components on the
surface of the support [15, 26].
18

In our case a detailed XPS analysis was carried out for all of our nine catalysts (3  Rh/RAC,
3  Rh/NAC, and 3  Rh/HAC); however, we should note that with respect to the surface
atomic composition, no Rh signal was detected. We ascribe this result to an intrinsically low
signal for Rh and that our Rh centres are extremely dispersed in our carbon matrix, with in
turn a very low exposed metal fraction for a meaningful detection, as also confirmed by
XRPD analysis (see section 3.4.5).
As a consequence, we focused our attention to XPS signals originating from the carbon
matrices only, and more specifically to the peak fitting of the C1s region. In fact, the samples
contain some SiO2, more specifically quartz, as detected also by XRPD (see paragraph 3.4.5),
which is not uncommon for activated carbons [67]. However, due to the relative high
intensity of Si2p (103.5 eV) in the XPS surveys, this would imply a significant O1s peaks at
ca. 533 eV [68], which fully overlaps with the O1s peaks originating from the carbon
matrices themselves ranging from 531 to 535 eV. As a consequence of these factors, and with
the aim to produce a meaningful data analysis, O1s has not been included and we focused to
the peak fitting of the C1s peak only.
The C1s signals were peak fitted (Fig. 3(A-I)) by using three components at: (i) 284.6 ± 0.2
eV assigned to C-C and C=C bonds [69], (ii) 286 ± 0.2 eV assigned to C-OH and C-O groups
[70], and (iii) 288.2 ± 0.2 assigned to C=O and O-C=O groups [71].
By carrying out this data treatment, the composition ratios of the three chemical groups were
calculated from the peak fitting (Table 5).
From these data a systematic effect is present: every catalyst obtained by pre-treating the
activated carbon with HNO3 has a greater relative amount of carbonyl groups, if compared to
catalysts obtained from untreated carbons. This shows that HNO3 is indeed oxidising the
carbon surface [16, 64]. And this also corroborates our initial hypothesis that if a higher
amount of acid (carbonyl) centres is present, then the smaller the particle size and the higher
activity, and this also well correlates with our activity trends and TEM data. For the catalyst
obtained from carbons pre-treated with HCl no clear trend is observed instead. However, this
also indirectly correlates with TEM data, where no clear trend with respect to
treatment/activity was observed for these catalysts.
3.5.4 X-ray powder diffraction of Rh/ACs catalysts
XRPD patterns were collected (Fig. 4(A-C)) to further complement TEM and XPS data
concerning the status of Rh in our materials (i.e. dispersed nanoparticles), and to identify
19

possible changes in the carbon structure, for example the presence or modifications of
graphitic carbon due to our acid pre-treatments.
Concerning rhodium species, no diffraction peaks for the Rh facets {111}, {002} and {022}
(expected at 41.07o, 47.79o and 69.89o 2 respectively) [72], nor diffraction peaks for the
Rh2O3 oxide facets {112}, {220} and {312} (expected at 34.49o, 48.80o and 62.07o 2
respectively) [73] were detected. In general terms, the lack of these diffraction peaks could
imply: (i) a particle size of below 4-5 nm, (ii) a thin layered metal structure or (iii) highly
dispersed metal species and in turn a very low metal exposed fraction [74, 75]. As TEM
revealed very large particles in the range of tens of nm, but XPS did not detect any Rh signal,
these methods rule out hypotheses (i) and (ii) and leave case (iii), a very low exposed
fraction, as the reason for the lack of detection of Rh signal in our patterns.
However, given the method that we have used to prepare our catalysts (see section 2.1), we
postulate our nanoparticles to be a combination of Rh and Rh2O3 species. The formation of
Rh2O3 species would be a consequence of the interaction of Rh with oxygenated groups over
the carbon surface responsible for the nucleation process (see sections 3.1 and 3.2), and the
formation of Rh centres a consequence of a reduction process induced by the carbon [11, 14,
76]. In this context we also prepared a Rh/RAC3 material comprising up to 10 wt% or Rh
(Fig. S19); however, also in this case, no obvious XRD reflections could be detected due to
the dispersion of our species into and on the carbon matrix.
Regarding a pattern analysis of the activated carbon supports instead, the only feature that
appears evident is the presence of inclusions of quartz in all the samples. Reflections at:
20.85o, 26.65o, 36.54o, 50.14o, and 59.95 o 2, for the facets {100}, {101}, {110}, {112} and
{211} respectively [77], thus fully matching XPS data and the presence of SiO2. These
inclusions are particularly noticeable in AC1 and AC3 containing materials, with only minor
amounts in AC2. This well reflects the origin of these activated carbons: lignite coal for
RAC1 (Darco-12) [54], bitumen for RAC2 (Norit GAC 1240) [555], and peat for RAC3
(Norit SA2) [56].
For all these samples though, no clear evidence of any graphitic carbon, for which a
characteristic reflection at 25.6o 2 would be expected for the {002} facet, is detected. Other
reflections, although less intense, that would be expected at 24.1o, 25.2o 2 [78], or in the
range of 43o 2, and characteristic of reflections associated to graphite basal planes [79]. (i.e.
20

