CHEMCATCHEM
FULL PAPERS
Comparison with the behavior of Pt/HS-MgF , which ap-
in which the alkane is dehydrogenated on a metal phase, and
the alkene formed is protonated at Brønsted acid sites, yielding
2
peared only slightly active in this reaction, clearly argues for
the importance of surface acidity for providing the bifunctional
character of n-pentane isomerization. HS-MgF2 is much less
[26,27]
an alkylcarbenium ion.
It must be mentioned that the clas-
sic bifunctional metal-acid catalysts show hydrogen reaction
orders between ꢁ1 and 0, hydrocarbon reaction orders close
to 1, and apparent activation energy between 100 and
acidic than HS-AlF thus it is assumed to be also less effective
3
in creating carbenium-like species operating in the reaction
mechanism. As no adsorption of pyridine was observed over
ꢁ
1 [28–36]
150 kJmol ,
and similar kinetic parameters were also
[
11]
HS-MgF2, this clearly indicates the absence of stronger acid
found in this study (Table 3).
sites on this support. Our previous results with palladium sup-
The effect of metal–acid balance in HS-AlF supported Pd
3
[
14]
ported on less acidic supports (SiO , Al O ), for which which
catalysts on the isomerization of n-pentane was preliminary
tested by using differently metal-loaded Pd catalyst prepared
by impregnation. Their basic physicochemical characteristics
are given in Table 2 and Figure 2, and catalytic results are
shown in Figure 6. It is observed that during initial 150 min of
reaction, the overall conversion levels of 0.5 wt% 2.0 wt% cata-
lysts are similar to that by 2 wt% catalyst prepared by the sol–
gel method (Figure 4). Again, a mild maximum in conversion is
observed, but it appears at the very initial reaction time. Short-
ly thereafter, the behavior of impregnated catalysts changes
and the conversion of n-pentane over impregnated catalysts
started to decrease. Much longer evolution of overall conver-
sion observed on the sol–gel Pd catalyst would be rationalized
by a more difficult accessibility of metal particles, partly em-
2
2
3
the catalytic performance is limited to the metal function, re-
vealed very low n-pentane conversion (ꢀ1%), and a rather
nonselective course of reaction (cracking, isomerization, and
cyclization). Only if the Pd/Al O catalyst was pretreated at very
2
3
high temperature (reduction by H at 6008C for several hours),
2
the selectivity toward isomerization increased significantly, in-
[
14]
dicating the effect of Lewis acidity development. In the pres-
ent case, the HS-AlF -surface is almost free of hydroxyl groups
3
as a result of the postfluorination and consists mainly of Lewis
acid–base pair sites with very strong Lewis acid character. As it
was shown very recently by one of us, these Lewis sites are
[24]
strong enough for forming hydride-like surface species.
Based on this assumption, the isomerization would start from
[
25]
hydride abstraction on Lewis acid sites. However, as further
arguments based acidity and kinetic data show, some contribu-
tion from alkene protonation associated with Brønsted acidity
should also be considered (see below).
bedded in the AlF gel (Figure 4). Therefore, carbon removal
3
from impregnated Pd catalysts must be much faster. It is also
observed that the 0.2 wt% metal loading is not sufficient to
play a satisfactory role in bifunctional catalysis of n-pentane
conversion. Its behavior echoes the performance of the metal-
In Figure 4 it is shown that the overall activity of Pt/HS-AlF3
slightly decreases with time on stream, with concomitant in-
crease in isomerization selectivity. On the other hand, the activ-
free HS-AlF (Figure 4 in Ref. [1]), although the selectivity for
3
isomerization is still high.
ity of Pd/HS-AlF passes through a gentle maximum. Whereas
TPD of ammonia showed the presence of strong acid sites.
It is seen that the postreaction samples exhibit similar loca-
tions of TPD peaks, as those observed for prior-to-reaction cat-
alysts (Figure 7). However, the intensity of the high tempera-
ture peak is reduced after reaction. It appears that some stron-
ger acid sites are blocked by carbonaceous residues (coke). It
is also observed that catalyst precalcination at 4008C reduces,
but not fully eliminates, the amount of highly acidic sites (not
shown). Such sites are still detected in the catalysts subjected
to a prolonged reaction.
3
the behavior of Pt/HS-AlF is typical for the mild deactivation
3
of acid-supported Pt catalysts and has been frequently report-
ed, the maximum in conversion observed for Pd/HS-AlF was
3
rather unexpected. We believe that this high conversion may
result from incomplete decontamination of palladium during
reduction with hydrogen at 3508C for 1 h. As discussed above,
the procedures of catalyst synthesis and postfluorination with
CHClF can lead to massive incorporation of carbon in palladi-
2
um bulk. The time-on-stream behavior of Pd/HS-AlF shown in
3
Figure 4 would suggest that approximately half a day on
stream at 3508C is needed to remove all carbon from
palladium.
As mentioned before, NH -TPD runs revealed that practically
3
all strong acid sites should still be blocked by adsorbed ammo-
nia at 3508C. Therefore, only weaker/medium acid sites would
serve for n-pentane rearrangement performed on the catalysts
pretreated this way. The effect of such catalyst pretreatment
on catalytic activity is shown in Figure 8. The overall conver-
sion of n-pentane decreased drastically on both catalysts after
Slow catalyst deactivation at 3508C is most probably caused
by coking of acid sites of the support and not by fluorine
leaching (HF release). Deactivation at lower H /pentane ratios
2
was expected to proceed slower, but a reverse correlation was
observed (see the Supporting Information).
blocking the strong acid sites by NH . Therefore, it is evident
3
It is recalled that although the HS-AlF exhibits very high
that these strong acid sites play an important role in shaping
3
acidity, it is not an efficient catalyst for n-pentane isomerization
the overall attractive behavior of HS-AlF supported metal cata-
3
(
Figure 4 in Ref. [1]). It should deactivate rapidly because a syn-
lysts. Together with the overall conversion, the selectivity to
isopentane is also decreased after NH3 treatment. However,
after short time on stream the selectivity begins to develop,
reaching a significant level of approximately 95% for Pd/HS-
AlF and >40% for Pt/HS-AlF . It seems that our expectation
ergistic action of metal and acid sites is needed for maintain-
ing the isomerization performance. This fact in combination
with low isomerization activity of platinum supported on
much less acidic magnesium fluoride indicates that the reac-
tion of n-pentane isomerization performed on Pd(Pt)/HS-AlF3
catalysts occurs through the classical bifunctional mechanism,
3
3
that ammonia effectively blocks the strong acid centers and
limits the overall catalytic performance to metal-only catalysis
ꢀ
2014 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ChemCatChem 2014, 6, 592 – 602 599