Mechanistic Insight into the Interaction between a Titanium Dioxide Photocatalyst and Pd Cocatalyst for Improved Photocatalytic Performance

Article abstract of DOI:10.1021/acscatal.6b00982

Understanding the cocatalyst/semiconductor interaction is of key importance for the design and synthesis of next generation photocatalytic materials for efficient hydrogen production and environmental cleanup applications. Here we investigate preformed Pd nanoparticles (NPs) supported on a series of anatase TiO₂ having well-controlled but varying degrees of crystallinity and crystallite size, and explore their photocatalytic performance for H₂ production and phenol decomposition. While tuning the anatase crystallite size significantly influences the photocatalytic performance, varying the TiO₂ crystallinity shows a negligible effect. Interestingly, the optimum quantum efficiency (～78%) for H₂ evolution is achieved with anatase having medium crystallite size (～16 nm), whereas for phenol decomposition, a promotional effect is only observed for anatase with larger crystallite sizes (>20 nm). Surface radical species and radical densities study reveal that the photogenerated charge carriers have been trapped at different sites depending on the crystallite size of anatase. While the excited electrons are only trapped in bulk lattice sites in small anatase (<16 nm), larger anatase particles provide extra surface sites for charge trapping, which benefit charge storage and transportation to Pd surface sites, leading to a more efficient utilization of charge carriers for photocatalysis. Additionally, Pd supported on medium sized anatase (～16 nm) hinders the formation of O₂^?- radicals on TiO₂ surfaces, thus preventing unwanted reoxidation of photogenerated H₂.

Full text of DOI:10.1021/acscatal.6b00982

Subscriber access provided by Western Michigan University
Article
Mechanistic Insight into the Interaction Between a Titanium Dioxide
Photocatalyst and Pd Co-catalyst for Improved Photocatalytic Performance
Ren Su, Nikolaos Dimitratos, Jinjia Liu, Emma Carter, Sultan Althahban, Xueqin
WANG, Yanbin Shen, Stefan Wendt, Xiaodong Wen, J. W. (Hans) Niemantsverdriet,
Bo B. Iversen, Christopher J. Kiely, Graham J. Hutchings, and Flemming Besenbacher
ACS Catal., Just Accepted Manuscript • DOI: 10.1021/acscatal.6b00982 • Publication Date (Web): 24 May 2016
Downloaded from http://pubs.acs.org on June 4, 2016
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a free service to the research community to expedite the
dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts
appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been
fully peer reviewed, but should not be considered the official version of record. They are accessible to all
readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered
to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published
in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just
Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor
changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers
and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors
or consequences arising from the use of information contained in these “Just Accepted” manuscripts.
ACS Catalysis is published by the American Chemical Society. 1155 Sixteenth Street
N.W., Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 30
ACS Catalysis
1
2
3
4
5
6
7
8
9
Mechanistic Insight into the Interaction Between a
Titanium Dioxide Photocatalyst and Pd Coꢀcatalyst
for Improved Photocatalytic Performance
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
†
,‡
╞,╫
‡,╪
╡
║
Ren Su , Nikolaos Dimitratos , Jinjia Liu , Emma Carter , Sultan Althahban , Xueqin
†
†,‡
†
‡,╪
‡,╬
Wang , Yanbin Shen , Stefan Wendt , Xiaodong Wen , J.W. (Hans) Niemantsverdriet , Bo
†
,╧
║
╞,╫
†
B. Iversen , Christopher J. Kiely , Graham J. Hutchings * and FlemmingBesenbacher *
†
Interdisciplinary Nanoscience Centre (iNANO), Aarhus University, DKꢀ8000 Aarhus C,
Denmark.
‡
SynCat@Beijing, SynfuelsChina Co. Ltd., Leyuan South Street II, No.1, Yanqi Economic
Development Zone C#, Huairou District, Beijing, 101407, China.
╞
Cardiff Catalysis Institute, School of Chemistry, Cardiff University, Cardiff, CF10 3AT, UK.
╫
The UK Catalysis Hub, Research Complex at Harwell, Rutherford Appleton Laboratory,
Oxfordshire, OX11 0FA, UK.
╡
School of Chemistry, Cardiff University, Cardiff, CF10 3AT, UK.
║
Department of Materials Science and Engineering, Lehigh University, 5 East Packer Avenue,
18015ꢀ3195, Bethlehem, Pennsylvania, USA.
ACS Paragon Plus Environment
1

ACS Catalysis
Page 2 of 30
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
╧
Department of Chemistry, Aarhus University, Langelandsgade 140, DKꢀ8000 Aarhus C,
Denmark.
╪
State Key Laboratory of Coal Conversion, Institute of Coal Chemistry, CAS, Taiyuan, China
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
╬
SynCat@DIFFER, Syngaschem BV, Eindhoven, The Netherlands
Understanding the coꢀcatalyst/semiconductor interaction is of key importance for the design and
synthesis of next generation photocatalytic materials for efficient hydrogen production and
environmental cleanꢀup applications. Here we investigate preformed Pd nanoparticles (NPs)
supported on a series of anatase TiO having wellꢀcontrolled but varying degrees of crystallinity
2
and crystallite size, and explore their photocatalytic performance for H production and phenol
2
decomposition. Whilst tuning the anatase crystallite size significantly influences the
photocatalytic performance, varying the TiO crystallinity shows a negligible effect.
2
Interestingly, the optimum quantum efficiency (~78%) for H evolution is achieved with anatase
2
having medium crystallite size (~16 nm), whereas for phenol decomposition, a promotional
effect is only observed for anatase with larger crystallite sizes (> 20 nm). Surface radical species
and radical densities study reveal that the photogenerated charge carriers have been trapped at
different sites depending on the crystallite size of anatase. Whilst the excited electrons are only
trapped in bulk lattice sites in small anatase (<16 nm), larger anatase particles provide extra
surface sites for charge trapping, which benefit charge storage and transportation to Pd surface
sites, leading to a more efficient utilisation of charge carriers for photocatalysis. Additionally, Pd
•ꢀ
supported on mediumꢀsize anatase (~16 nm) hinders the formation of O radicals on TiO
2
2
surfaces, thus preventing unwanted reꢀoxidation of photogenerated H2.
ACS Paragon Plus Environment
2

Page 3 of 30
ACS Catalysis
1
2
3
4
5
6
7
8
9
KEYWORDSPhotocatalysis, Metalꢀsemiconductor interaction, Electron spin resonance,
Hydrogen evolution, Phenol decomposition, TiO , Density functional theory
2
INTRODUCTION
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Investigations of advanced catalytic processes have received significant attention, as the
1
ꢀ3
demand for sustainable energy, purified water, and clean air keeps increasing. Photocatalysis
employing particulate semiconductor based materials has experienced tremendous research
activity due to the potential of using solar energy as the driving force for H production, CO2
2
4
ꢀ6
reduction, and water and air remediation. Numerous photocatalyst materials ranging from
7
8,9
10,11
12ꢀ14
simple oxides, nitrides, or sulfides
to complicated mixed phase materials
have been
developed for various applications, however their photocatalytic performance still needs to be
improved before implementation in industry can be anticipated.
The relatively poor performance of most photocatalyst materials can be directly linked to the
ꢀ
+
kinetics of photoꢀexcited electronꢀhole (e −h ) pairs within the semiconductor, where most of
them recombine rather than being separated and used for redox reactions with the surface
1
5ꢀ17
adsorbed reactants.
Engineering the electronic properties of the photocatalyst by tuning the
composition, structure, surface defects (impurities), surface acidity, and polymorph identity may
improve the charge transfer dynamics as well as the optical properties of the material to optimise
1
8ꢀ22
the photocatalytic performance.
Decorating the surface of the photocatalyst by metal (i.e., Au, Pd, Pt) nanoparticles (NPs) is
considered to be a promising and efficient approach to improve the photocatalytic
2
3,24
performance.
The presence of metal NPs as a coꢀcatalyst facilitates the spatial separation of
ꢀ
+
ꢀ
the e −h pairs by trapping the excited e that is transferred from the conduction band (CB) of the
ACS Paragon Plus Environment
3

