Importance of the Nature of the Active Acid/Base Pairs of Hydroxyapatite Involved in the Catalytic Transformation of Ethanol to n-Butanol Revealed by Operando DRIFTS

Article abstract of DOI:10.1002/cctc.201801880

Operando DRIFTS is used to identify the nature and the role of the surface sites of hydroxyapatites (HAps) involved in the catalytic transformation of ethanol to n-butanol. The surface processes occurring upon a first reaction step followed by a step under He flow greatly influence the reactivity of HAps in a subsequent second reaction step. Ethanol is found to be mostly activated by the basic OH^? groups of HAps, as indicated by the concomitant recovery of ethanol conversion and OH^? groups under He flow. The drastic changes in selectivity observed during the second reaction step reveal the key role of acidic sites cooperatively acting with basic sites for basic reaction steps. Once the POH groups are poisoned by extensive formation of polymeric carbon species and the Ca²⁺ sites are available, the production of acetaldehyde is drastically promoted at the expense of that of n-butanol. It is concluded that i) acetaldehyde acts as an intermediate in the formation of n-butanol, and ii) various active sites are involved in the key basic reaction steps such as Ca²⁺?OH^? and POH?OH^? acid-base pairs in the dehydrogenation of ethanol to acetaldehyde and the aldol condensation for n-butanol formation, respectively.

Full text of DOI:10.1002/cctc.201801880

Full Papers
DOI: 10.1002/cctc.201801880
1
2
3
4
5
6
7
8
9
Importance of the Nature of the Active Acid/Base Pairs of
Hydroxyapatite Involved in the Catalytic Transformation of
Ethanol to n-Butanol Revealed by Operando DRIFTS
Manel Ben Osman,[a] Jean-Marc Krafft,[a] Cyril Thomas,[a] Tetsuya Yoshioka,[b] Jun Kubo,[b] and
Guylène Costentin*[a]
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
Operando DRIFTS is used to identify the nature and the role of
the surface sites of hydroxyapatites (HAps) involved in the
catalytic transformation of ethanol to n-butanol. The surface
processes occurring upon a first reaction step followed by a
step under He flow greatly influence the reactivity of HAps in a
subsequent second reaction step. Ethanol is found to be mostly
activated by the basic OHÀ groups of HAps, as indicated by the
concomitant recovery of ethanol conversion and OHÀ groups
under He flow. The drastic changes in selectivity observed
during the second reaction step reveal the key role of acidic
sites cooperatively acting with basic sites for basic reaction
steps. Once the POH groups are poisoned by extensive
formation of polymeric carbon species and the Ca2+ sites are
available, the production of acetaldehyde is drastically pro-
moted at the expense of that of n-butanol. It is concluded that
i) acetaldehyde acts as an intermediate in the formation of n-
butanol, and ii) various active sites are involved in the key basic
reaction steps such as Ca2+À OHÀ and POHÀ OHÀ acid-base pairs
in the dehydrogenation of ethanol to acetaldehyde and the
aldol condensation for n-butanol formation, respectively.
Introduction
tion reaction to form unsaturated condensation products that
are eventually hydrogenated to n-butanol,[9] possibly via a
Meerwein-Ponndorf-Verley-like mechanism.[10] This pathway is
supported by recent kinetic studies,[4] in which adsorbed
acetaldehyde species have been revealed as intermediates.
Combining thermodynamic and kinetic data, Scalbert et al.
concluded that n-butanol is both a primary product resulting
from a “direct” bimolecular route (minor fraction) and a
secondary product.[7c] Such a “direct” bimolecular route was also
claimed for the KÀ Cu/MgCeOx system.[11] According to Scalbert
et al., the main “indirect” route to n-butanol would result from
the condensation of acetaldehyde and ethanol, rather than
from the self-condensation of acetaldehyde.[7c] Further insights
into the understanding of the mechanism of formation of n-
butanol from ethanol would require the identification of the
nature of the active sites involved in the transformation of
ethanol to n-butanol of the hydroxyapatite system. To date, it is
assumed that the Guerbet coupling reaction requires weak
acid-base bifunctional catalysts with a “proper balance of acid-
base site pairs”, that might explain the better catalytic perform-
ance of HAps compared to MgO.[4a] Finally, it was reported
recently that the active sites involved in the formation of
acetaldehyde and n-butanol are different.[4b]
Hydroxyapatite (Ca10(PO4)6(OH)2 generic formula) is a well-
known calcium phosphate compound that is also the main
mineral component of bones and teeth. This biocompatible,
ecofriendly and harmless material was recently found of interest
in heterogeneous bifunctional[12] and acid-base catalysis,[7a,13]
thanks to its synthesis-dependent tunable properties.[14] It was
found that changes in the Ca/P bulk ratio influence the surface
reactivity to a significant extent.[14a,15] Ca/P ratios lower than the
stoichiometric ratio, i.e. 1.67, would favor acidity, while higher
Ca/P ratios would rather promote basicity.[7a,16] Note, however,
In the context of growing production of ethanol from biomass,
its valorization into higher value chemicals such as ethylene,
1,3-butadiene and n-butanol has attracted much attention
recently.[1] Besides being an important industrial chemical for
paints and polymers, as well as a solvent,[2] n-butanol is also
considered as a potential gasoline fuel additive[1a,3] due to its
higher energy density and its lower miscibility in water relative
to ethanol.[4] Butanol is commonly produced via the hydro-
formylation reaction (oxo process) followed by a hydrogenation
step. Yet these processes require fuel-derived feedstocks and
severe operating conditions leading to high operating costs.[5]
Recently, it was shown that n-butanol can be alternatively
produced via the coupling of ethanol molecules over a wide
variety of catalysts,[3a,6] among which the hydroxyapatite (HAps)
system appeared to be very attractive due to its high activity
and selectivity.[4,6a,7]
Although still debated in the literature, it is commonly
reported that the ethanol coupling to n-butanol occurs via a
Guerbet-type mechanism.[7a,b,d] In the related Guerbet trans-
formation,[8] an ethanol molecule is firstly dehydrogenated into
acetaldehyde. Acetaldehyde then reacts via an aldol condensa-
[a] M. B. Osman, J.-M. Krafft, C. Thomas, G. Costentin
Laboratoire Réactivité de Surface, LRS
Sorbonne Université, CNRS
75005 Paris (France)
E-mail: guylene.costentin@upmc.fr
[b] T. Yoshioka, J. Kubo
Central Research Center
Sangi Co., Ltd.
Fudoinno 2745-1, Kasukabe-shi, Saitama 344-0001 (Japan)
Supporting information for this article is available on the WWW under
https://doi.org/10.1002/cctc.201801880
ChemCatChem 2019, 11, 1–15
1
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
��
These are not the final page numbers!