surface platelets with maximum electron conjugation), are also not detected Moreover, it also
appears that whereas there are some small differences among the various activated carbons,
there is no detectable difference within the same activated carbon upon acid treatment, either
comparing fresh carbon supports or materials comprising Rh. In summary, we can then
conclude that any effect induced by the acid pre-treatments involves only the surfaces of the
carbon matrices, but not their bulk structure.
3.6 Effect of catalyst grain size of the activated carbons on the catalytic activity
Finally, in order to complete the comparison for the activity of our catalysts, we took into
account the different granulometry of our materials as an additional factor that might
contribute to the catalytic trends we reported so far. In fact, the activated carbons we have
used, have the following average grain size (expressed as sieve diameter) of: 3.8 mm, 1.4 mm
and 0.18 mm for RAC1 (Darco-12), RAC2 (Norit GAC 1240) and RAC3 (Norit SA2). The
sieve diameter is the minimum width a granule can pass through a sieve of specified mesh if
orientated correctly [80]. These diameters, in principle, may sufficiently differ to contribute
to our catalytic results by affecting both conversion and selectivity [65] by diffusional
limitations.
In view of this, and to carry out a set of comprehensive control tests, all of the catalysts and
all the carbon supports, un-treated and pre-treated were ground and sieved by collecting a
fraction between 100 and 200 m. This fraction was selected because for slurry-type reactors
[81] (like in our case: a liquid and a gas that react over a solid), and that utilise small catalyst
particles, diffusion is considered negligible and this is also a grain size range that is the
closest to the smallest activated carbon we have used: RAC3 (Norit SA2).
These control tests (Tables S12-S14) showed that no significant change is detected for the
conversion nor for the selectivity, when comparing the results from original grain size
catalysts, as well as when comparing activated carbons only. All the trends we observed and
described in the previous paragraphs are preserved: Rh/AC1 catalysts and AC1 carbons are
those that induce the higher formation of by-products, whereas Rh/AC3 catalysts are those
more selective to benzaldehyde, although with the lower conversion. And Rh/AC2 are the
materials intermediate between these two. Any pre-treatment with HNO3 lead to a higher
catalytic activity and no actual trend is present from a HCl pre-treatment. In view of this, we
conclude all of the results reported so far as validated, and grain size effects are not operating
or negligible for our catalysts under our reaction conditions.
21

3. Conclusions
In the current work we have demonstrated that Rh can have the potential to be used for the
oxidation of benzyl alcohol to benzaldehyde. By using a straightforward impregnation
protocol, and activated carbons as supports, strong metal support interactions were able to
affect the final particle size and distribution of Rh nanoparticles, with effect on the final
activity of these materials. In particular, pre-treatments of the activated carbon supports with
HNO3 were capable to reduce the Rh particle size or to lead to a narrower particle size
distribution, whereas treatments with HCl either had no effect or sintered the metal
nanoparticles. In turn it was also possible to increase the activity of Rh/AC based materials
by pre-treatment with HNO3 while still preserving good selectivity towards benzaldehyde,
with the smaller the particle size, the higher the activity. We ascribe these effects to changes
in particle size to an increase or a site blocking of nanoparticle nucleation centres
respectively, with probably carboxylic acids as the most likely nucleation centre for Rh
species. By using these pre-treatments we also identified an upper limit for the size of Rh
nanoparticles to be active for the oxidation for benzyl alcohol to benzaldehyde in the range of
ca. 30 nm. We believe these data to be useful information for the development of oxidation
catalysts based on this metal, and contribute to add Rh in the toolkit of metals capable to
carry out oxidation reactions. In fact although our catalysts are not optimised yet, and this
study represent a proof of concept on how to transform a reducing catalyst to an oxidizing
species, we should nevertheless consider that our catalytic tests did not make use of any base
[82], nor pressurised O2 systems [83], nor at present, more sophisticated preparation methods
like sol immobilization [84, 85] which if applied to Rh should be able to lead to a narrower
and more uniform product distribution. Given these premises we consider our catalyst to be
competitive with more known Au, Pd or Pt based catalysts for this reaction [86-88].
Furthermore, this study may pave the way for the exploitation of this metal for alcohol
oxidation with implications beyond the current study.
Disclosure statement
The authors declare no competing financial interests.
Acknowledgements
22