ACS Catalysis
Page 4 of 30
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
25
semiconductor photocatalysts. Fundamental investigations of metalꢀsemiconductor systems
indicate that the elemental identity, size, and loading of the coꢀcatalyst all influence the kinetics
2
6ꢀ29
of the charge transfer process therefore affecting the photocatalytic performance.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Meanwhile, it has been shown that the nature of the semiconductor support (i.e., polymorph
composition, particle size) can also be used to manipulate the electronic properties of the metal
3
0,31
NPs and thus lead to tunable photoꢀreactivity and selectivity.
Whilst promotional effects have
3
1ꢀ36
37,38
been observed in most cases,
inhibition of the catalyst performance has been also noted.
This heterogeneity suggests that numerous details of the interplay between semiconductor
photocatalyst particles and the coꢀcatalyst NPs are still unclear. It has been well established that
wellꢀcontrolled and deliberately engineered coꢀcatalyst structures can have a significant positive
3
9,40
influence the photoꢀreactivity of the system.
However, the effect of modifying the
semiconductor support to alter the semiconductor/coꢀcatalyst interplay and thus engineering the
overall photoꢀreactivity remains largely unknown presumably because of the lack of a wellꢀ
defined system with tunable physical parameters. Recently, the influence of crystallinity and
crystallite size of the pristine anatase TiO as the photocatalyst has been investigated, indicating
2
that the reaction pathways of photocatalytic phenol decomposition can be tuned by adjusting the
41
anatasenanocrystallite size. Since varying the crystallinity and crystallite size can result in
changes in the surface properties of anatase, we consider that these parameters may be
deliberately controlled to modulate the interaction between metal NPs and semiconductor
materials and thus enabling fine tuning of the overall photocatalytic performance of the system.
From the perspective of the reaction mechanism, most photocatalytic processes involve radical
•
•
ꢀ
3+
ꢀ
+
16
species (i.e., OOH , OH , O , and Ti ) that are created from the photoꢀgenerated e −h pairs.
2
According to electron spin resonance spectrometry (ESR) analysis, the identity, concentration,
ACS Paragon Plus Environment
4

Page 5 of 30
ACS Catalysis
1
2
3
4
5
6
7
8
9
and lifetime of radicals are crucial parameters that dictate the photoꢀreactivity and selectivity of
3
0,42,43
the catalyst.
Surface science studies on model systems have shown that fine tuning of the
coꢀcatalyst/semiconductor interaction may alter the aforeꢀmentioned parameters to allow some
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
44,45
control over the reaction mechanism.
Unfortunately, such an investigation is still missing for
wellꢀdefined systems of real photocatalysts, limiting our understanding of photocatalytic
processes and hence the development of high performance photocatalysts.
Here we explore the coꢀcatalyst/semiconductor interaction using preformed Pd NPs supported
on wellꢀdefined nanoscale anatase TiO particles. The Pd NPs prepared by solꢀimmobilisation
2
46
were deposited onto anatasesynthesised using a supercritical methodology. By producing the
TiO this way we can systematically and independently control the anatase crystallinity
2
(
12%−82%) and mean crystallite size (6−27 nm). The photocatalytic performance of this matrix
of materials was investigated for photocatalytic H evolution and phenol decomposition reactions
2
to examine the interaction between the Pd coꢀcatalyst and anatase support particles. In particular,
we have probed the interaction mechanisms by following the identity and population density of
surface radicals by ESR spectrometry. Additionally, we also employed theoretical calculations to
rationalise the origins of the different radical species and to correlate the photocatalytic
performance with the electronic properties of the materials.
RESULTS AND DISCUSSION
Physical properties of the photocatalysts
The XRD patterns of samples after immobilisation of Pd NPs are shown in Fig. 1. Diffraction
patterns of Pd supported anatase samples with variable crystallinity (Fig. 1a) and crystallite size
(
Fig. 1b) remained almost identical compared to that of their pristine anatase equivalent (Fig.
ACS Paragon Plus Environment
5

ACS Catalysis
Page 6 of 30
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
41
S1), indicating that the solꢀimmobilisation process did not significantly alter the structure,
mean crystallite size or crystallinity of the anatase particles (Fig. 1c). The diffraction peaks of Pd
were not observed in all cases due to its ultraꢀsmall size and low loadings (1 wt%). However, we
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
o
do observe a very broad peak located at ~ 32 (yellow zone in Fig. 1(a) within low crystallinity
4
7
samples (i.e., 12.6%−60%), which can be tentatively assigned to PdO (101). Meanwhile, PdO
0
was not observed within the highlyꢀcrystalline samples (Fig. 1b), indicating the Pd sol supported
on a more disordered TiO surface tends to become more oxidised at ambient conditions.
2
Fig 1. (a) and (b): XRD patterns of asꢀprepared Pd NPs on anatase TiO particles having
2
different crystallinity and mean crystallite size, respectively. ‘A’ indicates the peak positions of
anatase. (c): Comparison of the mean crystallite size of anatase before and after
Pdimmobilisation as derived from Scherrer analysis of the XRD patterns. Based on comparisons
of the XRD patterns, the crystallinity and mean crystallite size of anatase particles was
41,48
considered to be unchanged after Pd addition.
ACS Paragon Plus Environment
6

Page 7 of 30
ACS Catalysis
1
2
3
4
5
6
7
8
9
A subꢀset of Pd/TiO catalyst samples was examined in some detail using an aberration
2
corrected STEM. A typical low magnification HAADF image from the Pd/TiO (Sample S6)
2
photocatalyst is shown in Fig. 2(a). It is clear that the dispersion of Pd NPs on TiO resulting
2
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
from the solꢀimmobilisation process is not particularly homogeneous, where some grains are
more highly decorated with Pd coꢀcatalyst NPs than others. Higher magnification HAADF
images (Fig. 2b) clearly show that the Pd NPs have a tendency to ‘wet’ the TiO , forming an
2
extended flat interface. In addition to the metallic Pd NPs, it was relatively easy to observe subꢀ
nm Pdꢀcontaining clusters dispersed on anatase that were much smaller than the original colloid
size.The formation of flattened interfaces and subꢀnm clusters suggests that some PVAꢀligand
o
disruptionand Pd diffusion has occurred during the drying process (120 C for 16 h). Similar Pd
coꢀcatalyst morphologies and dispersions were found in other Pd/TiO samples irrespective of
2
the degree of crystallinity or the mean crystallite size of TiO support (see Fig. S1ꢀS4 of sample
2
4
8
C1, S2, S4, and S6 in the Supporting Information). The particle size distributions of TiO in the
2
three 82% crystallinity samples (S2, S4, and S6) are plotted in Fig. 2(c) and show very good
agreement with the mean crystallite size analysis derived from the XRD measurements,
indicating that sintering and agglomeration of the TiO particles during Pdimmobilisation is
2
negligible. Although there is some degree of overlap between samples S2, S4 and S6, it is fair to
say that samples S2, S4 and S6 are mainly comprised of small to medium (5ꢀ15 nm), medium
(12ꢀ22 nm), and medium to large (17ꢀ30 nm) sized anatase particles respectively. The
corresponding Pd NP size distributions from these same samples are shown in Fig. 2(c) and have
mean values of 2.3ꢀ3.1 nm range which is consistent with the starting size of the PVA
stabilisedPdꢀcolloid.
ACS Paragon Plus Environment
7

