Identifying the role of N-heteroatom location in the activity of metal catalysts for alcohol oxidation

Article abstract of DOI:10.1002/cctc.201402951

This work focuses on understanding how the proximate location and bonding of N heteroatoms affect the stability and reactivity of Pd-based catalysts for the oxidation of alcohols in the solution. The results show that the simple adsorption of N groups, fr

Full text of DOI:10.1002/cctc.201402951

DOI: 10.1002/cctc.201402951
Full Papers
Identifying the Role of N-Heteroatom Location in the
Activity of Metal Catalysts for Alcohol Oxidation
Carine E. Chan-Thaw,[a] Alberto Villa,[a] Gabriel M. Veith,[b] and Laura Prati*[a]
This work focuses on understanding how the proximate loca-
tion and bonding of N heteroatoms affect the stability and re-
activity of Pd-based catalysts for the oxidation of alcohols in
the solution. The results show that the simple adsorption of N
groups, from the solution, has a detrimental effect on the cata-
lytic activity and stability. In contrast, chemically bound N moi-
eties within the carbon structure improve these properties,
which limits the leaching of metal and coarsening of metal
particles. Moreover, the benefits of N atoms are realized only if
the N atom is covalently bonded to the support and not di-
rectly bonded to the Pd nanoparticles.
Introduction
There is a critical and growing need to develop highly active
and selective heterogeneous catalysts that are suitable for low-
temperature liquid-phase oxidation reactions, such as the con-
version of oxygenated feedstocks.[1,2] The obtained products,
such as aldehydes, ketones, and acids, are valuable intermedi-
ates in the fine chemical industry.[1,2] One of the challenges is
the design of active and resistant materials, and to limit the
deactivation phenomenon owing to metal leaching, as well as
the dissolution of the support material, which is typical of
liquid-phase reactions. Numerous groups around the world
have focused on overcoming these challenges, especially in al-
cohol selective oxidation, which still represents a challenging
reaction.[2b,3]
the increased stability and the mediating role of N heteroa-
toms with respect to the catalytic activity are not yet under-
stood.
With this work, we sought to understand whether the N
functionality needs to be directly bound to the support at the
metal–support interface or whether a N species just needs to
be in close proximity to mediate its activity. To answer this
question, we formed a matrix of materials by using polymer-
protected Pd nanoparticles (Pd NPs) deposited on activated
carbon (AC) and N-functionalized AC. The AC was functional-
ized through the physisorption of pyridine from the solution
(ACpyr) or the formation of a CÀN bond by thermally reacting
NH3 with the AC support. In addition, two protecting groups
were used to form the Pd NPs: polyvinyl alcohol (PVA) and
polyvinyl pyrrolidone (PVP). PVA has no N functionalities,
whereas PVP has a pyrrole-type N species in close proximity to
the Pd NP. The use of these molecules prevents any agglomer-
ation during the generation of Pd NPs.[9] Moreover, both pro-
tective agents enable the precise control of the metal NP di-
ameters.[10] However, the presence of these protecting mole-
cules limits the reactant access to the active site(s), which de-
creases the catalytic activity.[10c] We assumed that this limitation
is similar in the cases of both PVA and PVP. Through this com-
bination of materials and using electron microscopy and X-ray
photoelectron spectroscopy (XPS), we have gained insights
into how the bonding of N is important to catalytic properties.
One approach is the introduction of Lewis base-type N het-
eroatoms. For example, covalent triazine framework[4] materials
derived from aromatic nitriles with high concentrations of N
functionalities were demonstrated to increase the stability and
activity of Pd- and Au-based catalysts in the liquid-phase oxi-
dation of glycerol[5] and benzyl alcohol.[6] For these catalysts,
performance was shown to be dominated by the concentra-
tion and type of N functionalities.[5,6] The addition of N atoms
to other carbon-based materials is similarly beneficial. For ex-
ample, N-doped carbon nanotubes promote better Pd disper-
sion and an increase in catalyst stability in the liquid phase oxi-
dation of benzyl alcohol and glycerol and in the direct synthe-
sis of H2O2.[7] Furthermore, the addition of N atoms to carbon-
based electrocatalysts produces materials with O reduction ac-
tivity rivaling Pt.[8] For these catalysts, and others, the origin of
Results and Discussion
To understand how N atoms mediate the catalytic activity and
stability of Pd NPs, we have adopted three strategies to intro-
duce N functionality: modification of the support material, use
of protective sol with N functionality, and combining support
modification with different protective sols. For N-free and N-
containing supports, PVA and PVP were used as protective
agents for Pd NPs. Our strategy is illustrated in Scheme 1. The
AC was treated with HNO3 and then with NH3 at high tempera-
[a] Dr. C. E. Chan-Thaw, Dr. A. Villa, Prof. L. Prati
Department of Chemistry
Universitꢀ degli Studi di Milano
via Golgi 19, 20133 Milano (Italy)
E-mail: laura.prati@unimi.it
[b] Dr. G. M. Veith
Materials Science and Technology Division
Oak Ridge National Laboratory
Oak Ridge, TN 37831 (USA)
ChemCatChem 0000, 00, 0 – 0
1
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
zation, probably owing to pore blocking in the case of ACpyr
and slight change in the morphology of N-AC. The porosity of
AC was not significantly affected by either the N-atom intro-
duction or the Pd NP loading (Table 1).
Carbon/hydrogen/nitrogen (CHN) analysis was performed to
quantify the total nitrogen content: ACpyr and N-AC materials
revealed the same N content (1.7 at%; Table 2). In the case of
bare AC, a negligible amount of N was detected (Table 2).
Table 2. Atomic percentage determined from XPS and elemental analy-
ses.
Sample
Elemental analysis:
C/H/N [at%]
XPS analysis:
C/N/Pd [at%]
AC
ACpyr
N-AC
PdPVA/AC
PdPVA/ACpyr
PdPVA/N-AC
PdPVP/AC
PdPVP 1:1/N-AC
PdPVP 1:0.5/N-AC
PdPVP 1:0.25/N-AC
PdPVA/N-AC used
PdPVP/N-AC used
85.1:1.1:0.1
85.7:1.4:1.7
86.5:0.9:1.7
86.3:2.1:0.2
84.7:1.3:1.2
82.9:1.3:1.3
81.1:1.4:0.8
83.1:1.2:2.2
81.6:1.2:1.5
82.1:1.6:1.3
83.1:1.6:1.1
83.4:1.5:2.0
–
–
–
Scheme 1. Representative schematic of the strategy adopted to introduce N
functionality. 1) Introduction of N functionalities inside the carbon structure.
