Production of renewable oleo-furan surfactants by cross-ketonization of biomass-derived furoic acid and fatty acids

Article abstract of DOI:10.1039/d0cy02349c

Synthesis of 2-dodecanoyl furan is a crucial step in the formulation of oleo-furan sulfonates as bio-surfactants from biomass-derived furans and vegetable-oil-derived molecules. Herein, cross-ketonization of 2-furoic acid and lauric acid is proposed to produce the bio-surfactant precursor. Among the commonly reported metal oxide ketonization catalysts, the inexpensive and abundant iron oxides are demonstrated as effective and recyclable catalysts, enabling up to 77% selectivity to 2-dodecanoyl furan at 56% lauric acid conversion. Catalyst characterization by X-ray diffraction, H₂temperature-programmed reduction, and X-ray photoelectron spectroscopy indicates that Fe₃O₄is the catalytically active and stable phase.¹³C isotopic tracing experiments suggest that cross-ketonization on Fe₃O₄proceedsviaa β-keto acid intermediate.

Full text of DOI:10.1039/d0cy02349c

Catalysis
Science &
Technology
View Article Online
View Journal | View Issue
PAPER
Production of renewable oleo-furan surfactants by
cross-ketonization of biomass-derived furoic acid
and fatty acids†
Cite this: Catal. Sci. Technol., 2021,
11, 2762
Hannah Nguyen,a Yunzhu Wang, b David Moglia,a Jiayi Fu,ab Weiqing Zheng,b
ab
ab
*
*
Marat Orazov
and Dionisios G. Vlachos
Synthesis of 2-dodecanoyl furan is a crucial step in the formulation of oleo-furan sulfonates as bio-
surfactants from biomass-derived furans and vegetable-oil-derived molecules. Herein, cross-ketonization
of 2-furoic acid and lauric acid is proposed to produce the bio-surfactant precursor. Among the commonly
reported metal oxide ketonization catalysts, the inexpensive and abundant iron oxides are demonstrated as
effective and recyclable catalysts, enabling up to 77% selectivity to 2-dodecanoyl furan at 56% lauric acid
conversion. Catalyst characterization by X-ray diffraction, H2 temperature-programmed reduction, and
X-ray photoelectron spectroscopy indicates that Fe3O4 is the catalytically active and stable phase. 13C
isotopic tracing experiments suggest that cross-ketonization on Fe3O4 proceeds via a β-keto acid
intermediate.
Received 6th December 2020,
Accepted 22nd February 2021
DOI: 10.1039/d0cy02349c
rsc.li/catalysis
water, i.e., in the presence of Ca2+, is much better than the
LAS's. Overall, OFSs enable the (1) utilization of renewable
Introduction
Recent research in biomass conversion has significantly
advanced the science and technology of converting the most
naturally abundant renewable source of carbon to chemicals.
Specifically, the non-edible biomass-derived furans can be
upgraded to valuable chemicals, such as plastics, rubber,
lubricants, and detergents.1–7 Examples include the synthesis
of para-xylene via cycloaddition of dimethyl furan and
lignocellulosic feedstock, (2) enhanced detergency in cold
water applications, and (3) improved stability in hard water,
eliminating the need for costly chelating agents used in
conventional LAS.
The OFSs preparation involves the formation of
2-alkoylfuran intermediate via Friedel–Crafts acylation of
furan with vegetable-oil-derived, long-chain (C8–C18)
a
a
ethylene,1
butadiene
via
“dehydra-decyclization”
of
carboxylic acid or fatty acid anhydride. Optionally,
hydrodeoxygenation of the ketone group may be used to
increase the hydrophobicity of the alkyl chain. Finally,
sulfonation introduces the hydrophilic head of the
surfactant. Among these steps, the C–C coupling reaction to
form 2-alkoylfuran is a bottleneck. In one route, furan is
coupled with a fatty acid anhydride (e.g., lauric acid C12
anhydride) over Al-SPP, a hierarchical porous Brønsted acid
zeolite, to yield up to 90% of the acylated furan
(Fig. 1, pathway A). Though selective, this pathway produces
an equimolar amount of lauric acid, with a high boiling point
tetrahydrofuran,4 C33–45 base-oil lubricants,5 and oleo-furan
sulfonates (OFSs), which possess excellent detergency. The
structure of OFSs (shown in Fig. 1) includes a central furan
moiety that links a hydrophobic hydrocarbon chain to a
hydrophilic sulfonate. Park et al. evaluated the surfactant
properties of OFSs using the critical micelle concentration
(CMC), defined as the minimum surfactant concentration for
micelle formation, the Kraft point (TKraft), below which
surfactants form solid crystals, and their stability in hard
water.7 OFSs with linear alkyl chains exhibit enhanced
surfactant performance as evidenced by their lower CMC and
TKraft than petroleum-derived alkylbenzene sulfonates (LAS).
