Insights into Ethanol Coupling over Hydroxyapatite using Modulation Excitation Operando Infrared Spectroscopy

Article abstract of DOI:10.1002/cctc.202000331

The coupling of biomass-derived ethanol to n-butanol is a topic of contemporary interest. Indeed, n-butanol can not only be used as a higher energy-density fuel additive, it is also a key component in perfumes and serves as a solvent for paints and dyes. Hydroxyapatite (HAP) emerged in the literature as a promising catalysts for this transformation, with n-butanol selectivity reaching ～75 percent at 10 percent ethanol conversion. However, the molecular-level mechanism for this reaction is still unclear and several mechanistic questions remain unanswered. Here, we use diffuse reflectance infrared Fourier Transform spectroscopy, coupled with mass spectrometry following a modulation excitation approach (ME-DRIFTS-MS) that enables us to better understand the dynamic processes involved. Our approach allows for a vibrational characterization of the active surface species and the formulation of a consistent mechanism. Based on our experimental observations, Ca²⁺/OH^? can be put forward as the main active site for the aldol condensation. POH/OH^? acid-base pair is proposed as the active site for the Meerwein-Ponndorf-Verley (MPV) direct hydrogen transfer of the aldol condensation product, crotonaldehyde.

Full text of DOI:10.1002/cctc.202000331

The European Society Journal for Catalysis
Supported by
Accepted Article
Title: Insights into Ethanol Coupling over Hydroxyapatite using
Modulation Excitation Infrared Spectroscopy
Authors: Shao-Chun Wang, Melissa Cendejas, and Ive Hermans
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: ChemCatChem 10.1002/cctc.202000331
Link to VoR: https://doi.org/10.1002/cctc.202000331
01/2020

10.1002/cctc.202000331
ChemCatChem
FULL PAPER
Insights into Ethanol Coupling over Hydroxyapatite using
Modulation Excitation Operando Infrared Spectroscopy
Shao-Chun Wang,[a] Melissa C. Cendejas,[a] Ive Hermans* [a] [b]
[a]
S.-C. Wang, M. C. Cendejas, Prof. Dr. I. Hermans
Department of Chemistry
University of Wisconsin-Madison
1101 University Avenue, Madison, WI, 53706 (USA)
E-mail: hermans@chem.wisc.edu
[b]
*
Prof. Dr. I. Hermans
Department of Chemical and Biological Engineering
University of Wisconsin-Madison
1415 Engineering Drive, Madison, WI, 53706 (USA)
hermans@chem.wisc.edu
Supporting information for this article is given via a link at the end of the document
Abstract: The coupling of biomass-derived ethanol to n-butanol is
a topic of contemporary interest. Indeed, n-butanol can not only be
used as a higher energy-density fuel additive, it is also a key
component in perfumes and serves as a solvent for paints and
Guerbet reaction offers one attractive alternative route for
producing n-butanol from ethanol through C-C bond formation
and hydrogenation.[4]
Several heterogeneous catalysts have been investigated for
the Guerbet reaction including metal oxides, zeolites,
hydroxyapatite, and supported metals.[5] Those studies indicate
that hydroxyapatite (Ca10(PO4)6(OH)2, HAP) shows promising
conversion (10%) and selectivity toward n-butanol (75%) under
atmospheric condition because of its favorable acid-base
properties.[6] The widely accepted description of the reaction
involves four main steps as depicted in Scheme 1.[4b, 5a, 6b, 7]
Ethanol first dehydrogenates/oxidizes into acetaldehyde,
followed by acetaldehyde self-aldol condensation. Under the
reaction conditions, the aldol addition product dehydrates very
rapidly to form crotonaldehyde. Crotonaldehyde is then
transformed to crotyl alcohol in a first hydrogenation. Finally,
crotyl alcohol is converted to n-butanol through a second
hydrogenation. This reaction pathway has been established
using conventional kinetics and steady-state isotopic transient
kinetic analysis (SSITKA). The latter revealed that
acetaldehyde, obtained from ethanol dehydrogenation, remains
on the surface to undergo consecutive reactions to generate n-
butanol.[8] However, there is still a debate on the second
hydrogenation step from crotyl alcohol to n-butanol. Two
different hydrogenation pathways have been suggested in the
literature and are summarized in Scheme 1. The first pathway
is a surface-mediated hydrogenation of the C=C bond of crotyl
alcohol, where the surface hydrogen comes from ethanol
dehydrogenation in the first step.[9] In an alternative mechanism,
crotyl alcohol first isomerizes and tautomerizes to butanal. The
C=O can then be directly hydrogenated by ethanol via the
Meerwein-Ponndorf-Verley (MPV) reaction to form n-butanol.[10]
We propose to take another approach to investigate ethanol
coupling over HAP to clarify the second hydrogenation step
and gain a deeper understanding of the reaction mechanism in
general that could potentially benefit future catalyst design.
Ho et al. reported Ca-O species to be responsible for ethanol
dehydrogenation and direct hydrogen transfer, while the self-
aldol condensation of acetaldehyde is proposed to occur on
dyes. Hydroxyapatite (HAP) emerged in the literature as
a
promising catalysts for this transformation, with n-butanol selectivity
reaching ~75% at 10% ethanol conversion. However, the
molecular-level mechanism for this reaction is still unclear and
several mechanistic questions remain unanswered. Here, we use
diffuse reflectance infrared Fourier Transform spectroscopy,
coupled with mass spectrometry following a modulation excitation
approach (ME-DRIFTS-MS) that enables us to better understand
the dynamic processes involved. Our approach allows for a
vibrational characterization of the active surface species and the
formulation of a consistent mechanism. Based on our experimental
observations, Ca2+/OH- can be put forward as the main active site
for the aldol condensation. POH/OH- acid-base pair is proposed as
the active site for the Meerwein-Ponndorf-Verley (MPV) direct
hydrogen transfer of the aldol condensation product,
crotonaldehyde.