2) PdPVA and 3) PdPVP particles were supported on N-AC. PdPVA on AC consti-
tutes our reference (REF), and 4) PdPVP on AC is N on Pd NP. 5) ACPyr repre-
sents the modification of AC through the physisorption of pyridine. 6) PdPVA
98.6:0.0:1.4
98.2:0.8:0.8
98.1:1.0:0.9
96.1:3.1:0.8
91.6:4.2:4.2
92.0:4.9:3.1
91.3:5.9:2.9
98.9:0.6:0.5
93.1:3.6:2.7
and 7) PdPVP NPs were immobilized on ACPyr
.
ture by using a procedure described elsewhere,[11] thus intro-
ducing N functionalities inside the carbon structure (1). On this
carbon, PdPVA (2) and PdPVP (3) particles were supported. PdPVA
on bare AC constituted our reference (REF), whereas PdPVP was
considered to have N functionalities only on Pd. Therefore, we
compared four catalysts that were expected to contain differ-
The N1s XPS data collected for the support materials
(Table 3) revealed the presence of a single N species for the
ACpyr sample with a binding energy of 398.8 eV, which was as-
signed to a pyridinic N group.[12] In contrast, the N1s spectra of
the thermally prepared N-ACs showed two distinct N species
with binding energies of 398.2 and 399.8 eV assigned to pyri-
dinic and pyrrolic N, respectively (42:58).[12] Notably, although
there are two basic sites on the N-AC sample, there is almost
the same number of total basic sites than that on ACpyr materi-
als.
ently allocated N groups: PdPVA/AC (REF), PdPVA/N-AC (2), PdPVP
/
N-AC (3), and PdPVP/AC (4) [i.e., N on reference (REF), N on sup-
port (2), N on Pd and support (3), and N on Pd (4)]. Further-
more, to investigate the effect not only of the N position but
also of the nature of N-containing groups, the behavior of AC
modified through the physisorption of pyridine (ACPyr) (5) was
studied. Also on the ACPyr support, PdPVA (6) or PdPVP (7) NPs
were immobilized.
The autogenerated pH values of the three support materials
(AC, ACpyr, and N-AC) suspended in 10À3 m KCl solution were
6.4, 8.6, and 8.3, respectively (Table 1). The increasing pH value
was consistent with the basic functionalities introduced. Acid/
base titrations with 0.1m HCl (Table 1) revealed almost the
same basic site density for ACpyr (1.5ꢁ10À4 molgÀ1) and N-AC
(1.3ꢁ10À4 molgÀ1), which was consistent with the similar
single-point pH measurements. Surface area measurement re-
vealed a slight decrease in the surface area after N functionali-
Pd NPs were generated in the presence of a capping agent,
either PVA or PVP. PVA contains only O functionalities, whereas
PVP contains both O and N atoms. In solution, the PVA- and
PVP-protected Pd NPs have an average diameter of 3.2 and
3.4 nm, respectively, as determined from TEM analysis (Table 4).
PdPVA and PdPVP were supported on different supports, as de-
picted in Scheme 1. The statistical median sizes of different Pd
NPs determined from TEM analysis are also listed in Table 4.
The representative images of PdPVA/AC and PdPVA/N-AC before
and after the reaction are shown in Figure 1. The NP size and
distribution of PdPVA are almost maintained if supported on
both bare AC and N-AC (3.9 and 3.5 nm, respectively, com-
pared with 3.2 nm for unsupported). In contrast, an increase in
the particle size and a wider distribution were observed after
the deposition of PdPVA on ACpyr (4.9 nm with s=1.5 vs 3.2 nm
with s=0.7 for unsupported). This observation could be attrib-
uted to a relaxation of the particle shapes or a slight agglom-
eration. The agglomeration of Pd NPs can be due to a reaction
between the physisorbed pyridine and the protecting mole-
cules, which implied a diminution of the protecting effect of
the latter. A similar growth in PdPVP is observed for particles
Table 1. Basicity and generated pH, surface area, pore size, and pore
volume of AC, ACpyr, and N-AC.
Sample
Number of Auto-
BET
Pore
Pore
basic sites
generated surface area size
volume
[molgÀ1
]
pH
[m2 gÀ1
]
[nm] [mLgÀ1
]
AC
ACpyr
N-AC
PdPVA/N-AC
–
6.4
8.6
8.3
8.3
1100
1000
1048
1035
2.1
2.2
2.4
2.4
1.40
1.38
1.36
1.34
1.5ꢁ10À4
1.3ꢁ10À4
1.3ꢁ10À4
&
ChemCatChem 0000, 00, 0 – 0
www.chemcatchem.org
2
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
Table 3. Binding energies (eV) and moiety fraction data (%) as deter-
Table 3. (Continued)
mined from XPS analysis.
Sample
Pyridinic
PdPVP/N-AC
[eV]
[%]
N1s
Pyrrolic
Pd3d
NO
Pd0
Pdd+
Sample
C 1s
C=O
CÀC,
CÀH
CÀO,
CÀN
C=C,
pÀp
C+,
C=N
–
–
400.0
93.4
403.0
6.6
335.3
71.5
336.8
28.5
PdPVA/AC
[eV]
[%]
284.9
63.1
286.3
23.9
287.9
3.8
289.4
6.0
291.2
3.2
Table 4. Statistical median and standard deviation of particle size for Pd
catalysts.