The furan moiety improves surfactant's solubility compared
to the benzene moiety. Moreover, the OFS function in hard
a Department of Chemical and Biomolecular Engineering, University of Delaware,
Newark, Delaware 19716, USA. E-mail: orazov@udel.edu, vlachos@udel.edu
b Catalysis Center for Energy Innovation, Newark, Delaware 19716, USA
† Electronic supplementary information (ESI) available. See DOI: 10.1039/
d0cy02349c
Fig. 1 Pathways to produce oleo-furan sulfonate bio-surfactant. R
represents the n-C11H23 chain; other chains could also be considered.
Path C is the one developed herein.
2762 | Catal. Sci. Technol., 2021, 11, 2762–2769
This journal is © The Royal Society of Chemistry 2021

View Article Online
Catalysis Science & Technology
Paper
(∼298 °C), rendering its separation and recycling difficult. An
alternative route utilizes trifluoroacetic anhydride to convert
lauric acid to lauric–trifluoroacetic anhydride, which then
acylates the furan (Fig. 1, pathway B). This pathway generates
a waste stream of corrosive trifluoroacetic acid.7 While direct
Friedel–Crafts acylation using acids and furan is highly
desirable, thermodynamic calculations show that the
chemistry is equilibrium limited for long chains, resulting in
low yields (see ESI† for equilibrium estimations).
Due to the disadvantages of the current approaches,
herein, we propose a new synthesis of 2-alkoyl furan via
cross-ketonization of 2-furoic acid and lauric acid using
heterogeneous catalysts (Fig. 1, pathway C). The cross-
ketonization strategy offers ease of separation and
recyclability of solid catalysts and produces only water and
carbon dioxide as byproducts. It also does not suffer from
the equilibrium limitations of acylation (see ESI† for
equilibrium estimations).
Ketonization is an established C–C coupling reaction to
prepare symmetrical dialkyl ketones or cyclic ketones from
aliphatic acids or diacids, respectively. Common examples
include the ketonization of acetic acid to acetone8,9 and
adipic acid cyclization to cyclopentanone.9 Two different
carboxylic acids lead to self- and cross-ketonization, with the
former reducing the selectivity to the desired cross-
ketonization product.10–12 Early work on ketonization of
acetic acid was facilitated by alkali and alkaline earth oxides
via decomposition of the acetate salts, which forms via the
reaction of the acid with the bulk oxides.13,14 Recent catalyst
developments show that ketonization occurs on the surface
of zeolites and high-lattice-energy metal oxides, such as TiO2,
ZrO2, CeO2, their mixed oxides, and transition metal (e.g., Pd,
Ru, Co)-doped oxides.15–19 Catalyst properties that may affect
ketonization activity include, but are not limited to, redox,
acid–base properties, and interaction of the carboxylic acid
with the oxide surface. Comprehensive reviews on the
catalysis and ketonization mechanism of short-chain
carboxylic acids have been reported elsewhere.20,21 Extensive
mechanistic research has shown that α-hydrogen (α-H) in at
least one of the carboxylic acid molecules is critical for the
surface reaction to take place.15,22 The reaction rate increases
monotonically with the number of α-H atoms in the absence
of steric hindrance. For instance, the self-ketonization of
propionic acid (with two α-H) occurs ten times faster than
2-methyl propionic acid (with one α-H).10 Accordingly, one
may expect a slower reaction rate for cross-ketonization since
2-furoic acid does not contain an α-H. While most prior
ketonization studies focused on small gas-phase molecules,
the high boiling point of lauric acid requires a liquid
reaction. Therefore, the proposed cross-ketonization
chemistry entails challenges, namely (1) slow rate of cross-
ketonization, (2) lack of mechanistic insights, and (3) limited
related literature. This report demonstrates the selective
production of 2-dodecanoyl furan bio-surfactant precursor
from biomass-derived 2-furoic acid and vegetable-oil-derived
lauric acid, using earth-abundant metal oxide catalysts. The
reaction mechanism and catalyst active sites are elucidated
to guide further research on catalyst design.
Experimental section
Materials
2-Furoic acid and n-dodecanoic (lauric) acid were purchased
from Sigma Aldrich. Solvents n-dodecane, cyclohexane, and
dimethyl sulfoxide (DMSO) were acquired from Fisher
Scientifics. The metal oxides, including iron(II) oxide (FeO),
iron(III) oxide (Fe2O3), and iron(II,III) oxide (Fe3O4), titanium
oxide (TiO2), zirconium oxide (ZrO2), gamma aluminum oxide
(γ-Al2O3), and cerium oxide (CeO2) were obtained from Sigma
Aldrich. All chemicals were used as received.
Catalytic reaction and analysis
Reactions were carried out in a 100 mL Parr reactor vessel,
with a glass liner, containing 30 mL of n-dodecane solvent,
2-furoic acid and lauric acid as reactants, 0.2 g metal oxide
catalyst, under 20 bar N2 pressure, with lauric acid as the
limiting reagent. N-Dodecane was chosen as the solvent due
to its inertness, low vapor pressure, and the ability to
solubilize lauric acid and the desired product, 2-dodecanoyl
furan. The reactor was heated up to the desired temperature
in
a fitted band heater, using an inhouse made PID
controller, on a magnetic stir bar. After a designated time,
the reactor was quenched in an ice bath to stop the reaction.