Introduction
With the increased corn-based bio-ethanol production in the
United States, the catalytic upgrading of ethanol to more
valuable chemicals such as higher alcohols and alkenes has
drawn renewed interest. Special attention is geared towards n-
butanol
– the simplest upgrading product, formed from
condensation of two ethanol molecules – because of its wide
range of applications such as a fuel additive, a solvent, and
additive in perfumes, amongst others.[1] When used as a fuel
additive, n-butanol is superior to ethanol because of its higher
energy density, and the fact that it is less corrosive and less
water soluble.[2] The conventional petrochemical route for
synthesizing n-butanol is hydroformylation of propylene to
butanal, and the subsequent hydrogenation of butanal to n-
butanol. The hydroformylation process to synthesize n-butanol
requires high pressures and uses homogeneous cobalt,
rhodium, palladium, or ruthenium catalysts,[3] which are
expensive and create separation issues. Hence, there exists an
interest to explore alternative approaches to produce n-butanol
under milder conditions using heterogeneous catalysts, and
preferably starting from a biomass-derived feedstock. The
3-
CaO/PO4 as an acid-base pair as identified in CO2 titration
experiments.[10a] On the other hand, the OH-/POH acid-base
pair was suggested to be the active site for ethanol coupling
over HAP by applying the amphoteric probe molecule,
acetylene, using IR spectroscopy.[11] Hill et al. also investigated
1
This article is protected by copyright. All rights reserved.

10.1002/cctc.202000331
ChemCatChem
FULL PAPER
the surface sites on HAP using CO2, pyridine, acetylene, and
glycine adsorption, and found that it is possible to have Ca2+
and POH as acid sites and OH- as a basic site on HAP.[12]
Recently, Ca2+/OH- and POH/OH- were reported as acid-base
pairs responsible for ethanol dehydrogenation and aldol
condensation, respectively, by using operando DRIFTS
measurement.[13] Although there are already various studies
that point out the importance of the acid-base pair on HAP
during ethanol coupling to n-butanol, the role of the specific
sites on HAP during each step, including hydrogenation, is still
unclear. Therefore, direct spectroscopic evidence from
operando experiments is needed to elucidate the relationship
between each step and surface site.
Results and Discussion
In this work we use modulation excitation (ME) with diffuse
reflectance infrared Fourier Transform spectroscopy coupled
with mass spectrometry (DRIFTS-MS). Modulation excitation
spectroscopy works similarly to a lock-in amplifier which is
used to enhance the signal-to-noise ratio and extract the
desired signals from noisy condition.[16-17] The same concept
can also apply to a spectroscopic technique by periodically
perturbing the system with one variable such as reactant
concentration, temperature, pressure, or radiation.[15a] When a
catalytic system is perturbed under quasi steady-state
conditions, reaction intermediates will oscillate with the same
frequency. After averaging several periods, the signal from
active species will be enhanced while the spectator species will
not respond to the stimulation and can thus be easily
distinguished. In addition, the signal-to-noise ratio will increase
because of the averaging. Moreover, by operating with phase
sensitive detection (PSD) to transform from time-domain to
phase-domain, different active species will show distinct phase-
delays compared to the starting reactant. These different phase
shifts contain kinetic information.
The transmission IR spectrum of the commercial HAP
sample dehydrated at 600oC is shown in Figure S1. The peak
at 3572 cm-1 corresponds to the columnar hydroxyl group (OH-)
on HAP. The band at 3648 cm-1 is attributed to bulk POH group;
3674 and 3706 cm-1 correspond to surface POH site which
coincide with previous reports. The combination bands and
overtone of P-O feature can also be observed at 2200-1900
cm-1. The bands between 1550 and 1250 cm-1 are attributed to
18]
carbonate species.[10a,
The wavenumbers of POH peaks
discovered here are slightly higher than in the literature.
Although the recording temperature may affect the IR peak
position, the room temperature HAP reference IR spectrum
(Figure S1) is similar to a previous study.[18b] In operando
condition, we recorded the IR spectra at 330oC compared to
350oC in previous report.[13] Given the closeness of the
recording temperatures, we attribute POH peak position
discrepancies mainly to the higher dehydration temperature
(600oC) in this study. The X-ray diffraction pattern is presented
in Figure S2 and is identical to previous literature and shows no
signs of impurities.[10a]
Scheme 1. Generally accepted pathways of ethanol coupling to n-butanol
over hydroxyapatite (HAP), MPV stands for Meerwein-Ponndorf-Verley
reaction.
Here, we investigate ethanol coupling to n-butanol over
hydroxyapatite (HAP) using diffuse reflectance infrared Fourier
transform spectroscopy coupled with mass spectrometry
(DRIFTS-MS) to simultaneously monitor the surface species
and the product distribution in the gas phase.[14] In addition, we
use modulation excitation (ME) to intensify the signal for active
species and enhance the signal-to-noise ratio.[15] Moreover, by
operating with phase-sensitive detection (PSD), the
spectroscopic signatures of consecutive reaction intermediates
can be revealed in the phase-domain, providing micro-kinetic
information.[16] Our operando DRIFTS-MS with ME experiments
show spectroscopic evidence to support the generally accepted
reaction mechanism of ethanol coupling over HAP via the
Guerbet pathway. In addition, Ca2+/OH- pair sites are proposed
as main active site for aldol condensation and POH/OH- pair
sites are proposed to be responsible for Meerwein-Ponndorf-
Verley (MPV) direct hydrogen transfer for hydrogenation.
