Probing chemistry and kinetics of reactions in heterogeneous catalysts

Article abstract of DOI:10.1039/c3sc51477c

Using benzene hydrogenation over Pt/SiO₂ as an industrially-relevant example, we show that state-of-the-art neutron total scattering methods spanning a wide Q-range now permit relevant time-resolved catalytic chemistry to be probed directly in situ within the pore of the catalyst. The method gives access to the reaction rates on both nanometric and atomic length scales, whilst simultaneously providing an atomistic structural viewpoint on the reaction mechanism itself.

Full text of DOI:10.1039/c3sc51477c

Chemical Science
View Article Online
View Journal | View Issue
EDGE ARTICLE
Probing chemistry and kinetics of reactions in
heterogeneous catalysts
Cite this: Chem. Sci., 2013, 4, 3484
Tristan G. A. Youngs, Haresh Manyar,b Daniel T. Bowron,a Lynn F. Gladdenc
a
*
b
*
and Christopher Hardacre
Using benzene hydrogenation over Pt/SiO2 as an industrially-relevant example, we show that state-of-the-
art neutron total scattering methods spanning a wide Q-range now permit relevant time-resolved catalytic
chemistry to be probed directly in situ within the pore of the catalyst. The method gives access to the
reaction rates on both nanometric and atomic length scales, whilst simultaneously providing an
atomistic structural viewpoint on the reaction mechanism itself.
Received 25th May 2013
Accepted 16th June 2013
DOI: 10.1039/c3sc51477c
www.rsc.org/chemicalscience
of the art pulsed neutron facilities is now able to obtain kinetic
information of reaction processes over multiple length scales
spanning the small angle and wide angle regimes
simultaneously.
Introduction
Catalysis in general, and heterogeneous catalysis in particular,
is critical to the production of chemicals worldwide. It has been
estimated that ꢀ80% of all man made materials at some point
in their manufacture use a catalyst1 and the majority of these
are heterogeneously catalyzed reactions. As a consequence of
their importance, it is thought that ꢀ35% of the world's GDP is
reliant on catalysis.2 Many of these reactions are performed in
the liquid phase; however, despite the importance of these
materials and processes, the understanding of multi-phase
interfaces and the impact of these on the overall rate process is
poor. Liquid phase heterogeneously catalyzed reactions are
commonly monitored using liquid sampling or gas uptake due
to the difficulty of obtaining in situ spatially- and chemically-
resolved information on the microscopic scale. Although
nuclear magnetic resonance techniques can obtain spatially-
resolved activity and diffusion information within packed-bed
systems,3,4 these methods are limited to micron resolution.
Neutron scattering is an excellent atomic probe of condensed
matter; however, until now, long data acquisition times have
prevented total scattering techniques from being used to
measure the kinetics of real systems. Small angle X-ray and
neutron scattering has been applied to time-resolved studies of,
for example, micelle aggregation5 and evolution,6 rheochaos
and ow instability,7 and polymerisation processes,8 but the
nature of these techniques necessarily means that information
in the atomistic and chemical regime is lacking. Herein, we
demonstrate that wide Q-range, total neutron scattering at state
In this study, the platinum-catalysed hydrogenation of
benzene to cyclohexane has been used as a proof of principle
that neutron methods can probe the reaction within the pore of
the catalyst. Benzene hydrogenation has been studied exten-
sively in both the liquid phase and vapour using a range of
catalysts including supported Ru,9 Fe,10 Pd,11 Ni12 and Pt13 with a
view to understand the mechanism of the reaction. This reac-
tion is of particular commercial importance due to the wide-
spread use of the product (cyclohexane) as a precursor in the
production of adipic acid and caprolactam en-route to the
eventual production of nylon as well as the drive for cleaner
fuels and the removal of benzene from gasoline.