PdPVA/ACpyr
[eV]
[%]
284.8
61.2
286.2
23.7
287.7
6.7
289.5
5.7
291.4
2.6
Catalyst
Statistical median
[nm]
Standard deviation
(s)
PdPVA/N-AC
[eV]
[%]
unsupported PdPVA NPs
unsupported PdPVA NPs
PdPVP/AC
PdPVP 1:0.5/AC
PdPVP 1:0.25/AC
PdPVA/AC
PdPVA/ACpyr
PdPVA/N-AC
PdPVA/N-AC used
PdPVA/AC used
PdPVP/N-AC
3.2
3.4
4.1
4.2
4.4
3.9
4.9
3.5
3.8
6.4
4.0
0.7
0.9
1.3
1.4
1.5
1.2
1.5
0.9
1.1
1.8
1.2
284.7
61.3
286.0
23.5
287.5
7.0
289.4
5.5
291.2
2.7
PdPVP/AC
[eV]
[%]
284.7
43.0
286.1
44.7
288.0
8.0
290.3
4.3
–
–
PdPVP 1:0.5/AC
[eV]
[%]
284.7
65.2
286.2
21.4
288.1
8.6
290.1
3.0
291.9
1.8
PdPVP 1:0.25/AC
[eV]
[%]
284.8
70.7
286.4
16.3
288.2
7.4
290.1
3.6
292.0
2.0
PdPVP/N-AC
[eV]
[%]
284.7
63.1
286.2
23.4
288.1
8.8
289.9
3.0
291.7
1.8
Sample
N1s
Pyrrolic
Pd3d
Pyridinic
–
NO
–
Pd0
–
Pdd+
PVP
[eV]
399.7
–
–
ACpyr
[eV]
398.8
–
–
–
N-AC
[eV]
[%]
398.2
42.0
399.8
58.0
–
–
–
–
–
–
PdPVA/AC
[eV]
[%]
–
–
–
–
–
–
335.9
65.9
337.4
34.1
PdPVA/ACpyr
[eV]
[%]
–
–
399.8
100
–
–
335.9
75.0
337.9
25.0
PdPVA/N-AC
[eV]
[%]
398.2
43.3
399.8
56.7
–
–
335.8
64.4
337.7
35.6
Figure 1. TEM images of a) PdPVA/AC and b) PdPVA/N-AC before the reaction
and the corresponding images of c) PdPVA/AC and d) PdPVA/N-AC after the re-
action. Scale bars=10 nm.
PdPVP/AC
[eV]
[%]
–
400.0
100
–
–
335.6
65.0
337.1
35.0
confined to N-AC and AC (4.0 and 4.1 nm, respectively). The
size distribution remained similar in all the cases (s=1.2–1.4;
Table 4). Only a slight increase in particle size was observed
with the increase in Pd/PVP ratio (i.e., with the decrease in PVP
amount) from the standard 1:1 to 1:0.5 and 1:0.25 (i.e., from
4.2 to 4.4 nm) (Table 4).
PdPVP 1:0.5/AC
[eV]
[%]
–
–
399.9
93.1
404.9
6.9
335.4
71.9
337.2
28.1
PdPVP 1:0.25/AC
[eV]
[%]
–
–
399.9
91.3
402.3
8.7
335.5
83.3
337.3
16.7
CHN analysis of the PdPVP/AC material revealed that a total
of 0.8 at% N was introduced by PVP. However, the XPS results
ChemCatChem 0000, 00, 0 – 0
www.chemcatchem.org
3
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
showed a higher N content (3.1 at%), which indicated that PVP
is confined to the catalyst surface (Table 2).
Further analysis of the XPS survey data indicated the pres-
ence of only Pd, C, N, and O species. There was no evidence of
residual Na or B from NaBH4. An analysis of the N1s spectra of
the PdPVP catalysts revealed the presence of single N species
with a binding energy of 400.0 eV, which was consistent with
the binding energy of N1s measured for PVP (399.7 eV) due to
a pyrrole group. A similar N1s species (binding energy=
399.8–400.1 eV) due to the pyrrole moiety from the PVP was
observed for all the PVP catalysts regardless of support chemis-
try (N-AC and AC). This is surprising, considering that the N-AC
material contained both pyridinic and pyrrolic N1s species
(Figure 2). This finding indicates that PVP or PdPVP either is
bound to pyridinic N or forms a thick condensed layer, which
blocks the emitted photoelectrons used for XPS analysis.
For the PVA-derived catalysts, elemental analysis indicated
that N was present in a negligible amount in PdPVA/AC (REF)
catalysts as expected, considering the absence of N in the pre-
cursors. In the cases of PdPVA/ACpyr (6) and PdPVA/N-AC (2), the
absolute N content slightly decreases relative to bare supports
(1.2 and 1.3, respectively, vs. 1.7) owing to the presence of N-
free PVA (Table 2), which dilutes the N concentration. There is
a commensurate decrease in the N1s signal in the XPS data
owing to the attenuating nature of the surface PVA, which
covers the surface N species. Interestingly, there appears to be
a change in the N1s spectra with Pd addition for the PdPVA
/
ACpyr (6) sample, which is not observed for the PdPVA/N-AC (2)
sample. The pyridinic N species observed for PdPVA/ACpyr (6)
evolves to become pyrrole-like N (399.8 eV). In contrast, pyri-
dinic and pyrrolic N species are observed for PdPVA/N-AC (2).
Care should be taken while interpreting this result because the
signal is very weak owing to the low initial N concentration
and PVA. However, this indicates that the physisorbed pyridine
reacts during the sol precipitation to form a pyrrole-like spe-
cies. The chemically bonded pyridinic N species observed on
the N-AC support may be more robust and resistant to reduc-
tion or PVA may not bind to this pyridinic N group, which is
different from the case of PVP. Given the low concentrations of
N in N-AC, additional studies are required to explore the bind-
ing of PVA and PVP to these groups.
The C1s XPS data collected for all the samples show a major
species (44–63% of the carbon species) with a binding energy
of 284.8 eV assigned to CÀH and CÀC bonds. The next most
significant species (23–44% of the carbon species) appears at
approximately 286.1–286.2 eV and is attributed to CÀN and CÀ
O species. Lower concentrations of C=O, C=C, and C=N species
are observed at approximately 288, 290, and 292 eV, respec-
tively. Together these results are consistent with a significant
concentration of CÀN, CÀO, CÀC, and CÀH species as expected
from the PVA and PVP molecules on the surface.
Figure 2. XPS data collected for PdPVA/N-AC.