ACS Catalysis
Page 8 of 30
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Fig. 2. (a, b) Representative BFꢀ and HAADFꢀSTEM images of Pd on TiO (sample S6).
2
(
c)Particle size distributions of the TiO particles and Pd NPs of samples S2, S4 and S6.
2
We further characterised the surface composition and valence states of all asꢀsynthesisedPd
NPs dispersed on various anatase particles, as shown in the XPS spectra (Fig. 3). Survey spectra
of the photocatalysts as a function of the TiO crystallinity (Fig. 3a) and mean crystallite size
2
(Fig. 3b) revealed that within the limit of detectability they only consisted of Ti, O, Pd, and
adventitious C. Quantitative analysis showed that the surface concentrations of Pd were ~2.5
wt% for all samples, which is slightly higher than the nominal 1 wt% bulk Pd concentration due
to the preferential location of Pd at the anatase surface.
ACS Paragon Plus Environment
8

Page 9 of 30
ACS Catalysis
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Fig. 3. Survey XPS spectra of Pd NPs supported on (a) 9 nm anatase having different degrees of
crystallinity and (b) 82% crystallinity with varying mean crystallite size, respectively. (c)−(f):
High resolution Ti 2p and Pd 3d spectra of sample sets with varying TiO crystallinity and
2
varying mean crystallite size, respectively. The solid and dashed lines are fitting results of the
4+
0
2+
raw data (dots). The binding energies of Ti , Pd , and Pd are indicated by dotted lines,
4
9,50
respectively.
All catalysts have a nominal Pd loading of 1 wt%.
High resolution Ti 2p spectra of the crystallinity series (Fig. 3c) and mean crystallite size
4+
series (Fig. 3d) suggest that, as expected, exclusively Ti is presented in all cases, again
confirming that the anatase TiO2 particles remained within the solꢀimmobilisation
ACS Paragon Plus Environment
9

ACS Catalysis
Page 10 of 30
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
processessentially unchanged. The Pd 3d spectra of all samples with different crystallinity and
crystallite size are shown in Fig. 3(e) and 3(f), respectively. The Pd 3d signals can be
0
2+
50
deconvoluted at least into two peaks, indicating the presence of both Pd and Pd species.
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
0
2+
However, the Pd /Pd ratios varied significantly depending on the degree of crystallinity
0
2+
exhibited by the anatase TiO particles. Whilst the Pd /Pd ratio increased from 2.4 to 6.7
2
following the increase of crystallinity from 12.6% to 82%, increasing the crystallite size of
0
2+
anatase showed negligible effect on the Pd /Pd ratio, which remained at ~7 (see Table S2 in
4
8
Supporting Information). In conjunction with the evidence from the corresponding XRD
patterns (Fig. 1a), we can conclude that a mixture of Pd/PdO NPs has been deposited on anatase
particles with low crystallinity, whereas for highly crystalline anatase, mainly metallic Pd NPs
are formed that are possibly covered with an ultraꢀthin PdO layer.
The effect of TiO structural parameters on the photocatalytic performance
2
We first studied the effect of the Pd−anatase interaction on photocatalytic H production, as
2
shown in Fig. 4. For comparative purposes, a solꢀimmobilisedPd coꢀcatalyst supported on a
commercial TiO (Degussa P25) was also tested. Time resolved H evolution (Fig. 4a and 4b)
2
2
suggests that all Pd/anatase variants can generate H upon UV irradiation, however, the
2
performance of these variants varied significantly depending on the nature of the anatase
particles. To better visualise the effects of varying TiO crystallinity and crystallite size, we
2
calculated the H2 production rate and apparent quantum efficiency (AQE) of all
48
Pd/TiO photocatalysts, as shown in Figs. 4(c) and (d), respectively. Whilst poorly crystalline 9
2
nm TiO particles (<60%) showed nearly identical performance for H evolution, a slight
2
2
increase in the performance was observed for those samples displaying a higher degree of
crystallinity (>60%). Interestingly, increasing the crystallite size of the most crystalline TiO2
ACS Paragon Plus Environment
10

Page 11 of 30
ACS Catalysis
1
2
3
4
5
6
7
8
9
particles (~82%) from 6.6 nm to 15.9 nm caused a significant enhancement in performance, but a
further increase of the crystallite size to 26.6 nm deactivated the photoꢀreactivity. Remarkably,
the Pd NPs supported on the 15.9 nm, 82% crystalline anatase particles presented an extremely
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
high AQE for H evolution (~78%), which was about double that found for Pd NPs with similar
2
particle size supported on commercial P25 TiO . Such an AQE value is even higher than that
2
reported previously for the AucorePdshell coꢀcatalyst on P25 (73%), which at the time of
publication was superior to that of any photocatalyst reported using renewable organic chemicals
2
7
as scavengers.
Fig. 4. Timeꢀresolved photocatalytic H evolution using 1 wt% Pd NPs supported on (a) 9 nm
2
anatase with different crystallinity and (b) 82% crystallinity and varying crystallite size,
respectively. A 1 wt% Pd NPs on Degussa P25 was tested for comparison. A 25 vol% of ethanol
solution was used for all tests. (c)−(d): Derived H production rates and AQE as a function of
2
varying crystallinity and varying crystallite size of the anatase particles. AQE =
ACS Paragon Plus Environment
11

ACS Catalysis
Page 12 of 30
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
2
n(H )/n(incident photons), where n(H ) and n(incident photons) are the numbers of generated
2
2
4
8
H and incident photons, respectively.
2
We further investigated the correlation of the Pd−anatase interaction and photocatalytic
oxidation performance through phenol decomposition reaction, as shown in Fig. 5. Previous
kinetic analysis based on inꢀsitu UVꢀvis and MS spectrometry suggested that the optimum
performance for full phenol oxidation was always observed for samples that produced minimum
fraction of phenolic species (i.e., hydroquinone and benzoquinone), as these redox couples
consume the photoꢀgenerated charge carriers continuously thus reducing the conversion of
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
39,41
phenol to CO2.
Therefore the decomposition rate of total phenolic species shown in Fig. 5
reflects the performance of the photocatalysts for the full oxidation of phenol. For 9 nm TiO2
with different degrees of crystallinity, surface decoration with Pd NPs showed negligible effects
on the phenol decomposition rate for the more crystalline TiO particles (>75%) but a decrease
2
in phenol decomposition for the less crystalline TiO (<60%) particles, respectively (Fig. 5a). We
2
did observe that the formation and decomposition of phenolic intermediates (hydroquinone and
benzoquinone) were slightly improved upon increasing the crystallinity of the TiO particles
2
4
6
(Fig. S8b and S8c), which resulted in a minor enhancement towards the full decomposition of
4
1
phenol using Pd/TiO compared to that of their pristine TiO counterparts.
2
2
ACS Paragon Plus Environment
12

Page 13 of 30
ACS Catalysis
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Fig. 5. Evolution of total phenolics during photocatalytic phenol decomposition using the 9 nm
pristine anatase (left column) and 1 wt% Pd NPs modified anatase (right column) with different
crystallinity (a), and using the 82% pristine anatase (left column) and 1 wt% Pd NPs modified
41
anatase (right column) with different mean crystallite size, (b).
Interestingly, Pd NPs supported on highly crystalline (82%) TiO with tunable crystallite size
2
presented a different phenomenon. Whilst Pd NPs supported on small anatase particles (<15.9
nm) showed identical performance in phenol decomposition compared to their Pdꢀfree
counterparts, a further increase in the mean size of anatase (>15.9 nm) boosted the photocatalytic
oxidation of phenol significantly (Fig. 5b). The evolution of hydroquinone and benzoquinone
using undecorated and Pdꢀdecorated TiO (82% crystallinity) with different mean crystallite
2
4
1,46
sizes are also presented in Figs. S8(f) and S8(g) respectively.
These plots clearly reveal that
the interaction of Pd with anatase can better promote the decomposition of both intermediates
when the mean size of the anatase crystallites exceeds ~15.9 nm, which eventually accelerates
the complete decomposition of phenol.
We can satisfactorily explain some of the observed phenomena using some prior experience
and knowledge. For example, amorphous TiO is generally considered to be poor for
2
ACS Paragon Plus Environment
13

