G Model
APCATA-15224; No. of Pages13
ARTICLE IN PRESS
J.A. Lopez-Ruiz et al. / Applied Catalysis A: General xxx (2015) xxx–xxx
3
2.5. Electron microscopy
composed of 95 wt% heptanoic acid and 5 wt% dodecane). In the
absence of catalyst, no conversion of heptanoic acid was observed.
However, when Norit or Vulcan carbon support was used, a back-
ground conversion of ≈2 and 1%, respectively, was detected. No
conversion was observed over the silica support.
Transmission electron microscopy (TEM) was performed to
determine the Pd particle size (DP) on an FEI Titan operated at
300 keV in bright field mode and on a JEOL 2010F operated at
200 KeV in STEM high-angle annular dark-field (HAADF) mode. Cat-
alyst samples were deposited on a SPI holey-carbon coated Cu grid
after being dispersed in ethanol [31]. For each fresh sample, four
TEM images were used to count Pd particle sizes and 40–50 par-
ticles were measured per image. The average Pd particle size was
calculated for each image followed by the mean and standard devi-
ation of the average sizes. The surface-weighted average Pd particle
2.7. Post mortem catalyst characterization
The spent samples were recovered from the reactor, washed
with 25 mL of methanol and sonicated in a methanol solution for
30 min to remove weakly adsorbed reactant and products from the
surface of the catalyst. After sonication, the solids were recovered
by filtration and dried overnight in air at 400 K. The spent sam-
ples were characterized by H2 chemisorption, N2 physisorption,
XRD, and TEM to determine the effect of reaction conditions on the
catalyst structure and adsorption capacity.
ꢀ
ꢀ
size was calculated using
D3P,i
/
D2P,i, and the volume-weighted
ꢀ
average Pd particle size was calculated using
D4P,i
/
D3P,i
.
2.6. Catalytic conversion of heptanoic acid
2.8. Calculations of rates
The catalytic decarbonylation/decarboxylation of heptanoic
acid (Sigma–Aldrich ≥99.50%) was performed in a fixed-bed tubu-
lar reactor described previously [4]. In summary, the reactor was
loaded with 50–500 mg of catalyst pellets supported on a glass
wool plug. Fresh catalyst was used for each experiment because
of the irreversible deactivation of the catalyst as described in our
previous work with Pt catalysts. After loading the reactor, the sys-
tem was purged for 60 min with 200 cm3 min−1 of N2 (GTS-Welco
99.999%) at atmospheric pressure to remove dioxygen from the
system and the pressure was increased to 45 bar to perform a leak
test and then depressurized. The temperature was then increased
at 1 K min−1 to the desired reduction temperature, 623 K, under
100 cm3 min−1 of H2 (GTS-Welco 99.999%) and held isothermally
for 3 h at atmospheric pressure. The reactor was finally cooled to
the desired reaction temperature, 573 K, and purged for 60 min in
200 cm3 min−1 of N2 (GTS-Welco 99.999%). Then, the pressure was
increased to the desired reaction condition with 200 cm3 min−1 of
N2 (GTS-Welco 99.999%). The reactor system was operated at two
different pressures, 40 bar for liquid-phase operation and 1 bar for
gas-phase operation.
For the “pure” acid experiments, the feed mixture was composed
of 95 wt% heptanoic acid (Sigma–Aldrich ≥99.50%) and 5 wt% dode-
cane (Sigma–Aldrich anhydrous 99.00%) as an internal standard.
Tetradecane (Sigma–Aldrich anhydrous 99.90%) was used as a dilu-
ent to prepare feeds with different heptanoic acid concentration
(50, 25, 10, 7.5, 5, and 2.5 mol%). The feed was then pumped in the
uid flow rates (0.01–1.0 cm3 min−1). The acid feed mixtures passed
through a heated zone before entering the catalytic reactor. A
schematic of the reactor can be found in the paper by Lopez-Ruiz
and Davis [4].