Full Papers
that the acid-base properties of HAps do not always follow the
above-mentioned guidelines based on the Ca/P ratio, especially
when comparing hydroxyapatite samples prepared under differ-
ent conditions.[14b] It must be stressed that the Ca/P ratio does
not only depend on the calcium content of the sample but also
on its phosphorus content as phosphates, which itself strongly
depends on the presence of carbonates. In addition, the Ca/P
bulk ratio does not account for the OH concentration. The latter
was shown to be influenced to a significant extent by the
presence of structural defects, as illustrated by the more general
formula of HAps: Ca10-x-B(PO4)6-x-B(HPO4)x(CO3)A+B(OH)2-x-2A-B, in
which A and B are related to carbonates located on hydroxyl
and phosphate structural sites, respectively.[17] The concentra-
tion of OH groups was thus found to be a more relevant
descriptor than the Ca/P ratio to account for the basic reactivity
of HAps.[14b] The discrimination between bulk and surface
species remains complex and investigations aiming at charac-
terizing the surface at a molecular level are needed to identify
the nature of the surface sites likely involved in acid-base
catalysis. Bulk hydroxyl (OH) and hydrogenophosphate (POH)
species could be discriminated recently from their surface
species by FTIR and NMR using isotopic labelling.[18] The infra-
red fingerprints and relative thermal stabilities of A- and B-type
oxygen atoms in these materials belong to phosphate and
hydroxyl groups, and the surface concentration in Ca2+ is
limited.[7e] In contrast, the potential involvement of surface
terminated acidic POH and basic OH groups has not been
considered yet to our knowledge, even if it was recently
reported that HAps are much more active and selective in n-
butanol compared to other calcium phosphates.[23]
The operando DRIFT studies reported in the present article
aim at investigating the nature and the role of the acidic and
basic surface sites involved in the transformation of ethanol to
n-butanol. For this purpose, hydroxyapatite samples with
various Ca/P ratios are investigated and the influence of the
activation temperature is also addressed for one particular
sample. The modifications occurring on the related hydroxyapa-
tite surfaces after their thermal activation are followed
operando (reaction 1). In order to ascribe the perturbations of
the surface vibrators to the catalytic active sites involved in the
conversion of ethanol into acetaldehyde and n-butanol and/or
to sites involved in deactivation processes, a further treatment
under He flow was carried out at the reaction temperature
before a second reaction step (reaction 2) was performed.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
bulk and surface carbonates have also been reported.[17] From Results and Discussion
the latter IR study, it was found that surface carbonates and
hydrogenocarbonates are formed upon CO2 adsorption, which
reveals the presence of potential Lewis basic sites such as PO4
and OH groups emerging from the channels running along the
c-axis.[17] However, when Brønsted basicity is involved, (ability to
interact with a proton donor molecule such as acetylene),[17] an
acid-base pair is required.[19] In such a case, only the OH groups
were found to be involved in the adsorption processes. This can
be explained by the fact that surface phosphate groups from
phosphate-rich terminations[20] are protonated to POH groups
to ensure the surface charge balance.[21] Such a conclusion is
supported by an earlier study in which Brønsted acidity was
revealed by means of the adsorption of CO, which is a basic
probe molecule.[7e] These surface-terminating POH groups were
found to be much more accessible than Ca2+ Lewis acid sites,
whose limited surface concentration was further confirmed by
complementary XPS and ISS measurements.[7e] Consequently,
the POH groups are likely involved as the acidic partners of the
acid-base pairs interacting with a proton donor molecule, such
as acetylene.[17] Moreover, increasing the relative POH to
calcium content of the top surface by modulation of the
synthesis parameters was found to be beneficial to the basic
reactivity of HAps[7e] as measured by a model reaction such as
the conversion of the 2-methyl-3-butyn-2-ol (MBOH).[22] Such an
increase in the relative POH to calcium content of the top
surface of HAps was also found to promote ethanol conversion
and this was assigned to an appropriate acid-base balance.[7e]
1. Initial Surface States
Absolute DRIFT spectra recorded at the end of the activation
step are shown in Figure 1. A corresponding representation of
the related hydroxyapatite surface is reported in Figure 2 A.
1.1 Carbonation
Figure 1 A shows that the HAp samples exhibit bulk A-type
carbonates (1414, 1444 and 1505 cmÀ 1) and few amounts of B-
type carbonates, (1456 and 1545 cmÀ 1).[17] Due to their low
thermal stability, surface carbonates formed on phosphate
groups and expected at 1485 and 1385 cmÀ 1[17] vanish upon
thermal pretreatment at 623 and 873 K. An additional band is
also observed at 1577 cmÀ 1, which relative intensity decreases
as the Ca/P ratio decreases. This band, which is ascribed to the
νOCO vibrator of surface unidentate calcium carbonates,[24]
vanishes upon activation at 873 K.
1.2 Hydroxylation
1.2.1 Hydroxyls from the Channels
All of the HAp samples exhibit an intense absorption band at
3566 cmÀ 1 (Figure 1 B). This band is attributed to the νOH
vibrator of both bulk hydroxyl groups located in the channels
running along the c-axis and to their top surface fraction.[18a]
Compared to spectra recorded at RT,[18a] the spectra recorded at
623 K reveal an additional νOH contribution at 3534 cmÀ 1. The
intensity of the band at 3566 cmÀ 1 decreases whereas the
3À
Ho et al. suggested that CaÀ O and CaO/PO4 pairs would be
involved in the transformation of ethanol to n-butanol as active
sites for the dehydrogenation and aldol condensation steps,
respectively.[4b] Such
a proposal remains uncertain when
considering the structural properties of HAps, for which no
evident CaÀ O-like species are present due to the facts that
ChemCatChem 2019, 11, 1–15
www.chemcatchem.org
2
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
��
These are not the final page numbers!

Full Papers
intensity of the shoulder at 3534 cmÀ 1 increases upon thermal
activation at 623 K (not shown). As shown in Figure 1 B (green
and red spectra), such a tendency is found to be promoted by
activation at 873 K. The band at 3534 cmÀ 1 results from a
dehydroxylation process in the channels, associated to ther-
mally activated proton migration.[25] This process leads to the
formation of few O2À ions inside the columns. Hence, as
supported by analogy with earlier studies performed on
fluorinated HAps,[26] the band at 3534 cmÀ 1 is assigned to νOH
vibrators H-bonded to the generated O2À species. It indirectly
indicates the presence of O2À species in the hydroxyapatite
channels, mostly as bulk species, but also possibly as surface
species emerging from the channels, as shown below.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
1.2.2 Terminated POH Groups
Besides the bands associated to the OH groups from the
channels, other low intensity νOH bands are detected at higher
wavenumbers (Figure 1 C). The bands at 3678, 3672, 3660 and
3626 cmÀ 1 are associated to terminating protonated phosphate
groups located on the surface, whereas that at 3650 cmÀ 1
originates from defective hydrogenophosphate groups present
in the bulk.[14b,18a] The slight modifications of the shape of the
spectra in this region are significant of different relative
intensities for the various contributions. The origin of such a
multiplicity of surface POH contributions is probably due to the
presence of phosphate rich terminations with slightly different
relaxation processes, and possibly to different protonation level
of these phosphates.
2. Reaction 1 Step
2.1 Gas Phase Analysis
It was shown earlier that catalytic performance measured in the
DRIFT cell reactor was representative of that measured in a
conventional reactor for similar contact times,[7e] even though
lower conversions may be achieved compared to the conven-
tional reactor probably due to differences in reactor geometry.
Ethanol conversion decreases drastically for all of the samples
within the 5 first minutes on stream (Figure 3 A) and then much
more slowly. Among the samples activated at 623 K, the
stoichiometric HAp-S623 and over-stoichiometric HAp-O623
Figure 1. Absolute DRIFT spectra of HAp-S623 K (blue), HAp-U623 K (black), HAp-
O623 K (green) and HAp-O873 K (red) recorded at 623 K at the end of the
activation step in the A) νCO, B) νOH and C) νPO-H regions.
K
K
samples exhibit similar initial ethanol conversion (after 2 min on
stream), whereas the under-stoichiometric HAp-U623 sample
K
appears to be less active. This ranking is maintained after
50 min on stream. Overall, after activation at 623 K, HAp-S623 K
seems to be the most active sample in agreement with earlier
studies.[7a] This ranking (same than that obtained with U-type
conventional reactor) is fully consistent with the ranking
obtained for the basic reactivity evaluated by means of the
MBOH model reaction[22] (SI 1 Figure S1a). The present study
therefore provides an additional support for a correlation
already evidenced between the MBOH and ethanol conversions,
as reported earlier.[7e] In these earlier studies, it was proposed
Figure 2. Schematic representation of the terminated surface of HAps (A)
freshly activated surface with up and down orientation for the protons of
the basic OH groups, protonation of phosphate groups and unidentate
calcium carbonates and (B) modified surface upon poisoning of the POH and
unidentate calcium carbonates groups by carbon polymeric species.
ChemCatChem 2019, 11, 1–15
www.chemcatchem.org
3
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
��
These are not the final page numbers!

