Synthesis, characterisation, and catalytic properties of halloysite-supported metal nanoparticles

Article abstract of DOI:10.1016/j.materresbull.2018.11.027

A facile synthetic method is reported for the immobilisation of metal nanoparticles on halloysite, an aluminosilicate material with a unique nanotubular structure. A one-step colloidal synthesis was developed, from which nine different catalysts were prepared. The materials were tested in three different catalytic hydrogenations, and the halloysite-supported catalysts were found to outperform alumina- or silica-supported catalysts prepared in the same fashion. Electron microscopy showed that the activity was closely linked to the spatial distribution of the supported metal particles, which in turn affected the shapes of certain surface-associated peaks in the infrared and nuclear magnetic resonance spectra. From this data insight is gained into the nature of the interaction between metal nanoparticles and the surfaces of these support materials.

Full text of DOI:10.1016/j.materresbull.2018.11.027

Accepted Manuscript
Title: Synthesis, Characterisation, and Catalytic Properties of
Halloysite-Supported Metal Nanoparticles
Authors: Jacob. E. Jaine, Michael. R. Mucalo
PII:
S0025-5408(17)34330-1
DOI:
Reference:
https://doi.org/10.1016/j.materresbull.2018.11.027
MRB 10283
To appear in:
MRB
Received date:
Revised date:
Accepted date:
19 November 2017
9 August 2018
20 November 2018
Please cite this article as: Jaine JE, Mucalo MR, Synthesis, Characterisation, and
Catalytic Properties of Halloysite-Supported Metal Nanoparticles, Materials Research
Bulletin (2018), https://doi.org/10.1016/j.materresbull.2018.11.027
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.

Synthesis, Characterisation, and Catalytic Properties of Halloysite-Supported Metal
Nanoparticles
Authors:
Jacob. E. Jaine (jacob.e.jaine@gmail.com)a,b
Michael. R. Mucalo (michael.mucalo@waikato.ac.nz, corresponding author)a
Affiliations:
a) Chemistry, School of Science, Faculty of Science and Engineering, University of
Waikato, New Zealand.
b) Analytica Laboratories, Ruakura Research Centre, Hamilton, New Zealand
Graphical Abstract
Highlights





Metal nanoparticles were immobilised onto halloysite by a facile synthetic method
The materials showed exceptional catalytic activity in several hydrogenations
Halloysite-based catalysts were more active than alumina or silica equivalents
The activity was found to be correlated with the spatial distribution of metal
Spectroscopic analyses gave insight into the binding sites of the metal
Abstract

A facile synthetic method is reported for the immobilisation of metal nanoparticles on halloysite, an
aluminosilicate material with a unique nanotubular structure. A one-step colloidal synthesis was
developed, from which nine different catalysts were prepared. The materials were tested in three
different catalytic hydrogenations, and the halloysite-supported catalysts were found to outperform
alumina- or silica-supported catalysts prepared in the same fashion. Electron microscopy showed that
the activity was closely linked to the spatial distribution of the supported metal particles, which in turn
affected the shapes of certain surface-associated peaks in the infrared and nuclear magnetic
resonance spectra. From this data insight is gained into the nature of the interaction between metal
nanoparticles and the surfaces of these support materials.
Abbreviations
TOF – Turnover frequency
HNC – Halloysite nanotubes
TEM – Transmission electron microscopy
SEM – Scanning electron microscopy
EDS – Energy dispersive spectroscopy
BSD – Backscatter detector/detection
XRD – X-ray diffraction
NMR – Nuclear magnetic resonance (spectroscopy)
FTIR – Fourier transform infrared (spectroscopy)
ICDD – International centre for diffraction data
FWHM – Full width at half maximum
Key Words
A. Nanostructures, A. Surfaces, B. Microstructure, C. Electron microscopy, D. Catalytic properties

1. Introduction
Halloysite nanoclay (abbreviated ‘HNC’) is a naturally occurring aluminosilicate clay with a
unique tubular microstructure. It is composed of alternate layers of octahedral alumina and
tetrahedral silica sheets, which have an intrinsic ‘lateral misfit’. This produces a stress across
the sheets, which is relieved by rolling into tubular crystals [1]. The resulting nano-sized
tubules have an alumina-rich interior, and a silica-rich exterior. Substitution of the structural
Al3+ ions for Fe3+ ions enlarges the octahedral alumina sheets, which relieves the stress and
produces halloysite particles with a flatter morphology [2]. The surface area of halloysite
ranges from 30 m2 g-1 to 170 m2 g-1 depending on the proportion of nanotubes, which account
for the majority of the surface area [3]. It forms naturally by hydrothermal alteration of other
clays throughout the world, though one of the purest sources is the Matauri Bay mine in
Northland, New Zealand [4].
A range of applications have been found for halloysite [5, 6], including drug delivery [7, 8] or
as a component of nanocomposites [9]. Importantly, its properties make it attractive as a
support material for catalysts, especially metal nanoparticles [10, 11]. Recent studies have
shown that halloysite is a viable support for many catalytic systems, with examples that
include decomposition of chlorophenols by Fe3O4/HNC [12], plasmonic photodegradation of
dyes by Au/HNC [13], reduction of nitrophenols by Rh/HNC [10, 14]_ENREF_8, synthesis of
propargylamines by Ag/HNC [15], hydrogen production by CdS/HNC [16], hydrogenation of
phenol by Ru/HNC [17], oxidation of benzyl alcohol by Au/HNC [18], and several others [19-
22].
A similar group of systems are metal nanoparticles supported on or confined within
zeolites, which have been used in a range of reactions including hydrogenations. For
example, Pd or Pt immobilised on zeolite has been used to hydrogenate olefins and nitro
compounds [23], while Pt and Rh encapsulated in zeolite has been used for the
chemoselective hydrogenation of cinnamaldehyde [24], Ru-loaded zeolite has been used for
the hydrolytic hydrogenation of cellulose to hexitols [25], and bimetallic Ni-Pt nanoparticles
immobilised on zeolite have been used for hydrogenating carbonyls [26]. While zeolites are a