The outlet of the reactor was connected to an air cooled
condenser maintained at room temperature to remove the
condensable products, such as hexane, 1-hexene, hexenes, 7-
tridecanone, dodecane, tetradecane, and unreacted heptanoic acid,
from the product stream. The liquid-phase sample was analyzed by
a gas chromatograph (GC 7890 A) equipped with a ZB-FFAP column
(length 45 m × 0.538 mm) and a flame ionization detector (FID). The
product gas was continuously removed from the condenser using
5 cm3 min−1 N2 (GTS-Welco 99.999%) as a purge gas. The outlet gas
of the condenser was fed to an on-line gas chromatograph (HP 5890
Series II) equipped with a packed column, ShinCarbon ST 80/100,
and a thermal conductivity detector (TCD) for detection of N2, CO,
CO2 and any other light hydrocarbons (if any).
As described in our previous work, the turnover frequency (TOF)
was calculated as the rate of formation of the products referred to
the number of surface metal atoms evaluated on a freshly prepared
catalyst [4]. The main components in the gas-phase were CO and
CO2, whereas 1-hexene, 2- and 3-hexenes (i-hexenes), hexane, 7-
tridecanone, and heptanoic acid were the main compounds in the
liquid-phase. Some minor amounts of heavier unknown products
were also detected in the liquid-phase products (which accounted
for ≈0.4% of the total detected liquid product).
The liquid-phase TOF was calculated using the rate of formation
of the products after 20 h of reaction, after which the liquid-phase
reaction was at a steady state, normalized to the number of sur-
face metal atoms evaluated on a fresh catalyst. The liquid-phase
TOF was therefore calculated as [(rate of production after 20 h of
reaction (mol s−1) hexane + 1-hexene + i-hexenes + 2 × moles of 7-
tridecanone)/(mol of surface metal on a fresh catalyst counted by
H2 chemisorption)].
During gas-phase operation, the catalyst deactivated exponen-
tially with time during the first 3 h of reaction. Thus, an initial TOF
was calculated by extrapolating the log of the reaction rate to t = 0 h.
The initial TOF was calculated as [(rate of production (mol s−1) of
hexane + 1-hexene + i-hexenes + 2 × moles of 7-tridecanone)/(mol
of surface metal on a fresh catalyst counted by H2 chemisorption)].
The deoxygenation conversion is defined here as the formation
rates of the major products divided by the feed rate of reagent.
Thus, the deoxygenation conversion was calculated as [(rate of pro-
duction (mol s−1) of hexane + 1-hexene + i-hexenes + 2 × moles of
7-tridecanone)/(moles of heptanoic acid fed (mol s−1))].
The product selectivity was defined as the moles of a product
formed divided by the total moles of products present in the same
phase (liquid or gas). Three different sets of product selectivity were
calculated, the selectivity to carbon oxides (CO and CO2), i.e. CO
selectivity [(rate of CO)/(rate of CO + rate of CO2)], the selectivity
of ␣-olefin relative to all olefin produced for 1-hexene and hex-
enes, i.e. relative ␣-olefin selectivity [(rate of 1-hexene)/(rate of
1-hexene + i-hexenes)], and the selectivity to deoxygenation prod-
ucts for mainly hexane, 1-hexene, i-hexenes, and 7-tridecanone,
i.e. selectivity of hexane [(rate of hexane)/(rate of hexane + 1-
hexene + i-hexenes + 2 × moles of 7-tridecanone)].
The carbon balance was defined as the moles of carbon
identified after reaction divided by the total moles of car-
bon fed to the reactor, i.e. carbon balance [(rate (mol s−1
)
of unreacted heptanoic acid × 7 + hexane × 6 + 1-hexene × 6 + i-
hexenes × 6 + 7-tridecanone × 13 + CO + CO2)/(moles of heptanoic
acid fed × 7 (mol s−1))]. For the experiments reported here, the
carbon balance was between 98 and 99%.
A series of control experiments was performed to determine the
background conversion of the system at our typical reaction condi-
tions (573 K, 40 bar for liquid-phase operation or 1 bar for gas-phase
operation, 250 mg of catalyst, and 0.01 cm3 min−1 of liquid feed