T.F. Berto et al. / Journal of Catalysis 338 (2016) 68–81
69
e.g., TiO
2
(band gap: 3.1 eV) [18,19]. The present contribution aims,
1 h in order to obtain the RhO
x
-decorated (Ga1ꢁxZn
x
)(N1ꢁx
O
x
) solid
therefore, to describe on a molecular level the elementary reac-
tions involved in photoreforming using ethylene glycol (EG) as
model reactant [20,21]. The simplicity of EG allows unambiguous
identification of the involved elementary steps. The reaction path-
way analysis and kinetics are based on quantitative gas and liquid
phase analysis linked in a kinetic model [22–24].
solution. After cooling to room temperature the sample was
washed to remove NaCl and dried in N at 383 K overnight.
Different procedures of co-catalyst preparation were chosen
in order to obtain photocatalysts with a maximum activity.
2
Rh/(Ga1ꢁxZn
RhO -decorated (Ga1ꢁxZn
photocatalytic activity. We hypothesize that H
x
)(N1ꢁx
O
x
), which was obtained by reducing
) in H flow, possessed poor
partially reduced
Zn -centers to create oxygen vacancies and, thus, increased the
electron–hole recombination rate. In contrast, the best RhO /TiO
x
x
)(N1ꢁx
O
x
2
2
2
+
2
. Experimental section
x
2
1
ꢁ
was obtained by treatment in synthetic air at 623 K (5 K min
)
2.1. Materials
ꢁ1
for 1 h and in H
2
at 623 K (5 K min ) for 1 h. H
2
treatment at
6
23 K may create oxygen vacancies on TiO
2
[25,26] and thus
All chemicals were obtained from commercial suppliers and
decrease electron hole recombination rates increasing photo-
catalytic performance [27]. As the reaction pathways and its
selectivities were not significantly altered by this treatment, this
phenomenon was not further studied.
used as provided: AEROXIDEÒ TiO
4
1
P
25 (Evonik, LOT:
162092398), sodium hexachlororhodate (III) (Alfa Aesar, Rh
7.1%), gallium oxide (ABCR, 99.99%, LOT: 1040437), zinc oxide
, BASF, 5.0, anhy-
drous), synthetic air (Westfalen), hydrogen (H , Westfalen, 5.0),
argon (Ar, Westfalen, 5.0), nitrogen (N , Westfalen, 5.0), ethanol
80 mg/100 mL, European Reference Materials), glyoxal trimer
2
(
ABCR, 99.7%, LOT: 1121535), ammonia (NH
3
2
2.3. Photocatalytic test
2
(
Photoreforming experiments. Photocatalytic reactions were car-
dihydrate (Fluka, P95%), EG (VWR-Chemicals, 99.9%), glycolalde-
hyde dimer (Aldrich), glycolic acid (Aldrich, 99%), glyoxylic acid
monohydrate (Aldrich, 98%), formaldehyde solution (Fluka,
ried out in a photo-reactor connected to a gas-tight gas circulation
ꢁ4
ꢁ1 ꢁ1
system (leakage rate <5 ꢀ 10 Pa s
L , V = 310 mL), the catalyst
being exposed to light via top-irradiation through a quartz-
window. 75 mg of photocatalyst was suspended in 100 mL of an
aqueous solution containing the reactant (typically 20 mM). The
reactor was kept at 288 K and the system was filled with Ar to
ꢁ1
1
1
(
000
lg mL
2
in H O, IC Standard), formic acid (Merck, 98–
00%), methanol (Aldrich, 99.8%, anhydrous), phloroglucinol
Aldrich, P99%), acetaldehyde solution (Aldrich, 35 wt.%), D
2
O
(
Euriso-Top, 98.85 atom%), deuterium chloride in D O (Acros
2
1
O
bar. The system was evacuated four times in order to remove
. Completeness of O removal was verified by GC analysis
mol O /detection limit). Subsequently, the suspension was
Organics, 1 M, 99.8 atom%), gallium ICP-standard (Merck, Certipur,
ꢁ
1
ꢁ1
2
2
1
000 mg L ), zinc ICP-standard (Merck, Certipur, 1000 mg L ), Rh
(
<0.3
l
2
ꢁ1
AAS-standard (Fluka, TraceCert, 999 ± 9 mg L ), Rh-foil (was pro-
vided by ESRF/BM25 station), and rhodium(III)oxide (Fluka, anhy-
drous, puriss).