Experimental details
Neutron diffraction
The structural quantity measured in a total scattering experi-
ment is the interference differential cross section, F(Q), where Q
is the magnitude of the momentum transfer vector in the
scattering process for neutrons of incident wavelength (l) and
scattering angle (2q), dened as:
4p
Q ¼
sinq
(1)
l
F(Q) is formed from a weighted sum of the partial structure
factors, Sij(Q) that reect the pair correlations between atoms of
type i and j:
aISIS Facility, STFC Rutherford Appleton Laboratory, Harwell Oxford, Didcot,
Oxfordshire, OX11 0QX, UK. E-mail: tristan.youngs@stfc.ac.uk; Tel: +44 (0)1235
445050
bCenTACat, School of Chemistry and Chemical Engineering, Queen's University
Belfast, Belfast, BT9 5AG, UK. E-mail: c.hardacre@qub.ac.uk; Tel: +44 (0)2890 974592
cDepartment of Chemical Engineering and Biotechnology, University of Cambridge,
Cambridge, CB2 3RA, UK
X
ꢀ
ꢁ
FðQÞ ¼
2 ꢁ dij cicjbibjðSijðQÞ ꢁ 1Þ
(2)
i; j
dij is the Kronecker delta to avoid double counting the pair
terms, and ci, cj and bi, bj are the atomic fractions and coherent
scattering lengths of the atom types. The interference
3484 | Chem. Sci., 2013, 4, 3484–3489
This journal is ª The Royal Society of Chemistry 2013

View Article Online
Edge Article
Chemical Science
differential cross section is related by Fourier transform to the stirred for 15 min and 9.4 g of triethyl orthosilicate (TEOS)
corresponding radial distribution function, G(r), weighted by (0.044 mol) was then added resulting in a gel with the molar
the atomic density of the system, r:
composition
(TEOS)(C16TMABr)0.3(NH3)11(H2O)144(EtOH)58.
Aer stirring for 2 h, the gel was transferred to a Teon-lined
autoclave and aged for 48 h at 378 K. The white precipitate
obtained was ltered, washed with 200 cm3 of ultra-pure water
and 200 cm3 of methanol and dried overnight at 363 K. The
white powder was calcined in air at 823 K for 5 h (heating rate
2 K minꢁ1). 5 wt% Pt/SiO2 was prepared by incipient wetness
impregnation using platinum nitrate (Johnson-Matthey) as the
platinum precursor. Aer impregnation, the material was dried
at 393 K for 12 h followed by calcination at 773 K for 4 h.
N
ð
1
sinQr
GðrÞ ¼
4pQ2FðQÞ
dQ
(3)
3
Qr
ð2pÞ r
0
G(r) is formed from a weighted sum of the partial pair
distribution functions, gij(r), in similar fashion to F(Q) (eqn (2)).
When investigating nanostructured materials, as in the present
case, it is oen preferable to use the differential correlation
function D(r), derived from the radial distribution G(r) as:14
D(r) ¼ 4prrG(r)
(4)
Batch liquid phase hydrogenation of benzene
This alternate form emphasizes the structural features at
longer distances.
Benzene hydrogenation experiments were carried out in a
100 cm3 Autoclave Engineers' high pressure reactor. In a typical
experiment, the reactor was charged with 20 cm3 of benzene
and 0.25 g of 5 wt% Pt/MCM-41 catalyst. The reactor was purged
three times with N2, the mixture was agitated at 1500 rpm and
heated to 323 K. Aer purging with H2 three times, the reactor
was pressurised to 5 bar, which corresponded to t ¼ 0. The
reaction was monitored by sampling at regular time intervals,
All neutron diffraction measurements were made on the
Near and InterMediate Range Order Diffractometer15 (NIMROD)
at the ISIS Second Target Station Facility of the Rutherford
Appleton Laboratory, Oxford, UK. NIMROD provides high ux
and a wide Q range accessing correlations between <0.1 nm and
ꢀ30 nm. Of critical importance is the instrument's high
neutron count-rate at molecular and intermediate order length
scales which allows short acquisition times to be employed,
thus facilitating the extraction of kinetic information. Benzene
hydrogenation was performed using Pt/SiO2 within the instru-
ment at 296 K and 0.25 bar D2(g). Therefore, to be precise, the
measurements probed the deuteriation of benzene-d6, Scheme
1, as the heavier isotope exhibits more favorable neutron scat-
tering characteristics. Before the neutron diffraction was
undertaken, the catalyst was prereduced by exposing the cata-
lyst to D2 in benzene-d6 at room temperature. This mixture was
then pumped from the system until the diffraction pattern of
the catalyst did not change and then the catalyst pore was
relled with benzene-d6 vapour and then D2.
