32
T. van Haasterecht et al. / Journal of Catalysis 319 (2014) 27–35
Beverskog et al. have calculated Pourbaix diagrams for the
Ni–water system at elevated temperatures (200–300 °C) and
shown that at a pH below neutral, bulk Ni0 can be oxidized by
water or acids [28]. In an acidic medium the oxidized Ni species
dissolve and Ni2+ ions are predominantly formed [29]. This
explains the leaching we observed during APR under standard con-
ditions, which, as a result of reaction products, took place in an
acidic medium.
EG concentration influences the nickel particle growth via the for-
mation of liquid phase products, we measured the concentrations
of the latter (Table S1). Increasing concentrations of ethanol
(0–21 mmol/L) and methanol (0–16 mmol/L) were observed as
functions of the EG feed concentration (1–50 wt.%). The concentra-
tion of glycolic acid, expected to have a negative effect on the sta-
bility, was between 1.6 and 2.0 mmol/L and was not a strong
function of the reactant concentration (see Fig. S2). Therefore we
speculate, in analogy with work of Sievers et al. [36], that coverage
of the catalyst by organic species (substrate and/or intermediates)
can play an essential role in stabilizing the Ni nanoparticles, in our
case by slowing down the Ostwald ripening process.
At a pH above neutral, oxidation of bulk Ni is prevented because
the redox potential of the H+/H2(g) couple decreases with increas-
ing pH (as follows from the Nernst equation, Eh = ꢁ0.1pH at 230 °C)
[30] and falls below that of the Ni/Ni2+ couple. While oxidation and
leaching of Ni are linked in acid media, this is not the case under
alkaline conditions. Even though the bulk Ni phase is thermody-
namically stable under alkaline conditions, surface oxidation can
still occur. As a result of water dissociation, Ni–OH or Ni–O species
form on the Ni surface or at low-coordination sites [31]. However,
under alkaline conditions, oxidized nickel species are not soluble
[28]. The prevention of Ni leaching is essential for the stability of
the Ni nanoparticles because the formation of mobile Ni species
is a prerequisite for Ostwald ripening, since they enable mass
transfer from the smaller to the larger Ni particles. We observed
that nickel particle growth occurred already at neutral pH, in con-
trast to what was expected from the previous discussion, and alka-
line conditions were required (pH > 8) to stabilize our catalyst
(Fig. 4). This is most likely because we used Ni nanoparticles,
which can be oxidized more easily than the bulk phase [32,33],
thus explaining the need for a higher pH than for the bulk phase.
In very alkaline solutions (pH > 13), the formation of anionic solu-
ble Ni species (e.g., Ni(OH)ꢁ3 ) becomes favorable [28], which can
explain the increase in particle size observed at very high pH
(Fig. 4).
3.2. Effects of gas atmosphere and pH on catalytic performance
The effect of reaction conditions on the catalytic performance of
Ni/CNF in reforming EG was evaluated using a slurry batch reactor.
The conversion over time was followed by taking intermittent
liquid phase samples, whereas the composition of the gas phase
was measured at the end of the reaction. In addition, the H2 and
CH4 productivity, expressed as turnover frequencies, was mea-
sured in a continuous flow reactor.