The authors thank Prof. Iain Coldham (University of Sheffield) for careful reading of the
manuscript and useful suggestions. Mr. Simon Thorpe and Mr. Neil Bramall for help and
support at the Mass Spectrometry Service and elemental analysis of the University of
Sheffield. TEM images were acquired at the University of Bradford, with TEM part of an
institutional EPSRC Capital Grant EP/L027011/1, for which P.D.C. is the principal
investigator. XPS data were acquired at the National EPSRC XPS Users’ Service (NEXUS),
an EPSRC Mid-Range Facility. The National Institute of Advanced Industrial Science and
Technology (AIST), Japan is acknowledged for the Spectral Database for Organic
1
Compounds, for H NMR and MS. This work is supported financially by the University of
Sheffield (grants CHM-313485 and CHM-316178) and the Royal Society (grant 148429).
23

References
[1] S. Cao, F. Tao, Y. Tang, Y. Li, J. Yu, Chem. Soc. Rev. 45 (2016) 4747–4765.
[2] S.C. Bourque, F. Maltais, W.-J. Xiao, O. Tardif, H. Alper, P. Arya, L.E. Manzer, J. Am.
Chem. Soc. 121 (1999) 3035–3038.
[3] R.J. Farrauto, R.M. Heck, Catal. Today 51 (1999) 351–360.
[4] S. Zhang, C. Guo, Q. Qian, G. Yuan, Catal. Commun. 9 (2008) 853–858.
[5] J.O. Weston, H. Miyamura, T. Yasukawa, D. Sutarma, C.A. Baker, P.K. Singh, M. Bravo-
Sanchez, N. Sano, P.J. Cumpson, Y. Ryabenkova, S. Kobayashi, M. Conte, Catal. Sci.
Technol. 7 (2017) 3985–3998.
[6] T. Mallat, A. Baiker, Annu. Rev. Chem. Biomol. Eng. 3 (2012) 11–28.
[7] T. Mallat, A. Baiker, Chem. Rev. 104 (2004), 3037–3058.
[8] F. Pinna, Catal. Today 41 (1998) 129–137.
[9] Z. Hou, N. Theyssen, W. Leitner, Green Chem. 9 (2007) 127–132.
[10] I.E. Beck, V.I. Bukhtiyarov, I.Y. Pakharukov, V.I. Zaikovsky, V.V. Kriventsov, V.N.
Parmon, J. Catal., 268 (2009) 60–67.
[11] M. Conte, A.F. Carley, C. Heirene, D.J. Willock, P. Johnston, A.A. Herzing, C.J. Kiely,
G.J. Hutchings, J. Catal. 250 (2007) 231–239.
[12] K. Köhler, R.G. Heidenreich, J.G.E. Krauter, J. Pietsch, Chem. Eur. J. 8 (2002) 622–
631.
[13] M. Conte, J.A. Lopez-Sanchez, Q. He, D.J. Morgan, Y. Ryabenkova, J.K. Bartley, A.F.
Carley, S.H. Taylor, C.J. Kiely, K. Khalid, G.J. Hutchings, Catal. Sci. Technol. 2 (2012)
105–112.
[14] M. Conte, A.F. Carley, G. J. Hutchings, Catal. Lett. 124 (2008) 165–167.
[15] M. Conte, C.J. Davies, D.J. Morgan, T.E. Davies, D.J. Elias, A.F. Carley, P. Johnston,
G.J. Hutchings, J. Catal. 297 (2013) 128–136.
[16] J.S. Noh, J.A. Schwarz, Carbon 28, (1990) 675–682.
[17] A. Obuchi, A. Ohi, M. Nakamura, A. Ogata, K. Mizuno, H. Ohuchi, Appl. Catal. B:
Environ. 2 (1993) 71–80.
[18] Y. Nagao, Y. Nakahara, T. Sato, H. Iwakura, S. Takeshita, S. Minami, H. Yoshida, M.
Machida, ACS Catal. 5 (2015) 1986–1994.
[19] R.A. Sheldon, I.W.C.E. Arends, A. Dijksman, Catal. Today 57 (2000) 157–166.
[20] L. Prati, P. Spontoni, A. Gaiassi, Top. Catal. 52 (2009) 288–296.
24