Due to its low solubility in n-dodecane at room temperature,
2-furoic acid was extracted from the post-reaction solution by
DMSO for further quantification. The samples were then
filtered for further analysis. In recyclability experiments, the
spent catalysts were collected, washed with cyclohexane,
which can dissolve dodecane and quickly evaporate in air,
and air-dried at room temperature overnight before used for
the next catalytic experiment. Due to the 10–15% typical loss
of spent catalysts during filtering, washing, and handling,
three repeated experiments (run 1) were conducted with a
fresh catalyst to collect enough spent catalysts for the second
recycle (run 2), and two repeated experiments were conducted
for run 2 to collect enough spent catalysts for the third
recycle (run 3). Products were identified and quantified with
a
gas chromatogram-mass spectrometer (GC-MS) system
(Agilent 7890B) and with gas chromatogram-flame
a
ionization detector (GC-FID) instrument, respectively. Both
the GC-MS and GC-FID were equipped with an HP-1 column
(Agilent). When available, calibration curves from commercial
standards were used to quantify the concentrations of
reactants and products. Because a commercial standard of
2-dodecanoyl furan was unavailable, this product was
quantified using the effective carbon number (ECN)
method.23 The ECN of 2-dodecanoyl furan was determined
from the ECN of the commercially available 2-acetyl furan
and hexane. The method was verified with the agreement
between experimentally measured and calculated calibration
slope for 2-hexanoyl furan (see Fig. S1 and Table S1†).
This journal is © The Royal Society of Chemistry 2021
Catal. Sci. Technol., 2021, 11, 2762–2769 | 2763

View Article Online
Paper
Catalysis Science & Technology
Catalyst characterization
Surface area and porosity were determined from N2
physisorption data at 77 K using the Micromeritics ASAP
2020 instrument. The crystallinity of the fresh and spent
catalysts were measured by a Bruker D8 X-ray diffraction
(XRD) instrument equipped with a monochromatic Cu Kα1
line (λ = 0.154 nm) at 40 kV and 40 mA, operating in 2θ range
of 5–80° at a scanning rate of 0.02° s−1. Temperature
programmed reduction (TPR) by hydrogen was performed on
a thermogravimetric analysis (TGA) instrument (TA Q600 HT)
using 50 mL min−1 of 5% H2 in N2 as the flowing gas in the
25–1200 °C temperature range. In addition, H2-TPR was also
conducted in a downward flow reactor. X-ray photoelectron
spectra (XPS) of the catalysts were collected on a Thermo-
Fisher K-alpha XPS, equipped with Al-Kα X-ray
monochromatic source with 400 μm spot size. The spent
catalysts were handled in an air-free glovebox and transferred
to a vacuum transfer module (Thermo Scientific) before XPS
measurements to eliminate air exposure. Iron species
concentration in the post-reaction filtrates was measured by
ICP. One milliliter of the filtrate was heated in a vacuum oven
at 200 °C to evaporate the dodecane. The remaining solids
were dissolved in a 20 mL mixture containing 50 v/v%
ethanol, 2.5 v/v% HNO3 and 47.5 v/v% water, and sonicated
in a 50 °C water bath for 30 min. The mixture was further
diluted 50 times using a 5 v/v% HNO3 solution for ICP
analysis. Six standards containing 1 ppb, 10 ppb, 50 ppb, 100
ppb, 500 ppb, and 1000 ppb of Fe (SPEX CertiPrep) were used
for calibration. The Fe content in each sample was measured
Fig. 2 Performance of various commercial metal oxides in the reaction
of 2-furoic acid with lauric acid in n-dodecane. Reaction conditions:
0.05 M lauric acid, 0.25 M 2-furoic acid, 0.2 g catalyst, 316 °C, 90 min,
20 bar N2, 800 rpm. Pressure at reaction temperature = 47 bar.
90 min. Among these, the highest selectivity (up to 77% at
316 °C and 84% at 300 °C) to the cross-ketonization product,
based on the lauric acid conversion and 2-dodecanoyl yield,
was on iron oxides (FexOy). This demonstrates the feasibility
of cross-ketonization for the oleo-furan sulfonate production.
However, the 2-dodecanoyl furan yield is still lower than the
90% yield of the pathway A.7 The yield to 2-dodecanoyl over
iron oxides for three consecutive cycles at 316 °C (Fig. 4) was
within experimental error. These recyclability experiments
were conducted at near to full conversion of furoic acid to
assess yields. A detailed catalyst stability/deactivation study
in future work should be carried out at low conversions.