Spectroscopic evidence for the isomerization of crotyl alcohol
supports direct hydrogen transfer as the second hydrogenation
step during ethanol coupling. Interactions between surface
species and different reaction intermediates are discussed to
guide further catalyst design for ethanol upgrading.
Modulation experiments with ethanol
We first introduce the starting reactant, ethanol, to the
system to understand the formation of reaction intermediates.
Before starting the modulation experiment, HAP was treated
under pure Ar flow and heated to 330 °C, followed by periodic
switching between ethanol and pure Ar. The outlet is monitored
by mass spectrometry. The chosen m/z values correspond to
unique MS fragments of each intermediate and product and
some signals are scaled for better reading (Figure 1a). When
ethanol is introduced into the system, the signals from the
reaction intermediates and product such as acetaldehyde
(m/z=44), crotyl alcohol (m/z=57), and n-butanol (m/z=56) can
be observed. During ethanol coupling, 1,3-butadiene (m/z=54)
occurs as the side product obtained via the dehydration of
crotyl alcohol by Lewis acid site on HAP.[12] Only trace amounts
of butanal (m/z=72) and crotonaldehyde (m/z=70) are observed,
implying that hydrogenation of the carbonyl group on both
species occurs rapidly. The product distribution (i.e. the
selectivity) is in line with
a
previous kinetic study.[10a]
2
This article is protected by copyright. All rights reserved.

10.1002/cctc.202000331
ChemCatChem
FULL PAPER
Figure 1. Modulation experiment between ethanol (5.0%) in Ar and pure Ar (20 mL min-1) over 5 mg hydroxyapatite (HAP) at 330oC, the total flow rate for both
channel are 20 mL min-1 which Ar behaves as balance gas (a) selected m/z signals from mass spectrometry response (b) Time-domain DRIFT spectra and the
line at 122 sec. indicates gas channels switch.
In the first half period of the experiment, the surface is
covered by ethanol as indicated by the features at 3666, 2972,
2933 and 2903 cm-1 for υ(O-H), υ(CH3, asymmetric stretching),
υ(CH2, asymmetric stretching), and υ(CH3, symmetric
stretching), respectively (Figure 1b). This observation
corroborates the conclusion of a previous study that shows
high ethanol coverage on HAP comparing to MgO at reaction
temperature by using steady-state isotopic kinetic transient
analysis (SSITKA).[8] Aside from the intense ethanol signals in
the spectra, a small band at 1758 cm-1 indicates the formation
of acetaldehyde, an intermediate in ethanol coupling following
the Guerbet pathway. In contrast, a band at 1577 cm-1
corresponds to aromatic coke acting as spectator species that
does not respond to the modulation, showing the capability of
modulation excitation to discriminate active and spectator
species. Moreover, to the best of our knowledge, acetaldehyde
has not been directly observed before in ethanol coupling using
in-situ spectroscopy. The intensity of columnar hydroxyl group
(OH-) at 3572 cm-1 on HAP[18b, 19] (Figure S1) decreases during
the reaction (Figure 1b) suggesting an interaction of ethanol
with the OH-. The importance of the columnar hydroxyl group is
supported by a previous kinetic study which shows that
HAP(Ca10(PO4)6(OH)2) with OH- is more selective to n-butanol
than beta-tricalcium phosphate (β-Ca3(PO4)2) and fluoride-
substituted phosphate (Ca10(PO4)6F2),[20] since the OH- group
could be responsible for base-catalyzed aldol condensation.
line with previous observations that the rate of C-C bond
formation increases linearly with co-fed acetaldehyde.[9] In
addition to the increase of crotonaldehyde, other intermediates
and side products including butanal (m/z=72), crotyl alcohol
(m/z=57) and 1,3-butadiene (m/z=54) are observed. The final
product, n-butanol (m/z=56), appears with 17 seconds time-
delay in the gas phase compared to acetaldehyde and other
intermediates, suggesting that desorption of product may be
slow over HAP.
The time-domain infrared spectra show a negative peak at
2980 cm-1 indicating partial desorption and/or reaction of
surface-bound ethanol. The other significant peaks in the C-H
region at 2736, 3026, and 2960 cm-1 correspond to
acetaldehyde, crotyl alcohol, and n-butanol, respectively,
showing a sequential time-delay in each species, which is in
line with the proposed Guerbet pathway (Figure 2b). Notably,
the co-fed acetaldehyde (2736 cm-1) can be first seen at 8 sec.
and the consecutive intermediate crotyl alcohol (3026 cm-1)
appears at 25 sec. suggesting aldol condensation and direct
hydrogen transfer from ethanol via Meerwein-Ponndorf-Verley
(MPV) pathway proceed rapidly. The strong signal at 1761-
1720 cm-1 is due to υ(C=O) from gas phase acetaldehyde. The
emergence of a feature at 1645 cm-1 corresponding to υ(C=C)
from the aldol product, crotonaldehyde, shows a time-delay
compared to υ(C=O) from acetaldehyde. The features from the
other intermediate butanal are barely observed in the IR
spectrum, which could be attributed to the fast desorption
and/or reaction, or they are hindered by peaks from
acetaldehyde, as there is a notable amount of crotonaldehyde
(m/z=70) and butanal (m/z=72) in the MS response (Figure 2a).
The decreasing signal of columnar hydroxyl group at 3572 cm-1
indicates that the OH- group may behave as basic site to
abstract α-hydrogen on acetaldehyde to form the enolate for
aldol condensation.