illuminated with a 300 W Xenon lamp, equipped with a cold-
mirror 1 (CM1) and a water filter tempered at 303 K. High EG con-
version experiments were performed over Rh/TiO
2
using high
power UV-LEDs (365 nm) instead of the 300 W Xe-lamp. The
evolved gases were analyzed by an online gas chromatograph
2.2. Catalyst preparation
(
Shimadzu, GC 2010 Plus with Ar as carrier gas and a Chromosorb
Synthesis of 1.0 wt.% RhO
x
/TiO
2
. TiO
2
was dried under static air at
101 column connected with a MS-5Å column), equipped with a
TCD, FID and a methanizer. The concentrations of dissolved gases,
4
5
73 K for 2 h prior to impregnation. The support (BET surface area:
2
ꢁ1
ꢁ1
3 m g and pore volume of 0.11 mL g ) was treated with an
RhCl O via incipient wetness impreg-
ꢀ12H
nation. The reddish powder was kept at 383 K (5 K min ) for 1 h,
in particular CO
2
, were accounted for by applying Henry’s law. H
2
aqueous solution of Na
3
6
2
production rates were determined by dividing the difference of H
2
ꢁ1
amounts between two adjacent data points by the corresponding
time interval. Concentrations in the liquid were determined by
ꢁ1
and heat treated in synthetic air at 623 K (5 K min ) for 1 h
ꢁ1
1
(
100 mL min ). After cooling to room temperature, the RhO
TiO material was treated by heating with an increment of
K min to 623 K in H
perature. Subsequently, the sample was washed thoroughly to
remove NaCl and dried in N at 383 K overnight. Within the study
used, parent TiO was subjected to identical treatments such as
RhO /TiO . 1.0 wt.% Pt/TiO was synthesized using H PtCl as Pt
precursor, following the synthesis procedure used to prepare RhO
TiO
Synthesis of 1.0 wt.% RhO
x
/
quantitative H NMR spectroscopy. Liquid samples were with-
2
drawn via a sample valve, filtered with a nylon syringe filter and
ꢁ
1
ꢁ1
5
2
(100 mL min ) and cooled to room tem-
analyzed.
1
H NMR analysis. A sample of 400 lL was mixed with 400 lL of
2
pH-adjusted internal standard (20 mM 1,3,5-trihydroxybenzene in
2
D O, pH adjusted with DCl to 2.7). All experiments were performed
2
x
2
2
2
6
at 305 K using an Avance III 500 System (Bruker Biospin, Rheinstet-
ten, Germany) with an UltraShield 500 MHz magnet and a
x
/
)
1
13
2
.
SEI 500 S2 probe head (5 mm, inverse H/ C with Z-gradient).
The measurements were conducted at a magnetic field of
x
x x x
/(Ga1ꢁxZn )(N1ꢁxO ). RhO /(Ga1ꢁxZn
x
1
(
N
1ꢁx
O
x
) was synthesized according to a procedure reported by
11.75 T. The resonance frequency of H was 500.13 MHz. For all
1
Domen et al. [19]. A physical mixture of 0.6 g (3.2 mmol) Ga
and 0.520 g (6.4 mmol) ZnO was treated at 1098 K (10 K min
2
O
3
samples, the H NMR spectra were acquired using the one-
ꢁ
1
)
dimensional NOESY sequence ‘‘noesygppr1d.comp” with presatu-
ration of the residual water signal during the relaxation delay
and the mixing time using spoil gradients. The relaxation delay
was 26 s, and the acquisition time was 4.1 s. Spectra were the
result of 64 or 128 scans, with data collected into 32 k data points.
Each FID was zero-filled to 64 k data points. Prior to Fourier trans-
formation, an exponential window function with a line broadening
factor of 0.2 Hz was applied. The resulting spectra were manually
phased, baseline corrected, and integrated using Mestre-C 7.1.0
software package. Chemical shifts were referenced to the internal
ꢁ1
for 16 h under NH
tion was cooled to room temperature under NH
3
flow (200 mL min ). The obtained solid solu-
flow and was sub-
sequently treated at 873 K (5 K min ) for 1 h under synthetic
3
ꢁ1
ꢁ1
airflow (50 mL min ). This procedure is defined as post-
calcination in agreement with Ref. [19]. Post-calcined (Ga1ꢁxZn
) (x = 0.14) was modified with RhO (denoted as a co-
catalyst) by wet impregnation. In an evaporating dish 0.2 g of
x
)
(
N
1ꢁx
O
x
x
(
Ga1ꢁxZn
= 18.2 M
2H O. The stirred suspension was evaporated to dryness. The
obtained powder was kept in static air at 623 K (5 K min ) for
x
)(N1ꢁx
O
x
) was dispersed in 1 mL of bidistilled water
(r
X
cm) containing the appropriate amount of Na RhCl
3
6
ꢀ
1
2
1
standard. T , the longitudinal relaxation time, was determined by
ꢁ
1
the inversion recovery pulse sequence method. The sum of