The post-reaction solutions were yellowish, and the total
GC carbon balance for 2-furoic acid was lower than 80%,
indicating the formation of organic deposits (see Fig. S5† for
carbon balance analysis). Di-furfuryl ketone was not detected
from 2-furoic acid self-ketonization, consistent with the
critical role of α-H in the ketonization and the fact that the
reaction takes place on the catalyst surface instead of in the
bulk, using alkali, alkaline earth, and rare earth metal
oxides,13,20 where an α-H is not required.
using
a Thermo Scientific iCAP triple quad-inductively
coupled plasma-mass spectrometer (TQ-ICP-MS). The results
were averaged from triplicates. 45Sc (VHG Lab product # LIS2-
100) was used as an internal standard to monitor the
machine drift throughout the measurement. 45Sc was used
due to its m/z ratio close to the target element Fe and was
presumably not present in the samples. Results with recovery
rates between 80 to 120% were considered valid.
Interestingly, the filtered solutions from the FeO-catalyzed
reactions turned reddish upon exposure to air, and iron
species were determined with ICP analysis (Table S2†). After
removing the spent FeO particles in the first cycle by
filtration, fresh 2-furoic acid and lauric acid were added to
the iron-containing filtrate solution, and the mixture was
brought to reaction conditions. 2-Dodecanoyl furan formed
Results and discussion
The catalytic activity of metal oxides for cross-ketonization of
2-furoic acid and lauric acid
We screened various commercial metal oxides with high
lattice energies, which are active for surface-catalyzed
ketonization,8,15,24,25 for the reaction of 2-furoic acid and
lauric acid in dodecane at 300 °C (Fig. S2†) and 316 °C
(Fig. 2), with the molar ratio of 5 (see ESI† for tests at other
2-furoic acid to lauric acid molar ratios (Fig. S3†), and test
for external mass transfer limitations in Fig. S4†).
GC-detected products included long-chain ketones,
2-dodecanoyl furan (desired product), and 12-tricosanone,
resulting from the cross- and self-ketonization of the lauric
acid, respectively, and furan, from the decarboxylation of
2-furoic acid, as proposed in Fig. 3. While the furan yield,
based on 2-furoic acid, was >30% on all catalysts, ketones
only formed over iron oxides (FexOy), ceria, and titania after
Fig. 3 Proposed reaction network for the reaction of 2-furoic acid
and lauric acid in n-dodecane by iron oxide catalyst. Only paths to
detectable products are shown. R represents the n-C11H23 group.
Other chains could also be used.
2764 | Catal. Sci. Technol., 2021, 11, 2762–2769
This journal is © The Royal Society of Chemistry 2021

View Article Online
Catalysis Science & Technology
Paper
00-02501402). After the reaction, the spectrum of Fe3O4
remained unchanged, the aforementioned additional XRD
peaks in the spent Fe2O3 disappeared, and the diffraction
pattern of the used FeO transformed to that of Fe3O4. In
short, all FexOy catalysts displayed the crystallinity pattern of
magnetite, suggesting reduction of Fe2O3 and oxidation of
FeO during the reaction. The same phenomenon was
reported for iron oxides supported on silica after acetic acid
ketonization.28 The drastic change in the iron oxides' surface
area and porosity may be attributed to their corresponding
oxidation state transformation.
Fig. 4 Lauric acid conversion (a) and 2-dodecanoyl furan yield (b) in
dodecane for three consecutive uses of the catalyst. Reaction conditions:
0.05 M lauric acid, 0.25 M 2-furoic acid, 0.2 g catalyst, 316 °C, 90 min, 20
bar N2, 800 rpm. Pressure at reaction temperature = 47 bar.
To better understand the reduction of the FexOy,
temperature-programmed reduction by H2 coupled with
thermogravimetric analysis were performed on the fresh iron
oxides. Standard reduction of Fe2O3 was previously
at 15% yield (Fig. S6†), suggesting that homogeneous iron
species (likely carboxylate complexes), created during the
reaction over FeO, are active for cross-ketonization. No
homogeneous iron species were detected in Fe2O3 and Fe3O4;
their post-reaction filtrates exhibited no activity. CeO2 is
promising but gives higher carbon loss of lauric acid (Fig.
S5†). Given their promising performance at the test
conditions, the commercial iron oxides were down-selected
to provide further insights into the cross-ketonization of fatty
acids and furoic acid.
postulated to take place via a 3-step mechanism: Fe2O3
→
Fe3O4 → FeO → Fe.29 However, the TPR–TGA profiles of the
commercial iron oxides indicate that the reduction is more
complex and involves multiple convoluted steps, in
agreement with other literature30–32 (Fig. S7†). Specifically,
partial reduction of Fe2O3 to Fe3O4 was observed in the 240–
416 °C temperature range. Total reduction to Fe0 metal
occurred in the range of 416–743 °C. Fe3O4 reduction took
place at 343–457 °C first, and complete reduction to Fe0
occurs at 457–984 °C. The cross-ketonization reaction
temperature is 316 °C is in the temperature range of Fe2O3
reduction to Fe3O4, as indeed observed by XRD. However, the
reaction temperature is lower than the reduction temperature
of Fe3O4, explaining the stability of the spent Fe3O4. Lastly,
the reduction of FeO was minimal for temperatures lower
than about 600 °C. It is well documented that FeO is
thermally unstable at temperatures lower than 570 °C and
the disproportionation reaction of FeO to Fe metal and Fe3O4
takes place in the temperature range of 250–600 °C without
hydrogen consumption,31 explaining the apparent oxidation
of FeO to Fe3O4 during the ketonization reaction. The
metallic Fe formed by disproportionation likely reacted with
the acid reagents to generate homogeneous iron(II)
carboxylates. These then oxidized to iron(III) to turn reddish
after exposure to air and were active for cross-ketonization,
as described above. Homogeneous iron species observed in
the filtrate after the second and third runs of FeO indicate
that the disproportionation was not complete during the
reactions; some FeO nanoparticles under the detection limit
of XRD (<5 nm) were still present in the spent FeO. Other
iron(II) salts, e.g., ferrocene (see Fig. S2†), are effective for
cross-ketonization.