[21] N. Dimitratos, J.A. Lopez-Sanchez, D. Morgan, A.F. Carley, R. Tiruvalam, C.J. Kiely,
D. Bethell, G.J. Hutchings, Phys. Chem. Chem. Phys. 11 (2009) 5142–5153.
[22] J.M. Hoover, B.L. Ryland, S.S. Stahl, J. Am. Chem. Soc. 135 (2013) 2357–2367.
[23] B. Wang, M. Lin, T.P. Ang, J. Chang, Y. Yang, A. Borgna, Catal. Commun. 25 (2012)
96–101.
[24] M. Conte, A.F. Carley, G. Attard, A.A. Herzing, C.J. Kiely, G.J. Hutchings, J. Catal. 257
(2008) 190–198.
[25] G.D. Robertson Jr., D.M. Mason, W.H. Corcoran, J. Phys. Chem. 59 (1955) 683–690.
[26] M. Conte, C.J. Davies, D.J. Morgan, T.E. Davies, A.F. Carley, P. Johnston, G.J.
Hutchings, Catal. Sci. Technol. 3 (2013) 128–134.
[27] O.A. Ekpete, M. Horsfall Jr., Res. J. Chem. Sci. 1 (2011) 10–17.
[28] H.E. Gottlieb, V. Kotlyar, A. Nudelman, J. Org. Chem. 62 (1997) 7512–7515.
[29] M.J. Frisch, G.W. Trucks, H.B. Schlegel, G.E. Scuseria, M.A. Robb, J.R. Cheeseman, G.
Scalmani, V. Barone, B. Mennucci, G.A. Petersson, H. Nakatsuji, M. Caricato, X. Li, H.P.
Hratchian, A.F. Izmaylov, J. Bloino, G. Zheng, J.L. Sonnenberg, M. Hada, M. Ehara, K.
Toyota, R. Fukuda, J. Hasegawa, M. Ishida, T. Nakajima, Y. Honda, O. Kitao, H. Nakai, T.
Vreven, J.A. Montgomery Jr., J.E. Peralta, F. Ogliaro, M.J. Bearpark, J. Heyd, E.N. Brothers,
K.N. Kudin, V.N. Staroverov, R. Kobayashi, J. Normand, K. Raghavachari A.P., Rendell,
J.C. Burant, S.S. Iyengar, J. Tomasi, M. Cossi, N. Rega, N.J. Millam, M. Klene, J.E. Knox,
J.B. Cross, V. Bakken, C. Adamo, J. Jaramillo, R. Gomperts, R.E. Stratmann, O. Yazyev,
A.J. Austin, R. Cammi, C. Pomelli, J.W. Ochterski, R.L. Martin, K. Morokuma, V.G.
Zakrzewski, G.A. Voth, P. Salvador, J.J. Dannenberg, S. Dapprich, A.D. Daniels, O. Farkas,
J.B. Foresman, J.V. Ortiz, J. Cioslowski, D.J. Fox, Gaussian 09, Revision D.01; Gaussian,
Inc.: Wallingford, CT, 2009.
[30] M.M. Conradie, J. Conradie, Inorg. Chim. Acta, 362 (2009) 519–530.
[31] J. Narbutt, W.P. Oziminski, Dalton Trans. 41 (2012) 14416–14424.
[32] P. Stephens, F. Devlin, C. Chabalowski, M.J. Frisch, J. Phys. Chem. 98 (1993) 11623–
11627
[33] T. Ahrens, M. Ahrens, T. Braun, B. Braun, R. Herrmann, Dalton Trans. 45 (2016),
4716–4728.
[34] D.N. Briggs, K.H. Lawrence, A.T. Bell, Appl. Catal. A: Gen. 366 (2009) 71–83.
[35] C. Sepúlveda, R. García, P. Reyes, I.T. Ghampson, J.L.G. Fierro, D. Laurenti, M. Vrinat,
N. Escalona, Appl. Catal. A: Gen. 475 (2014) 427–437.
25