Modulation experiment with ethanol/acetaldehyde and
ethanol
To investigate the aldol condensation reaction, we added
acetaldehyde to the reagent stream while keeping the ethanol
concentration constant at 2.5%. The HAP surface was first
saturated with 2.5% ethanol in Ar and the background was
taken before starting periodic modulation in which the first half
period was ethanol/acetaldehyde with 3:1 molar ratio and
ethanol was maintained at 2.5% in the feed. After averaging
and operating phase sensitive detection (PSD) for five periods
of MS responses and DRIFT spectra, the changing signals are
attributed to the perturbation by acetaldehyde. The response
MS signals show an enhancement for the aldol condensation
product, crotonaldehyde (m/z=70) (Figure 2a). The result is in
By using phase-sensitive detection (PSD), phase-domain IR
spectra show three distinct species in the C-H region:
acetaldehyde with a phase delay of 10o (or 7 s) at 2736 cm-1,
crotyl alcohol with a phase delay of 40o (or 28 s) at 3026 cm-1,
and n-butanol with a phase delay of 70o (or 49 s) at 2960 cm-1
(Figure 3). We emphasize that the phase delay is corrected by
30o due to the inherent dead volume of the DRIFTS cell and
3
This article is protected by copyright. All rights reserved.

10.1002/cctc.202000331
ChemCatChem
FULL PAPER
Figure 2. Modulation experiment between ethanol (2.5%)/AA with 3:1 molar mixture in Ar as carrier gas and ethanol in Ar over 5 mg hydroxyapatite (HAP) at
330oC, the total flow rate for both channel are 20 mL min-1 which Ar behaves as balance gas. (a) selected m/z signals from mass spectrometry response (b) Time-
domain DRIFT spectra and the line at 122 sec. indicates gas channels switch.
Modulation experiment with ethanol/crotonaldehyde and
ethanol
To study the role of crotonaldehyde and hydrogenation
products such as crotyl alcohol, butanal, and n-butanol, we co-
fed crotonaldehyde with ethanol as described previously for
acetaldehyde. In the experiment shown in Figure 4a, HAP was
first equilibrated in ethanol feed, followed by periodically
switching to a 3:1 molar ratio ethanol/crotonaldehyde flow. In
the gas phase, acetaldehyde (m/z=44) increases dramatically
when crotonaldehyde is co-fed. Crotyl alcohol (m/z=57),
butanal (m/z=72) and n-butanol (m/z=56) are enhanced as well.
The increased production of acetaldehyde in the beginning of
first half period indicates that direct hydrogen transfer from
ethanol to crotonaldehyde occurs rapidly. An undesired side
product, 1,3-butadiene (m/z=54) also increases with co-fed
crotonaldehyde, suggesting dehydration of crotyl alcohol may
compete with the further hydrogenation step. Moreover, n-
butanol (m/z=56) and water (m/z=18) appear with 11 seconds
time-delay in the MS response, implying slow desorption from
the HAP surface.
A similar result can also be seen in the time-domain IR
spectra, (Figure 4b) where υ(C=O) from acetaldehyde emerges
as a side band at 1758 cm-1 at the beginning, along with the
Figure 3. Phase-sensitive detection (PSD) DRIFT spectra in C-H region with
υ(C-H) for crotyl alcohol at 3026 cm-1 with co-fed
modulation experiment between ethanol (2.5%)/AA with 3:1 molar mixture in
crotonaldehyde, indicating MPV hydrogen transfer is fast. More
Ar as carrier gas and ethanol in Ar flow over 5 mg HAP, the total flow rate for
both channel are 20 mL min-1 which Ar behaves as balance gas. The phase-
delay at 10o corresponds to acetaldehyde, 40o correlates to crotyl alcohol,
and 70o indicates n-butanol. A full period (250s) is identical to 360o.
importantly, the n-butanol signal at 2960 cm-1 persists into the
second half period when crotonaldehyde is removed from the
feed, which is in line with MS response that the desorption rate
of n-butanol is slow. In the phase-domain IR spectra, the co-fed
crotonaldehyde can be observed at 2730 and 2815 cm-1 with
10o (7 s) phase delay, while crotyl alcohol appears at 3026 cm-1
with 40o (28 s) phase delay. The final product, n-butanol, can
be seen at 2960 cm-1 with 70o (49 s) phase delay (Figure S7).
This result mirrors the behavior we observed earlier when
modulating between ethanol/acetaldehyde and ethanol. The
phase-domain IR spectra demonstrate that the aldol
condensation product subsequently undergoes hydrogenation
to form n-butanol. Although several studies show that the C=O
that phase delays of various species should therefore only be
used relative to one another. The observed consecutive
reaction intermediates and product agree with the generally
accepted hypothesis that ethanol coupling to n-butanol on HAP
at 330oC proceeds via the Guerbet pathway.
4
This article is protected by copyright. All rights reserved.

10.1002/cctc.202000331
ChemCatChem
FULL PAPER
Figure 4. Modulation experiment between ethanol (2.5%)/CA with 3:1 molar mixture in Ar as carrier gas and ethanol in Ar over 5 mg hydroxyapatite (HAP) at
330oC, the total flow rate for both channel are 20 mL min-1 which Ar behaves as balance gas. (a) selected m/z signals from mass spectrometry response (b) Time-
domain DRIFT spectra and the line at 122 sec. indicates gas channels switch.