Determination of catalyst active
centers
The BET surface area and pore volumes of the fresh iron
oxides increased in the order of FeO < Fe3O4 < Fe2O3 and
were overall the lowest (<35 m2 g−1 and <0.08 cm3 g−1)
among the screened catalysts, indicating that the materials
are almost nonporous and the catalytic reactions take place
mostly on their external surfaces (Table S3†). After the third
reaction cycle, the surface area and pore volume substantially
increased, remained unchanged, and decreased for FeO,
Fe3O4, and Fe2O3, respectively. The unexpected enhancement
of surface area by more than 2-orders of FeO indicated a
significant alteration in the catalyst morphology. XRD
patterns of the fresh iron oxides displayed their expected
polymorphs (Fig. 5). The XRD patterns of γ-Fe2O3
(maghemite) in the tetragonal structure and Fe3O4
(magnetite) are not distinguishable because both oxides have
the inverse spinel-type structure.26 However, some
maghemites contain other cations in the octahedral Fe
vacancies,27 giving additional XRD peaks in the fresh
commercial Fe2O3 of this work (matched with maghemite-Q-
In addition to XRD and H2 TPR–TGA techniques, we
employed XPS to investigate the changes of the catalyst
surfaces based on the binding energies of Fe 2p in fresh and
spent catalysts (Fig. 6). Interestingly, the binding energy of Fe
2p photoelectrons of all the fresh iron oxide catalysts is the
same at 711.0 eV for Fe 2p3/2 and 724.4 eV for Fe 2p1/2
,
corresponding to Fe3+ according to literature values.33 The
observed satellite peak at 719.4 eV, which is characteristic of
Fe2O3, suggests the oxidation of the Fe on the surface of the
Fig. 5 XRD pattern of the fresh and spent iron oxides FeO (a), Fe3O4
(b), Fe2O3 (c) with assigned diffraction peaks for FeO (■), Fe3O4 or
γ-Fe2O3 (*), and additional octahedral Fe cation peaks (♦).
This journal is © The Royal Society of Chemistry 2021
Catal. Sci. Technol., 2021, 11, 2762–2769 | 2765

View Article Online
Paper
Catalysis Science & Technology
yield. The XPS spectrum of Fe 2p of the pre-reduced Fe3O4 is
similar to that of the spent Fe3O4 after a standard reaction
experiment (Fig. S9†). Specifically, the Fe 2p3/2 binding
energy shifted to a lower value of 710.6 eV, with a shoulder at
709 eV, assigned to Fe2+ 2p3/2. The absence of a satellite peak
confirmed the generation of Fe3O4 on the surface. The
enhancement of Fe3O4 catalytic activity after H2 pretreatment
indicates that the newly formed Fe2+ centers on the catalyst
surface are responsible for cross-ketonization. The
improvement in the ketonization activity of Fe2O3 in the
presence of H2 has been reported for acetic acid
ketonization.34,35 We hypothesize that even though the
standard reaction conditions were under inert N2, either the
reagents, intermediates, or the solvent in Fig. 3 act as
reducing agents during the induction time, thereby reducing
the Fe3+ centers to generate a catalytically active surface for
ketonization. To better understand this finding, further
experiments were conducted. When the catalysts were
pretreated with a dodecane solution of 2-furoic acid or lauric
acid individually at 316 °C for 30 min prior to the catalytic
reaction, similar enhancements in catalytic activity were
observed at 20 min (Fig. 7).
Interestingly, an increased 2-dodecanoyl furan yield was
also noted when the catalyst was pretreated in dodecane
solvent alone, although the effect was not as pronounced as
in the cases of using lauric acid, 2-furoic acid, or H2 gas. XPS
measurements of all the pretreated Fe3O4 catalysts displayed
the Fe 2p spectra of Fe3O4 instead of the oxidized surface of
the fresh magnetite (Fig. S9†). After pretreating the catalyst in
dodecane at reaction conditions, traces of 2, 3, and
5-dodecanone were detected in GC-MS, suggesting oxidation
of dodecane. We propose that the solvent dodecane is
dehydrogenated and releases H2 or surface species that
reduce the catalyst surface. Lauric acid or 2-furoic acid in
dodecane probably promotes the formation of reducing
agents, such as H2 and CO, via dehydrogenation/
decarbonylation of the acid reactants. As a result, the cross-
ketonization activity when the catalysts are pre-reduced with
lauric acid or 2-furoic acid solution is higher than that with
pure dodecane. Iron-based oxides are effective in
dehydrogenation and oxidation,29,36 corroborating this
hypothesis. The detailed mechanism of reducing agent
formation during the reaction is outside the scope of this
work. H2, CO, or CO2, if formed, were not detected in GC-MS
analysis, likely due to their trace concentration in the inert
N2 headspace. A schematic of the changes in the oxidation
state of the Fe3O4 catalyst surface after pretreatment and
reaction is shown in Fig. S10.† The enhanced catalytic-activity
correlation with the catalyst surface reduction indicates that
a reduced iron oxide surface facilitates ketonization, and Fe2+
is the likely catalytically active center.