ACS Catalysis
Page 14 of 30
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
41
photocatalysis. Furthermore the optimised photocatalytic H production noted for medium
2
sized crystalline anatase decorated with Pd can be considered as a deliberate tradeꢀoff between
surface area effects and quantum size effects. However, the underlying mechanism is not
sufficiently well understood to fully explain the following issues: (i), the fact that increasing the
degree of crystallinity of the anatase does not seem to significantly affect the photocatalytic H2
production and phenol decomposition; and (ii), the optimised crystallite size required for
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
photocatalytic H production and phenol decomposition is markedly different (i.e., 15.9 nm
2
versus 26.6 nm). Surface science studies on model TiO photocatalysts revealed that surface
2
defects (i.e., vacancies and adatoms) and bulk defects affect the charge transfer and trapping
kinetics, which in turn control the surface adsorbed species that influence the reaction pathways
1
5,51
and determine catalytic performance.
However, the above conclusions were drawn from data
collected using the surface science approach based on single crystal TiO (normally a reduced
2
rutile phase) under ultraꢀhigh vacuum conditions, which are not directly comparable to the TiO2
NPs (normally pure anatase or mixed anatase/rutile polymorphs [P25]) that are used in real
photocatalytic reactions under ambient conditions.
To rationalise the Pd−anatase interaction under different reaction conditions, we performed
solid state ESR analysis on a selected subꢀset of anatase samples and their corresponding Pd
decorated counterparts, as shown in Fig. 6. When pristine TiO samples were irradiated under
2
ꢀ5
3+
─
high vacuum (Fig. 6a, Pcell = 10 mbar), well characterised Ti centres, CB electrons (e CB), and
•─
─
4+
3+
2─
+
•─
O were formed as expected due to charge separation (e + Ti → Ti , O + h → O ). Note
─
CB
that e
showed a g value of 2.003, which can effectively be considered as “free electrons” (g =
2
.0023). The ESR signal at g=2.003 could be also induced by the defect states at particleꢀparticle
52
interfaces on reduced anatase, but such contribution should be negligible in our case as all
ACS Paragon Plus Environment
14

Page 15 of 30
ACS Catalysis
1
2
3
4
5
6
7
8
9
samples consist of stoichiometric TiO . However, the relative concentrations of different radical
2
3
+
species varied significantly as the crystallinity changed. Whilst mainly Ti in regular bulk lattice
3+
positions (bulk Ti , g = 1.992, g = 1.960) were detected for TiO with poor crystallinity or
/
/
2
┴
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
5
3
3+
3+
small sizes (C1, C3, C5, and S3), Ti in a disordered environment (nearꢀsurface Ti , g =
─
•─
1
.93), e CB, and O were the dominant species found for wellꢀcrystalline TiO with larger
2
54
3+
crystallite sizes (S4 and S6). Because the most intense surface Ti features were observed on
3
+
samples S6, and the extensive bulk Ti features were observed on samples C1, C3, C5, and S4,
3
+
we surmise that large anatase is responsible for the surface Ti and small anatase originates the
3+
bulk Ti . Moreover, increasing the crystallinity from 12.6% (C1) to 60.3% (C3) and then to
3
+
─
CB
82% (C5 and S3) solely resulted in an increase of the bulk Ti , whereas most of e
converted
3+
to nearꢀsurface Ti centres when the TiO size grew from 16.4 nm (S4) to 27 nm (S6). Therefore,
2
─
CB
the e
should be solely generated on medium sized anatase as the corresponding ESR peak was
significantly observed on sample S4 and visible on sample S6. Interestingly, the signals of excess
charges vanished almost completely in all samples when preformed Pd NPs were immobilised
onto the TiO particles (see Fig. 6c), indicating most of excited electrons were successfully
2
3+
transferred from TiO and localised at Pd surface sites. In comparison, Ti species in regular
2
bulk lattice positions were predominantly detected for pure P25 after irradiation under
high vacuum, which is a similar situation to that of sample S3 (Fig. S10a). These trapped
46
electrons can also be transferred to Pd surface sites (Fig. S10c).
ꢀ1
We also performed similar experiments under low vacuum conditions (Pcell = 10 mbar) to
examine the effect of oxygen and water upon radical formation under UV irradiation, as shown
in Figs. 6(b) and 6(d) for pristine TiO and Pd/TiO , respectively. Surprisingly, whilst surface
2
2
3
+
•─
•─
•─
Ti , O , and O were observed for large anatase particles (S4 and S6), O was the only
2
2
ACS Paragon Plus Environment
15

ACS Catalysis
Page 16 of 30
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
radical species detected for poorly crystalline or small grain TiO samples (C1, C3, C5, and S3),
2
and the intensity was much weaker compared to that observed for samples S4 and S6. The
spectra changed considerably in the presence of Pd NPs on TiO particles, as shown in Fig. 6(d).
2
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3+
Whilst the characteristic nearꢀsurface Ti signals observed for large pristine anatase vanished
2
+
due to charge transfer from TiO to Pd, weak fingerprints of Pd signals were observed in the
2
55
low crystallinity sample (C1 and C3), which agrees well with our XRD and XPS observations.
•─
Noticeably, the number of surface O2 radicals decreased when increasing the crystallinity from
12.6% (C1) to 60.3% (C3) and to 82% (C5 and S3), as well as increasing the crystallite size from
9.3 nm (C5) to 11.3 nm (S3) and to 15.9 nm (S4), and reached a minimum with a mean size of
•ꢀ
1
5.9 nm (S4). The concentration of O2 radical shown in Fig. 6(d) can also be associated with
the divergence in the content of medium sized anatase between samples S3, S4, and S6. Surface
3+
•─
Ti and O2 were the dominant species when pure P25 was irradiated under low
•─
vacuum, and the intensity of the O2 signal was similar to that of S4 (Fig. S10b).
3+
However, Ti cations were still evident in the ESR spectra of Pd on P25, indicating the
interfacial charge transfer between P25 and Pd to be relatively poor compared to that of
46
the Pdꢀanatase system (Fig. S10d).
ACS Paragon Plus Environment
16

Page 17 of 30
ACS Catalysis
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Fig. 6. ESR spectra of selected pristine TiO and 1 wt% Pd/TiO samples that have been UV
2
2
irradiated for 30 min at 77 K under high vacuum (a and c) and low vacuum (b and d),
respectively. (e) an (f): Integrated intensities of all ESR spectra recorded under high vacuum and
low vacuum, respectively. Lines in (e) and (f) are guidelines only.
We have further estimated the total numbers of active paramagnetic species by integrating the
ESR signals, as shown in Fig. 6(e) and 6(f). Under high vacuum conditions (Fig. 6e), small TiO2
particles with low crystallinity (C1, C3, and C5) generated far fewer active paramagnetic
speciescompared to that of other TiO particles. However, for large TiO particles with high
2
2
crystallinity (S3, S4, and S6), the number of Pd NPs seems to become insufficient for trapping
ACS Paragon Plus Environment
17