group comprising a relatively broad range of structures, much of this work has been performed
with the easily-synthesisable zeolite Y.
Though it is clear that natural aluminosilicates like halloysite have a promising role as
catalyst supports, the benefits of halloysite over traditional support materials like silica or
alumina are seldom investigated. Therefore, this communication describes the
characterisation and activity of catalysts prepared using halloysite sourced from New
Zealand’s Matauri Bay Mine. A simple one-pot synthetic method was used to immobilise Rh,
Pd, or Pt nanoparticles onto the surface of the halloysite, as well as the traditional support
materials alumina and silica. The resulting materials were tested in the catalytic hydrogenation
of three olefins, and then characterised by a suite of instrumental methods. Structure-activity
relationships are drawn from the data, and used to derive insight into the binding of the metal
particles.
2. Experimental
2.1 Preparation of Catalysts
Methods for synthesising supported catalysts have become increasingly complex and
specialist [27], so we sought to a develop a simple wet chemical method for this purpose. This
involved reducing appropriate metal salts in a suspension of the support material, followed by
precipitation of the colloidal material using a flocculant.
For the research described herein, 2.97 g of either SiO2 (>99.9%, Merck), Al2O3 (>99.5%,
BDH), or halloysite (Imerys) was suspended in solutions of either RhCl3 (>99%, Precious
Metals Online), Na2PdCl4 (>99%, Aldrich), or H2PtCl6 (>99.0%, BDH). Concentrations of the
metal salts were set such that in 100 mL of solution there was 30 mg of each metal, so that
when reduced, the catalysts would have a loading of 1.0 wt%. Reduction of the metal salts to
their finely divided metallic form was brought about by dropwise addition of 5.00 mL of 0.05
mol L-1 NaBH4 solution (>99%, Ajax). During this period, the nascent nanoparticles could
adsorb to the support materials, producing a supported catalyst.
After reduction, the suspensions were stirred for a further 30 minutes, followed by addition
of 100 μL of 0.1 mol L-1 LaCl3 solution (>99.9%, Ajax) to destabilise any material remaining in
colloidal suspension [28]. The resulting materials were gravity filtered, washed free of chloride
and other residual salts, dried at 100°C in air, and then stored in vacuo until further analysis.

2.2 Hydrogenations
The catalytic activity of the materials were tested using olefin hydrogenation as a probe
reaction. Three different olefins were hydrogenated, namely cyclohexene (>99% AppliChem),
1,5-cyclooctadiene (‘1,5-COD’, >99%, Aldrich), and 1-octene (>97%, Merck). All
hydrogenations were performed in a glass stirred-tank batch reactor under 1 atm of H2
(>99.99%, BOC) at a temperature of 20 ± 0.2 °C. In each reaction ca. 100 mg of catalyst was
added to the reactor along with 5.00 mL of organic substrate, and 45.0 mL of n-heptane as a
solvent, chosen due to its favourable wetting properties [29]. Reactions were typically run for
90 minutes, and progress was monitored every 10 minutes by withdrawing 100 µL aliquots of
the mixture through a rubber septum, diluting to 0.2 mg mL-1 with acetone, and analysing by
gas chromatography. All volumes, masses, and times were chosen to operate the reactions at
high substrate concentrations and low conversion rates as appropriate for kinetic
measurements.
All gas chromatographic analyses were performed using a Hewlett Packard 6890 gas
chromatograph with 5972 mass-selective detector. Samples were injected (1 μL) onto a
Zebron ZB-5 capillary column at 50°C using a 1:100 split, and run isothermally using helium
carrier gas at 1.0 mL min-1, with a total programme length of 3 minutes. Normalisation was
performed relative to the heptane peak area, and absolute concentrations obtained by
interpolation from a standard curve.
2.3 Instrumental Characterisation
The pristine support materials and the metal-containing catalysts were analysed by a suite
of instrumental methods. The spectra were then compared to determine the physical and
chemical effects of immobilising metal particles onto the supports, and related to the catalytic
activity of the systems to explore structure-activity relationships.
Morphological and textural characteristics of the materials were evaluated by scanning
electron microscopy (SEM). All analyses were performed using a Hitachi S-4700 SEM
equipped with a Thermo energy dispersive spectrometer (EDS) and a Hitachi backscatter
detector (BSD). Powders were mounted on double-sided carbon tape on an aluminium stub
and analysed in BSD mode, then sputtered with Pt and re-analysed in secondary electron
mode.