Synthesis
All chemicals, unless otherwise stated, were purchased from
Aldrich or Qmx and used without further purication. The
mesoporous SiO2 was synthesized by modication of a previ-
16
¨
ously reported procedure by Grun et al. 5 g of n-hexadecyl-
trimethylammonium bromide (C16TMABr, 0.014 mol) was
dissolved in 100 g of ultra-pure water (distilled, deionised
>18 MU) by heating at 318 K, and 26.4 g of aqueous ammonia
(32 wt%, 0.5 mol) and 120.0 g of absolute ethanol (EtOH,
2.6 mol) were added to the surfactant solution. The solution was
Fig. 1 HR-TEM images of the prepared catalyst, acquired in bright field using a
Scheme 1 Deuteriation of benzene-d6 to cyclohexane-d12
This journal is ª The Royal Society of Chemistry 2013
.
Tecnai 200 kV F20 Transmission Electron Microscope with a Field Emission Gun.
Chem. Sci., 2013, 4, 3484–3489 | 3485

View Article Online
Chemical Science
Edge Article
ꢁ1
peak gives a centre of 0.190 A with FWHM of 0.022 Aꢁ1, cor-
˚
˚
˚
˚
responding to a pore spacing of 33.1 A (s.d. 1.48 A). Direct
Fourier transform of the measured empty Pt/SiO2 sample is
given in Fig. 3, which show the differential correlation func-
˚
tions, D(r). The spacing of the pores in the material (d ¼ 33 A) is
clearly visible in the Pt/SiO2 data, with oscillations of this period
˚
extending to beyond 100 A. X-ray diffraction measurements on
˚
the same material suggest a d spacing of 34.6 A, in good
agreement with the neutron value.19
Results and discussion
Reactant/product characterisation
Following measurement of the evacuated catalyst, benzene-d6
was absorbed via capillary condensation of the vapour into the
pores. Aer ve minutes, saturation of the pores was achieved
and no further changes in the structure factor observed (Fig. 2).
This data shows a decrease in the intensity of the rst diffrac-
tion peak together with an increase in complexity of the signal
Fig. 2 Total differential cross scattering data (F(Q)) of dry Pt/SiO2, the system
after exposure to benzene-d6 vapour, and the final system after 9 h reaction time.
Error bars not shown since symbol height is comparable to line thickness.
with analysis of the samples by a Perkin Elmer GC equipped
with a DB-1 capillary column and a FID detector. Under these
conditions, 98% conversion of benzene was obtained aer 24 h.
beyond 1 Aꢁ1. The former is consistent with the pores becoming
˚
lled with a material possessing a scattering length density
(SLD), i.e. benzene-d6, which offers less contrast to the silica
than the evacuated pore. For deuteriated benzene, DSLD ¼
Catalyst characterisation
1.93 ꢂ 10ꢁ6
A
ꢁ2 compared with DSLD ¼ 3.47 ꢂ 10ꢁ6
A
ꢁ2 for the
˚
˚
High resolution TEM images of the prepared catalyst, Fig. 1,
show the expected regular hexagonal array of uniform parallel
channels, while Pt nanoparticles (<1 nm in size) can be seen as
dark spots dispersed within the channels.
In order to interpret the data of the complete system, careful
textural characterization of the catalyst is critical. Traditionally,
adsorption isotherms, e.g. BET, are employed to measure the
porosity of these materials; however, these methods provide
only a rough characterization.17 Neutron methods are advanta-
geous as they provide more comprehensive characterization
with a single probe.18
empty pore thus reducing the contribution of pore–pore
featꢁu1res in the F(Q). The increased signal complexity above
˚
1 A is consistent with additional atomic-scale correlations
corresponding to the absorbed benzene-d6. In real space, the
decrease in intensity of the rst diffraction peak is matched by a
‘washing out’ of the periodic pore correlation (Fig. 3) – the
oscillations at long r are almost completely damped because of
contrast matching.
Aer exposure to D2(g) for 9 h, the F(Q) data (Fig. 2) show an
increase in the rst diffraction peak which is consistent with the
Fig. 2 shows the F(Q) obtained from evacuated Pt/SiO2 from
which a direct characterization of the porous framework can be
obtained. Fitting a Lorentzian prole to the rst diffraction
Fig. 4 Direct Fourier transforms of diffraction data in Fig. 2 along with those
from measurements of neat benzene-d6 and cyclohexane-d12, focusing on the
atomic pair correlations in the short r (chemical structure) region. Curves have
˚
Fig. 3 Direct Fourier transforms of the data shown in Fig. 2. d corresponds to the
pore spacing in the material. The underlying feature at large r apparent in the
pore-filled systems is an artefact of the discrete Fourier transform process, arising
been offset vertically for clarity. Positions of typical C–D (1.09 A), aromatic C–C
˚
˚
(1.397 A), and aliphatic C–C (1.54 A) are shown to aid interpretation. Note the
˚
˚
decreased intensity at 1.397 A and increased intensity at 1.09 and 1.54 A in the
system after 9 h reaction, indicating the hydrogenation of benzene.
from the unavoidable step function at the first data point (Q ¼ 0.01 Aꢁ1).