The Pd3d XPS data collected for all the catalysts (Figure 2)
reveal the presence of at least two Pd species. The main spe-
cies has a binding energy of approximately 335.6Æ0.4 eV and
is attributed to Pd. The second Pd species has a binding
energy of approximately 337.5Æ0.5 eV and is attributed to an
oxidized Pd species. On the basis of the binding energy, we es-
timated the oxidation state of this second species to be ap-
proximately 2+. The oxidized Pd makes up a minority of Pd in
the samples (15–35%) and is likely due to surface oxidation
from the reaction with O2. The small range of the ratio of re-
The O1s XPS data collected for PVA-derived catalysts are
consistent with the results collected for the PVP-stabilized cata-
lysts. Again, there is evidence of CÀO and C=O species on the
surface along with low concentrations of CÀN. The exact com-
position is difficult to elucidate owing to Pd3p species, which
overlap with the O1s signals at approximately 533 eV.
&
ChemCatChem 0000, 00, 0 – 0
www.chemcatchem.org
4
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
duced and oxidized Pd is consistent with similarly sized Pd
NPs.
1) PdPVP/AC (4), in which N is coordinatively bound to Pd; N
groups are mostly pyrrolic in nature.
To evaluate the binding of the protecting group to the Pd
NPs, IR spectroscopic studies were performed before immobili-
zation on the support. In the case of pure PVA (Figure 3a), two
bands of interest are observed at 1735 and 1375 cmÀ1. These
bands are ascribed to CÀO vibration (and residual C=O by in-
2) PdPVA/N-AC (2), in which N is strongly bound within the
carbon structure; however, N groups are both pyridinic and
pyrrolic in nature.
3) PdPVA/ACPyr (6), in which the nature of N groups appeared
to be mostly pyrrolic but differs in the bonding configura-
tion compared to PdPVP/AC (4).
4) PdPVP/N-AC (3) and PdPVP/ACpyr (7) were studied, highlighting
any possible addition effect.
These observations are summarized in Table 5.
The catalytic activity, expressed as converted moles of alco-
À1 À1
hol (in mol h ), was tested in the liquid-phase oxidation of
Pd
benzyl alcohol; the results are summarized in Table 6. By com-
paring PVA- and PVP-protected Pd NPs on bare carbon, we
found a 30% decrease in activity for PdPVP/AC despite the simi-
lar particle size (3.9 nm for PdPVA/AC vs. 4.1 nm for PdPVP/AC;
Table 4).
Table 5. Location, nature, and amount of N funcationalities.
Sample
Nature and amount [%]
Location
Pyridinic
Pyrrolic
PVP
N-AC (1)
ACpyr (5)
PdPVA/AC (REF)
PdPVA/N-AC (2)
PdPVP/AC (4)
PdPVP/N-AC (3)
PdPVA/ACpyr (6)
–
42.2
100
–
43.3
–
–
–
100
58.8
–
Pd
carbon
carbon
–
carbon
Pd
–
56.7
100
100
100
Pd+carbon
carbon
Figure 3. IR spectra of a) pure PVA and PdPVA and b) pure PVP and PdPVP
.
Table 6. Comparison of activities of Pd catalysts for benzyl alcohol oxida-
tion.[a]
complete hydrolysis of acetate groups) and CÀH vibration, re-
spectively.[13] In the corresponding spectrum of PdPVA, CÀO/C=
O stretching is redshifted, which indicates weakening of the
CÀO bond; this result is consistent with the coordination of O
atoms of PVA to Pd NPs. Thus, we can conclude that PVA inter-
acts with Pd NPs through the PdÀO interaction.
Pathway
Catalyst
Catalytic activity[b]
Selectivity [%][c]
[molÀPd1 hÀ1
]
Tol
BAl
BAc
BB
(REF)
(4)
(3)
(6)
(2)
PdPVA/AC
PdPVP/AC
2756
1953
2354
940
3784
2208
2409
25
23
22
13
20
24
25
74
76
77
83
70
75
73
0
0
0
1
4
0
1
1
1
1
3
6
1
1
PdPVP/N-AC
PdPVA/ACpyr
PdPVA/N-AC
PdPVP 1:0.5/AC
PdPVP 1:0.25/AC
PVP chemisorption on Pd NPs was previously studied by
Xian et al.,[14] showing that the coordination of the PVP moiety
on the Pd surface depends on the PVP/Pd ratio and the size of
the metal particles. The FTIR spectra of our materials (Fig-
ure 3b) show that the C=O stretching vibrational band at
1659 cmÀ1 for pure PVP is only slightly shifted to 1650 cmÀ1 in
PdPVP NPs whereas the NÀC stretching vibrational bands for NÀ
C=O and NÀCÀC of the ring (1290 and 1651 cmÀ1, respectively)
almost disappear. This indicates that Pd binds selectively to
the pyrrole ring of PVP, which forms a strong bond. Because of
the above-mentioned results, we considered the prepared cat-
alysts as representatives of different N-modified materials rela-
tive to our reference catalyst PdPVA/AC (REF):
(4)
(4)
[a] Reaction conditions: alcohol/metal=5000, T=808C, pO2 =2 atm,
1.25m benzyl alcohol in cyclohexane; [b] Catalytic activity (expressed as
converted moles of alcohol per mole of Pd per h) calculated after 15 min
of reaction; [c] Selectivity calculated at 60% conversion. BAc=benzoic
acid; BAl=benzaldehyde; BB=benzyl benzoate; Tol=toluene.
If PdPVA is immobilized on N-AC, we observed a marked in-
crease in the catalytic activity of approximately 30% with re-
spect to our reference catalyst (PdPVA/AC) (activity of 3784 and
2756 hÀ1 for PdPVA/N-AC and PdPVA/AC, respectively). Therefore,
it is apparent that despite the similar nature of N groups (pyri-
dinic and pyrrolic in PdPVA/N-AC and only pyrrolic in PdPVP-AC),
ChemCatChem 0000, 00, 0 – 0
www.chemcatchem.org
5
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
the N atom chemically bound to Pd has a detrimental effect
on the activity.