ACS Catalysis
Page 18 of 30
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
3+
excess Ti . Under low vacuum conditions (Fig. 6f), the densities of surface active paramagnetic
speciesincreased sharply only when the crystallinity and size of TiO reached certain thresholds
2
•─
(
> 80% and > 15.9 nm), and a minimum concentration of surface O2 species was observed for
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Pd NPs supported on TiO with a crystallite size of 15.9 nm (S4).
2
Fig. 7. (a) The three layer anatase (101) surface model with five different oxygen vacancies
1
4
(
V ). (b) The surface DOS of Ti3d with VO and V . The arrow indicates the midꢀgap state.
O O
1
4
O
(cand d) LUMO of the anatase (101) surface with VO and V , respectively.Ti atom: gray, O
atom: red.
Since the density and identity of surface radicals are a reflection of the thermodynamics and
kinetics of the charge carriers under reaction conditions, it is crucial to obtain a deep
understanding of the charge carriers. In this connection we note that various defects such as OH
5
6ꢀ58
species, lowꢀcoordinated ions, and oxygen vacancies (V ) may exist in pure TiO ,
turning it
O
2
into an nꢀtype semiconductor, where the electrons and holes are the majority and minority charge
carriers, respectively. However in the present study all samples were stoichiometric anatase (see
Fig. 3) that should be close to behaving like intrinsic semiconductors. Nevertheless, excess
ACS Paragon Plus Environment
18

Page 19 of 30
ACS Catalysis
1
2
3
4
5
6
7
8
9
charge may occur due to minority defects. In the following we focus on the excited electrons for
H evolution with ethanol as the hole scavenger. As the phenol decomposition is more
2
complicated, we would have to consider both electrons and holes since phenol and O have to be
2
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
oxidised and reduced simultaneously.
In order to reveal the effect of charge carriers on the photocatalytic performance, we have
modelled a threeꢀlayer anatase TiO (101) surface with a V placed at various surface and subꢀ
2
O
surface (bulk) sites (Fig. 7a), and calculated the surface DOS of the system with different V and
O
+
the LUMO distributions, as shown in Fig. 7. Our focus here is the creation of Ti3 in the system
rather than the structure of the defects. Here, introducing one V corresponds to the creation of
O
3
+
4+
two Ti , which is equivalent to two trapped electrons at the Ti sites.
1
O
2
O
3
O
X
O
Among these five possible V sites shown in Fig. 7(a), V , V , V , and V can be
O
4
assigned to different variants of surface V , whereas VO is a subsurface V defect. The
O
O
1
2
O
3
X
calculated formation energies derived from the model with one single V , V , V , V , and
O
O
O
4
O
46
V defect are 3.99, 4.90, 4.57, 4.27, and 3.88 eV at 0 K, respectively (see Fig. S11), indicating
that oxygen vacancies are most stable on site 4 (i.e., subsurface) followed by site 1 (surface)
rather than on the other sites considered.
1
4
The surface DOS shown in Fig. 7(b) reveals the effect of VO and VO on the Ti 3d orbitals,
which constitute the conduction band of TiO . Interestingly, a midꢀgap state close to the
2
1
unoccupied Ti 3d orbitals was observed for anatase with a VO type defect (see arrow), whereas
4
such a feature was not found for anatase with a VO defect. This Ti 3d midꢀgap level induced by
the surface V provides additional trapping sites for the excited electrons, which could be
O
transferred to the Pd NPs for use in subsequent reduction reactions. Meanwhile, the location of
46
V has a negligible effect on the O 2p orbitals in the valence band (Fig. S12), thus the exact
O
ACS Paragon Plus Environment
19

ACS Catalysis
Page 20 of 30
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
location of V does not influence the hole species significantly. The relatively high intensity
O
•─
ESR signals of O observed in samples S4 and S6 in Fig. 6(a) are probably due to the enhanced
charge separation caused by the midꢀgap states that originate from surface defects. We further
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
O
4
O
mapped the LUMO of the anatase (101) surface with V and V , as shown in Fig. 7(c) and (d).
1
O
Obviously the LUMO located at the surface Ti sites (adjacent to the V site) is more feasible for
charge trapping and subsequent charge transfer than the LUMO located at the subsurface Ti sites
4
O
(
adjacent to the V site). Previous surface science studies and DFT calculations showed that the
5
7
subsurface and bulk V ’s are significantly more stable on highly reduced anatase (101).
O
However, in our case the samples are all very close to stoichiometric TiO (see Fig. 3), TiO2
2
particles are used instead of single crystals, and we are also not working under ultraꢀhigh vacuum
conditions. Hence, other surface defects can also potentially be envisaged to be present on our
3+
57
samples that lead to Ti excess charge. Although all calculation results were derived at 0 K,
the electrons that have been trapped at the surface sites should be thermodynamically much
easier to utilisefor charge transfer even at room temperature. Moreover, the additional
1
O
unoccupied Ti 3d states observed for the model with V should be temperature independent.
We can now correlate the photocatalytic performance with the charge carriers and the
properties of the photocatalyst materials by combining the ESR results with the insight of the
─
theoretical calculations. For the H evolution reaction, “free electrons” (e CB) and surface
2
3
+
trapped electrons (nearꢀsurface Ti ) are thermodynamically and kinetically more favourable to
+
reduce H to H compared to electrons trapped within the subsurface or in the bulk (i.e., lattice
2
3
+
3+
Ti ). Ti excess charge is available at the surface of large anatase crystals having high
crystallinity, resulting in the formation of surface midꢀgap states below the conduction band, thus
improving the charge trapping and transfer efficiency. For anatase with low crystallinity and
ACS Paragon Plus Environment
20

Page 21 of 30
ACS Catalysis
1
2
3
4
5
6
7
8
9
3+
small crystallite size, Ti excess charge is available predominantly in the subsurface or bulk,
which results in a relatively poor charge trapping efficiency. Moreover, the unoccupied states are
located at the subsurface or bulk that are unfavourable for interfacial charge transfer, leading to a
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
•─
relatively poor photocatalytic performance. Furthermore, a lower density of surface O2 radicals
in medium sized anatase is ideal to inhibit reꢀoxidation (i.e., back reactions) of the freshly
4
0
synthesised H during the photocatalytic process. Accordingly, an optimum level of
2
photocatalytic H evolution was observed for highly crystalline TiO (82.6%) with moderate
2
2
crystallite size (15.9 nm). The relatively poor performance of Pd on P25 can mainly be
associated with the bulk trapped electrons that require extra energy for charge transfer to
the Pd NPs.For phenol decomposition, however, a situation is favourable where a dense
population of surface radicals exist, which makes breaking of the aromatic rings more feasible.
Thus the best performance for phenol decomposition was observed for highly crystalline TiO2
(86.3%) of large size (26.6 nm).
CONCLUSIONS
We have synthesisedphotocatalysts comprised of Pd NPs supported on TiO (anatase)
2
materials having well controlled crystallinity and crystallite size. Whilst mainly metallic Pd NPs
were deposited on highly crystalline TiO regardless of the anatase crystallite size, about
2
2+
2
0−30% of the Pd was presented in the Pd form (i.e.,PdO) when using poorly crystalline TiO
2
particles. We have investigated the interaction of the Pd coꢀcatalyst with TiO for photocatalytic
2
H production and phenol decomposition reactions. Tuning the crystallinity of the TiO particles
2
2
showed a negligible effect on the photocatalytic performance of both reactions in comparison to
varying the crystallite size. For H production, a remarkable apparent quantum efficiency of
2
~
78% was obtained when using anatase particles with a medium crystallite size of ~16 nm.
ACS Paragon Plus Environment
21