˚
3486 | Chem. Sci., 2013, 4, 3484–3489
This journal is ª The Royal Society of Chemistry 2013

View Article Online
Edge Article
Chemical Science
formation of cyclohexane-d12 from the hydrogenation of with the porous catalyst. The data also display a marked increase
benzene. Cyclohexane-d12 has a SLD which gives comparable in the intensity of the C–D peak aer reaction indicating
contrast with the bulk SiO2 to that of the empty pore, DSLD ¼ complete conversion of benzene. This was conrmed by
3.24 ꢂ 10ꢁ6 cf. 3.47 ꢂ 10ꢁ6
A
ꢁ2, thus promoting reappearance of extracting the discharged catalyst with n-hexane and analysis by
˚
the rst diffraction peak. Consequently, the long r oscillations in GC-MS which showed only the presence of cyclohexane-d12. This
the D(r) are somewhat restored, Fig. 3. In addition, the presence is consistent with the batch liquid phase reactions undertaken
ꢁ1
of a strong feature at Q ¼ 1.2 A suggests the presence of a using this catalyst as well as literature data21 which showed little
species which is different from benzene and not found at t ¼ 0. formation of the intermediates cyclohexadiene or cyclohexene,
Fig. 4 shows the direct Fourier transform of the initial benzene- and is in agreement with theoretical studies which suggest that
loaded system, the average of the last six datasets collected20 these molecules are not part of the major reaction pathway.22
corresponding to the nal 0.5 h of the study where little further
˚
reaction occurs and those from the pure liquids. Both the
benzene-loaded and pure benzene systems show C–C correla-
Time resolved data
˚
tions at 1.397 A but these are absent aer reaction. Similarly, the The evolution of F(Q) as a function of time, Fig. 5a, illustrates
˚
new feature at ꢀ1.55 A is consistent with the cyclohexane C–C the changes within structural features across atomic, molec-
˚
distance which overlaps with the Si–O peak at 1.59 A associated ular, and mesoscopic length scales simultaneously. By taking
Fig. 5 (a) F(Q)/time domain data over the course of the reaction. (b) Slices taken through (a) at specific Q values, and corresponding exponential fits to the data. Curves are
offset vertically for clarity. Fit equations are: F(Q ¼ 0.19) ¼ (a0 + a1) ꢁ a0exp(ꢁk0t) ꢁ a1exp(ꢁk1t), where a0 ¼ 3.584, a1 ¼ 2.234, k0 ¼ 2.146 hꢁ1, and k1 ¼ 0.136 hꢁ1; F(Q s
0.19) ¼ an – anexp(ꢁknt), where a2 ¼ 0.332 and k2 ¼ 0.368 hꢁ1 (Q ¼ 1.22 Aꢁ1), a3 ¼ 0.143 and k3 ¼ 0.361 hꢁ1 (Q ¼ 3.1 Aꢁ1), and a4 ¼ 0.074 and k4 ¼ 0.333 hꢁ1 (Q ¼
˚
˚
4.3 Aꢁ1). (c) Sample temperature as measured at the top and bottom extremes of the TiZr cell. (d) Pressure of D2 gas present in 4 L supply reservoir.
˚
This journal is ª The Royal Society of Chemistry 2013
Chem. Sci., 2013, 4, 3484–3489 | 3487

View Article Online
Chemical Science
Edge Article
slices at specic Q values with respect to time, the kinetics the pore of a catalyst. In the present case, the fact that pore
corresponding to different length scales in the system can be liquid diffusion is critical in determining the rate of reaction is
obtained, as shown in Fig. 5b (together with correspondiꢁn1g important in understanding how the catalyst design may be
˚
exponential ts). The rst diffraction peak (Q ¼ 0.19 A
)
developed. Improvements in the surface reaction process, i.e.