Fig. 6 XPS spectra of Fe 2p on the iron oxide catalyst surfaces (a)
before and (b) after the cross-ketonization reaction.
commercial magnetite and wustite, probably due to extended
exposure to air. After the reaction, clear shifts in Fe 2p3/2
peaks of the used catalysts to 710.6 eV are seen, a value in
between Fe 2p3/2 binding energy of Fe2O3 (711.0 eV) and FeO
(709.0 eV).
A shoulder at 709.0 eV, assigned to Fe 2p3 of FeO, was
observed in all XPS spectra of the spent catalysts. Therefore,
the catalyst surfaces of all FexOy catalysts were reduced from
Fe3+ to
a
mixture of Fe3+ and Fe2+ post-reaction. The
disappearance of the satellite peaks in the spent catalysts
provides further support that the catalyst surfaces contain
Fe3O4. Therefore, the initially oxidized catalyst surfaces of all
iron oxide catalysts are reduced to Fe3O4 during the reaction,
in agreement with the XRD results of the bulk FeO/Fe2O3.
Interestingly, a closer look at the reaction evolution over Fe3O4
shows no 2-dodecanoyl furan production during the first 20
min of the reaction followed by rapid formation (Fig. S8†).
This induction period suggests that the initial Fe2O3
surface (as measured by XPS) transformed to the catalytically
active form – potentially Fe3O4. To verify the hypothesis, we
reduced the commercial Fe3O4 with H2 at 316 °C for 30 min
in the dodecane solvent before using it for the cross-
ketonization reaction. The resulting material exhibited much
higher activity, i.e., 36% of 2-dodecanoyl furan yield (Fig. 7),
whereas the starting material resulted in no measurable
Fig.
7 Cross ketonization activity of Fe3O4 catalysts exposed to
different pretreatment conditions: (a) catalyst as received; catalyst
pretreated in a dodecane solution of (b) pure dodecane, (c) 0.05 M
lauric acid, (d) 0.25 M 2-furoic acid, and (e) 5 bar of H2. Pretreatment
conditions: 316 °C, 30 min, 0.2 g fresh Fe3O4. Reaction conditions:
0.05 M lauric acid, 0.25 M 2-furoic acid, 316 °C, 20 min, 20 bars of N2,
0.2 g Fe3O4.
In addition to the redox properties, the role of acidity of
the metal oxides was evaluated using pyridine adsorption IR
measurements on the catalysts (see Fig. S11 and ESI† for
more details). The iron oxides exhibited little acidity while
having the highest cross-ketonization activity. In contrast,
2766 | Catal. Sci. Technol., 2021, 11, 2762–2769
This journal is © The Royal Society of Chemistry 2021

View Article Online
Catalysis Science & Technology
Paper
Fig. 8 Proposed reaction mechanism of cross-ketonization of 2-furoic acid and lauric acid on Fe3O4.
TiO2 showed the highest acid density with no cross-
ketonization activity. Therefore, we proposed that the catalyst
acid density does not substantially affect the cross-
ketonization rate under our conditions.
carboxylates prior to the α-H abstraction from a basic oxygen
site. C–C coupling between the adsorbed carboxylates
generates the β-keto acid, thermally decomposing to the final
product, 2-dodecanoyl furan. Coordinately unsaturated metal
cations are necessary for surface carboxylate formation.15,18,40
As the density of the surface Fe2+ increases, so does the
probability of two adjacent carboxylates to form the β-keto
acid. A similar reaction mechanism has been proposed for
the vapor phase ketonization of propanoic acid to 3-penta-
none over CeO2–Mn2O3.10 In addition, the roles of redox
properties and surface oxygen vacancy on maintaining
ketonization activity have been demonstrated for other metal
oxides, such as CeO2, ZrO2, and TiO2.41,42 Numerous studies
have utilized metal doping of metal oxides and/or catalyst
pretreatment to create oxygen vacancies and unsaturated
metal cations and by doing so to ultimately increase
ketonization activity,10,18,43 suggesting a potential strategy for
improved activity in this system.
Reaction mechanism
Metal oxides facilitate surface-catalyzed ketonization via two
possible mechanisms.