Figure 5. Modulation experiment between ethanol (2.5%)/butanal with 3:1 molar mixture in Ar as carrier gas and ethanol in Ar over 5 mg hydroxyapatite (HAP) at
330oC, the total flow rate for both channel are 20 mL min-1 which Ar behaves as balance gas. (a) selected m/z signals from mass spectrometry response (b) Time-
domain DRIFT spectra and the line at 122 sec. indicates gas channels switch.
on crotonaldehyde will be hydrogenated by MPV direct
hydrogen transfer from ethanol to form crotyl alcohol, the
mechanism of the second hydrogenation step is still not well
understood. Indeed, the hydrogenation can occur either to the
Modulation experiment with ethanol/crotyl alcohol and
ethanol
When crotyl alcohol is co-fed into the system during ethanol
coupling, the dehydration side products, 1,3-butadiene (m/z=54)
and water (m/z=18), increase directly in MS response. On the
other hand, the desired product, n-butanol (m/z=56), increases
slightly compared to 1,3-butadiene when co-feeding crotyl
alcohol (m/z=57) into the system (Figure S8a). The
enhancement of 1,3-butadiene could be due to the moderate
Lewis acidic site on HAP[11-12] which favors the dehydration
pathway under crotyl alcohol co-feed condition. Interestingly,
while 1,3-butadiene seems to increase with crotyl alcohol
addition, the final product, n-butanol, does not increase
substantially. This difference in response suggests that, n-
butanol could predominately come from MPV direct
C=C bond of crotyl alcohol through
a surface-mediated
pathway[9] or it can occur at the C=O bond of butanal, which is
isomerized from crotyl alcohol, through the MPV direct
hydrogen transfer from ethanol to obtain n-butanol[10] (Scheme
1). In a subsequent ME experiment, we perturbed the system
with crotyl alcohol and butanal in order to obtain insights into
this second hydrogenation step.
5
This article is protected by copyright. All rights reserved.

10.1002/cctc.202000331
ChemCatChem
FULL PAPER
hydrogenation of C=O on butanal, obtained from crotyl alcohol
isomerization; hydrogenation on C=C of crotyl alcohol through
surface mediated hydrogen seems unlikely.
suggested that the isomerization can occur on basic sites or
Ca-O sites on HAP surface,[10a] there has been no
spectroscopic support for the hypothesis. Here, the peak at
3059 cm-1 observed in IR spectra corresponds to a crotyl
alcohol isomer with a higher frequency feature corresponding
to a terminal alkene υ(=CH2). Furthermore, because of the poor
dehydrogenation ability of HAP,[10b] the main source of
acetaldehyde may come from direct hydrogen transfer
(Scheme 1); thus, the second hydrogenation most likely
proceeds via the MPV direct hydrogenation pathway. Besides
dehydrogenation of ethanol to obtain the very first
acetaldehyde to proceed the reaction, the trace amount of
dissolved O2 in ethanol may also cause oxidation of ethanol to
acetaldehyde, since there are barely redox properties on HAP
compared to transition metal catalysts.
In the time-domain IR spectra (Figure S8b), crotyl alcohol
becomes the dominant surface species on HAP with several
significant features at 3026, 2923, 2879, 1676 cm-1, in-line with
the MS response that the increased ethanol in the first half
period is due to competitive adsorption with crotyl alcohol (viz,
ethanol desorbs from HAP surface because of the addition of
crotyl alcohol). The competitive adsorption can explain the
inhibition of direct hydrogen transfer due to a lower abundance
of ethanol on the HAP surface, therefore enhancing the
dehydration on Lewis acidic sites. The evidence for 1,3-
butadiene from dehydration of crotyl alcohol can also be
observed at 1817 cm-1. As a result, ethanol coupling over HAP
at high crotyl alcohol concentration tends to undergo the
dehydration pathway due to Lewis acidity on HAP[20] and
competitive adsorption with ethanol.
Modulation experiment with ethanol/butanal and ethanol
When co-feeding butanal during ethanol coupling over HAP,
the MS responses show that n-butanol (m/z=56) increases
dramatically as a main product and the ethanol (m/z=46) signal
decreases in the first half period, indicating that direct hydrogen
transfer from ethanol to butanal occurs rapidly (Figure 5a). The
result implies that n-butanol stems from butanal as the last
reactive intermediate by direct hydrogen transfer. The
sacrificial ethanol will then dehydrogenate to acetaldehyde and
Scheme 2. Isomerization pathway for crotyl alcohol to butanal.
Modulation experiment with crotonaldehyde and ethanol
To better understand the hydrogenation process, we first
saturated the HAP surface and reached steady state in an
ethanol feed and subsequently switched to crotonaldehyde
feed without ethanol. Before switching to the crotonaldehyde
flow, ethanol is the predominant surface species on HAP. After
switching to crotonaldehyde flow, acetaldehyde and C4
intermediates such as crotyl alcohol and butanal increase
immediately as well as n-butanol, indicating that direct
hydrogen transfer from ethanol occurs rapidly (Figure 6a).
More importantly, once the surface ethanol is fully consumed
by direct hydrogen transfer, the reaction stops at crotyl alcohol
and butanal. The final hydrogenation products, n-butanol and
acetaldehyde stop producing at 30 seconds after switching to
the crotonaldehyde feed. Crotyl alcohol formation also reaches
its maximum around 40 seconds because of a lack of surface
ethanol to hydrogenate crotonaldehyde. It is worth noting that
after consuming surface ethanol species, butanal increases
slightly until the end of the first half period, while crotyl alcohol
slightly decreases at the same time. This finding also indicates
that isomerization from crotyl alcohol to butanal takes place on
HAP surface.
In the time-domain IR spectra, the negative peaks at 2977
and 2903 cm-1 correspond to υ(CH3, asymmetric stretching),
and υ(CH3, symmetric stretching) from surface ethanol
consumption which corroborates the finding that ethanol mostly
behaves as a hydrogen donor (Figure 6b). The features from
crotonaldehyde at 2815, 2730, 1720, 1697, 1645 cm-1 gradually
dominate the HAP surface, indicating the reaction stops without
adding ethanol. While the bands at 1720 and 1645 cm-1
correspond to υ(C=O) and υ(C=C), respectively, from gas
phase crotonaldehyde, the emergence of 1697 cm-1 could be
adsorption of carbonyl group on acid site such as Ca2+ or POH.