exhibits both a fast (rate constant k0 ¼ 2.146 hꢁ1) and a slow increasing the number or reactivity of the active sites, will have
(k1 ¼ 0.138 hꢁ1) component, while at Q ¼ 1.2, 3.1, and 4.3 A
little inuence on the catalyst performance whereas addition of
ꢁ1
˚
the ts reveal an intermediate third rate (k2,3,4 z 0.35 hꢁ1). We a solvent, to increase the rate of diffusion, is likely to have a
have already established that the increase in intensity of the signicant promoting effect. This proof of concept study has the
rst diffraction peak is related to the change in contrast potential to enable the temporal structure/spatial changes and
between the mesoporous substrate and absorbed material; the reaction kinetics to be correlated in the future, and there-
however, a second process also contributes to this change on a fore, obtain critical information pertaining to the catalytic
ꢁ1
˚
different timescale. Inspection of the data (1 < Q < 2p A
)
process itself.
reveals no other changes which occur at a rate similar to the fast
component (k0). This rst process is, therefore, unlikely to be
related to benzene and is attributed to the dissociative
adsorption of D2. This is consistent with the rapid and relatively
large increase in temperature over the rst 30 minutes following
introduction of the D2, Fig. 5c. Similarly, the slow component
(k1) does not reect a chemical change and is likely to be related
to the mass transport of the productsꢁw1ithin the pore, i.e. pore
Conclusions
The kinetics within the pores of a heterogeneous catalyst during
the liquid phase reduction of benzene have been probed
directly using neutron diffraction, potentially providing a
method of examining structure-reactivity correlations for these
complex systems in detail, and thus allowing the effects of mass
transport within the catalyst and surface reaction to be decou-
pled. Moreover, the technique takes a step to allow both struc-
tural and spatial data to be tied effectively with the kinetics of
the underlying processes.
˚
diffusion. The change at Q ¼ 1.22 A reects nearest-neigꢁh1-
˚
bour molecular interactions, while those at Q ¼ 3.1 and 4.3 A
are associated with atomistic chemical changes within the
system. These three features evolve with similar time constants
and are correlated with the reduction process and the formation
of the cyclohexane product. Moreover, these time constants
suggest that the overall process is likely to be limited by liquid
diffusion (k1), as this is the slowest rate observed, whilst the
Acknowledgements
We thank the EPSRC and Johnson Matthey for funding under the
CasTech project; M. Kibble and P. Hawkins for support during
neutron diffraction experiments. Experiments at the ISIS Pulsed
Neutron and Muon Source were supported by a beamtime allo-
cation from the Science and Technology Facilities Council
(experiment RB1220486, DOI: 10.5286/ISIS.E.24089729).
reaction itself is governed by the hydrogenation process (k2,3,4
)
rather than the dissociation of D2 (k0). From these results we are
able to elucidate the following process scheme for the reaction,
at conditions of room temperature and 250 mBar D2:
D2(g) / 2D(ads), k0 ¼ 2.146 hꢁ1
Notes and references
1 Z. Ma and F. Zaera, Heterogeneous Catalysis by Metals in
Encyclopedia of Inorganic Chemistry, John Wiley, 2006.
2 F. Zaera, Catal. Lett., 2012, 142, 501.
C6D12(l) pore diffusion, k1 ¼ 0.136 hꢁ1
3 E. H. L. Yuen, A. J. Sederman and L. F. Gladden, Appl. Catal.,
A, 2002, 232, 29.
4 D. Weber, D. J. Holland and L. F. Gladden, Appl. Catal., A,
2011, 392, 192.
Discussion
The necessity for more detailed information with respect to the
kinetic evaluation of catalysts has been known for many years.23
For example, there is a need to replace the ubiquitous concept
5 J. Adelsberger, So Matter, 2013, 9, 1685.
´
6 F. Michaux, N. Baccile, M. Imperor-Clerc, L. Malfatti,
of ‘turnover frequency’ (TOF) with
a
more quantitative
N. Folliet, C. Gervais, S. Manet, F. Meneau, J. S. Pedersen
and F. Babonneau, Langmuir, 2012, 28, 17477.
measure24 preferably with one whereby the rate equation for the
system is known.25 Such characterization has, in the past, been
difficult, especially for liquid phase heterogeneously catalyzed
processes since the rate law commonly contains several
processes including both surface reaction and mass transport.
Each has distinct rate constants, and no single technique has
been able to comprehensively provide the necessary informa-
tion. With this study total neutron scattering has been shown to
be able to provide simultaneously the kinetic information
associated with chemical reaction and mass transport within
7 L. Gentile, B. F. B. Silva, S. Lages, K. Mortensen,
J. Kohlbrecher and U. Olsson, So Matter, 2013, 9, 1133.
8 R. Motokawa, T. Taniguchi, Y. Sasaki, Y. Enomoto,
F. Murakami, M. Kasuya, M. Kohri and T. Nakahira,
Macromolecules, 2012, 45, 9435.