ACS Catalysis
Page 22 of 30
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Meanwhile, an improvement in phenol decomposition was only observed for anatase particles
with larger crystallite size (>20 nm). ESR analysis revealed that larger anatase particles (>16 nm)
offer more nearꢀsurface sites for charge trapping, whereas solely bulk trapping sites were
observed for smaller anatase. Such additional surface sites benefit both charge storage and
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
charge transfer from TiO to the Pd coꢀcatalyst surface sites, and thus provide a more efficient
2
utilisation of charge carriers to enhance H evolution and phenol decomposition. Additionally,
2
•─
the presence of Pd on mediumꢀsize anatase particles (~16%) limited the formation of O2
radicals on TiO surfaces, preventing unwanted reꢀoxidation of photoꢀgenerated H . DFT
2
2
calculations on surface DOS and LUMO distribution illustrate that the surface trapping
sites for photoexcited electrons in medium and large grain anatase can indeed be
associated with different locations of defects such as V species. Irrespective of the exact
O
3+
nature of the defects, it is clear that surface Ti sites are beneficial for photocatalytic
reactions, since they facilitate interfacial charge transfer from TiO to the Pd NPs.Our
2
results suggest that varying the size of the semiconductor is an efficient method to tune
the electronic interaction between metal NPs and semiconductor to improve the
0
2+
photocatalytic performance.Further investigation on the Pd size effect, Pd : Pd ratio, and
Pd/TiO interface may further tune the electronic properties of the material thus
2
improving the photocatalytic performance.
EXPERIMENTAL SECTION
Sample preparation
Synthesis of TiO particles:A supercritical synthesis approach using a continuous flow
2
46
reactor was employed to prepare TiO nanoparticles. Isopropanol mixed with deionized
2
(
DI) water and titanium isopropoxide (TTIP, ACROS, 98%) were used as the
ACS Paragon Plus Environment
22

Page 23 of 30
ACS Catalysis
1
2
3
4
5
6
7
8
9
supercritical solvent and the reactant, respectively. The crystallinity and crystallite size of
pure anatase TiO were controlled by tuning the concentration of TTIP, composition of
2
the solvent, flow rate, temperature, and pressure, as listed in Table S1 in the Supporting
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
48
Information. The suspensions of asꢀprepared anatase particles were centrifuged, washed
with DI water, and dried overnight at 120 °C. A series of samples were prepared where
the crystallinity of anatase was tuned independently from 12% to 82% while maintaining
a constant crystallite size of ~9 nm (samples C1ꢀC5). A second series was made where
the anatase crystallite size was varied from ~6 nm to ~27 nm while keeping a constant
41,48
crystallinity of ~82% (samples S1ꢀS6), as shown in Fig. S1.
Deposition of Pd NPs on TiO particles:A standard sol immobilisation method was utilised
2
3
2
to deposit 1 wt% of Pd NPs on the various anatase particles described above. Aqueous
solutions of PdCl (Johnson Matthey), poly vinyl alcohol (PVA) (1 wt% aqueous solution,
2
Aldrich, MW=10000, 80% hydrolyzed), and NaBH (0.1 M) were freshly prepared and mixed to
4
generate the Pd colloid for deposition. Immobilisation of the Pd colloid was then performed by
adding the anatase support into the fresh sol (acidified to pH 1ꢀ2 by sulfuric acid) under vigorous
stirring conditions. The catalyst slurry was then filtered, washed thoroughly with DI water, and
dried at 120 °C overnight. For comparative purposes, a 1 wt% Pd supported on commercial TiO2
(
Degussa P25) was also prepared by sol immobilisation.
Materials characterisation
Powder Xꢀray diffraction (XRD) was used to acquire crystallographic information of the
samples by using an Xꢀray diffractometer (SmartLab, Rigaku) operating with CuꢀK radiation. A
α
ꢀ1
scan rate of 0.04 degꢁs and integration time of 10.0 s was used for all measurements. Rietveld
refinement was applied to determine the peak area thus the crystallite size from the diffraction
ACS Paragon Plus Environment
23

ACS Catalysis
Page 24 of 30
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
41,59
patterns.
The crystallinity of the sample is defined as the percentage of crystallites in a
mixture of crystalline and amorphous material, thus can be measured by mixing the sample with
a CaF reference (1:1 weight ratio), which is 100% crystalline. The crystallinity of the sample
2
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
was calculated by comparing the integrated areas of the diffraction peaks of anatase (101) with
6
0
CaF (111) by the following equation:
2
ꢀ
ꢁꢂꢃꢄꢅꢆꢆꢇꢈꢇꢄꢂ = 0.763 × ꢉꢊꢋꢊꢌꢊꢍꢎ(ꢏꢐꢏ)/ꢉ
× 100%
(1)
ꢑꢊꢒ (ꢏꢏꢏ)
ꢓ
Highꢀangle annular darkꢀfield (HAADF) images of some of the photocatalyst samples were
acquired using a 200 kV JEOL ARM 200CF scanning transmission electron microscope (STEM)
equipped with a CEOS probe aberration corrector. The samples for STEM analysis were
prepared by dry dispersing the catalyst powder onto a holey carbon TEM grid.
A Kratos Axis Ultra spectrometer equipped with a monochromatic Al K source (10 mA, 15
α
kV) was employed to perform Xꢀray photoelectron spectroscopy (XPS) analysis. Pass energies
of 160 eV and 40 eV were used for survey scans and high resolution scans, respectively.
Calibration of the binding energy was referenced to adventitious carbon, which has a binding
energy of 284.8 for the C1s peak.
Photocatalytic performance measurements
Photocatalytic H production and phenol decomposition reactions were both performed to
2
evaluate the performance of each material. Prior to the experiment, fresh photocatalyst (50 mg)
was dispersed in DI water and subsequently cleaned by UV irradiation (Optimax 365, 365 nm
1
7
ꢀ1
LED, photon flux: 4 x 10 photonsꢁs ) for 2 h to remove the PVA ligand.
For H production, 6.25 mL of ethanol (99 vol%) was added to the reactor to form a 25 vol%
2
ethanolꢀcatalystꢀwater suspension (25 mL), which was transferred to a leakꢀtight reactor that was
connected to a quadruple mass spectrometer (QMS, Hiden HPRꢀ20) for the analysis of evolved
H .The reactor was then evacuated using a byꢀpass pump until the dissolved O level was
2
2
ACS Paragon Plus Environment
24

Page 25 of 30
ACS Catalysis
1
2
3
4
5
6
7
8
9
reduced to below 5 µM. UV irradiation of 2 h duration was then employed to photocatalytically
evolve H under conditions of continuous stirring at room temperature using the same LED light
2
ꢀ
source as described above. The partial pressures of m/e = 2 (H ), 18 (H O), 28 (N ), 32 (O ), and
2
2
2
2
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
4
4 (CO ) species were monitored inꢀsitu.
2
For phenol decomposition, 1 mL of phenol solution (20 mM) was added to the reactor to form
a 400 µM phenolꢀcatalystꢀwater suspension (50 mL). An adsorption/desorption equilibrium was
achieved by keeping the suspension in the dark for 1 h. Then UV irradiation was commenced
using the aforementioned LED light source to initiate the photocatalytic decomposition of
phenol. A 1.5 mL aliquot of the suspension was collected at given time intervals, centrifuged,
and analysed by UVꢀvis spectrometry (UVꢀ1800, Shimadzu, JP).
Further details of the photoꢀreactivity measurements and related calculations are summarised
28,39,48
in the Supporting Information and can also be found elsewhere.
ESR characterisation of the photocatalysts
An Xꢀband Bruker EMX spectrometer equipped with a highꢀsensitive cavity (ER 4119HS) was
utilised to record the ESR spectra of pure anatase and the various Pd/anatase samples that have
been treated under different oxygen partial pressures. This has been realised by evacuating the
ꢀ
5
ꢀ1
cell using a turbo pump system (P : 1e mbar) or a rotary pump (P : 1e mbar), respectively.
cell
cell
Before evacuation, the samples (10 mg) were cleaned by UV irradiation for 2 h at room
o
temperature. The samples were then evacuated under dynamic vacuum at 80 C (for low
o
vacuum) and 120 C (for high vacuum) to remove excess surface adsorbed water. The evacuated
cell was then irradiated using the same UV light for 30 min at 77 K, and then transferred into the
cavity which was held at 120 K to record ESR spectra. The modulation field and microwave
ACS Paragon Plus Environment
25