Notably, a band emerges at 3059 cm-1 which is similar to the
feature in butanal co-feed IR spectra above (Figure 5b),
subsequently
undergo
self-aldol
condensation
to
crotonaldehyde which can be hydrogenated by direct hydrogen
transfer to finish the autocatalytic cycle. This finding supports a
previous study which stated that acetaldehyde is generated
mainly from direct hydrogen transfer and ethanol is mainly a
hydrogen source for direct hydrogen transfer.[10b] Furthermore,
there is no significant increase in crotonaldehyde (m/z=70),
rendering further support for the fast hydrogenation of C=O to
crotyl alcohol (m/z=57) which we can clearly observe in MS
response.
In the time-domain IR spectra, the peaks at 2960 cm-1
correspond to υ(C-H) from n-butanol which remains on the
surface into the second half period compared to butanal at
2707 and 1743 cm-1 (υ(C-H) and υ(C=O), respectively) (Figure
5b). The slow desorption of n-butanol can also be observed in
the MS response, where n-butanol (m/z=56) shows a time-
delay of 6 seconds relative to the introduction of butanal.
Although there is only a trace amount of crotonaldehyde
observed in MS response, the peak at 1653 cm-1 might be the
evidence of υ(C=C) in crotonaldehyde surface species from
self-aldol condensation of acetaldehyde coming from direct
hydrogen transfer. Interestingly, a peak appears at 3059 cm-1
during the butanal co-feed experiment with a 25 seconds time-
delay compared to the signal from butanal. This vibrational
feature is too high a frequency for υ(=C-H) in crotyl alcohol, but
similar to the terminal alkene υ(=CH2) signal in 1-butene.[21]
Thus, the emergence of the peak at 3059 cm-1 could
correspond to 3-buten-1-ol having terminal alkene feature from
crotyl alcohol isomerization (Scheme 2). However, the other
isomer, 1-buten-1-ol, cannot be found in the IR spectra; the
less stable isomer, 1-buten-1-ol, will most likely undergo fast
tautomerization to butanal. This finding supports that
isomerization and tautomerization can occur on the HAP
surface to generate butanal which can be further hydrogenated
by direct hydrogen transfer to n-butanol. Although it has been
6
This article is protected by copyright. All rights reserved.

10.1002/cctc.202000331
ChemCatChem
FULL PAPER
Figure 6. Modulation experiment between CA (0.8%) and ethanol (2.5%) in Ar as carrier gas over 5 mg hydroxyapatite (HAP) at 330oC, the total flow rate for both
channel are 20 mL min-1 which Ar behaves as balance gas. (a) selected m/z signals from mass spectrometry response (b) Time-domain DRIFT spectra and the
line at 122 sec. indicates gas channels switch.
suggesting a terminal alkene-like species which could form
from crotyl alcohol isomerization. This inference is supported
by the MS responses, as butanal increases while crotyl alcohol
decreases once the sacrificial ethanol is consumed. The
decreased peaks at 3572 and 3674 cm-1 correspond to OH-
and POH on HAP which are likely responsible for direct
hydrogen transfer and isomerization. The same decreasing
peaks for OH- and POH can also be found in the butanal co-
feed experiment (Figure 5b).
The isomerization of crotyl alcohol to 1-butene-1-ol can occur
on both acid and base site, since Bronsted acid site such as
POH can protonate to C=C and subsequently, the adjacent
carbon releases a proton to form a new C=C bond to finish
isomerization.[21] The basic OH- site can also isomerize crotyl
alcohol to form 1-butene-1-ol. This isomerization is facile over
22]
metal oxide catalysts like MgO and CaO.[10a,
In the
experiment where we feed crotonaldehyde over an ethanol-
saturated system, a decreased peak of POH at 3674 cm-1 may
indicate deprotonation of POH group and a decrease peak at
3572 cm-1 suggests the abstraction of hydrogen by the OH-
group to eliminate the columnar hydroxyl group signal (Figure
6b). Therefore, both POH and OH- can be able to catalyze
crotyl alcohol isomerization and the isomer, 1-butene-1-ol, will
subsequently tautomerize to butanal and be hydrogenated via
direct hydrogen transfer to n-butanol.
The possible acid and base sites on HAP have been
characterized by using pyridine, CO2, acetylene, and glycine as
probe molecules.[11-12] The proposed acid sites are Ca2+ and
protonated phosphate groups, POH, while the columnar OH-
group is the most significant basic site on the HAP surface.
Therefore, Ca2+/OH- and POH/OH- acid-base pairs are
potential active sites for ethanol coupling over HAP surface.
The significant decreased band at 3572 cm-1 corresponding to
OH- group on HAP surface in acetaldehyde co-feed experiment,
indicates that OH- is one of the active sites for base-catalyzed
aldol condensation (Figure 2b). The importance of OH- for C-C
coupling is further supported by the increased selectivity
towards n-butanol for HAP compared to calcium phosphate
(Ca3(PO4)2) without OH- group.[20] Since we do not observe a
significant signal corresponding to the POH group in
acetaldehyde co-feed IR spectra, Ca2+/OH- may likely act as
acid-base pair for aldol condensation over HAP. We emphasize
that the background of co-feed experiments are collected under
standard reaction condition when 2.5% ethanol in the feed at
330oC. Therefore, the changes in the IR and MS spectra are
the responses due to the co-feed of different intermediates.
When co-feeding butanal and modulating between
crotonaldehyde and ethanol (Figure 5b and Figure 6b) there is
a decrease in the POH/OH- acid-base pair signal. This
suggests that POH/OH- is the active site for Meerwein-
Ponndorf-Verley (MPV) direct hydrogen transfer. (Scheme 3)
The findings corroborate the work by Osman et al. where they
found no production of n-butanol if the POH site has been
poisoned.[13]
Scheme 3. Meerwin-Ponndorf-Verley (MPV) direct hydrogen transfer on
POH/OH- acid-base pair.