[36] M. Horgnies, I. Dubois-Brugger, N.J. Krou, I. Batonneau-Gener, T. Belin, S. Mignard,
Eur. Phys. J. Spec. Top. 224 (2015) 1985–1994.
[37] S.L. Barbosa, M. Ottone, M.C. Santos, G.C. Junior, C.D. Lima, G.C. Glososki, N.P.
Lopes, S.I. Klein, Catal. Commun. 68 (2015) 97–100.
[38] N. Candu, M. Florea, S.M. Coman, V.I. Parvulescu, Appl. Catal. A: Gen. 393 (2011)
206–214.
[39] M. Conte, H. Miyamura, S. Kobayashi, V. Chechik, Chem. Commun. 46 (2010) 145–
147.
[40] J.-P. Morel, N. Marmier, C. Hurel, N. Morel-Desrosiers J. Colloid Interf. Sci. 298 (2006)
773–779.
[41] E. Cao, M. Sankar, E. Nowicka, Q. He, M. Morad, P.J. Miedziak, S.H. Taylor, D.W.
Knight, D. Bethell, C.J. Kiely, A. Gavriilidis, G. J. Hutchings, Catal. Today, 203 (2013) 146–
152.
[42] C. Della Pina, E. Falletta, Michele Rossi, J. Catal. 260 (2008) 384–386.
[43] R. T. Kumar, N. C. S. Selvam, C. Ragupathi, L. J. Kennedy, J. J. Vijaya, Powder
Technol. 224 (2012) 147–154.
[44] C. Ragupathi, J. J, Vijaya, R. T. Kumar, L. J. Kennedy, J. Mol. Struct. Volume 1079
(2015) 182–188.
[45] M. Martin-Martinez, M.F.F. Barreiro, A.M.T. Silva, J.L. Figueiredo, J.L. Faria, H.T.
Gomes, Appl. Catal. B: Environ. 219 (2017) 372–378.
[46] H.T. Gomes, S.M. Miranda, M.J. Sampaio, J.L. Figueiredo, A.M.T. Silva, J.L. Faria,
Appl. Catal. B: Environ. 106 (2011) 390–397.
[47] C. Gabaldón, P. Marzal, J. Ferrer, A. Seco, Wat. Res. 30 (1996) 3050–3060.
[48] F. Rodríguez-Reinoso, Carbon 36 (1998) 159–175.
[49] T. Gong, L. Qin, W. Zhang, H. Wan, J. Lu, H. Feng, J. Phys. Chem. C, 119 (2015)
11544−11556.
[50] S.R. Joshi, K. L. Kataria, S. B. Sawant, J. B. Joshi, Ind. Eng. Chem. Res. 44 (2005) 325–
333.
[51] J. P. Chen and S. Wu, Langmuir 20 (2004) 2233–2242.
[52] A. Borgna, F. Le Normand, T.F. Garetto, C.R. Apesteguía, B. Moraweck, Stud. Surf.
Sci. Catal. 126 (1999) 179–186.
[53] K. An, G.A. Somorjai, ChemCatChem 4 (2012) 1512–1524.
[54] K. Wilson, H. Yang, C.W. Seo and W.E. Marshall, Bioresour. Technol. 97 (2006) 2266–
2270.
26