Size distributions of the supported metal particles were measured by high-resolution
transmission electron microscopy (TEM). All analyses were performed using a Philips CM30
operated at 250 kV. Samples were prepared by dispersing 10 mg of sample in 10 mL of water
and sonicating for 10 seconds, followed by depositing 10 µL on carbon/formvar-coated copper
grids (ProSciTech Pty. Ltd.) and drying in air. Magnification calibrations were performed using
bovine catalase (EMS Ltd.).
Crystallographic properties of the supports and catalysts were analysed by X-ray diffraction
(XRD). All analyses were performed using a PANalytical Empyrean diffractometer with Ni-
filtered Cu radiation. Diffractograms were each collected over an eight-hour period, covering
the 2θ range 10°-90° in 0.013° steps. Samples were prepared by backloading, and spun at 1
Hz during acquisition. Pattern matching was performed using HighScore Plus. As a secondary
particle size measurement method, line broadening analysis was carried out on the patterns of
the catalysts using the Scherrer equation, which relates the breadth of a diffraction peak to the
size of the crystallites from which it originates. Using HighScore Plus, backgrounds were
manually subtracted from the patterns using a cubic spline, and Voigt peaks fitted to the
resulting profiles. The breadths were corrected for instrumental broadening by subtracting the
peak widths of a CeO2 standard prepared according to the literature [30]. All visible metal
peaks were used during fitting, which typically included the (111), (200), (220), and (311)
diffraction peaks, and a shape factor (K) of 0.9 was used in the calculation.
1
The bulk chemical structure of the materials were investigated by H, 27Al, and 29Si solid-
state nuclear magnetic resonance spectroscopy (NMR). All analyses were performed on a
Bruker AVANCE 300 MHz spectrometer with a standard 4 mm H/X solids probe. Spectra were
typically recorded with 90° pulses, spectral widths of 200 ppm, 30 ms acquisition times and
varied recycle delays and numbers of transients depending on the nucleus being examined.
Finally, the bulk and surface vibrational properties of the materials were investigated by
Fourier transform infrared spectroscopy (FTIR). Samples were prepared as KBr discs and
analysed using a PerkinElmer Spectrum 200 FT spectrometer. Additionally, samples were
analysed as neat powders by infrared microspectroscopy using a PerkinElmer Spotlight 200
FT microscope (yielding so-called ‘MicroFTIR’ or ‘transflection’ spectra). By virtue of this
particular optical configuration it was considered that this method may be more surface-

sensitive than the traditional transmission/KBr disc method, and thus appropriate for studying
surface-active systems like those described herein.
3. Results
3.1 Catalytic Activity
When the catalysts were loaded into the reactor under hydrogen, conversion of the
substrate molecules began immediately and proceeded linearly throughout the 90-minute
reactions. This was observed for all of the systems containing metal nanoparticles, but not for
any of the support materials without metal nanoparticles. The conversion rates observed
varied depending on the metal, the support material, and the substrate, as shown in Table 1.
Table 1: Measured catalytic activity of the thirty-six systems studied
herein, expressed as the number of micromoles of substrate converted
to product per hour per gram of catalyst. All rates are given to two
significant figures.
Measured Activity (µmol hr-1 g-1)
Catalyst
Cyclohexen
e
1-Octene
1,4-Cyclooctadiene
Rh/SiO2
Rh/Al2O3
Rh/HNC
Pd/SiO2
Pd/Al2O3
Pd/HNC
Pt/SiO2
50
180
87
36
17
66
34
30
66
14
6.0
24
0.0
0.0
0.0
12
60
36
26
98
50
17
170
0.0
250
130
200
340
130
63
Pt/Al2O3
Pt/HNC
300
0.0
0.0
0.0
SiO2 (pristine)
Al2O3 (pristine) 0.0
HNC (pristine) 0.0
The hydrogenation of cyclohexene proceeded in all cases directly to cyclohexane, with no
side products observed. In all cases the reaction rates were linear, and in all cases were
largest with halloysite-supported metals. In the case of Pt/HNC, a ten-fold greater

hydrogenation rate was observed compared to Pt/Al2O3 and nearly four-fold higher than
Pt/SiO2. The same trend was observed with supported Rh or Pd, though to a lesser extent.
These conversion rates compare favourably with previously reported cyclohexene
hydrogenation catalysts [31-34], especially as the reactions reported herein were performed
under standard conditions.
The hydrogenation of 1-octene also proceeded linearly, though significant isomerisation
was observed with the supported palladium catalysts (ca. 10-50% relative to the rate of
hydrogenation depending on the system). The hydrogenation rates were generally several
times higher than for cyclohexene hydrogenation, with a maximum conversion rate of 340
µmol hr-1 g-1 for the Pd/HNC catalyst. These rates are again significantly higher than for the
equivalent alumina- or silica-supported catalysts, and are higher than any activity reported in
the literature with the exception of the study of Adlim et al. [35].
Lesser activities were observed in the hydrogenation of 1,5-COD, though once again the
halloysite-supported catalysts were several times more active than the equivalent silica- or
alumina-supported catalysts. The production of cyclooctane was linear in all observed cases,
the maximum conversion rate observed being 66 µmol hr-1 g-1 with the Pd/HNC catalyst, which
was twice the rate observed for the Al2O3- or SiO2-supported systems. These represent some
of the highest activities for 1,5-COD hydrogenation by heterogeneous platinum-group
catalysts, with the exception of the remarkably high activity of the Pd/pumice catalysts
reported by Deganello et al. [36]. Like with 1-octene, some degree of isomerisation was
observed during the hydrogenation, with the products at 30 minutes containing an average of
5.0% 1,5-COD isomers for the Pt catalysts, 6.4% for the Rh catalysts, and a much higher
33.5% for the Pd catalysts. These trends are consistent with known differences in the
substrate absorption mechanism between the three metals [37]. Comparatively the choice of
support material had little effect on the amount of isomerisation, the only obvious trend being
that the silica-supported catalysts showed the greatest amount of isomerisation amongst all
three metals tested. These same trends were also true of the octene hydrogenation described
previously.
3.2 Electron Microscopy