Conclusion
In this study, the mechanism for ethanol coupling to n-butanol
over HAP is investigated using operando DRIFTS-MS with
modulation excitation (ME) and phase sensitive detection
(PSD). The enhanced signal-to-noise ratio allows us to observe
the IR features from intermediates and help to elucidate the
mechanism. MS results and the phase-domain spectra show
the consecutive formation of reaction intermediates, supporting
the Guerbet pathway as the overall mechanism for ethanol
coupling over HAP. The phase-domain IR spectra from
7
This article is protected by copyright. All rights reserved.

10.1002/cctc.202000331
ChemCatChem
FULL PAPER
detection (PSD) transforms the DRIFT spectra from time-domain to
phase domain by using the equations as follows.
acetaldehyde co-feed experiments demonstrate consecutive
reaction intermediates during ethanol coupling. The first
hydrogenation step in which crotonaldehyde converts to crotyl
alcohol is attributed to direct hydrogen transfer from ethanol by
the result of crotonaldehyde co-feed experiment. From the
result of modulating between crotonaldehyde and ethanol
experiment, the emergence of the peak at 3059 cm-1
corresponds to terminal-like alkene υ(=CH2) indicating the
appearance of 3-butene-1-ol, the isomer of crotyl alcohol. This
finding implies that crotyl alcohol undergoes isomerization and
tautomerization to butanal over the HAP surface and the
second hydrogenation subsequently occurs on butanal by
direct hydrogen transfer to produce n-butanol.
푇
2
퐴푘�φ푃푘푃푃� = � 퐴 푡 sin�푘푘푡 + φ푃푘푃푃� 푑푡
( )
푇
0
T is the length of a period, ω is the stimulation frequency, φk is the
phase delay, k is the demodulation index (k = 1 in the study), A(t) is the
active species response in the time-domain, and Ak is the response in
the phase-domain. The transformation to the phase-domain leads to a
dependence of the vibrational signals on the phase angle φ instead of
the time. For example, 250 sec period can be converted to 360o. The
analysis of the spectra was processed using MATLAB codes.
We also show that the Ca2+/OH- acid-base pair may be
mainly responsible for aldol condensation based on
acetaldehyde co-feed experiment. In addition, direct hydrogen
transfer occurs on POH/OH- acid-base pair, since hydrogen
donor, ethanol, can be stabilized on HAP surface on both sites
based on ethanol adsorption IR spectrum and operando DRIFT
spectra. Finally, the active site for isomerization can be either
POH or OH-, because both sites show interaction in modulation
DRIFT spectra between crotonaldehyde and ethanol. The
competitive adsorption between ethanol and crotyl alcohol
suggests that isomerization may be critical for the overall
performance, since direct hydrogen transfer occurs rapidly.
This puts forward the hypothesis that eliminating the Lewis acid
site could avoid the undesired dehydration side-reaction, while
increasing the POH and OH- site density could enhance the
isomerization.
Acknowledgements
Financial support from the University of Wisconsin—Madison,
as well as the Donors of the American Chemical Society
Petroleum Research Fund under Grant No. PRF 56136-ND5, is
kindly acknowledged.
Keywords: modulation excitation spectroscopy • Guerbet
reaction • ethanol coupling • hydroxyapatite • butanol • DRIFTS
[1]
C. Jin, M. Yao, H. Liu, C.-f. F. Lee, J. Ji, Renew. Sust. Energ.
Rev. 2011, 15, 4080-4106.
[2]
[3]
N. Savage, Nature 2011, 474, S9-S11.
a) C. E. O'Rourke, P. R. Kavasmaneck, R. E. Uhl, in
Monohydric Alcohols, Vol. 159, AMERICAN CHEMICAL
SOCIETY, 1981, pp. 71-85; b) G. M. Torres, R. Frauenlob, R.
Franke, A. Börner, Catal. Sci. Technol 2015, 5, 34-54.
a) S. Veibel, J. I. Nielsen, Tetrahedron 1967, 23, 1723-1733;
b) D. Gabriels, W. Y. Hernandez, B. Sels, P. Van Der Voort,
A. Verberckmoes, Catal. Sci. Technol 2015, 5, 3876-3902; c)
A. J. O'Lenick Jr., J. Surfactants Deterg 2001, 4, 311-315; d)
J. Scalbert, F. Thibault-Starzyk, R. Jacquot, D. Morvan, F.
Meunier, J. Catal. 2014, 311, 28-32.
[4]
Experimental Section
Hydroxyapatite (Ca10(PO4)6(OH)2, HAP) was purchased from Sigma-
Aldrich and was calcined at 600oC with 5oC/min ramp rate for 2 hours
under dry air before operating ethanol coupling reaction. Ethanol was
dehydrated by molecule sieves (3Å) before using to eliminate the
interference in IR spectra.
[5]
[6]
a) J. T. Kozlowski, R. J. Davis, ACS Catal. 2013, 3, 1588-
1600; b) X. Wu, G. Fang, Y. Tong, D. Jiang, Z. Liang, W.
Leng, L. Liu, P. Tu, H. Wang, J. Ni, X. Li, ChemSusChem
2018, 11, 71-85.
a) T. Tsuchida, S. Sakuma, T. Takeguchi, W. Ueda, Ind. Eng.
Chem. Res. 2006, 45, 8634-8642; b) T. Tsuchida, J. Kubo, T.
Yoshioka, S. Sakuma, T. Takeguchi, W. Ueda, J. Catal. 2008,
259, 183-189; c) L. Silvester, J.-F. Lamonier, J. Faye, M.
Capron, R.-N. Vannier, C. Lamonier, J.-L. Dubois, J.-L.