3.2.1. Slurry batch experiments
Fig. 5 displays the conversion profiles for EG reforming in a
slurry batch reactor. Batch reaction time t = 0 is defined as the
moment the reactor reached 230 °C; however, during the heating
period of approximately 30 min, substantial conversion was real-
ized. With standard APR conditions, 80% conversion was realized
after 1 h. When the initial gas phase was changed to H2, only
48% conversion was obtained in the first hour, indicating that the
initial rate for EG conversion decreases under H2. The decreased
activity can be explained by a negative effect of H2 on the reaction
kinetics [37,38]. Fig. 5 further shows that the rate of EG conversion
is considerably increased by the addition of KOH, as full conversion
is now reached within 1 h. A similar observation was made by Liu
et al. [41], who found that during APR of glycerol the activity pro-
gressively increased by up to a factor of 4 when the KOH concen-
tration increased from 0 to 1.6 M. This could be related to the
shift in redox potential, which predicts that Ni is less likely to be
oxidized under these conditions and the Ni particles are thus more
metallic in nature under alkaline conditions. Also, KOH can act as a
co-catalyst, and the mechanism might be different at high pH, as
can be inferred from the work of Zope et al. [39] on the alkaline-
enhanced oxidative dehydrogenation of glycerol. The authors
showed that in alkaline media the oxidative dehydrogenation of
glycerol is facilitated over a Pt surface. They elaborate by means
of DFT calculations that the activation barrier for O–H dissociation
and subsequent activation of C–H bonds is significantly reduced in
alkaline media over a Pt catalyst, resulting in enhanced reaction
rates. Finally, the addition of base leads to capture of CO2 (Table 2)
and thus may enhance kinetics if CO2 inhibits the APR over nickel.
The effect of the reaction conditions on the gas phase composi-
tions is shown in Table 2. The main components detected were H2,
CO2, and CH4, together with small amounts of CO and higher
alkanes (C2H6 and C3H8). As proposed by Dumesic [10], the APR
of EG proceeds through reforming,
Besides the pH, the partial hydrogen pressure also affects the
oxidation and solubility of Ni. In acidic media, the extent of particle
size growth was shown to be diminished with an H2 gas atmo-
sphere compared to that with an inert gas (Table 1, entries 2 and
3). Lower levels of Ni leaching were measured for reactions with
1 wt.% EG under H2 versus Ar, 3.8 ppmw and 6.4 ppmw, respec-
tively. As expected with the use of H2 gas, the Ni-ion solubility
(Ni(s) + 2H+ (aq) M Ni2+(aq) + H2(g)) was suppressed [34], resulting
in a lower Ni crystallite growth rate.
Finally, the presence of organic species also plays an important
role in the Ni particle growth mechanism. Fig. 4 shows that Ni crys-
tallites are more stable in EG mixtures than in water. Also, without
KOH, in acidic media, a decrease in the extent of particle growth
was accomplished by increasing EG concentration from 1 to
50 wt.% (Table 1, entries 10–14). However a straightforward rela-
tion between the concentration of dissolved Ni species and the ini-
tial EG concentration was not observed. The amount of leached Ni
increased from 3.8 for 1 wt.% EG to 7.1 and 6.6 ppmw for 10 and
20 wt.%, respectively, and decreased again for 50 wt.% solutions
(1.6 ppmw). Shabaker et al. also showed that leaching of Ni from
Sn-modified Raney Ni catalyst during APR can be reduced by a fac-
tor of 5 when the feed concentration is changed from 5 to 63 wt.%
EG. This resulted in higher catalyst stability, and the authors con-
cluded that the presence of excess water is partly responsible for
the deactivation [35]. Increasing the reactant concentration might
have a stabilizing effect by decreasing the surface coverage of
adsorbed water and increasing the coverage of the reactant and
products, thereby minimizing oxidation and slowing down the
Ostwald ripening process. This is in agreement with our observa-
tion that the stabilizing effect of externally added H2 pressure
becomes less significant with increasing feed concentration (see
Fig. 2). In addition, Sievers et al. [36] have shown that increased
coverage of the catalyst by organic species (substrate and/or inter-
mediates) has a stabilizing effect. To investigate whether the initial
C2H6O2ðlÞ ! 2COðgÞ þ 3H2ðgÞ;
ð1Þ
ð2Þ
ð3Þ
and subsequent WGS (water-gas shift) reaction,
COðgÞ þ H2OðlÞ $ CO2ðgÞ þ H2ðgÞ;
while especially for Ni-based catalysts, methanation of CO,
COðgÞ þ 3H2ðgÞ ! CH4ðgÞ þ H2OðlÞ;
or CO2,