Analysis of the pristine alumina by scanning electron microscopy showed that it was
composed of spherical agglomerates, with diameters of 63±31 µm and no pore system.
Catalysts prepared from the alumina shared the same general morphological characteristics,
but were speckled with clusters of material not present in the pristine support. Under
backscatter conditions these clusters showed up as bright points, and point analysis by EDS
showed strong emission peaks corresponding in energy to the Kα emission lines of Rh, Pd, or
Pt. Further analysis by transmission electron microscopy showed the clusters were composed
of spherical nanoparticles 3.3- 9.7 nm in diameter, as shown in Table 2. In addition to these
clusters, inspection of the edges of the support material showed it was uniformly coated in
particles with a surface coverage in excess of ca. 50%. Micrographs for these systems, and
others, are given in Figures 1 and 2.
Table 2: Sizes of the metal nanoparticles in the nine catalyst systems
studied, as measured by TEM (mean ± standard deviation) and XRD
(mean only).
Particle Size [nm]
System
TEM
XRD
4.0
7.3
6.3
<3
9.2
7.3
3.0
5.3
5.2
Rh/SiO2
Pd/SiO2
Pt/SiO2
Rh/Al2O3
Pd/Al2O3
Pt/Al2O3
Rh/HNC
Pd/HNC
Pt/HNC
3.5 ± 0.5
5.0 ± 0.8
3.6 ± 0.6
3.3 ± 0.6
9.7 ± 4.1
4.0 ± 1.0
3.7 ± 0.8
8.3 ± 2.8
3.2 ± 0.6

Figure 1: Transmission electron micrographs representing the systems studied
herein. The top row shows the pristine support materials, while the remaining three
rows show the catalysts containing immobilised nanoparticles.

Figure 2: Scanning backscatter electron micrographs representing the twelve
different systems studied herein. The top row shows the pristine support materials,
while the remaining three rows show the catalysts containing immobilised
nanoparticles.
Similar analyses of the pristine silica showed that it was composed of angular mesoporous
particles with diameters of 94±40 µm, which were present as discrete particles rather than

aggregates. Under backscatter conditions the same bright flecks were observed, as well as
the same sets of emission lines via EDS. Analysis by transmission electron microscopy
showed the clusters were also composed of spherical nanoparticles, though of smaller
diameters of 3.5 – 5.0 nm. Most of the nanoparticles were present as large aggregates, and
few were adsorbed on the surface of the silica. In this sense, the current catalysts may be
considered ‘mixed agglomerated catalysts’, as described by Perego and Villa [38].
Analysis of the halloysite showed it shared characteristics with both of the previous support
materials. The particles were small clusters 12.7±7.6 µm in diameter, each composed of a
mixture of tubules and platelets. Relatively few macroscopic aggregates of metal were
observed, with most of the nanoparticles present distributed around the edges of the support
particles which had the platelet morphology. All of the nanoparticles were spherical in
morphology. Few particles were found adsorbed on the tubules, and those which did were
found exclusively on their outsides. The average particle size ranged from 3.2 nm – 8.3 nm
depending on the metal, though no such trend was observed in the spatial distribution of the
particles. The size and clustering of the particles is similar to that reported by Rostamzadeh et
al. who prepared Au/HNC catalysts using a similarly simple synthetic method, though the
locations of the nanoparticles are different between their study and ours [39]; while their
catalysts contained many particles within the interior of the tubules, ours contained none.
3.3 X-Ray Powder Diffraction
Powder diffraction analysis of the alumina produced a pattern with a low, flat background,
and a series of narrow diffraction peaks corresponding in position to α-Al2O3 (ICDD PDF 01-
077-2135). In the case of the alumina-supported catalysts, the patterns additionally contained
five diffraction peaks corresponding to metallic rhodium (ICDD PDF 00-001-1214), palladium
(ICDD PDF 01-071-3757), or platinum (ICDD PDF 01-071-3756). Line broadening analysis of
the metal peaks gave average particle sizes of 7.3 nm – 9.2 nm, except for the Rh/Al2O3
system for the diffraction peaks were too wide to be reasonably analysed. These values are
shown in Table 2, along with the sizes derived from TEM data. This illustrates that at least
some of the metal within the systems is in the zero-valent metallic form, rather than as oxides
such as PdO, Rh2O3, or PtO2.

Analysis of the silica-based systems all gave similar results, exhibiting a weak and broad
background peaking at 22° 2θ and tailing off toward higher angles. The systems containing
nanoparticles clearly exhibited the same metal peaks as observed in the alumina-based
systems, but were significantly narrower and could be easily resolved from the background.
The particle sizes calculated from line broadening analysis for these systems ranged from 4.0
nm – 7.3 nm.
The diffraction pattern of the pristine halloysite was different again from both the silica and
alumina, exhibiting a series of slightly broadened peaks characteristic of partially disordered
clay minerals, and similar to previously reported halloysite diffraction patterns [40-42]. The
diffraction patterns of the halloysite-supported metals contained the same identical sets of
peaks as the support, but once again contained the additional peaks corresponding to metallic
Rh, Pd, and Pt. The calculated particle sizes in these instances ranged from 3.0 – 5.2 nm.
3.4 Solid-State Nuclear Magnetic Resonance Spectroscopy
1
The pristine silica, as received, exhibited three resonances in the H-29Si CPMAS spectrum
at -91.3 ppm, -101.2 ppm, and -110.6 ppm, corresponding respectively to silicon atoms with
two attached silanols (‘Q2 sites’), one attached silanol (‘Q3 sites’), and no attached silanols (‘Q4
sites’). This pattern is typical of silica gels reported in the literature, especially those of high
surface area [43]. The 1H spectrum exhibited a broad resonance at 4.4 ppm (adsorbed water)
and two sharp resonances at 3.6 ppm (H-bonded silanols) and 1.1 ppm (free silanol groups)
[44]. No significant changes were observed in the 29Si spectra after immobilisation of
1
nanoparticles, though the H spectra showed several key changes: the water peak was
significantly weaker and narrower than for the pristine support, and the peaks for the surface
silanols disappeared entirely. This is illustrated in Figure 3. This suggests that the preparative
procedure largely dehydrates the surface of the silica, while silanols present in the bulk remain
unaffected.