Couturier, C. Calais, F. Dumeignil, Catal. Sci. Technol 2015,
5, 2994-3006; d) M. Ben Osman, S. Diallo Garcia, J.-M.
Krafft, C. Methivier, J. Blanchard, T. Yoshioka, J. Kubo, G.
Costentin, PCCP 2016, 18, 27837-27847.
The modulation excitation DRIFTS-MS setup allows for the introduction
and vaporization of liquid substrates, and switching between two
different flows, effectively modulating the concentration of reactants
during experiment. In short, argon was used as carrier gas and
connected to two different mass flow controllers to generate two
independent feed. Besides of Ar as carrier gas, Ar is also a balance gas
to maintain total flow rate at 20 mL min-1.Two syringe pumps were used
to introduce liquids that are evaporated in a heated spiral before
reaching an electronically controlled two-position-four-way valve.
Depending on the position of this valve, either flow A or B enters into
the DRIFTS accessory. The catalyst, HAP, is filled in ta ceramic cup in
DRIFTS cell with approx. 5 mg. Finally, the gas-phase composition is
monitored with an online mass spectrometer. By periodically switching
between flow A and B, the influence of either component on the
reaction can be analyzed in a transient manner. A single period of the
complete modulation is 250 seconds in this study, unless further
mentioned. The detail for ME-DRIFTS-MS setup can be found in
previous study.[14]
[7]
a) S. Ogo, A. Onda, Y. Iwasa, K. Hara, A. Fukuoka, K.
Yanagisawa, J. Catal. 2012, 296, 24-30; b) S. Ogo, A. Onda,
K. Yanagisawa, Appl. Catal. A: Gen. 2011, 402, 188-195.
S. Hanspal, Z. D. Young, H. Shou, R. J. Davis, ACS Catal.
2015, 5, 1737-1746.
[8]
[9]
T. Moteki, D. W. Flaherty, ACS Catal. 2016, 6, 4170-4183.
[10] a) C. R. Ho, S. Shylesh, A. T. Bell, ACS Catal. 2016, 6, 939-
948; b) Z. D. Young, R. J. Davis, Catal. Sci. Technol 2018, 8,
1722-1729.
[11] S. Diallo-Garcia, M. B. Osman, J.-M. Krafft, S. Casale, C.
Thomas, J. Kubo, G. Costentin, J. Phys. Chem. C. 2014, 118,
12744-12757.
All IR measurements are collected by a Bruker Vertex 70 spectrometer
with mercury cadmium telluride (MCT) detector. Each spectrum is
obtained from 64 scans with 8 cm-1 resolution and 4 seconds as
temporal resolution in modulation experiment. The DRIFTS accessory
was purchased from PIKE Technologies (DiffusIR) and approximately 5
mg HAP can be loaded in the ceramic crucible. The m/z responses are
obtained from ThermoStar mass spectrometer, Pfeiffer Vacuum,
attached on the outlet of DRIFTS accessory. The phase sensitive
[12] I. M. Hill, S. Hanspal, Z. D. Young, R. J. Davis, J. Phys.
Chem. C. 2015, 119, 9186-9197.
[13] M. B. Osman, J.-M. Krafft, C. Thomas, T. Yoshioka, J. Kubo,
G. Costentin, ChemCatChem 2019, 11, 1765-1778.
[14] a) P. Müller, S. P. Burt, A. M. Love, W. P. McDermott, P.
Wolf, I. Hermans, ACS Catal. 2016, 6, 6823-6832; b) P.
Müller, S. C. Wang, S. P. Burt, I. Hermans, ChemCatChem
2017, 9, 3572-3582.
8
This article is protected by copyright. All rights reserved.

10.1002/cctc.202000331
ChemCatChem
FULL PAPER
[15] a) P. Müller, I. Hermans, Ind. Eng. Chem. Res. 2017, 56,
1123-1136; b) P. D. Srinivasan, B. S. Patil, H. Zhu, J. J.
Bravo-Suárez, React. Chem. Eng. 2019, 4, 862-883.
[16] A. Urakawa, T. Bürgi, A. Baiker, Chem. Eng. Sci. 2008, 63,
4902-4909.
Osman, J.-M. Krafft, S. Boujday, C. Guylène, Catal. Today
2014, 226, 81-88.
[19] M. I. Kay, R. A. Young, A. S. Posner, Nature 1964, 204, 1050.
[20] S. Hanspal, Z. D. Young, J. T. Prillaman, R. J. Davis, J. Catal.
2017, 352, 182-190.
[17] a) R. M. Hexter, J. Opt. Soc. Am. 1963, 53, 703-709; b) U. P.
Fringeli, H. H. Günthard, Appl. Opt. 1971, 10, 819-824.
[18] a) Z. H. Cheng, A. Yasukawa, K. Kandori, T. Ishikawa,
Langmuir 1998, 14, 6681-6686; b) S. Diallo-Garcia, M. Ben
[21] J. N. Kondo, K. Domen, J. Mol. Catal. A: Chem. 2003, 199,
27-38.
[22] Y. Schächter, H. Pines, J. Catal. 1968, 11, 147-158.
9
This article is protected by copyright. All rights reserved.

10.1002/cctc.202000331
ChemCatChem
FULL PAPER
Entry for the Table of Contents
Mechanistic study of ethanol coupling to n-butanol over hydroxyapatite is investigated by DRIFTS and mass spectrometry following a
modulation excitation approach. The approach confirms the Guerbet pathway and suggests the Ca2+/OH- acid-base pair as the main
active site for aldol condensation and POH/OH- as the major acid-base pair for direct hydrogen transfer and isomerization.
Institute and/or researcher Twitter usernames: @hermansgroup ; @UWMadisonChem ; @UWMadison
10
This article is protected by copyright. All rights reserved.