Full Papers
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
Figure 3. Reaction 1 step: (A) Evolution of the ethanol conversion upon time on stream for HAp-S623 K (blue), HAp-U623 K (black), HAp-O623 K (green) and HAp-
O873 K (red), (B) evolution of the ethanol conversion and the selectivity in the main detected products upon time on stream for the HAp-S623 K sample.
that similar acid-base pairs were involved in the rate determin-
ing step of the two reactions,[7e] which is likely to be the aldol
condensation in the case of the ethanol condensation reac-
tion.[4b,7b] As already reported for HAps samples[4,6a,7c–e] and for
the present samples tested in a U-type conventional reactor,[7e]
Figure 3 B shows that, beside acetaldehyde, which selectivity
remains quite low (~7%), and very few amount of ethylene, the
main reaction product is found to be n-butanol. Its production
results from a quite complex reaction network including
sequential steps that also lead to the formation of various
products: other C4- compounds such as 3-butene-2-ol, 3-
butene-1-ol, 2-methyl-2-propene-1-ol, 1,3-butadiene and diethyl
ether and C6- compounds (2-ethylbutanol and hexanol).
be dependent on the OHÀ concentration[14b] and that the
increase of the activation temperature favors dehydroxylation
leading to the formation of O2À species (Figure 2 B), these
results raise the question of the role of OHÀ versus O2À in the
catalytic transformation of ethanol to n-butanol. Ho et al. have
only considered O2À species as active sites,[4b] thus neglecting
the eventual involvement of OHÀ species. Note also that in the
case of the cobalt or vanadium hydroxyapatite system, specific
active species derived from OHÀ groups were proposed to favor
the activation of the CÀ H bonds of propane.[27]
2.2 Surface Analysis by DRIFT
As illustrated for HAp-S623 K, n-butanol selectivity rapidly
increases within the 5 first minutes on stream (Figure 3 B) and
then stabilizes around 25%. The lower n-butanol selectivity
compared to the earlier studies performed in a conventional
reactor[7e] (~50–60% depending on the HAps stoichiometry)
may be attributed to the low flow rate that had to be used in
the present study due to the limited amount of sample
introduced in the DRIFT cell. Such operating conditions were
shown to favor secondary reactions.[4a] Here, they even hinder
the desorption of heavy products being responsible of the
decrease of n-butanol and other heavy products evolved in gas
phase, and hence accounting for the gas-phase carbon deficit
observed in the present operando experiments. Due both to
these phenomena and to the different EtOH conversion levels
measured for the different HAps samples (Figure 3A), the
different selectivities in C4- compounds obtained for HAps of
various stoichiometries is rather difficult to comment.
Despite a lower surface area, HAp-O873 K is the most active
sample with the greatest n-butanol yield and the least sensitive
to deactivation (Figure 3 A). Such a positive effect of the high
activation temperature has not been reported to date in the
transformation of ethanol to our knowledge. Similar trends
were also obtained for HAp-S and when characterizing these
samples by means of the MBOH model reaction (SI 1 Fig-
ure S1a). Given that MBOH conversion on HAps was shown to
The difference DRIFT spectra exhibit positive and negative
contributions ascribed to the formation of newly adsorbed
species and to the perturbation and/or the disappearance of
pre-existing bands, respectively, as detailed below for the HAp-
S623 K. The assignments of the related IR bands are summarized
in Table 1.
2.2.1 Newly Adsorbed Species
Figure 4 A shows the rather rapid appearance of several
contributions at 2963, 2930, 2880 and 2862 cmÀ 1 under time on
stream. These contributions are assigned to the vibrations of
CÀ H groups originating from adsorbed reactant and/or prod-
ucts.[28] CÀ O contributions of low intensity are observed at
1390–1470 cmÀ 1 together with an intense multicomponent
bands around 1500–1670 cmÀ 1 gathering two main νCO contri-
butions at 1554 and 1536 cmÀ 1 and
a less intense νCC
contribution at 1596 cmÀ 1 (Figure 4 B). These bands are
assigned to carboxylates[28c] possibly in heavy polymeric com-
pounds.[29] The increasing formation of these adsorbates
contribute to the gas-phase carbon deficit (Figure 3 B) and to
the deactivation process, as shown by the correlation between
the evolution of the area of the 1500–1650 cmÀ 1 contributions
ChemCatChem 2019, 11, 1–15
www.chemcatchem.org
4
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
��
These are not the final page numbers!

Full Papers
Table 1. Assignment of the IR bands related to vibrators originating from hydroxyapatite sites and adsorbates.
1
2
Species
Location
Vibrator Wavenumber Reference
[cmÀ 1
]
3
[17]
4
5
6
7
8
9
Carbonates
bulk type A
νCO
1414
1444
1505
1456
1545
1485
1385
1577
3566
3534
bulk type B
νCO
νCO
3À
surface (in interaction with PO4
)
[24]
unidentate calcium carbonates
νOCO
νO-H
νO-H
[18a]
[26a}
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
Hydroxyls
bulk and surface OH of the columns
OHÀ À O2À
columnar OHÀ in H-bonding interaction with neighboring columnar O2À (bulk
and surface)
[17]
split into 2 surface OH contributions revealed upon interaction of C2H5OH
with the surface
terminal surface POH
νO-H
3568
3560
3678
3672
3660
3626
3650
2963
2930
2880
2862
[14b,18a]
Protonated phosphates
Ethanol and products
νPO-H
[18a]
bulk POH
adsorbed on the surface
νPO-H
νC-H
[28a,c]
[28c]
[28c]
[29]
Carboxylates/heavy polymeric com- adsorbed on the surface
pounds
νsCOO
νasOCO
νC-C
1390–1470
1536, 1554
1596
[a] by analogy with νO-H at 3546 cmÀ 1 in hydroxyapatites modified by fluoride assigned to an OHÀ À FÀ contribution.
and the loss of ethanol conversion upon time on stream
(Figure 5). The strong adsorption of polymeric compounds is
also supported by the yellow-brownish color observed for the
used catalysts. Note that there is no evidence for the formation
of chemisorbed water (band expected ~1640 cmÀ 1). The
absence of adsorbed acetaldehyde (bands expected at 1709–
1711 cmÀ 1),[28a,b] is indicative that acetaldehyde is rapidly
desorbed or transformed.
ethanol and/or its products reveals two perturbed contributions
at 3560 and 3568 cmÀ 1(Figure 4 C). Such a splitting was also
observed for the interaction of acetylene with the HAp surface
[17]
and following the MBOH reaction under operando con-
ditions (SI 1 Figure S1b). Several explanations can account for
such a surface OH differentiation (Figure 3 A). Firstly, the
existence of two different surface terminating hydroxyl groups
with up and down orientations of the protons with respect to
oxygen was recently evidenced in a NMR study.[18b] Such a
speciation, which cannot be seen in the absolute IR spectrum
due to the much higher concentration of bulk OH species
compared to surface OH species,[18a] might be more easily
observable in the difference spectra that are only sensitive to
surface modifications. Such different up and down configura-
tions might lead to slightly different basicity or accessibility and
thus to slightly different reactivity. Secondly, in the context of
basic reactivity, acid-base pairs are involved and one may also
consider the involvement of basic OHÀ species with different
acidic partners (Ca2+ versus POH, and/or different POH species
as suggested by the various surface νPO-H contributions pointed
out in Figure 1 C and reported in Table 1) with different
strengths resulting in the differentiation of OHÀ species (Fig-
ure 2 A). Note that as reported earlier for identical materials, the
accessibility of Ca2+ ions as potential Lewis acid sites was
shown to be quite limited compared to that of POH from CO-
FTIR adsorption measurements at 77 K.[17] In the present study,
this lack of accessibility may be explained by the presence of
stable surface unidentate calcium carbonates. The negative
contributions at 3678, 3672 and 3660 cmÀ 1 attributed to the
perturbation of the surface POH vibrators (Figure 4 C) clearly
2.2.2 Perturbation of Pre-Existing Surface Species
The band at 1577 cmÀ 1 is perturbed from the first minutes on
stream of the reaction 1 step (Figure 4 B). The increase in
intensity of this perturbation with time on stream is in line with
the involvement of an interaction of the related unidentate
calcium carbonate species with adsorbates, rather than with
their sudden vanishing under reactant flow.
The surface contribution of hydroxyl groups emerging from
the channels is also modified (Figure 4 C). The band at
3534 cmÀ 1, which appears after the thermal activation process
(Figure 1 B), is only slightly modified under reaction conditions.
This is consistent with the main assignment of this contribution
to OH groups H-bonded to bulk O2À species. This contribution
also indicates the presence of a limited fraction of the O2À
species on the surface and of their involvement in the catalytic
reaction, as discussed below. Whereas only one contribution at
3566 cmÀ 1 could be identified from the absolute spectra of the
thermally-activated surface (including both bulk and surface
species) (Figure 1 B),[18a] the exposure of the HAp surface to
ChemCatChem 2019, 11, 1–15
www.chemcatchem.org
5
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
��
These are not the final page numbers!