Understanding the correlation between the catalytic activity
and the N-containing support constituted the next step. We,
therefore, turn our attention to N heteroatoms on the support
material. The data show large variations in the observed activ-
ity depending on the support chemistry for both PVA- and
We assumed that any decrease in activity, because of
a shielding effect due to the presence of a capping agent, is
similar in both cases, as well as the effect of the support and
particle sizes. From the IR spectra, a strong coordination to Pd
was detected for PVP through N atoms whereas for PVA only
O atoms were involved.
PVP-based catalysts (Table 6). These data show that the PdPVA
/
N-AC (2) catalyst has an activity of approximately 30% higher
than the activity of PdPVA/AC (REF) (activity of 3784 hÀ1 vs.
2756 hÀ1). However, there is a significant decrease in activity
for PdPVA NPs on the ACpyr support (6). The increase in PdPVA
NP size when immobilized on ACpyr compared to AC (from 3.9
to 4.9 nm; Table 4) cannot justify such a large difference in
terms of catalytic activity (activity of 2756 and 940 hÀ1 for
PdPVA/AC and PdPVA/ACPyr, respectively). Following the same
reasoning, PdPVA NPs immobilized on N-AC are smaller than
those immobilized on bare AC (particle size: 3.5 nm vs. 3.9 nm
for PdPVA/N-AC and PdPVA/AC). This variance in particle size
cannot justify the definite increase in activity (activity of 3784
and 2756 hÀ1 for PdPVA/N-AC and PdPVA/AC, respectively). Keep-
ing in mind that 1) PdPVA/N-AC contains pyridinic and pyrrolic
groups within the carbon structure, 2) PdPVA/ACpyr has only
pyrrolic groups, and 3 PdPVP/AC is only made of pyrrolic
groups coordinatively bonded to Pd (Table 5), we could argue
that the most important feature in determining a positive cat-
alytic effect is the presence of pyridinic moieties within the
carbon structure (case 1). To confirm the benefits of bound N
groups, we investigated PdPVP immobilized on the N-AC sup-
port (Scheme 1, pathway 3). A better catalytic performance
was obtained with PdPVP/N-AC (activity of 2354 hÀ1; Table 6)
compared with that obtained with PdPVP supported on AC (ac-
tivity of 1953 hÀ1) (Scheme 1, pathway 4). Because N-AC con-
sists of both pyridinic and pyrrolic groups and PdPVP/N-AC
contains only pyrrolic groups (Table 5), the pyrrolic groups are
both within the structure and coordinatively bonded to the
metal. It is therefore possible to conclude that the location
and bonding of N groups are more important than the nature
of the groups.
It reasonably appears that the active sites can be firmly coor-
dinated to PVP, which results in a restricted access to the cata-
lytic sites. A similar effect of PVP on catalytic properties of Pt
NPs has been observed in CO oxidation.[15] However, the coor-
dination of PVP could be dependent on the Pd/PVP ratio as re-
ported in Ref. [14]. Therefore, we investigated the effect of the
Pd/PVP ratio by decreasing the PVP amount. We considered
only those Pd/PVP ratios that ensured a negligible variation in
particle size distribution obtained with the standard 1:1 ratio.
We also investigated the behavior of Pd/PVP (1:0.5 and 1:0.25;
Table 4). These ratios led to the formation of supported Pd NPs
with the mean size, which was similar to the one obtained
with a 1:1 original ratio (4.2 and 4.4 nm vs. 4.1 nm for the Pd/
PVP ratio of 1:0.5, 1:0.25, and 1:1, respectively; Table 4). The re-
sults indicated (Table 5) that the activity decreases with the in-
crease in PVP amount (activity of 2409, 2208, and 1953 hÀ1 for
the Pd/PVP ratio of 1:0.25, 1:0.5, and 1:1 respectively). This re-
duction in activity cannot be correlated with any changes in
the nature of surface species. The N1s, C1s, and Pd3d XPS
data are remarkably consistent with each sample (Table 3). The
only exception is the apparent increase in the concentration of
reduced Pd, which may be attributed to the lower concentra-
tion of PVP bound to Pd. In addition to similar oxidation states
(Table 3), Pd/PVA and Pd/PVP (1:1) supported on AC have a dif-
ferent catalytic activity (Figure 4). We can thus presume that
the Pd oxidation state cannot dominate the catalytic per-
formance. Therefore, it is apparent that the detrimental effect
on the activity of PVP is likely due to the increase in PVP on
the Pd surface. A similar blocking of Pd active sites was ob-
served for Pd–N NPs, and it may indicate that N groups have
a beneficial effect only in close proximity but not directly
bound to the Pd NPs.[16]
Moreover, following the reaction profiles an evident deacti-
vation was observed in the case of PdPVA/ACPyr (Figure 4) after
6 h of reaction. Therefore, we investigated the possibility of
a poisoning effect of the pyridine moiety by testing the cata-
lytic activity of PdPVA/AC directly impregnated with pyridine.
The catalytic test of pyridine-treated PdPVA/AC showed a drastic
drop of the activity. Thus, this finding supports the detrimental
effect of adsorbed pyridine on the catalytic activity, probably
owing to the stronger coordination bond between pyridine
and Pd than between pyrrole and Pd. Notably, PdPVA/ACPyr is
the only case in which an increased selectivity to benzalde-
hyde at the expense of toluene formation was observed
(Table 6). Toluene can be produced through two pathways: dis-
proportion of two molecules of benzaldehyde and reoxidation
of the intermediate Pd hydride by benzyl alcohol instead of
O.[17] In both cases, it can be envisaged that Pd active sites
could be involved. The change in selectivity in the case of
PdPVA/ACPyr indicates that the Pd active sites differ from those
present in PdPVA/AC and PdPVA/N-AC. This result could be attrib-
uted to a progressive leaching of pyridine from the support
Figure 4. Reaction profile of Pd catalysts for benzyl alcohol oxidation.
&
ChemCatChem 0000, 00, 0 – 0
www.chemcatchem.org
6
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
and a consecutive readsorption on Pd NPs, which thus modi-
fies the active site exposure.
N-AC remained stable along the six runs. The inductively cou-
pled plasma (ICP) analysis of the collected solution of six runs
resulted in a loss of 21 wt% of total Pd in the case of PdPVA/AC
but only 3 wt% in the case of PdPVA/N-AC.