A
B
C
D
12.0
10.0
8.0
6.0
4.0
2.0
0.0
-2.0
1H Chemical Shift, δ [ppm]
Figure 3: 1H solid-state NMR spectra of the silica-supported catalysts,
corresponding to Rh/SiO2 (A), Pd/SiO2 (B), Pt/SiO2 (C), and pristine
SiO2 (D)
The 27Al DPMAS spectra of the pristine alumina exhibited a single broad symmetric
resonance at 9.8 ppm (FWHM 12.4 ppm) corresponding to octahedral AlO6 environments [45].
Spectra of the alumina-supported catalysts produced similar spectra, but with some subtle
changes to the line shapes at ca. -75 ppm, the region of spectral space corresponding to
tetrahedral AlO4 units adjacent to octahedral AlO6 units. As this is not a structural motif which
α-Al2O3 formally contains, it is most likely associated with regions in the material at which
significant crystallographic distortions may occur, for example the surfaces of the particles.
Adsorption of nanoparticles to these surfaces would then produce changes to the coordination
of the alumina atoms, and changes to the structure of the resonances. This is precisely what
was observed, and is shown in Figure 4.
A
B
C
D
-100
-75
-50
-25
0
25
50
27Al Chemical Shift, δ [ppm]

Figure 4: 27Al NMR spectra of the alumina-based systems. The four
plots correspond to Rh/Al2O3 (A), Pd/Al2O3 (B), Pt/Al2O3 (C), and pristine
Al2O3 (D). All spectra are scaled to equal relative intensity.
1
The halloysite, as received, exhibited two overlapping resonances in the H spectrum of
nearly equal intensity, corresponding to silanol and aluminol resonances. A single resonance
at -28 ppm (19.6 ppm FWHM) was observed in the 27Al DPMAS spectrum, tailing toward
higher field. This is similar to previously reported spectra, and originates from halloysite’s
structural AlO6 units47. The 1H-29Si CPMAS spectrum contained only a single resonance at -
92.8 ppm (2.5 ppm FWHM), while the 29Si DPMAS spectrum contained an additional
resonance at -108.9 ppm (1.2 ppm FWHM) with a relative intensity of ca. 5%. Immobilisation
of nanoparticles on the halloysite produced an array of effects. Though no significant changes
were observed in either of the 29Si CPMAS or DPMAS spectra, the 1H DPMAS spectra
showed a clearly diminished aluminol resonance, and the 27Al DPMAS spectra showed the
appearance of a weak shoulder at ca. 25 ppm, as shown in Figure 5. This suggests that the
interaction of the metal particles with the halloysite occurs at the alumina-rich surface of the
particles, rather than the silica-rich surface.
A
B
C
D
60
40
20
0
-20
-40
-60
-80
27Al Chemical Shift, δ [ppm]
Figure 5: 27Al solid-state NMR spectra of the halloysite-supported
catalysts, corresponding to Pt/HNC (A), Pd/HNC (B), Rh/HNC (C), and
pristine HNC (D). The resonance at 25 ppm, a shoulder off the main
peak, is significantly different in A-C compared to D.

3.5 Fourier Transform Infrared Microspectroscopy
Spectra of the pristine silica contained exhibited a set of absorbance bands broadly
consistent with those in the literature [46-48], but with slightly different peak shapes than are
observed in traditional transmission spectra. This is due to the more complex optical path
taken during infrared transflection [49, 50]. Example spectra are shown in Figure 6.
At the high energy end of the spectrum, a broad peak at 3400 cm-1 is observed which
corresponds to hydrogen-bonded surface hydroxyls, with a shoulder at 3740 cm-1
corresponding to isolated hydroxyls. The broad peak is slightly narrowed when nanoparticles
are adsorbed, while the shoulder remains unaffected. This can be interpreted as a reduction in
the degree of hydration at the surface of the silica, which is consistent with the observations
made by NMR. At the low-energy end of the spectrum there are two regions which are subtly
different between the pristine silica and the catalysts. At 700-1000 cm-1 there is change in the
shape of the peaks, which is the spectral region corresponding to Si-O-H bending and Si-O
stretching modes [51]. This is unlikely to be due to an interaction with the nanoparticles as
these systems have a low surface coverage, and is thus more probably associated with the
reduction in surface hydration.
A
B
C
D
4000
3500
3000
2500
2000
1500
1000
500
Wavenumber, ν (cm-1)
Figure 6: MicroFTIR spectra of the silica-supported catalysts,
corresponding to Pt/SiO2 (A), Pd/SiO2 (B), Rh/SiO2 (C), and pristine
SiO2 (D). The Y axis is shown in absorbance mode, and spectra are
artificially offset for clarity.