Full Papers
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
Figure 5. Correlation between the loss in ethanol conversion and the
increase of the area of the contribution in the 1500–1650 cmÀ 1 range for the
HAp-S623 K sample.
2.3 Comparison of the Various HAp Samples
Overall, similar tendencies are observed for all of the HAp
samples. This can be seen upon reaction time on stream in the
case of the most active sample (HAp-O873 K) (SI 2 Figure S2).
Consequently, only the spectra recorded at the end of the
reaction 1 step are reported here for all of the samples. The
changes in intensity of the positive CÀ H contributions (Figure 6
A) on the difference spectra can account for the differences in
conversion and deactivation profiles observed for the various
samples (Figure 3 A). The intensity of the negative contributions
at 1577, 3678, 3672 cmÀ 1 are in line with the intensity of these
bands on the absolute spectra recorded before reaction
(Figure 6 B, C versus Figure 1 A, C). The contribution at 1500–
1670 cmÀ 1 (Figure 6 B) may be attributed to the presence of
various types of carbon adsorbates. By comparing the samples
with various stoichiometries activated at 623 K, it is found that
the higher the perturbation of the bands related to OH
emerging from the channels is, the higher the EtOH conversion
is. This behavior is also valid when considering the perturbation
of the POH areas. However, the intensity of the perturbation of
the whole POH area recorded during reaction 1 does not match
the absolute intensities of the POH area recorded after
activation step (see HAp-O623 versus HAp-S623 in Figure 1C).
K
K
Figure 4. Difference DRIFT spectra recorded during reaction 1 for the HAp-
S623 K sample. A) νCH, B) νCO and νCC and C) νOH regions.
Such a different behavior can be attributed to the fact that, as
mentioned in the former section, the contribution at 3678 cmÀ 1,
particularly intense for HAp-S623
is comparatively more
K
perturbed than the POH contributions at lower wavenumbers.
In the case of the HAp-O873 K sample (SI 2, Figure S2), beside the
absence of a negative signal at 1577 cmÀ 1 that confirms the
release of unidentate carbonates at high temperature (red
spectrum in Figure 6 B), the fraction of perturbed OH vibrators
is not only much greater than that for HAp-O623 K (comparison
of green and red spectra in Figure 6 C), but also the most
intense among all of the investigated samples. This confirms
the correlation between the intensity of the perturbation of the
OH contributions and the EtOH conversion. In addition, for
HAp-O873 K, a broadening of the negative signal is also observed
reveal their involvement in the reaction. Note that the intensity
of these perturbations (quite similar for the three components)
does not match that observed on the absolute spectra shown
in Figure 1 C (the intensity of the bands increases in the order
3678<3672!3660 cmÀ 1). Similar features were obtained using
acetylene and CO probes and were ascribed to different POH
acidic strengths, accessibility and/or proximity with the neigh-
boring OH surface basic sites.[17] Interestingly, similar features
are also observed following the MBOH reaction under operando
conditions (SI 1 Figure S1b).
ChemCatChem 2019, 11, 1–15
www.chemcatchem.org
6
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
��
These are not the final page numbers!

Full Papers
3. Evolution under He Flow
1
2
In order to evaluate the reversibility of the surface processes
occurring during the reaction, ethanol was suddenly removed
from the flow at 623 K.
3
4
5
6
7
8
3.1 Desorption of the Products Detected in Gas Phase
9
As shown in the case of the HAp-S623 K sample in Figure 7, the
amounts of acetaldehyde, n-butanol and ethanol detected in
the gas phase progressively decrease. n-butanol is no longer
observed after 35 min under He flow, whereas ethanol and
acetaldehyde are still observed after 55 min on stream. Note
also that few unidentified peaks, probably related to the
desorption of some unidentified heavy products, were also
detected.
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
3.2 State of the HAp Surfaces Monitored by DRIFT
3.2.1 Evolution of the Adsorbates
For all of the HAp samples, the intensity of the contributions
associated to νC-H (2800–3000 cmÀ 1) decreases under helium
flow.
After 2 hours under He flow, only about half of the CÀ H
adsorbates formed during reaction 1 were released (Figure 8 A).
In contrast, the intensity of the multi-components band at
1500–1650 cmÀ 1 increases slightly under helium flow (Figure 8
B). This indicates that the corresponding carboxylates, which
formation started during reaction 1 step, are strongly adsorbed.
Moreover, even in the case of the HAp-O873 sample (SI 2
K
Figure S2) that is the only sample not exhibiting the character-
istic band ascribed to unidentate calcium carbonates,
a
significant increase in the intensity of the 1596–1587 cmÀ 1
signal assigned to νCC vibrators is observed (Figure 9 B), which
indicates an increasing formation of carbon polymeric species
under helium flow.
Figure 6. Difference DRIFT spectra recorded at the end of the reaction 1 step
for the HAp-S623 K (blue), HAp-U623 K (black), HAp-O623 K (green) and HAp-O873 K
samples (red) in the (A) νCH (B) νCO and (C) νOH regions.
with a perturbation of the contribution at 3534 cmÀ 1 (indirectly
related to the formation of O2À species). This is in line with the
relative increase of the 3534 cmÀ 1 contribution compared to
that at 3566 cmÀ 1 observed upon increasing the temperature of
the thermal pretreatment (see green and red spectra in Figure 1
B). This strongly suggests that, in addition to basic surface OHÀ
groups, the corresponding fraction of surface O2À species
(expected to be more basic) are also involved in the catalytic
process. This might also account for the higher reactivity of the
sample activated at the higher temperature for the MBOH
reaction (SI 1 Figure S1a).
Figure 7. Evolution of the EtOH and the products concentrations in gas
phase under helium flow for the HAp-S623 K sample.
ChemCatChem 2019, 11, 1–15
www.chemcatchem.org
7
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
��
These are not the final page numbers!

Full Papers
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
Figure 8. Difference DRIFT spectra recorded upon increasing exposure to He
flow for the HAp-S623 K sample in the (A) νCH (B) νCO and (C) νOH regions.
Figure 9. Difference DRIFT spectra recorded after 2 hours under He flow for
HAp-S623 K (blue), HAp-U623 K (black), HAp-O623 K (green) and HAp-O873 K (red) in
the (A) νCH (B) νCO and (C) νOH regions.
3.2.2 Poisoning and Regeneration of the Surface Sites?
For all of the samples, neither the unidentate calcium
carbonates (when initially present) (Figures 8 B, 9 B) nor the
surface POH groups (Figures 8 C, 9 C) are regenerated under He
flow. Compared to the spectrum recorded at the end of the
reaction 1 step, the POH vibrators are even slightly more
perturbed over time under He flow (Figure 8 C). Such a
perturbation of the POH groups under He flow is indicative of
their strong interaction with the adsorbates formed under the
reaction 1 step. This also shows that all of the POH groups have
not been fully poisoned during the reaction 1 step and
therefore remained available for the reaction.
Regarding the OH region, the perturbation of the OH bands
occurred during reaction 1 decreases upon time under He flow
(Figure 8 C and Figure 9 C), but at different extents depending
ChemCatChem 2019, 11, 1–15
www.chemcatchem.org
8
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
��
These are not the final page numbers!