The long-term stability of PdPVA/N-AC relative to PdPVA/AC
was also investigated. PdPVA/N-AC represents the most active
catalyst (activity of 3784 and 2756 hÀ1 for PdPVA/N-AC and
PdPVA/AC, respectively; Table 6). However, the synergistic inter-
actions between Pd and N atoms could have a positive effect
on the stability of the catalyst itself. Recycling tests were per-
formed on PdPVA/AC and PdPVA/N-AC by filtering the catalyst
and reusing it without any further purification for the next run.
The results highlight the excellent stability of PdPVA/N-AC. As
The TEM analysis revealed that Pd NP distribution on N-AC
appeared almost the same as in the fresh catalyst whereas ag-
gregation can be observed for PdPVA/AC (Figure 6). In the case
of PdPVA/AC, an increase of more than 50% is observed in the
Pd NP size (3.9 and 6.4 nm for the fresh and the used catalyst,
respectively), whereas in the case of PdPVA/N-AC, an increase of
only 10% is observed (Table 4).
shown in Figure 5, PdPVA/AC deactivated quickly whereas PdPVA
/
Conclusions
From the reported data, we conclude that simply having N in
close proximity to the Pd catalysts is not sufficient to enhance
the catalytic activity. The N atom has to be confined in such
a way that it does not interfere with the Pd active sites. By
binding N species directly to Pd through the pyrrolic groups
that polyvinyl pyrrolidone contains, the catalytically active sites
are not accessible. Furthermore, simply having N in a solution
or weakly bound to the carbon surface in the form of pyridine
is detrimental to the catalytic activity and stability. In contrast,
if N groups are covalently bonded to the support, a significant
positive catalytic effect can be detected and no desorption
phenomenon occurs. Moreover, the stability in terms of activity
Figure 5. Stability tests using PdPVA/AC and PdPVA/N-AC.
Figure 6. Particle distribution of a) PdPVA/AC, b) PdPVA/AC used, c) PdPVA/N-AC, and d) PdPVA/N-AC used.
ChemCatChem 0000, 00, 0 – 0
www.chemcatchem.org
7
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

Full Papers
ing of 1 wt% was obtained. The obtained catalysts were labeled
according to the nature of Pd sol used (PdPVA or PdPVP) and the sup-
port (AC, ACpyr, or N-AC). For example, PdPVP 1:0.5/AC corresponded
to a sol generated with a Pd/PVP ratio of 1:0.5 wt/wt and support-
ed on AC whereas PdPVP/AC indicated a Pd sol generated with
a Pd/PVP ratio of 1:1 wt/wt and immobilized on AC. The Pd/PVA or
Pd/PVP ratio was always 1 wt/wt if not specifically indicated. The
ICP analysis of the filtrate confirmed the actual loading on the sup-
port (1 wt%).
has been shown upon recycling of the PdPVA/N-AC catalyst,
which maintained activity as well as selectivity to benzalde-
hyde for six runs. The N-modified support limited the leaching
of Pd and the coarsening of Pd NPs, which is more evident in
the case of bare support. Further studies are underway to ex-
plore the effect of different kinds of bound N functionalities,
that is, pyridinic, pyrrolic, and quaternary, on the observed cat-
alytic activity.
Experimental Section
Catalytic test
The reactions were performed in a thermostated glass reactor
(30 mL) agitated with an electronically controlled magnetic stirrer
connected to a large reservoir (5000 mL) containing O at 2 atm
(1 atm=101.3 kPa). The O2 uptake was followed by a mass flow
controller connected to a personal computer through an analog/
digital board. The oxidation experiments were performed in cyclo-
hexane (1.25m substrate, substrate/Pd=5000 mol/mol, T=808C,
pO2 =2 atm). The reaction was monitored by analyzing periodically
withdrawn samples. In the analysis, mass balances were always
(98Æ3)%. Analyses were performed with a HP 7820A gas chroma-
tograph equipped with a capillary column HP-5 (30 mꢁ0.32 mm,
0.25 mm film; Agilent Technologies). Authentic samples were ana-
lyzed to determine separation times. Quantitative analyses were
performed by using an external standard method (n-octanol).
Materials
Na2PdCl4·2H2O was from Aldrich (purity 99.99%), and AC was from
Norit (GSX; surface area=1100 m2 gÀ1; pore volume=1.4 mLgÀ1).
NaBH4 (purity >96%) and HNO3 were from Fluka. PVA (MW
ꢀ13000–23000; 87–89% hydrolyzed), PVP (MWꢀ10000; purity
95%), and pyridine (purity >99%) were from Sigma–Aldrich. Gas-
eous oxygen (purity 99.99%) and gaseous ammonia (purity
99.70%) were from SIAD.
Support functionalization
N-AC
AC was functionalized using the procedure described elsewhere.[11]
O-containing carbon was obtained by treating pristine AC with
HNO3 (65 wt%, 20 gAC LHNO À1) at 1008C for 2 h under continuous
Recycling test
3
stirring. The material was rinsed with distilled water, filtered,
washed several times, and finally dried at 708C for 8 h. A portion
of this carbon was thermally treated with NH3 (0.2 LminÀ1) at
6008C for 4 h and labeled as N-AC.
Each run was performed under the same conditions (alcohol/
metal=5000, T=808C, pO2 =2 atm, 1250 rpm, 50 vol% cyclohex-
ane, t=2 h for PdPVA/N-AC and t=7 h for PdPVA/AC). The catalyst
was recycled in the subsequent run after filtration without any fur-
ther treatment.
ACPyr
A second procedure to introduce N functionality consisted in im-
pregnating the AC with pyridine at RT. AC (5 g) was poured into
pyridine (100 mL) at RT and stirred for 8 h. After filtration, carbon
was dried at 808C for 4 h. The so-obtained functionalized AC was
labeled as ACpyr. The amount of adsorbed pyridine was quantified
by using thermal gravimetric analysis performed with a PerkinElmer
TGA7 thermogravimetric analyzer at a temperature ranging from
50 to 5008C (heating rate=38CminÀ1). CHN analyses were per-
formed with a PerkinElmer 2400 Series II CHNS/O Elemental Ana-
lyzer.