Spectra of the pristine alumina were also similar to those in the literature, and showed
significant differences between those collected via transmission or transflection methods.
Transmission spectra were mostly featureless except below 1000 cm-1, where overlapping
bands corresponding to AlO6 bending, stretching, and mixed surface modes are observed but
are largely irresolvable from one another [52, 53]. Comparatively, the transflection spectra
showed clearer definition of these peaks, especially at 1050 cm-1 (Al-O-H bend), 820 cm-1
(surface mode), and 630 cm-1 (Al-O6 stretch) [52]. The enhancement of these features in the
transflection spectra is most likely due to their association with the surface of the alumina
particles, which has been observed in a previous study on infrared reflectance [54].
When nanoparticles were immobilised onto the alumina there significant changes evident in
the transflection spectra. Several features disappeared, including a broad band centred
around 3400 cm-1 (water O-H stretch), a sharper peak at 3295 cm-1 (surface AlO-H stretches),
and the two bands around 1500 cm-1 (adsorbed water bends) [55]. At the same time, some
subtle distortions are evident around 650 cm-1, which together suggest that the surface of the
alumina is becomes dehydrated and slightly strained. Given the high surface coverage of
nanoparticles this is unsurprising, and is supported by the NMR data. Example spectra are
shown in Figure 7.
A
B
C
4000
3500
3000
2500
2000
1500
1000
500
Wavenumber, ν (cm-1)
Figure 7: FTIR spectra of pristine Al2O3 by KBr transmission (A),
pristine Al2O3 by microspectroscopy (B) and Pd/Al2O3 by
microspectroscopy (C). The Y axis is shown in absorbance mode, and
spectra are artificially offset for clarity.

The spectra of halloysite were far more complex than either of the other materials, though
with similarities to both. At the high energy end of the spectra three sharp peaks were
observed at 3700 cm-1, 3628 cm-1, and 3552 cm-1, corresponding to O-H stretches of three
different types of surface hydroxyl [41]. Progressing toward lower energy, two sets of peaks
are evident at 1500-2000 cm-1 (adsorbed water) analogous to both alumina and silica, before
reaching a complex ‘fingerprint region’ below 1200 cm-1. Immobilisation of nanoparticles to the
halloysite caused an array of distortions in the fingerprint region of the spectrum, especially in
the region around 1000 cm-1 assigned in alumina as being due to surface modes, but no
changes to any of the other features in the spectra. This suggests that the interaction between
the halloysite and the nanoparticles is not mediated through surface hydroxyls, but via an
interaction with the alumina-rich edges of the material. Example spectra are shown in Figure 8
A
B
C
D
4000
3500
3000
2500
2000
1500
1000
500
Wavenumber, ν (cm-1)
Figure 8: MicroFTIR spectra of the halloysite systems, Pt/HNC (A),
Pd/HNC (B), Rh/HNC (C), and pristine HNC (D). The Y axis is shown
in absorbance mode, and spectra are artificially offset for clarity.
4. Discussion
It is clear from a cursory analysis of the data that these materials exhibit significant catalytic
activity, beyond that of the pristine support materials. This activity is not surprising, as precious
metal nanoparticles are known as potent hydrogenation catalysts. Of significant interest is

difference in the extent of activity between the systems, and the correlations which can be
drawn with the spectroscopic data.
By far the greatest catalytic activity observed in this study was with the halloysite-supported
systems. In the case of octene hydrogenation, conversion rates of several hundred µmol hr-1
g-1 were maintained by the Pd/HNC catalyst for more than an hour, which corresponded to
100% substrate conversion during this period. If the number of active sites in the catalysts are
estimated using the method of Borodziński and Bonarowska [56], and then used to calculate
turnover frequencies [57], activities in excess of 30,000 hr-1 are obtained. A full tabulation of
TOFs calculated in this fashion is given in Table 3. These compare favourably to what is
reported in the literature, especially as the current reactions were performed under ambient
conditions. While these are a limited set of results, they further support that halloysite is a
viable support material to use for the preparation of supported catalysts. Comparatively, the
alumina and silica-supported catalysts displayed activities of less than half of these values. As
no mass-transfer tests were conducted in this set of experiments, it is possible that these may
be conservative estimates and that selected values could be increased further by optimising
the experimental conditions.
Table 3: Calculated turnover frequencies (TOF) of the thirty-six
systems studied herein. All rates are given to two significant figures.
Calculated TOF (hr-1)
Catalyst
Cyclohexen
e
1-Octene
1,4-Cyclooctadiene
Rh/SiO2
Rh/Al2O3
Rh/HNC
Pd/SiO2
Pd/Al2O3
Pd/HNC
Pt/SiO2
4,400
16,000
7,200
3,200
1,500
6,100
2,800
6,200
11,000
1,200
610
1,000
5,600
2,900
5,400
16,000
4,400
1,700
13,000
24,000
10,000
40,000
55,000
12,000
6,500
Pt/Al2O3
Pt/HNC
25,000
2,000

The most likely cause of the differences in activity between systems is the morphological
properties of the supported metal particles. Other factors being equal, a greater activity should
be obtained when a greater surface area of metal is exposed, which arises with small particle
sizes as well as a homogeneous spatial distribution of metal particles across the support. In
the case of the presently reported systems, no statistically significant difference in particle size
was detected between the three support materials using ANOVA (F(2,6)=0.29, P=0.76),
suggesting particle size is not the cause of the differences.
Spatial distribution data obtained via electron microscopy suggested the metal distribution
was the most homogeneous in the halloysite-supported catalysts, and the least homogeneous
in the silica catalysts, which correlates with the observed activities. The question thus arises
why the metal distribution is better throughout the halloysite catalysts. As the immobilisation
occurred selectively at the edges of the particles, the mechanism must be related to structural
motifs uniquely present at these sites. These are the positions at which lattice defects would
most likely occur, producing absorption sites where interactions between the halloysite
support particles and the metal nanoparticles could occur. These may be acid-base or
electrostatic type interactions, and further work would be required to elucidate the relative
contributions of the two [1, 58].
Though the observations of metal distribution are qualitative, they are supported by
changes observed in the spectroscopic data. The alumina-based systems showed evidence of
surface perturbation in both the FTIR and NMR spectra upon the immobilisation of
nanoparticles, which is consistent with the high degree of surface coverage observed by
electron microscopy. Conversely, the silica-based systems showed less evidence of surface
changes aside from dehydration, which is also consistent with the low degree of surface
coverage of metal particles. The halloysite-based systems exhibited some of the most
significant changes to the surface chemical resonances or adsorption bands, which once
more, is consistent with the good surface coverage and low levels of aggregation observed by
electron microscopy.
Interestingly, the spectroscopic data for the halloysite catalysts showed changes to the
alumina-specific regions of the FTIR spectra and the 27Al NMR spectra, though no such
changes were observed to the silica-specific spectra. This suggests that this method results in