Full Papers
on the samples. In particular, the HAp-U623 and HAp-O623
K
K
1
2
3
4
5
6
7
8
9
samples do not show any negative contributions around 3570–
3560 cmÀ 1 (flat baseline on the black and green difference
spectra in Figure 9 C). This suggests that the OHÀ groups
previously present on the freshly activated samples have been
fully regenerated under helium flow. Hence, contrary to surface
POH groups and possibly surface unidentate calcium carbo-
nates, the surface OHÀ groups must not directly be involved in
the formation of the carbon polymeric species.
Moreover, the possible full regeneration of the OHÀ groups
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
(Figure 9 C for HAp-U623 and HAp-O623 K) in the absence of
K
ethanol in the feed may indicate their involvement in the
interaction with ethanol. In contrast, only a limited fraction of
the perturbed surface OHÀ groups at 3568 and 3560 cmÀ 1 has
been recovered under helium flow for HAp-S623 K and HAp-O873 K
samples (blue and red spectra in Figure 6 C and Figure 9 C).
This is also the case for the 3534 cmÀ 1 contribution that is
indirectly characteristic of the involvement of the O2À species
Figure 10. Comparison of the initial ethanol conversions between reaction 1
and reaction 2 steps for the HAp samples.
for HAp-O873
.
K
OHÀ (and O2À ) sites were only partially
must be directly involved in the activation of ethanol.
Interestingly, the fraction of ethanol conversion recovered
during the reaction 2 step appears to be directly related to the
fraction of surface OHÀ groups recovered upon the He step.
Indeed, HAp-U623 and HAp-O623 show a full regeneration of
regenerated under He flow in the case of the most reactive
samples toward MBOH (HAp-O873 and HAp-S623 K), but fully
K
regenerated for the less active ones (HAp-O623 K and HAp-U623 K).
The regeneration ability of surface OHÀ sites of these hydrox-
yapatite samples must therefore be correlated to the strength
and/or density of the basic surface OHÀ groups evaluated by
the conversion of MBOH into acetone and acetylene (SI 1).
K
K
the surface OHÀ groups upon the He step (Figure 9 C) together
with a full recovery of the initial ethanol conversion in reaction
2 step. In contrast, HAp-S623 and HAp-O873 show a partial
K
K
regeneration of the OHÀ (and derived O2À ) groups after the He
step (Figure 9 C) and a decrease in the initial ethanol conversion
in reaction 2 step (25% initial ethanol conversion in the reaction
4. Reaction 2 Step
step 2 versus 35% in the reaction step 1 over HAp-S623
,
K
The initial surface state associated with the reaction 2 step (end
of He flow step) is different from that associated with reaction 1
step (end of activation step). As seen above, modifications on
the related surfaces occurred between these two steps (Fig-
ure 2). Unidentate calcium carbonates and POH groups are still
present on the surface at the end of He flow step but no more
accessible since they are now interacting strongly with
adsorbed carbon polymeric species (Figure 2 B). Compared with
the end of the activation step, basic OHÀ sites are present either
Figure 10). Hence, this shows that the conversion of ethanol on
the HAp surface is clearly governed by the density and/or the
strength of the surface basic OHÀ (and derived O2À species)
groups. Such an involvement of the OHÀ sites as active sites of
the conversion of ethanol over HAp is also supported by the
much lower activity and n-butanol selectivity obtained for other
calcium phosphate phases that do not exhibit basic OHÀ
groups, such as beta tricalcium phosphate, fluorapatite[23] and
octacalcium phosphate[30] that also exhibit poor basicity
compared to HAps.[23,30]
in similar amounts (for HAp-U623 and HAp-O623 K) or in lower
K
amounts (for HAp-S623 and HAp-O873 K). Let us examine how
K
such modifications influence the surface reactivity.
4.1.2 Impact on the Selectivity
4.1 Gas Phase Analysis
Regardless of whether the ethanol conversion has been fully
recovered or not, the products selectivities in reaction 2 step
were found to be greatly modified compared to reaction 1 step.
For all of the samples, as illustrated for HAp-S623 K and HAp-U623 K
in Figure 11 A and B, the production of n-butanol and
acetaldehyde in the reaction 2 step are greatly modified.
From the beginning of reaction 2 step, n-butanol is hardly
produced, while the production of acetaldehyde is drastically
increased compared to reaction 1 step. Such a sharp increase in
acetaldehyde production during reaction 2 step is the most
4.1.1 Ability to Recover the EtOH Conversion
As reported in Figure 10, the ability of the samples to recover
an initial conversion of ethanol during reaction
2 step
comparable to that measured during the reaction 1 step is
different from one sample to another. Ethanol conversion is
fully recovered for HAp-U623
and HAp-O623 K, whereas a
K
significant loss of conversion is observed for HAp-S623 and
K
HAp-O873 K. Given that both POH groups and unidentate
carbonates (when present for samples activated at 623 K) are
poisoned for all of the samples, none of these surface sites
pronounced for HAp-U623
by about a two-fold increase
K
compared to other samples (Figure 11 B). In parallel, hydrogen
concentration remained quite limited during reaction 1 step, as
ChemCatChem 2019, 11, 1–15
www.chemcatchem.org
9
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
��
These are not the final page numbers!

Full Papers
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
Figure 11. Evolution of the concentrations in ethanol, n-butanol and
acetaldehyde evolved during the successive steps for A) HAp-S623 K and B)
HAp-U623 K samples.
expected from its production upon the dehydrogenation step
of ethanol to acetaldehyde, whereas a sharp increase in its
concentration was observed during reaction 2 step (not shown).
4.2. Nature of the Acidic Sites Involved in the Key Reactions
Steps
Such changes in the selectivity profiles between reaction steps
1 and 2 indicate that different active sites must be involved in
the relevant key reaction steps, i.e. dehydrogenation for
acetaldehyde formation and subsequent aldol condensation for
n-butanol formation. Such a conclusion is in accordance with
that already drawn by Ho et al..[4b]
On HAp-S623 K, the intensity of νCH contributions (Figure 12
A) increases during the reaction 2 step and remains of lower
intensity compared to those achieved at the end of the reaction
1 step. This lower concentration of adsorbed products is
consistent with the lower ethanol conversion recorded during
reaction 2 step compared to reaction 1 step (Figures 10 and 11
A). In contrast, contributions associated to the presence of
carbon polymeric species (1500–1650 cmÀ 1) (Figure 12 B)
increase continuously. The perturbed bands related to POH
sites are not modified upon the introduction of ethanol during
the reaction 2 step (Figure 12 C). This confirms that the POH
groups have already been fully poisoned during the He step, as
mentioned in the previous section. Hence, in the absence of
Figure 12. Difference DRIFT spectra recorded for HAp-S623 K sample from the
beginning of reaction 2 step in the (A) νCH (B) νCO and (C) νOH regions.
regeneration of the surface POH acid sites (not recovered under
He flow), it is likely that the few Ca2+ cations accessible on the
surface still contribute to the reaction together with the
regenerated surface basic OHÀ groups, as illustrated by the
perturbation observed at ~3568 cmÀ 1 in Figure 12 C, finally
promoting the formation of acetaldehyde. Whereas two neg-
ative contributions of similar intensity were observed at 3568
and 3560 cmÀ 1 at the end of reaction 1 step (Figure 4 C),
Figure 12 C shows that the band at 3568 cmÀ 1 is much more
ChemCatChem 2019, 11, 1–15
www.chemcatchem.org
10
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
��
These are not the final page numbers!