A ketene (R2CCO) forms via
dehydration of the first carboxylic acid that contains an α-H,
which then couples with the second adsorbed carboxylate to
create the final ketone product.25,37 Alternatively, other
studies have provided evidence that a β-keto-acid, which has
a ketone group at the β position of the carboxylic acid and
can readily decarboxylate to form the ketone and CO2, is the
intermediate of ketonization.10,22,38,39
In the latter pathway, the α-H containing adsorbed
carboxylate undergoes α-H abstraction by a surface metal site
to form an anion radical, which attacks the adjacent second
adsorbed carboxylate to give
a β-ketoacid intermediate.
Conclusions
Accordingly, the carbonyl group of the ketone product is
inherited from the parent carboxylic acid of ketene in the
ketene-like mechanism or from the second adsorbed
carboxylate in the β-keto acid mechanism. Since only lauric
acid contains an α-H, isotopically labeled lauric acid with 13C
at the C1 position was employed to distinguish between the
two reaction paths. GC-MS fragment patterns of the 12-
tricosanone product show one unit shift to a higher m/z ratio
(Fig. S12b†) compared to that in the unlabeled lauric acid
(Fig. S12a†), suggesting a 13C in the structure of self-
ketonization of lauric acid. The GC-MS fragmentation of the
cross-ketonization product and the standard 2-dodecanoyl
furan are the same, indicating no 13C incorporated into the
2-dodecanoyl furan; the carbonyl group arises from the
2-furoic acid (Fig. S.12c†). The results support the β-keto acid
intermediate acid mechanism. Consequently, we postulate
the catalytic pathway illustrated in Fig. 8. Accordingly, the
surface Lewis acid Fe2+ sites facilitate the initial adsorption
of 2-furoic acid and lauric acid to their corresponding surface
In this work, we introduced a method to produce oleo-furan
surfactant precursors via cross-ketonization of biomass-
derived furoic acid and vegetable-oil-derived lauric acid to
2-dodecanoyl furan. The volatile byproducts and solid
catalysts can easily be separated from the desired product
stream. In addition, one can use fatty acids instead of their
anhydrides, without the typical thermodynamic limitations of
the acylation reaction. This aspect is expected to minimize
process waste.
Among the tested catalysts, earth-abundant, commercially-
available iron oxides are active and selective towards cross-
ketonization and were employed for further mechanistic
studies. Detailed catalyst characterization, structure–activity
relationships, and elucidation of reaction pathways provide
insights into the active catalyst centers, reaction mechanism,
and stability of the catalysts. XRD and XPS measurements of
the fresh and spent catalysts reveal that the bulk and surface
of all tested iron oxides transform to a stable Fe3O4 phase
This journal is © The Royal Society of Chemistry 2021
Catal. Sci. Technol., 2021, 11, 2762–2769 | 2767

View Article Online
Paper
Catalysis Science & Technology
during the reaction. Enhancement in the pre-reduced catalyst
surface reactivity leads us to hypothesize that surface Fe2+ is
critical for the reaction. 13C isotopic labeling indicates that
adjacently adsorbed laurate and furoate couple on the Fe2+
sites to form the corresponding β-ketoacid, which is then
decarboxylated to 2-dodecanoyl furan. The iron oxide
catalysts maintain activity for three consecutive cycles tested
herein. The insights obtained here can guide future work on
process and catalyst optimization for this new path to
renewable surfactants.
7 D. S. Park, K. E. Joseph, M. Koehle, C. Krumm, L. Ren, J. N.
Damen, M. H. Shete, H. S. Lee, X. Zuo and B. Lee, ACS Cent.
Sci., 2016, 2, 820–824.
8 M. Gliński, J. Kijeński and A. Jakubowski, Appl. Catal., A,
1995, 128, 209–217.
9 O. Neunhoeffer and P. Paschke, Ber. Dtsch. Chem. Ges.,
1939, 72, 919–929.
10 O. Nagashima, S. Sato, R. Takahashi and T. Sodesawa, J. Mol.
Catal. A: Chem., 2005, 227, 231–239.
11 B. Boekaerts and B. F. Sels, Appl. Catal., B, 2020, 283,
119607.
12 E. V. Fufachev, B. M. Weckhuysen and P. C. Bruijnincx, ACS
Sustainable Chem. Eng., 2020, 8, 11292–11298.
13 Y. Yamada, M. Segawa, F. Sato, T. Kojima and S. Sato, J. Mol.
Catal. A: Chem., 2011, 346, 79–86.
Author contributions
Dionisios G. Vlachos: conceptualization, supervision, writing
–
review
&
editing; Marat Orazov: conceptualization,
supervision, writing – review & editing; Hannah Nguyen:
methodology, validation, investigation, data curation, writing
– original draft; Yunzhu Wang: investigation, data curation;
David Moglia: investigation, data curation; Jiayi Fu:
investigation; Weiqing Zheng: methodology, validation.
14 V. Yakerson, L. Lafer and A. Rubinshtein, Bull. Acad. Sci.
USSR, Div. Chem. Sci., 1966, 15, 280–285.