preferential adsorption of nanoparticles to the alumina-rich faces of the halloysite particles.
This was essentially confirmed by electron microscopy, which showed that the metal particles
were primarily located at the edges of the flat halloysite platelets, rather than on the interior of
the tubules. In the future, it is therefore important to consider which morphology is preferential
– though the use of nanotubules is attractive, it may not necessarily be the most effective as a
support material. Most studies using halloysite as a support material, including nearly all of
those referenced in this manuscript, use the tubular variety of halloysite, though our study
suggests that better results may be obtained using other morphologies, depending on the
nature of the synthetic method used.
Another possibility which may have affected the activities is the presence of residual
lanthanum in the systems, which was used to destabilise the colloidal material present.
Lanthanum [59], along with many other lanthanides [60-62], are known to promote catalytic
hydrogenations. It is likely that a similar effect has occurred in the catalysts described in this
research, though the effect is most pronounced in the halloysite catalysts due to their better
spatial distribution of metal than the other catalysts studied. The concept is exciting
regardless, as it suggests that the activity of the catalysts could be further improved by varying
the identity, concentration, or distribution of the lanthanide within the catalysts.
In the future, it would be of interest to develop methods to improve the distribution of the
metal particles, both on halloysite and other support materials. Clearly the synthetic method
used herein is viable, but produces catalysts with varying degrees of metal aggregation
depending on the nature of the support. This in turn carries through to the observable catalytic
activity of the systems, making comparisons between systems difficult. Regardless, taken “as
is”, it can give rise to halloysite-supported catalysts with exceptional activity.
5. Conclusions
We have shown for the first time that highly active halloysite-supported metal catalysts can be
prepared by a simple one-pot reduction method. The resulting catalysts displayed excellent
activity in the hydrogenation of a range of olefins at standard temperature and pressure, with
conversion rates of up to 340 µmol hr-1 g-1 (or 55,000 hr-1 as a TOF) for the hydrogenation of 1-
octene by halloysite-supported palladium nanoparticles. In all instances the halloysite-

supported catalysts displayed superior catalytic and properties to equivalent catalysts
prepared using the more traditional support materials alumina and silica. Characterisation of
the materials showed that the halloysite-supported catalysts had the most favourable textural
properties, with the most even dispersion of metal across their surfaces, though a tendency
toward immobilisation on the alumina-rich edges of particles. This shows that while
nanotubular halloysite may seem the most attractive option as a catalyst support, better metal
distribution may be obtained with other morphologies. Further work should be undertaken to
optimise the activity of these catalysts, and investigate their activity in a range of other
catalytic reactions.
6. Acknowledgements
JEJ and MRM thanks Imerys for the generous donation of halloysite. JEJ additionally thanks
the University of Waikato for the provision of a doctoral scholarship during the course of this
work.

References
1.
2.
3.
E. Joussein et al., Clay Miner., 40(4), 2005, 383-426
M. Soma, G. J. Churchman, B. K. G. Theng, Clay Miner., 27(4), 1992, 413-421
G. J. Churchman, T. J. Davy, L. A. G. Aylmore, R. J. Gilkes, P. G. Self, Clay
Miner., 30(2), 1995, 89-98
4.
R. L. Brathwaite, A. B. Christie, K. Faure, M. G. Townsend, S. Terlesk, Miner.
Deposita, 47(8), 2012, 897-910
5.
6.
7.
8.
G. Lazzara et al., Curr. Opin. Colloid Interface Sci., 35, 2018, 42-50
Y. Zhang, A. Tang, H. Yang, J. Ouyang, Appl. Clay Sci., 119, 2016, 8-17
S. R. Levis, P. B. Deasy, Int. J. Pharm., 243(1–2), 2002, 125-134
Y. M. Lvov, D. G. Shchukin, H. Möhwald, R. R. Price, ACS Nano, 2(5), 2008
814-820
,
9.
M. Du, B. Guo, D. Jia, Polym. Int., 59(5), 2010, 574-582
M. Zou et al., Mater. Res. Bull., 61(0), 2015, 375-382
M. Massaro et al., J. Mater. Chem. A, 5(26), 2017, 13276-13293
T. Tsoufis et al., Chem. Eng. J, 313, 2017, 466-474
L. Gómez, J. L. Hueso, M. C. Ortega-Liébana, J. Santamaría, S. B. Cronin,
Catal. Commun., 56(0), 2014, 115-118
10.
11.
12.
13.
14.
15.
X. Zeng, Q. Wang, H. Wang, Y. Yang, J. Mater. Sci., 52(14), 2017, 8391-8400
M. Malmir, M. M. Heravi, S. Sadjadi, T. Hosseinnejad, Appl. Organomet.
Chem., 32(4), 2018, e4291
16.
17.
18.
V. A. Vinokurov et al., ACS Sustain. Chem. Eng., 5(12), 2017, 11316-11323
V. Vinokurov et al., Ind. Eng. Chem. Res., 56(47), 2017, 14043-14052
A. Philip, J. Lihavainen, M. Keinänen, T. T. Pakkanen, Appl. Clay Sci., 143,
2017, 80-88
19.
20.
21.
22.
23.
M. Massaro et al., RSC Advances, 6(60), 2016, 55312-55318
Q. Wei, R. Shi, D. Lu, Z. Lei, RSC Advances, 6(35), 2016, 29245-29253
C. Li, X. Li, X. Duan, G. Li, J. Wang, J. Colloid Interface Sci., 436, 2014, 70-76
Y. Zhou et al., J. Clean. Prod., 103, 2015, 925-933
S. Mandal, D. Roy, R. V. Chaudhari, M. Sastry, Chem. Mater., 16(19), 2004
3714-3724
,
24.
25.
P. Gallezot, A. Giroir-Fendler, D. Richard, Catal. Lett., 5(2), 1990, 169-174
J. Geboers, S. Van de Vyver, K. Carpentier, P. Jacobs, B. Sels, Chem.
Commun., 47(19), 2011, 5590-5592
26.
R. V. Malyala, C. V. Rode, M. Arai, S. G. Hegde, R. V. Chaudhari, Appl. Cat.
A: Gen., 193(1), 2000, 71-86