Full Papers
intense than that at 3560 cmÀ 1 in reaction 2 step. Even if DFT
calculations would be required to assess the origin of these two
surface OHÀ contributions, this result may be an indication that
the 3568 cmÀ 1 contribution is characteristic of surface basic OHÀ
groups acting cooperatively with Ca2+ ions as an acidic partner
(whereas that at 3560 cmÀ 1 may correspond to OHÀ groups
acting together with POH groups).
1
2
3
4
5
6
7
8
9
Such a trend is also observed for the other samples
(Figure 13) (SI 2), except for HAp-U623 (see black spectra in
K
Figure 13 B and C). For this latter sample, as the reaction 2 step
starts, most carbon polymeric species (bands in the 1500–
1670 cmÀ 1 region) suddenly vanished unexpectedly, may be
due to an uncontrolled exothermic process. Simultaneously, the
underlying surface POH groups were also suddenly recovered
(flat baseline in the corresponding area of the black spectrum in
Figure 13 C) and the unidentate calcium carbonate contribution
(band at 1577 cmÀ 1) disappeared, as shown on the absolute
DRIFT spectrum reported in Figure 14 and the absence of the
corresponding negative contribution in Figure 14 B. More
importantly, despite the full regeneration of the POH and OHÀ
groups for HAp-U623 K, the product selectivities related to
reaction 2 step once more differ from those obtained during
reaction 1 step, with a significant increase in the acetaldehyde
production (Figure 11 B). The IR contribution related to surface
basic OHÀ sites is once again perturbed, with a single negative
contribution at 3568 cmÀ 1 (Figure 13 C) and the surface POH
species remain as spectators species during the reaction 2 step,
as illustrated by the flat baseline observed in the related regions
in Figure 13 C. Such an increase in acetaldehyde production in
reaction 2 step over HAp-U623 K (Figure 11 B) can be explained
by the prevailing involvement of Ca2+ cations whose accessible
amount was strongly increased due to the release of the
corresponding unidentate carbonates. Hence, Ca2+ cations
acting together with OHÀ groups (either because POH groups
have been previously poisoned or because their relative
amount is increased) favor the formation of acetaldehyde. Such
ability of the Ca2+À OHÀ pair to favor the formation of
acetaldehyde is supported by the previous work of Tsuchida
et al.[3a] in which it was shown that the prevailing formation of
acetaldehyde from ethanol was favored on CaO (that also
exposes Ca2+À OHÀ pairs under the related operating conditions
[31]). In contrast, n-butanol can be formed as long as a fraction of
POH groups keeps on working (reaction 1 step).
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
How can it be considered that different acidic partners, Ca2+
and POH species would be involved in the dehydrogenation
and aldol condensation key reaction steps? Firstly, the strength
of the acidic site partner of the deprotonating basic site plays a
key role in basic reactions for the stabilization of the anionic
intermediate (here, Ca2+ is a stronger acidic site than POH sites).
More specifically in the case of the aldol condensation, it was
underlined that a peculiar acid-base balance is required.[32]
Secondly, besides the relative strength, the respective Lewis
and Brønsted nature of the two acidic sites needs to be taken
into consideration. On the one hand, in the case of dehydro-
genation of primary or secondary alcohol (such as isopropanol),
both a deprotonating basic site and a Lewis acid site are
required to form the alcoholate intermediate and to abstract
Figure 13. Difference DRIFT spectra recorded for HAp-S623 K (blue), HAp-O623 K
(green), HAp-O873 K (red) and HAp-U623 K (black) samples from the beginning
of reaction 2 step in the (A) νCH (B) νCO and (C) νOH regions.
the hydride ion, respectively.[22b] On the other hand, Brønsted
acid sites were proposed to be involved in the formation of
crotonaldehyde from acetaldehyde.[33]
Compared to other basic catalytic systems, hydroxyapatite
is the only catalyst to exhibit both Lewis (Ca2+) and Brønsted
(POH) acid sites likely to act as acidic partners of the acid-base
ChemCatChem 2019, 11, 1–15
www.chemcatchem.org
11
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
��
These are not the final page numbers!

Full Papers
conversion during a subsequent second reaction step. This
clearly shows that the availability of the basic OHÀ groups
governs the activation of ethanol. The promoting effect of the
high activation temperature (873 K) of HAp on the ethanol
conversion measured at 623 K is ascribed to the presence of
few strong O2À basic sites derived from the dehydroxylation of
columnar OHÀ groups that are involved in the activation of
ethanol. The existence of two surface OH contributions may
refer to the involvement of two types of surface OHÀ groups,
with up and down orientations of their protons, that would
react differently with their neighboring acidic partners, namely
Ca2+ and POH species. The nature of the acidic sites acting
cooperatively with these OHÀ groups within acid-base pairs
required for any basic reaction steps is determinant to control
the selectivity in the formed products: 1) n-butanol is formed
only as long as POH sites are still available, 2) once the POH
sites have been poisoned (upon intermediate treatment under
He) and a higher number of Ca2+ species is accessible (resulting
from the release of unidentate carbonates), only Ca2+ cations
are involved as an acidic partner of the basic OHÀ groups.
Hence, the acid-base pair involved in the transformation of
ethanol switches to Ca2+À OHÀ during reaction 2 step, and the
reaction leads mainly to the formation of acetaldehyde. This not
only confirms that acetaldehyde acts as an intermediate in the
formation of n-butanol, but also indicates that various active
sites are involved in the relevant key reaction steps: Ca2+À OHÀ
and POHÀ OHÀ as acid-base pairs in the dehydrogenation of
ethanol to acetaldehyde and the aldol condensation for n-
butanol formation, respectively. The simultaneous availability of
two weak acid-base pairs on a single material, also including
Lewis acid sites favoring hydride abstraction to dehydrogenate
ethanol and Brønsted acid sites favoring the aldol condensation
reaction, makes hydroxyapatite a unique catalytic system able
to transform selectively ethanol to n-butanol without the need
of any metallic function.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
Figure 14. Absolute spectra recorded for HAp-U623 K sample at the end of the
activation step and at the beginning of the reaction 2 step in the νCO region.
pairs involved in the two first steps of n-butanol formation,
namely the dehydrogenation of ethanol and the aldol con-
densation steps. In combination with the basic sites, mainly
OHÀ groups under the present operating conditions, but also
the few derived O2À species present when high thermal
treatment is performed, these acidic sites provide two acid-base
pairs able to form acetaldehyde and to further transform it via
hydride aldol condensation. Finally, another key reaction step
to produce n-butanol from ethanol is the hydrogenation of
crotonaldehyde into n-butanol.[7a] It remains unclear at present
whether the recombination of the species (stored on Ca2+ ions)
with protons (particularly available for this proton conductor
material[25b]) may explain the peculiar ability of hydroxyapatite
to achieve by itself this last hydrogenation step, whereas other
traditional basic systems always require the addition of a
metallic function to selectively produce n-butanol. If the
involvement of MPV-like mechanism is proposed,[10] the role of
the proton mobility on the catalytic behavior of hydroxyapatite
will also be discussed in a forthcoming paper.
Experimental Section
Sample Preparation and Characterization
Conclusions
Hydroxyapatite samples were prepared according to the procedure
described previously:[30] the co-precipitation of Ca(NO3)2 and
(NH4)2HPO4 solutions beforehand adjusted at pH=10 was per-
formed at 353 K under a N2 flow to limit carbonation of the
materials. As detailed elsewhere,[14b] depending on the pH control
procedures, even with a nominal Ca/P ratio of 1.67 of the
precursors in the solutions, hydroxyapatite samples can be
obtained with different Ca/P ratios. The three HAp samples studied
in this work have been obtained by periodic, continuous or without
addition of NH4OH during the precipitation step, respectively. The
obtained precipitates were then matured at the same temperature
for 4 hours under reflux. The washed precipitates were then dried
overnight at 373 K and thermally treated under Ar flow
(150 mLminÀ 1) up to 623 K (5 KminÀ 1) for 90 min. In one case, a
thermal treatment was also carried out at 873 K. The Ca/P ratios
measured by ICP-AES analysis (“Service Central d’Analyse” of the
CNRS, Solaize, France) were 1.67, 1.77 and 1.64, for the samples
prepared with periodic, continuous or without addition of NH4OH,
Operando DRIFT study was carried out to follow the catalytic
transformation of ethanol into acetaldehyde and n-butanol, and
to identify the nature and the role of the active sites. Apart
from the formation of positive IR contributions in difference
spectra attributed to the presence of strongly-bound carbon
polymeric species, responsible for the observed deactivation
process, peculiar attention was devoted to the changes in the
negative contributions related to the perturbation of surface
vibrators. The ability of such sites to be recovered under an He
atmosphere and their consequence on a second reaction step
were investigated. Among the surface sites, unidentate calcium
carbonates, terminated POH and OH groups emerging from the
columns, only the basic OH groups could be partially or fully
(for the most basic ones) regenerated under the He step,
resulting in the partial or full recovery of the initial ethanol
respectively. The samples will be hereafter referred to as HAp-S623 K
,
ChemCatChem 2019, 11, 1–15
www.chemcatchem.org
12
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
��
These are not the final page numbers!