[55] R.L. Vaughan Jr., J. Yang, L.E. Le Mire, B.E. Reed, Adsorpt. Sci. Technol. 25 (2007)
295–310.
[56] B. Kordić, B. Jović, J. Tričković, M. Kovačević, J. Mol. Liq. 259 (2018) 7–15.
[57] H. Masuda, K. Gotoh, Adv. Powder Technol. 10 (1999) 159–173.
[58] C. Yu, S. Koh, J.E. Leisch, M.F. Toney, P. Strasser, Faraday Discuss. 140 (2009) 283–
296.
[59] Z. Jian-Min, M. Fei, X. Ke-Wei, Chinese Phys. 13 (2004) 1082–1090.
[60] R. Fisker, J.M. Carstensen, M.F. Hansen, F. Bødker, S. Mørup, J. Nanopart. Res. 2
(2000) 267–277.
[61] L.B. Kiss, J. Söderlund, G.A. Niklasson, C.G. Granqvist, Nanotechnology, 10 (1999)
25–28.
[62] M. Barberio, P. Antici, Sci. Rep. 7 (2017) 41372.
[63] A.M. Maley, G.J. Lu, M.G. Shapiro, R.M. Corn, ACS Nano 11 (2017) 7447–7456.
[64] J. Li, L. Ma, X. Li, C. Lu, H. Liu, Ind. Eng. Chem. Res. 44 (2005) 5478–5482.
[65] D. Lennon, D.T. Lundie, S.D. Jackson, G.J. Kelly, S.F. Parker, Langmuir 18 (2002)
4667–4673.
[66] A.M. Puziy, O.I. Poddubnaya, R.P. Socha, J. Gurgul, M. Wisniewski, Carbon 46 (2008)
2113–2123.
[67] O. Senneca, P. Russo, P. Salatino, S. Masi, Carbon 35 (1997) 141–151.
[68] M.L. Miller R.W. Linton, Anal. Chem. 57 (1985) 2314–2319.
[69] A.P. Dementjev, A. de Graaf, M.C.M. van de Sanden, K.I. Maslakov, A.V. Naumkin,
A.A. Serov, Diamond Relat. Mater. 9 (2000) 1904–1907.
[70] S. Akhter, K. Allan, D. Buchanan, J.A. Cook, A. Campion, J.M. White, Appl. Surf. Sci.,
36 (1988-89) 241–258.
[71] E. Desimoni, B. Brunetti, Chemosensors 3 (2015) 70–117.
[72] H.S. Swanson, R.K. Fuyat, G.M. Ugrinic, US Natl. Bur. Stand. Circ. (1953), in:
International Centre for Diffraction Data, Powder Diffraction File, Entry 5-685, 1996.
[73] K. Neininger,W. Eysel, Grant-in-Aid Heidelberg, (1990) in: International Centre for
Diffraction Data, Powder Diffraction File, Entry 41-541, 1996.
[74] J. Liu, Z. Zhao, J. Wang, C. Xu, A. Duan, G. Jiang, Q. Yang, Appl. Catal. B: Environ.
84 (2008) 185–195.
[75] R.T. Kumar, P. Suresh, N.C.S. Selvam, L.J. Kennedy, J.J. Vijaya, J. Alloy. Compd. 522
(2012) 39–45
27

[76] H.S. Swanson, R.K. Fuyat, G. M. Ugrinic, US Natl. Bur. Stand. Circ. (1954) in:
International Centre for Diffraction Data, Powder Diffraction File, Entry 5-490, 1996.
[76] F. Rodriguez-Reinoso, Carbon. 36 (1998) 159–175.
[77] H.S. Swanson, R.K. Fuyat, G. M. Ugrinic, US Natl. Bur. Stand. Circ. (1954) in:
International Centre for Diffraction Data, Powder Diffraction File, Entry 5-490, 1996.
[78] A. Dankekar, R.T.K. Baker, M.A. Vannice, Carbon 36 (1998) 1821–1831.
[79] A.S. Aricó. V. Antonucci, M. Minutoli, N. Giordano, Carbon 27 (1989) 337–347.
[80] L.M. Cohen, M.R. Denison, N. Gat, A.B. White, Combust. Sci. Technol. 27 (1981) 79–
82.
[81] M. Mandić, B. Todić, L. ꢀivanić, N. Nikačević, D. B. Bukur, Ind. Eng. Chem. Res. 56
(2017) 2733−2745.
[82] M. Besson, P. Gallezot, Catal. Today, 57 (2000) 127−141.
[83] C. Keresszegi, D. Ferri, T. Mallat, A. Baiker, J. Phys. Chem. B 109 (2005) 958−967.
[84] N. Dimitratos, J.A.Lopez-Sanchez, D. Morgan, A.F. Carley, R. Tiruvalam, C.J. Kiely, D.
Bethell, G.J. Hutchings, Phys. Chem. Chem. Phys. 11 (2009) 5142–5153.
[85] C. Della Pina, E. Falletta, M. Rossi, J. Catal. 260 (2008) 384–386.
[86] J. Muzart, Tetrahedron, 59 (2003) 5789–5816.
[87] C. Parmeggiani, F. Cardona, Green Chem. 14 (2012) 547–564.
[88] P. Haider, B. Kimmerle, F. Krumeich, W. Kleist, J.-D. Grunwaldt, A. Baiker, Catal.
Lett. 125 (2008) 169–176.
28

Appendix A. Supplementary Data file available: 1H-NMR spectra, GC/MS chromatograms
and fragmentation patterns, XPS spectra, particle size distributions. Details of product
characterization, and analysis of the reaction mixtures. Figs. S1 to S19, Tables S1 to S14 and
equations S1 to S13.
29