Characterization
The intrinsic acidity of the carbon atoms was measured with a Met-
rohm 718 STAT Titrino. Typically, the sample (100 mg) was dis-
persed in KCl solution (10À3 m, 50 mL). The mixture was stirred
overnight at RT, and the pH of the solution was measured. To
quantify the concentration of basic sites, titration was performed
with an HCl solution (0.01m) on a suspension of carbon (100 mg)
in deionized water (100 mL). Before measurement, the mixture was
degassed under Ar for at least 1 h until the pH value was constant.
The UV/Vis spectra of sols were recorded on HP8452 and HP8453
spectrophotometers in water between 190 and 800 nm in a quartz
cell.
Catalyst preparation
The metal content of the catalyst was determined from the ICP
analysis of the filtrate or directly of the catalyst after burning off
the carbon by using a Jobin Yvon JY24 spectrometer. For TEM anal-
yses, the catalyst samples were ultrasonically dispersed in ethanol
and mounted onto copper grids covered with a holey carbon film.
A Philips CM200 LaB6 transmission electron microscope, operating
at 200 kV and equipped with a Gatan charge-coupled device
camera, was used. The mean size of Pd NPs was calculated from
the TEM data of 300 particles from different areas. XPS data were
collected with a PHI 3056 spectrometer equipped with an Al
anode source operating at 15 kV and an applied power of 350 W.
The energy scale calibration of the instrument was checked by
using a sputter-cleaned piece of Ag. The samples were manually
pressed between two pieces of the In foil; the piece of the In foil
Pd sol preparation
The Na2PdCl4·2H2O (0.043 mmol) salt and a freshly prepared PVA or
PVP solution (1 wt%) were added to water (130 mL; Pd/PVA ratio=
1:1 wt/wt; Pd/PVP ratio=1:1, 1:0.5, and 1:0.25 wt/wt). After 3 min,
the NaBH4 solution (0.1m; Pd/NaBH4 ratio=1:8 mol/mol) was
added to the yellow-brown solution under vigorous magnetic stir-
ring. The brown Pd0 sol was immediately formed. A UV/Vis spec-
trum of Pd sol was recorded to ensure the complete reduction of
Pd2+. Within a few minutes from its generation, the suspension
was acidified at pH 2 by H2SO4 and the support was added under
vigorous stirring. The catalyst was filtered and washed several
times with distilled water. The samples were dried at 808C for 2 h.
The amount of the support was calculated, and a final metal load-
&
ChemCatChem 0000, 00, 0 – 0
www.chemcatchem.org
8
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!

Full Papers
J. Hartmann, A. Thomas, M. Antonietti, Adv. Mater. 2009, 21, 897–901;
c) A. Thomas, Angew. Chem. Int. Ed. 2010, 49, 8328–8344; Angew. Chem.
2010, 122, 8506–8523.
with the sample on it was then mounted to the sample holder
with a piece of carbon tape (Nisshin EM Co., Ltd.). In was used as
a support because the most intense C1s binding energy, from the
support material, overlapped with the C signal from the tape. The
lowest C1s binding energy was always measured at 248.6 eV, but
the peaks were shifted to 284.8 eV for comparison to literature
values. There were no treatments (i.e., cleaning, heating, and proc-
essing) to the samples before the XPS measurements. High-resolu-
tion scans were acquired with a 23.5 eV pass energy and 0.05 eV
energy steps a minimum of 50 times; survey scans were measured
with a 93.9 eV pass energy and 0.5 eV energy steps a minimum of
25 times. Surface concentrations were calculated by integrating
the peak areas and using standard atomic sensitivity factors sup-
plied by the equipment manufacturer. The spectra were deconvo-
luted by using Gaussian–Lorentzian functions and a Shirley-type
background.
[5] C. E. Chan-Thaw, A. Villa, P. Katekomol, D. Su, A. Thomas, L. Prati, Nano
Lett. 2010, 10, 537–541.
[6] C. E. Chan-Thaw, A. Villa, L. Prati, A. Thomas, Chem. Eur. J. 2011, 17,
1052–1057.
[7] a) A. Villa, D. Wang, P. Spontoni, R. Arrigo, D. S. Su, L. Prati, Catal. Today
2010, 157, 89–93; b) L. Prati, A. Villa, C. E. Chan-Thaw, R. Arrigo, D.
Wang, D. S. Su, Faraday Discuss. 2011, 152, 353–365; c) R. Arrigo, S.
Wrabetz, M. E. Schuster, D. Wang, A. Villa, D. Rosenthal, F. Girsgdies, G.
Weinberg, L. Prati, R. Schlçgl, D. S. Su, Phys. Chem. Chem. Phys. 2012,
14, 10523–10532; d) S. Abate, M. Freni, R. Arrigo, M. E. Schuster, S. Pera-
thoner, G. Centi, ChemCatChem 2013, 5, 1899–1905; e) R. Arrigo, M. E.
Schuster, S. Abate, S. Wrabetz, K. Amakawa, D. Teschner, M. Freni, G.
Centi, S. Perathoner, M. Hꢂvecker, R. Schlçgl, ChemSusChem 2014, 7,
179–194.
[8] M. Lefꢃvre, E. Proietti, F. Jaouen, J. P. Dodelet, Science 2009, 324, 71–74.
[9] a) P. S. Roy, J. Bagchi, S. K. Bhattacharya, Trans. Met. Chem. 2009, 34,
447–453; b) T. Umegaki, J. M. Yan, X. B. Zhang, H. Shioyama, N. Kuriya-
ma, Q. Xu, Int. J. Hydrogen Energy 2009, 34, 3816–3822; c) R. Narayanan,
M. A. El-Sayed, J. Am. Chem. Soc. 2003, 125, 8340–8347; d) C. D. Pina, E.
Falletta, M. Rossi, Chem. Soc. Rev. 2012, 41, 350–369.