Full Papers
HAp-O623 K, HAp-O873 K and HAp-U623 K, respectively, where S, O and
U stand for stoichiometric, over-stoichiometric and under-stoichio-
metric Ca/P ratios, respectively, and 623 and 873 K refer to as the
temperatures of final thermal treatment applied for the various
samples (the same temperature is further applied for activation
step performed prior to catalytic measurements).
light and heavy products, respectively. EtOH conversion and
selectivities were calculated as follows: convEtOH (%)=
1
2
3
4
5
°
°
°
(P EtOHÀ PEtOH)×100/P EtOH, Seli (%)=αi Pi ×100/(P EtOHÀ PEtOH), where
°
P
is the reference EtOH pressure measured before reaction,
EtOH
PEtOH and Pi are partial pressures of EtOH and i product during the
reaction and αi the 1= of the number of carbon in the i product.
2
6
7
8
9
X-ray powder diffractograms were recorded with a D8 Advance
diffractometer equipped with a Copper anode generator (λ=
Diffuse reflectance infrared spectra were recorded in the 4000–
1200 cmÀ 1 range (4 cmÀ 1 resolution, 256 scans/spectrum, MCT
detector) using a Brüker IFS 66 V spectrometer every 150 s. A
reference spectrum was recorded with KBr (Fluka, purity >99.5%)
under the same operating conditions. The absolute spectra
recorded at the end of the activation step (623 or 873 K) are
reported in log 1/R with R=IHAp/IKBr. The difference DRIFT spectra
reported for the successive steps of the sequence are reported in
log 1/R’ with the relative reflectance R’=IHAp sequence/IHAp (spectrum
recorded during the successive steps of the sequence/spectrum
recorded at 623 or 873 K at the end of the activation step).[35] The
gas phase contribution of ethanol was systematically subtracted for
spectra related to the two reaction steps. No significant modifica-
tion of the shape nor intensity of the PÀ O combination bands
(2200–1950 cmÀ 1) were observed during the successive investi-
gated steps performed at 623 K, which indicates that the optical
properties of the samples are not modified during the reaction
processes. For comparison purpose, when spectra related to differ-
ent samples are reported on a same Figure, normalization using the
PÀ O combination bands has been systematically applied.
°
1.5418 Å). Diffractograms were recorded by 0.02 steps in the 2θ
°
range of 10–85 . The crystalline hydroxyapatite structure of the
samples was identified from comparison with the ICDD pattern 01-
074-9780(A). Specific surface areas were measured by adsorption of
N2 at 77 K on a Micromeritics (ASAP 2010) apparatus and calculated
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
from the BET method. After thermal treatment at 623 K, HAp-S623 K
,
HAp-O623 and HAp-U623 exhibit specific surface areas of 41, 42
K
K
and 31 m2 gÀ 1, respectively. Thermal treatment of sample HAp-O at
873 K (HAp-O873 K) leads to a decrease in the specific surface area
down to 33 m2 gÀ 1. The in situ characterization of the acidity of
these materials was detailed elsewhere[7e] from the adsorption of
CO at 77 K monitored by infra-red. It was shown that the
accessibility of Ca2+ ions exposed at the top surface of HAp-S623 K
,
HAp-U623 K, HAp-O623
and HAp-O873
remained quite limited
K
K
compared to that of terminal POH groups.
Operando DRIFT Experiments
The HAp samples (40 mg) were placed inside a heated crucible
located in a Thermo Spectra-Tech high-temperature cell equipped
with ZnSe windows and with appropriate gas inlet and outlet
connections as to pass the gas flow through the catalytic bed and
to ensure homogenous contact between the sample and the
flowing gas.[32a,34] Successive steps were considered, as summarized
in Figure 15. After activation, a first reaction step (reaction 1) was
carried out for one hour at 623 K. The reactant feed was then
switched to He flow for 2 hours, maintaining the temperature at
623 K. Finally, a second reaction step (reaction 2) was carried out at
the same temperature.
Acknowledgements
The authors are grateful to Vincent Lhosinho from the Laboratoire
de Réactivité de Surface, Sorbonne Université, for his help in the
maintenance of the EtOH and MBOH reaction set-up.
Conflict of Interest
The activation of the samples was carried out under helium flow
(20 mLminÀ 1), heating the sample (5 KminÀ 1) up to 623 or 873 K, and
maintaining this temperature for 90 min, before switching from the
inert flow to the reactant feed. The reactant feed consisted of ethanol
diluted in He. Helium (20 mLminÀ 1) was bubbled in ethanol (Aldrich,
99.9%) at 278 K to achieve a contact time of 37 gcat.h/molethanol. A flow of
20 mLminÀ 1 was maintained during the intermediate step under He.
The authors declare no conflict of interest.
Keywords: POH/OHÀ and Ca2+/OHÀ acid base pairs · DRIFT ·
hydroxyapatites · operando
The gas phase composition at the exit of the DRIFT cell was
analyzed every 5 min using a two channels micro-gas chromato-
graph (Varian, CP4900) equipped with PPQ and 5CB columns
allowing the separation and the catharometric quantification of
[1] a) G. R. M. Dowson, M. F. Haddow, J. Lee, R. L. Wingad, D. F. Wass,
Angew. Chem. Int. Ed. 2013, 52, 9005–9008; Angew. Chem. 2013, 125,
9175–9178; b) J. Sun, Y. Wang, ACS Catal. 2014, 4, 1078–1090.
[2] D. L. Carvalho, R. R. d. Avillez, M. T. Rodrigues, L. E. P. Borges, L. G.
Appel, Appl. Catal. A 2012, 415–416, 96–100.
[3] a) T. Tsuchida, T. Yoshioka, S. Sakuma, T. Takeguchi, W. Ueda, Ind. Eng.
Chem. Res. 2008, 47, 1443–1452; b) N. Savage, Nature 2011, 474, S9–S11.
[4] a) S. Hanspal, Z. D. Young, H. Shou, R. J. Davis, ACS Catal. 2015, 5, 1737–
1746; b) C. R. Ho, S. Shylesh, A. T. Bell, ACS Catal. 2016, 6, 939–948.
[5] K. Koda, T. Matsu-ura, Y. Obora, Y. Ishii, Chem. Lett. 2009, 38, 838–839.
[6] a) T. Tsuchida, S. Sakuma, T. Takeguchi, W. Ueda, Ind. Eng. Chem. Res.
2006, 45, 8634–8642; b) J. T. Kozlowski, R. J. Davis, ACS Catal. 2013, 3,
1588–1600; c) D. Gabriëls, W. Y. Hernández, B. Sels, P. Van Der Voort, A.
Verberckmoes, Catal. Sci. Technol. 2015, 5, 3876–3902.
[7] a) T. Tsuchida, J. Kubo, T. Yoshioka, S. Sakuma, T. Takeguchi, W. Ueda, J.
Catal. 2008, 259, 183–189; b) S. Ogo, A. Onda, Y. K. , Appl. Catal. A 2011,
402, 188–195; c) J. Scalbert, F. Thibault-Starzyk, R. Jacquot, D. Morvan, F.
Meunier, J. Catal. 2014, 311, 28–32; d) L. Silvester, J. F. Lamonier, Faye,
J. , M. Capron, R. N. Vannier, C. Lamonier, J. L. Dubois, J. L. Couturier, C.
Calais, F. Dumeignil, Catal. Sci. Technol. 2015, 5, 2994–3006; e) M. Ben
Osman, S. Diallo Garcia, J. M. Krafft, C. Methivier, J. Blanchard, T.
Figure 15. Sequence of successive steps implemented under operando
conditions (the dashed line refers to as the pretreatment achieved in the
case of HAp-O873 K).
ChemCatChem 2019, 11, 1–15
www.chemcatchem.org
13
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
��
These are not the final page numbers!