ACS Catalysis
Page 26 of 30
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
power employed were 100 kHz and 10 mW, respectively. The value of g tensor for each
4
3,61
measurement was determined using a 2,2ꢀdiphenylꢀ1ꢀpicrylhydrazyl (DPPH) standard.
Theoretical calculations
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Density functional theory (DFT) was used to calculate the formation energy of the photoꢀ
excited electrons that are trapped at surface and subsurface sites by creating surface and
subsurface oxygen vacancies (V ). Subsequently, the impact of various surface and subsurface
O
V defects on the surface density of states (DOS) and the distribution of the lowest unoccupied
O
molecular orbitals (LUMO) were also evaluated. A three OꢀTiꢀO layer anatase model (144
2
atoms) was used to simulate the (101) surface (surface area 10.44×15.57 Å ). Vienna AbꢀInitio
Simulation Package (VASP) with the frozenꢀcore projectedꢀaugmented wave (PAW) method
was used with application of the generalised gradient approximation (GGA) of PedewꢀBurkeꢀ
62ꢀ65
Ernzerhof (PBE).
The Ti3d states were described by GGA + U with a U value of 3.5 eV to
6
6,67
correct for the onꢀsite Coulomb interactions.
The energy and the electronic properties of the
anatase (101) surface were computed using a plane wave cutoff of 500 eV and a Monkhorst–
68
Pack grid of (2×2×1) kꢀpoints. The computational resources for the project were supplied by
the Tianheꢀ2 in Lvliang, Shanxi province. More details of the calculation parameters can be
4
6
found in the Supporting Information.
AUTHOR INFORMATION
Corresponding Author
*
hutch@cardiff.ac.uk (Graham Hutchings), fbe@inano.au.dk (FlemmingBesenbacher)
Author Contributions
ACS Paragon Plus Environment
26

Page 27 of 30
ACS Catalysis
1
2
3
4
5
6
7
8
9
The manuscript was written through contributions of all authors. All authors have given approval
to the final version of the manuscript.
ACKNOWLEDGMENT
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
We thank the financial support from the iNANO Centre through the Danish Strategic Research
Council and the Carlsberg Foundation, and the Centre for Materials Crystallography (Denmark
National Research Foundation, DNRF93). We also acknowledge the EPSRC (EP/K014854/1)
and NSFC (project number: 21503257) for financial support. R.S. acknowledges financial
support from Syngaschem BV and SynfuelsChina Technology Co. Ltd. for his stay at Cardiff
University in 2014. RS would like to thank Prof. Damien Murphy at Cardiff University for EPR
training and fruitful discussions and Dr. Jose Gracia at SynCat@Beijing for discussions in
calculations.
Supporting Information.The sample preparation &characterisation, apparent quantum
efficiency, evolution of the phenolic intermediates, additional ESR spectra, and additional
calculation details are available. This material is available free of charge via the Internet at
http://pubs.acs.org.
REFERENCES
(
1)
B. Sci. Rep.2014, 4, 4503ꢀ4511.
2) Tedsree, K.; Li, T.; Jones, S.; Chan, C. W. A.; Yu, K. M. K.; Bagot, P. A. J.; Marquis, E.
A.; Smith, G. D. W.; Tsang, S. C. E. Nat. Nano2011, 6, 302ꢀ307.
(3) Wong, M. S.; Alvarez, P. J. J.; Fang, Y.ꢀl.; Akçin, N.; Nutt, M. O.; Miller, J. T.; Heck, K.
N. J. Chem. Technol. Biotechnol.2009, 84, 158ꢀ166.
4) Liu, J.; Liu, Y.; Liu, N.; Han, Y.; Zhang, X.; Huang, H.; Lifshitz, Y.; Lee, S.ꢀT.; Zhong,
J.; Kang, Z. Science2015, 347, 970ꢀ974.
5) Sastre, F.; Puga, A. V.; Liu, L.; Corma, A.; García, H. J. Am. Chem. Soc.2014, 136,
6798ꢀ6801.
Jin, F. M.; Zeng, X.; Liu, J. K.; Jin, Y. J.; Wang, L. Y.; Zhong, H.; Yao, G. D.; Huo, Z.
(
(
(
(
(
6)
7)
Chong, M. N.; Jin, B.; Chow, C. W. K.; Saint, C. Water Res.2010, 44, 2997ꢀ3027.
Li, R.; Weng, Y.; Zhou, X.; Wang, X.; Mi, Y.; Chong, R.; Han, H.; Li, C. Energy
Environ Sci2015, 8, 2377ꢀ2382.
ACS Paragon Plus Environment
27

ACS Catalysis
Page 28 of 30
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
(
8)
Wang, X.; Maeda, K.; Thomas, A.; Takanabe, K.; Xin, G.; Carlsson, J. M.; Domen, K.;
Antonietti, M. Nat. Mater.2009, 8, 76ꢀ80.
9) Wang, L.; Dionigi, F.; Nguyen, N. T.; Kirchgeorg, R.; Gliech, M.; Grigorescu, S.;
Strasser, P.; Schmuki, P. Chem. Mater.2015, 27, 2360ꢀ2366.
10) Zhang, Y. C.; Du, Z. N.; Li, K. W.; Zhang, M.; Dionysiou, D. D. ACS Appl. Mater.
Interfaces2011, 3, 1528ꢀ1537.
(
(
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
(11) Bao, N.; Shen, L.; Takata, T.; Domen, K. Chem. Mater.2007, 20, 110ꢀ117.
(12) Moniz, S. J. A.; Shevlin, S. A.; Martin, D. J.; Guo, Z.ꢀX.; Tang, J. Energy Environ.
Sci.2015, 8, 731ꢀ759.
(13) Nasalevich, M. A.; Becker, R.; RamosꢀFernandez, E. V.; Castellanos, S.; Veber, S. L.;
Fedin, M. V.; Kapteijn, F.; Reek, J. N. H.; van der Vlugt, J. I.; Gascon, J. Energy Environ.
Sci.2015, 8, 364ꢀ375.
(14) GarcíaꢀLópez, E.; Marcì, G.; Megna, B.; Parisi, F.; Armelao, L.; Trovarelli, A.; Boaro,
M.; Palmisano, L. J. Catal.2015, 321, 13ꢀ22.
(15) Henderson, M. A. Surf. Sci. Rep.2011, 66, 185ꢀ297.
(16) Hoffmann, M. R.; Martin, S. T.; Choi, W. Y.; Bahnemann, D. W. Chem. Rev.1995, 95,
6
9ꢀ96.
(17) Tachikawa, T.; Fujitsuka, M.; Majima, T. J. Phys. Chem. C2007, 111, 5259ꢀ5275.
(18) Su, R.; Christensen, M.; Shen, Y. B.; Kibsgaard, J.; Elgh, B.; Vang, R. T.; Bechstein, R.;
Wendt, S.; Palmqvist, A.; Iversen, B. B.; Besenbacher, F. J. Phys. Chem. C2013, 117, 27039ꢀ
27046.
(
19) HernandezꢀAlonso, M. D.; Fresno, F.; Suarez, S.; Coronado, J. M. Energy Environ.
Sci.2009, 2, 1231ꢀ1257.
20) Martin, D. J.; Qiu, K.; Shevlin, S. A.; Handoko, A. D.; Chen, X.; Guo, Z.; Tang, J.
(
Angew. Chem. Int. Ed.2014, 53, 9240ꢀ9245.
(21) Scanlon, D. O.; Dunnill, C. W.; Buckeridge, J.; Shevlin, S. A.; Logsdail, A. J.; Woodley,
S. M.; Catlow, C. R. A.; Powell, M. J.; Palgrave, R. G.; Parkin, I. P.; Watson, G. W.; Keal, T.
W.; Sherwood, P.; Walsh, A.; Sokol, A. A. Nat. Mater.2013, 12, 798ꢀ801.
(22) Kozlov, D. V.; Vorontsov, A. V. J. Catal.2008, 258, 87ꢀ94.
(23) Yang, J.; Wang, D.; Han, H.; Li, C. Acc. Chem. Res.2013, 46, 1900ꢀ1909.
(24) Rawalekar, S.; Mokari, T. Adv. Energy. Mater.2013, 3, 12ꢀ27.
(25) Takai, A.; Kamat, P. V. ACS Nano2011, 5, 7369ꢀ7376.
(26) Subramanian, V.; Wolf, E. E.; Kamat, P. V. J. Am. Chem. Soc.2004, 126, 4943ꢀ4950.
(27) Cozzoli, P. D.; Curri, M. L.; Agostiano, A. Chem. Commun.2005, 25, 3186ꢀ3188.
(28) Su, R.; Tiruvalam, R.; Logsdail, A. J.; He, Q.; Downing, C. A.; Jensen, M. T.;
Dimitratos, N.; Kesavan, L.; Wells, P. P.; Bechstein, R.; Jensen, H. H.; Wendt, S.; Catlow, C. R.
A.; Kiely, C. J.; Hutchings, G. J.; Besenbacher, F. ACS Nano2014, 8, 3490ꢀ3497.
(
29) Priebe, J. B.; Radnik, J.; Lennox, A. J. J.; Pohl, M.ꢀM.; Karnahl, M.; Hollmann, D.;
Grabow, K.; Bentrup, U.; Junge, H.; Beller, M.; Brückner, A. ACS Catal.2015, 5, 2137ꢀ2148.
30) Tsukamoto, D.; Shiraishi, Y.; Sugano, Y.; Ichikawa, S.; Tanaka, S.; Hirai, T. J. Am.
(
Chem. Soc.2012, 134, 6309ꢀ6315.
(31) Murdoch, M.; Waterhouse, G. I. N.; Nadeem, M. A.; Metson, J. B.; Keane, M. A.; Howe,
R. F.; Llorca, J.; Idriss, H. Nat. Chem.2011, 3, 489ꢀ492.
(
32) Su, R.; Forde, M. M.; He, Q.; Shen, Y.; Wang, X.; Dimitratos, N.; Wendt, S.; Huang, Y.;
Iversen, B. B.; Kiely, C. J.; Besenbacher, F.; Hutchings, G. J. Dalton Trans.2014, 43, 14976ꢀ
4982.
1
ACS Paragon Plus Environment
28