[10] a) Z. Kꢄnya, V. F. Puntes, I. Kiricsi, J. Zhu, P. Alivisatos, G. A. Somorjai,
Catal. Lett. 2002, 81, 137–140; b) H. Tsunoyama, N. Ichikuni, H. Sakurai,
T. Tsukuda, J. Am. Chem. Soc. 2009, 131, 7086; c) A. Villa, D. Wang, G. M.
Veith, F. Vindigni, L. Prati, Catal. Sci. Technol. 2013, 3, 3036–3041.
[11] a) R. Arrigo, M. Hꢂvecker, R. Schlçgl, D. S. Su, Chem. Commun. 2008,
4891–4893; b) R. Arrigo, M. Haevecker, S. Wrabetz, R. Blume, M. Lerch,
J. McGregor, E. P. J. Parrott, J. A. Zeitler, L. F. Gladden, A. Knop-Gericke, R.
Schloegl, D. Su, J. Am. Chem. Soc. 2010, 132, 9616–9630.
The FTIR spectra were recorded on a JASCO FT/IR 410 spectrome-
ter equipped with a mercury cadmium telluride detector. The spec-
tra with 4 cmÀ1 resolution, 64 scans, and
a scan speed of
0.20 cmsÀ1 were recorded at RT by using KRS5 Thallium Bromo-
Iodide windows in the range of 4000–1000 cmÀ1. KBr pellets (press-
ed at 10 MPa for 10 min) were impregnated with a solution con-
taining only the capping agent or the Pd NPs protected by PVA or
PVP.
Surface area and pore volume were determined by performing
low-temperature N2 adsorption with an SA 3100 Coulter apparatus.
Before measurements, samples were outgassed at 808C for 3 h.
[12] a) J. F. Moulder, W. F. Stickle, P. E. Sobol, K. D. Bomben in Handbook of X-
ray Photoelectron Spectroscopy, PerkinElmer Corporation, Eden Prairie,
1992, pp. 86–89; b) J. P. Boudou, M. Chehimi, E. Broniek, T. Siemieniew-
ska, J. Bimer, Carbon 2003, 41, 1999–2007; c) F. Kapteijn, J. A. Moulijn,
S. Matzner, H. P. Boehm, Carbon 1999, 37, 1143–1150; d) K. Staczyk, R.
Dziembaj, Z. Piwowarska, S. Witkowski, Carbon 1995, 33, 1383–1392;
e) R. J. J. Jansen, H. van Bekkum, Carbon 1995, 33, 1021–1027.
[13] a) M. N. Nadagouda, R. S. Varma, Macromol. Rapid Commun. 2007, 28,
465–472; b) H. S. Mansur, C. M. Sadahira, A. N. Souza, A. A. P. Mansur,
Mater. Sci. Eng. C 2008, 28, 539–548; c) L. Shao, W. Ji, P. Dong, M. Zeng,
C. Qi, X.-M. Zhang, Appl. Catal. A 2012, 413, 267.
Acknowledgements
This research was partially supported by the U.S. Department of
Energy, Basic Energy Sciences, Materials Sciences and Engineering
Division (to G.M.V.).
Keywords: carbon · nanoparticles · nitrogen · oxidation ·
palladium
[14] J. Xian, Q. Hua, Z. Jiang, Y. Ma, W. Huang, Langmuir 2012, 28, 6736–
6741.
[15] J. Y. Park, C. Aliaga, J. R. Renzas, H. Lee, G. A. Somorjai, Catal. Lett. 2009,
129, 1–6.
[16] G. M. Veith, A. R. Lupini, L. Baggetto, J. F. Browning, J. K. Keum, A. Villa,
L. Prati, A. B. Papandrew, G. A. Goenaga, D. R. Mullins, S. E. Bullock, N. J.
Dudney, Chem. Mater. 2013, 25, 4936–4945.
[1] T. Mallat, A. Baiker, Catal. Today 1994, 19, 247–283.
[2] a) A. Corma, H. Garcia, Chem. Soc. Rev. 2008, 37, 2096–2126; b) C. P.
Vinod, K. Wilson, A. F. Lee, J. Chem. Technol. Biotechnol. 2011, 86, 161–
171.
[3] a) A. Villa, M. Schiavoni, L. Prati, Catal. Sci. Technol. 2012, 2, 673–682;
b) L. Prati, A. Villa, A. R. Lupini, G. M. Veith, Phys. Chem. Chem. Phys.
2012, 14, 2969–2978; c) N. Dimitratos, J. A. Lopez-Sanchez, G. J. Hutch-
ings, Chem. Sci. 2012, 3, 20–44; d) S. E. Davis, M. S. Ide, R. J. Davis,
Green Chem. 2013, 15, 17–45; e) X. Y. Liu, A. Wang, T. Zhang, C.-Y. Mou,
Nano Today 2013, 8, 403–416.
[17] a) C. Keresszegi, D. Ferri, T. Mallat, A. Baiker, J. Phys. Chem. B 2005, 109,
958–967; b) S. Meenakshisundaram, E. Nowicka, P. J. Miedziak, G. L.
Brett, R. L. Jenkins, N. Dimitratos, S. H. Taylor, D. W. Knight, D. Bethell,
G. J. Hutchings, Faraday Discuss. 2010, 145, 341–356.
[4] a) P. Kuhn, M. Antonietti, A. Thomas, Angew. Chem. Int. Ed. 2008, 47,
3450–3453; Angew. Chem. 2008, 120, 3499–3502; b) P. Kuhn, A. Forget,
Received: November 26, 2014
Published online on && &&, 0000
ChemCatChem 0000, 00, 0 – 0
www.chemcatchem.org
9
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
&
These are not the final page numbers! ÞÞ

FULL PAPERS
C. E. Chan-Thaw, A. Villa, G. M. Veith,
L. Prati*
Perfect bonding: Differently bound ni-
trogen groups have a significant effect
on catalytic activity. Simply adding ni-
trogen groups using solution chemistry
is insufficient to improve catalytic prop-
erties. Instead, a strongly bound nitro-
gen moiety within the activated carbon
structure enhances the activity and
stability of the Pd nanoparticles.
&& – &&
Identifying the Role of N-Heteroatom
Location in the Activity of Metal
Catalysts for Alcohol Oxidation
&
ChemCatChem 0000, 00, 0 – 0
www.chemcatchem.org
10
ꢀ 0000 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
ÝÝ These are not the final page numbers!