9 J. Bu, J. L. Liu, X. Y. Chen, J. H. Zhuang, S. R. Yan, M. H. Qiao,
H. Y. He and K. N. Fan, Catal. Commun., 2008, 9, 2612;
J. L. Liu, L. J. Zhu, Y. Pei, J. H. Zhuang, H. Li, H. X. Li,
M. H. Qiao and K. N. Fan, Appl. Catal., A, 2009, 353, 282.
3488 | Chem. Sci., 2013, 4, 3484–3489
This journal is ª The Royal Society of Chemistry 2013

View Article Online
Edge Article
Chemical Science
¨
10 K. J. Yoon, P. L. Walker, L. N. Mulay and M. A. Vannice, Ind. 16 M. Grun, I. Lauer and K. K. Unger, Adv. Mater., 1997, 9, 254.
Eng. Chem. Prod. Res. Dev., 1983, 22, 519; K. J. Yoon and 17 D. T. Bowron and A. K. Soper, Neutron News, 2011, 22, 12.
M. A. Vannice, J. Catal., 1983, 82, 457.
18 A. K. Soper, J. Phys.: Condens. Matter, 2012, 24, 064107.
11 U. K. Singh and M. A. Vannice, AIChE J., 1999, 45, 1059; 19 X-ray diffraction measurements were made with Cu Ka
˚
P. Chou and M. A. Vannice, J. Catal., 1987, 107, 140;
P. Chou and M. A. Vannice, J. Catal., 1987, 107, 119;
M. A. Vannice and W. C. Neikam, J. Catal., 1971, 23, 401;
P.-H. Jen, Y.-H. Hsu and S. D. Lin, Catal. Today, 2007, 123, 133.
12 K. C. Metaxas and N. G. Papayannakos, Chem.–Eur. J., 2008,
140, 352; M. Peyrovi and M. Toosi, React. Kinet. Catal. Lett.,
2008, 94, 115; P. G. Savva, K. Goundani, J. Vakros,
radiation (1.5405 A) on a PANalytical X'PERT PRO MPD
diffractometer equipped with reection geometry, a NaI
scintillation counter,
chromator and a nickel lter. The scattered intensities were
collected from 1.5ꢃ to 6ꢃ (2q) by scanning at 0.017ꢃ (2q) steps
with a counting time of 0.5 s at each step. No peaks due to
Pt were observed due to the small Pt crystallite size.
a curved graphite crystal mono-
K. Bourikas, C. Fountzoula, D. Vattis, A. Lycourghiotis and 20 Direct Fourier transform of the individual ve minute
C. Kordulis, Appl. Catal., B, 2008, 79, 199.
13 D. Poondi and M. A. Vannice, J. Catal., 1996, 161, 742;
S. D. Lin and M. A. Vannice, J. Catal., 1993, 143, 539;
datasets is challenging, owing to the reduced statistics in
the high Q regime, therefore, several datasets have been
summed to increase the signal to noise ratio.
V. V. Pushkarev, K. An, S. Alayoglu, S. K. Beaumont and 21 For example, T. Bera, J. W. Thybaut and G. B. Marin, Ind. Eng.
G. A. Somorjai, J. Catal., 2012, 292, 64.
Chem. Res., 2012, 50, 12933.
14 D. A. Keen, J. Appl. Crystallogr., 2000, 34, 172.
15 D. T. Bowron, A. K. Soper, K. Jones, S. Ansell, S. Birch,
22 M. Saeys, M.-F. Reyniers, M. Neurock and G. B. Marin, J.
Phys. Chem. B, 2005, 109, 2064.
J. Norris, L. Perrott, D. Riedel, N. J. Rhodes, 23 B. W. Wojciechowski, Chem. Innovation, 2000, 30, 47.
S. R. Wakeeld, A. Botti, M.-A. Ricci, F. Grazzi and 24 S. Kozuch and J. M. L. Martin, ACS Catal., 2012, 2, 2787.
M. Zoppi, Rev. Sci. Instrum., 2010, 81, 033905.
25 G. Lente, ACS Catal., 2013, 3, 381.
This journal is ª The Royal Society of Chemistry 2013
Chem. Sci., 2013, 4, 3484–3489 | 3489