Full Papers
Yoshioka, J. Kubo, G. Costentin, Phys. Chem. Chem. Phys. 2016, 18, 2783–
27847.
[8] M. Guerbet, C. R. Chim. 1909, 149, 129.
[9] a) S. Veibel, J. I. Nielsen, Tetrahedron 1967, 213, 1723–1733; b) A. J.
O’Lenick, J. Surfactants Deterg. 2001, 4, 311–315; c) C. Carlini, M.
Di Girolamo, A. Macinai, M. Marchionna, M. Noviello, A. M. Raspolli
Galletti, G. Sbrana, J. Mol. Catal. 2003, 200, 137–146.
[10] Z. D. Young, R. J. Davis, Catal. Sci. Technol. 2018, 8, 1722–1729.
[11] M. J. L. Gines, E. Iglesia, J. Catal. 1998, 176, 155–172.
[12] a) S. Sugiyama, Y. Iguchi, T. Minami, H. Hayashi, J. B. Moffat, Catal. Lett.
1997, 46, 279–285; b) S. Sugiyama, E. Nitta, H. Hayashi, J. B. Moffat,
Catal. Lett. 1999, 59, 67–72; c) S. Sugiyama, T. Osaka, Y. Hirata, K. I.
Sotowa, Appl. Catal. A 2006, 312, 52–58.
[21] a) R. M. Wilson, J. C. Elliot, S. E. P. Dowker, J. Solid State Chem. 2003, 174,
132–140; b) M. Jarlbring, D. E. Sandström, O. N. Antzutkin, W. Forsling,
Langmuir 2006, 22, 4787–4792.
[22] a) H. Lauron-Pernot, F. Luck, J. M. Popa, Appl. Catal. A. Gen 1991, 78,
213–225; b) H. Lauron-Pernot, Catal. Rev. 2006, 48, 315–361.
[23] S. Hanspal, D. Young, J. T. Prillaman, R. J. Davis, J. Catal. 2017, 352, 182–
190.
[24] R. Philipp, K. Fujimoto, J. Phys. Chem. 1992, 96, 9035–9038.
[25] a) C. Rey, J. C. Trombe, G. Montel, J. Inorg. Nucl. Chem. 1978, 40, 27–30;
b) S. Nakamura, H. Takeda, K. Yamashita, J. Appl. Phys. 2001, 89, 5386;
c) M. Yashima, Y. Yonehara, H. Fujimori, J. Phys. Chem. C 2011, 115,
25077–25087; d) N. Horiuchi, M. Nakamura, A. Nagai, K. Katayama, K.
Yamashita, J. Appl. Phys. 2012, 112, 074901–074906.
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
[26] a) N. W. Cant, J. A. S. Bett, G. R. Wilson, W. K. Hall, Spectrochim. Acta
1971, 27A, 425–439; b) F. Freund, R. M. Knobel, J. C. S. Dalton 1977,
1136–1140.
[27] a) S. Sugiyama, H. Hayashi, Int. J. Mod. Phys. B 2003, 17, 1476–1481; b) S.
Sugiyama, T. Shono, D. Makino, T. Moriga, H. Hayashi, J. Catal. 2003,
214, 8–14.
[28] a) T. W. Birky, J. T. Kozlowski, R. J. Davis, J. Catal. 2013, 298, 130–137;
b) H. Idriss, Platinum Met. Rev. 2004, 48, 105–115; c) R. M. B. Faria, D. V.
Cesar, V. M. M. Salim, Catal. Today 2008, 133–135, 168–173.
[29] J. Datka, Z. Sarbak, R. P. Eischens, J. Catal. 1994, 145, 544–550.
[30] S. Diallo-Garcia, D. Laurencin, J. M. Krafft, S. Casale, M. E. Smith, H.
Lauron-Pernot, G. Costentin, J. Phys. Chem. C 2011, 115, 24317–24327.
[31] H. Petitjean, J. M. Krafft, M. Che, H. Lauron Pernot, G. Costentin, Phys
Chem Phys 2010, 12, 14740–14748
[32] a) J. F. Groust, G. Costentin, J. M. Krafft, P. Massiani, Phys. Chem. Chem.
Phys. 2010, 12, 937–946; b) Y. C. Chang, A. N. Ko, Appl. Catal. A 2000,
190, 149–155; c) V. V. Ordomsky, V. L. Sushkevich, I. I. Ivanova, J. Mol.
Catal. A 2010, 333, 85–93.
[13] a) S. J. Joris, C. H. Amberg, J. Phys. Chem. 1971, 75, 3167–3171; b) H.
Monma, J. Catal. 1982, 75, 200–203; c) M. Zahouily, Y. Abrouki, B.
Bahlaouan, A. Rayadh, S. Sebti, Catal. Commun. 2003, 4, 521–524; d) S.
Sebti, A. Solhy, R. Tahir, A. Smahi, Appl. Catal. A 2002, 235, 273–281;
e) A. Solhy, J. H. Clark, R. Tahir, S. Sebti, M. Larzek, Green Chem. 2006, 8,
871–874.
[14] a) T. Tsuchida, J. Kubo, T. Yoshioka, S. Sakuma, T. Takeguchi, W. Ueda, J.
Jap. Petrol. Inst. 2009, 52; b) M. Ben Osman, J. M. Krafft, Y. Millot, F.
Averseng, T. Yoshioka, J. Kubo, G. Costentin, Eur. J. Inorg. Chem. 2016,
2016, 2709–2720.
[15] a) S. Ogo, A. Onda, K. Yanagisawa, Appl. Catal. A 2008, 348, 129–134;
b) L. Silvester, J. F. Lamonier, R. N. Vannier, C. Lamonier, M. Capron, A. S.
Mamede, F. Pourpoint, A. Gervasini, F. Dumeignil, J. Mater. Chem. A
2014, 2, 11073–11090; c) B. Yan, L. Z. Tao, Y. Liang, B. Q. Xu, ACS Catal.
2014, 4, 1931–1943; d) V. C. Ghantani, S. T. Lomate, M. K. Dongare, S. B.
Umbarkar, Green Chem. 2013, 15, 1211–1217.
[16] a) C. L. Kibby, K. W. Hall, J. Catal. 1973, 31, 65–73; b) A. Yasukawa, S.
Ouchi, K. Kandori, T. Ishikawa, J. Mater. Chem. 1996, 6, 1401–1405; c) N.
Cheikhi, M. Kacimi, M. Rouimi, M. Ziyad, L. F. Liotta, G. Pantaleo, G.
Deganello, J. Catal. 2005, 232, 257–267.
[17] S. Diallo-Garcia, M. Ben Osman, J. M. Krafft, S. Casale, C. Thomas, J.
Kubo, G. Costentin, J. Phys. Chem. C 2014, 118, 12744–12757.
[18] a) S. Diallo-Garcia, M. Ben Osman, J. M. Krafft, S. Boujday, G. Costentin,
Catal. Today 2014, 226, 81–88; b) M. Ben Osman, S. Diallo-Garcia, V.
Herledan, T. Yoshioka, K. Kubo, Y. Millot, G. Costentin, J. Phys. Chem. C
2015, 119, 23008–23020.
[19] C. Chizallet, M. L. Bailly, G. Costentin, H. Lauron-Pernot, J. M. Krafft, P.
Bazin, J. Saussey, M. Che, Catal. Today 2006, 116, 196–205.
[20] C. A. Ospina, J. Terra, A. J. Ramirez, M. Farinac, D. E. Ellis, A. M. Rossi,
Colloids Surf B: Biointerfaces 2012, 89, 15–22.
[33] A. I. Biaglow, J. Sepa, R. J. Gorte, D. White, J. Catal. 1995, 151, 373–384.
[34] C. Drouilly, J. M. Krafft, F. Averseng, H. Lauron-Pernot, D. Bazer-Bachi, C.
Chizallet, V. Lecocq, Catal. Today 2013, 205, 67–75.
[35] J. Sirita, S. Phanichphant, F. C. Meunier, Anal. Chem. 2007, 79, 3912–
3918.
Manuscript received: November 16, 2018
Revised manuscript received: January 14, 2019
Accepted manuscript online: January 15, 2019
Version of record online: ■■■, ■■■■
ChemCatChem 2019, 11, 1–15
www.chemcatchem.org
14
© 2019 Wiley-VCH Verlag GmbH & Co. KGaA, Weinheim
��
These are not the final page numbers!

Full Papers
FULL PAPERS
1
2
3
4
5
M. B. Osman, J.-M. Krafft, C. Thomas,
T. Yoshioka, J. Kubo, G. Costentin*
6
7
1 – 15
Importance of the Nature of the
Active Acid/Base Pairs of Hydrox-
yapatite Involved in the Catalytic
Transformation of Ethanol to n-
Butanol Revealed by Operando
DRIFTS
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
Drifting away. Atypical active sites
were revealed on hydroxyapatite
catalysts: Ca2+À OHÀ and POHÀ OHÀ
are proposed to act as acid-base pairs
in the dehydrogenation of ethanol to
acetaldehyde and the aldol condensa-
tion for n-butanol formation, respec-
tively.