27.
28.
P. Munnik, P. E. de Jongh, K. P. de Jong, Chem. Rev., 115(14), 2015, 6687-
6718
D. N. Furlong, A. Launikonis, W. H. F. Sasse, J. V. Sanders, J. Chem. Soc.
Faraday Trans., 80(3), 1984, 571-588
29.
30.
31.
32.
33.
34.
35.
J. E. Jaine, M. R. Mucalo, Powder Technol., 276(0), 2015, 123-128
D. Balzar et al., J. Appl. Crystallogr., 37(6), 2004, 1-14
E. E. Gonzo, M. Boudart, J. Catal., 52(3), 1978, 462-471
O. Domínguez-Quintero et al., J. Mol. Catal A Chem., 197(1-2), 2003, 185-191
L. M. Rossi et al., Green Chem., 9(4), 2007, 379-385
M. Boudart, D. J. Sajkowski, Faraday Discuss., 92(0), 1991, 57-67
M. Adlim, M. Abu Bakar, K. Y. Liew, J. Ismail, J. Mol. Catal A Chem., 212(1–2),
2004, 141-149
36.
37.
G. Deganello, D. Duca, A. Martorana, G. Fagherazzi, A. Benedetti, J. Catal.,
150(1), 1994, 127-134
R. L. Augustine, F. Yaghmaie, J. F. Van Peppen, J. Org. Chem., 49(11), 1984
1865-1870
,
38.
39.
40.
41.
42.
C. Perego, P. Villa, Catalysis Today, 34(3–4), 1997, 281-305
T. Rostamzadeh et al., Langmuir, 33(45), 2017, 13051-13059
D. Banaś et al., Radiat. Phys. Chem., 93(0), 2013, 129-134
B. Szczepanik et al., J. Mol. Struct., 1084(0), 2015, 16-22
G. Tarı, I. Bobos, C. S. F. Gomes, J. M. F. Ferreira, J. Colloid. Interf. Sci.,
̀
210(2), 1999, 360-366
43.
44.
D. W. Sindorf, G. E. Maciel, J. Am. Chem. Soc., 105(6), 1983, 1487-1493
C. E. Bronnimann, R. C. Zeigler, G. E. Maciel, J. Am. Chem. Soc., 110(7),
1988, 2023-2026
45.
K. J. D. MacKenzie, M. E. Smith, Multinuclear Solid-State NMR of Inorganic
Materials (Elsevier Science Ltd., Oxford, 2002)
46.
47.
48.
A. J. Van Roosmalen, J. C. Mol, J. Phys. Chem., 83(19), 1979, 2485-2488
A. J. Van Roosmalen, J. C. Mol, J. Phys. Chem., 82(25), 1978, 2748-2751
E. I. Kamitsos, A. P. Patsis, G. Kordas, Phys. Rev. B, 48(17), 1993, 12499-
12505
49.
50.
51.
52.
53.
54.
B. J. Davis, P. S. Carney, R. Bhargava, Anal. Chem., 82(9), 2010, 3487-3499
B. J. Davis, P. S. Carney, R. Bhargava, Anal. Chem., 82(9), 2010, 3474-3486
H. A. Benesi, A. C. Jones, J. Phys. Chem., 63(2), 1959, 179-182
A. Boumaza et al., J. Solid State Chem., 182(5), 2009, 1171-1176
L. Favaro et al., J. Solid State Chem., 183(4), 2010, 901-908
A. S. Barker, Physical Review, 132(4), 1963, 1474-1481

55.
56.
57.
58.
59.
P. Jackson, G. D. Parfitt, Trans. Faraday Soc., 67(0), 1971, 2469-2483
A. Borodziński, M. Bonarowska, Langmuir, 13(21), 1997, 5613-5620
M. Boudart, Chem. Rev., 95(3), 1995, 661-666
P. Yuan, D. Tan, F. Annabi-Bergaya, Appl. Clay. Sci., 112-113, 2015, 75-93
Y.-z. Xiang, L.-n. Kong, C.-s. Lu, L. Ma, X.-n. Li, React. Kinet. Mech. Catal.,
100(1), 2010, 227-235
60.
H. Imamura, M. Yoshinobu, T. Mihara, Y. Sakata, S. Tsuchiya, J. Mol. Catal.,
69(2), 1991, 271-280
61.
62.
J. T. Lee, H. Alper, J. Org. Chem., 55(6), 1990, 1854-1856
H. Imamura, K. Igawa, Y. Kasuga, Y. Sakata, S. Tsuchiya, J. Chem. Soc.,
Faraday Trans., 90(14), 1994, 2119-2124