Page 29 of 30
ACS Catalysis
1
2
3
4
5
6
7
8
9
(
(
(
33) Paramasivalm, I.; Macak, J. M.; Schmuki, P. Electrochem. Commun.2008, 10, 71ꢀ75.
34) Chen, X. B.; Liu, L.; Yu, P. Y.; Mao, S. S. Science2011, 331, 746ꢀ750.
35) Maeda, K.; Teramura, K.; Lu, D. L.; Saito, N.; Inoue, Y.; Domen, K. Angew. Chem. Int.
Ed.2006, 45, 7806ꢀ7809.
(36) Mizukoshi, Y.; Sato, K.; Konno, T. J.; Masahashi, N. Appl. Catal. B2010, 94, 248ꢀ253.
(37) Primo, A.; Corma, A.; Garcia, H. Phys. Chem. Chem. Phys.2011, 13, 886ꢀ910.
(38) Mogyorósi, K.; Kmetykó, Á.; Czirbus, N.; Veréb, G.; Sipos, P.; Dombi, A. React. Kinet.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
Catal. Lett.2009, 98, 215ꢀ225.
(39) Su, R.; Tiruvalam, R.; He, Q.; Dimitratos, N.; Kesavan, L.; Hammond, C.; Lopezꢀ
Sanchez, J. A.; Bechstein, R.; Kiely, C. J.; Hutchings, G. J.; Besenbacher, F. ACS Nano2012, 6,
6
(
284ꢀ6292.
40) Li, Y. H.; Xing, J.; Chen, Z. J.; Li, Z.; Tian, F.; Zheng, L. R.; Wang, H. F.; Hu, P.; Zhao,
H. J.; Yang, H. G. Nat. Commun.2013, 4, 2500ꢀ2507.
41) Wang, X. Q.; Sø, L.; Su, R.; Wendt, S.; Hald, P.; Mamakhel, A.; Yang, C. X.; Huang, Y.
D.; Iversen, B. B.; Besenbacher, F. J. Catal.2014, 310, 100ꢀ108.
42) Hurum, D. C.; Agrios, A. G.; Gray, K. A.; Rajh, T.; Thurnauer, M. C. J. Phys. Chem.
B2003, 107, 4545ꢀ4549.
(
(
(43) Carter, E.; Carley, A. F.; Murphy, D. M. J. Phys. Chem. C2007, 111, 10630ꢀ10638.
(44) Matthey, D.; Wang, J. G.; Wendt, S.; Matthiesen, J.; Schaub, R.; Lægsgaard, E.;
Hammer, B.; Besenbacher, F. Science2007, 315, 1692ꢀ1696.
(45) Porsgaard, S.; Jiang, P.; Borondics, F.; Wendt, S.; Liu, Z.; Bluhm, H.; Besenbacher, F.;
Salmeron, M. Angew. Chem. Int. Ed.2011, 50, 2266ꢀ2269.
(
46) Hald, P.; Becker, J.; Bremholm, M.; Pedersen, J. S.; Chevallier, J.; Iversen, S. B.;
Iversen, B. B. J. Solid State Chem.2006, 179, 2674ꢀ2680.
47) Ganji, S.; Bukya, P.; Vakati, V.; Rao, K. S. R.; Burri, D. R. Catal. Sci. Technol.2013, 3,
409ꢀ414.
48) Details for sample synthesis and photocatalytic tests, additional TEM and XPS
characterisations, and theoretical calculations can be found in Supporting Information.
49) Su, R.; Bechstein, R.; So, L.; Vang, R. T.; Sillassen, M.; Esbjornsson, B.; Palmqvist, A.;
(
(
(
Besenbacher, F. J. Phys. Chem. C2011, 115, 24287ꢀ24292.
(50) Noack, K.; Zbinden, H.; Schlögl, R. Catal. Lett.1990, 4, 145ꢀ155.
(51) Henderson, M. A.; Lyubinetsky, I.Chem. Rev., 2013, 113, 4428ꢀ4455.
(
52) Baumann, S. O.; Elser, M. J.; Auer, M.; Bernardi, J.; Hüsing, N.; Diwald, O.
Langmuir2011, 27, 1946ꢀ1953.
53) Chiesa, M.; Paganini, M. C.; Livraghi, S.; Giamello, E. Phys. Chem. Chem. Phys2013,
5, 9435ꢀ9447.
(
1
(54) Barolo, G.; Livraghi, S.; Chiesa, M.; Paganini, M. C.; Giamello, E. J. Phys. Chem.
C2012, 116, 20887ꢀ20894.
(
55) Vedyagin, A. A.; Volodin, A. M.; Stoyanovskii, V. O.; Mishakov, I. V.; Medvedev, D.
A.; Noskov, A. S. Appl. Catal. B2011, 103, 397ꢀ403.
56) Mi, J.ꢀL.; Jensen, K. M. O.; Tyrsted, C.; Bremholm, M.; Iversen, B. B.
CrystEngComm2015, 17, 6868ꢀ6877.
57) He, Y.; Dulub, O.; Cheng, H.; Selloni, A.; Diebold, U. Phys. Rev. Lett.2009, 102,
06105.
(
(
1
(
(
58) Di Valentin, C.; Pacchioni, G.; Selloni, A. J. Phys. Chem. C2009, 113, 20543ꢀ20552.
59) Rietveld, H. J. Appl. Crystallogr.1969, 2, 65ꢀ71.
ACS Paragon Plus Environment
29