15 R. Pestman, R. Koster, A. Van Duijne, J. Pieterse and V.
Ponec, J. Catal., 1997, 168, 265–272.
16 C. A. Gaertner, J. C. Serrano-Ruiz, D. J. Braden and J. A.
Dumesic, J. Catal., 2009, 266, 71–78.
17 H. Idriss and M. A. Barteau, in Advances in Catalysis,
Elsevier, 2000, vol. 45, pp. 261–331.
18 T. N. Pham, D. Shi, T. Sooknoi and D. E. Resasco, J. Catal.,
2012, 295, 169–178.
Conflicts of interest
There are no conflicts to declare.
19 A. Gumidyala, T. Sooknoi and S. Crossley, J. Catal.,
2016, 340, 76–84.
20 T. N. Pham, T. Sooknoi, S. P. Crossley and D. E. Resasco,
ACS Catal., 2013, 3, 2456–2473.
21 B. Boekaerts and B. F. Sels, Appl. Catal., B, 2021, 283, 119607.
22 A. Pulido, B. Oliver-Tomas, M. Renz, M. Boronat and A.
Corma, ChemSusChem, 2013, 6, 141–151.
23 J. T. Scanlon and D. E. Willis, J. Chromatogr. Sci., 1985, 23,
333–340.
24 K. Kim and M. Barteau, J. Catal., 1990, 125, 353–375.
25 S. D. Randery, J. S. Warren and K. M. Dooley, Appl. Catal., A,
2002, 226, 265–280.
26 A. M. Jubb and H. C. Allen, ACS Appl. Mater. Interfaces,
2010, 2, 2804–2812.
Acknowledgements
This work was supported as part of the Catalysis Center for
Energy Innovation, an Energy Frontier Research Center
funded by the US Dept. of Energy, Office of Science, Office of
Basic Energy Sciences under award number DE-SC0001004.
This research used Thermo Scientific iCAP TQ-ICP-MS in the
Advanced Materials Characterization Lab (AMCL) at the
University of Delaware. The authors want to acknowledge
Chin-Chen Kuo for ICP measurements and suggestions on
ICP sample preparations, Dr. Stavros Caratzoulas for assisting
with the Gaussian calculations of the thermochemistry
equilibrium, and Prof. Raul Lobo for useful discussions.
27 R. M. Cornell and U. Schwertmann, The iron oxides:
structure, properties, reactions, occurrences and uses, John
Wiley & Sons, 2003.
28 J. A. Bennett, C. M. Parlett, M. A. Isaacs, L. J. Durndell, L.
Olivi, A. F. Lee and K. Wilson, ChemCatChem, 2017, 9,
1648–1654.
Notes and references
1 N. Nikbin, S. Feng, S. Caratzoulas and D. G. Vlachos, J. Phys.
Chem. C, 2014, 118, 24415–24424.
2 T. A. Brandvold, US Pat., US20100331568A1, Honeywell UOP
LLC, 2012.
3 M. Peters, J. Taylor, D. E. Henton, L. E. Manzer, US Pat.,
EP2401307A1, Gevo Inc, 2012.
4 O. A. Abdelrahman, D. S. Park, K. P. Vinter, C. S. Spanjers, L.
Ren, H. J. Cho, D. G. Vlachos, W. Fan, M. Tsapatsis and P. J.
Dauenhauer, ACS Sustainable Chem. Eng., 2017, 5,
3732–3736.
5 M. Balakrishnan, E. R. Sacia, S. Sreekumar, G. Gunbas, A. A.
Gokhale, C. D. Scown, F. D. Toste and A. T. Bell, Proc. Natl.
Acad. Sci. U. S. A., 2015, 201508274.
29 F. G. E. Nogueira, J. H. Lopes, A. C. Silva, R. M. Lago, J. D.
Fabris and L. C. A. Oliveira, Appl. Clay Sci., 2011, 51,
385–389.
30 M. J. Tiernan, P. A. Barnes and G. M. Parkes, J. Phys. Chem.
B, 2001, 105, 220–228.
31 W. K. Jozwiak, E. Kaczmarek, T. P. Maniecki, W. Ignaczak
and W. Maniukiewicz, Appl. Catal., A, 2007, 326, 17–27.
32 H.-Y. Lin, Y.-W. Chen and C. Li, Thermochim. Acta, 2003, 400,
61–67.
6 D. Jadhav, A. M. Grippo, S. Shylesh, A. A. Gokhale, J.
Redshaw and A. T. Bell, ChemSusChem, 2017, 10, 2527–2533.
33 T. Yamashita and P. Hayes, Appl. Surf. Sci., 2008, 254,
2441–2449.
2768 | Catal. Sci. Technol., 2021, 11, 2762–2769
This journal is © The Royal Society of Chemistry 2021

View Article Online
Catalysis Science & Technology
Paper
34 A. A. Taimoor, A. Favre-Reguillon, L. Vanoye and I. Pitault,
Catal. Sci. Technol., 2012, 2, 359–363.
35 E. Karimi, A. Gomez, S. W. Kycia and M. Schlaf, Energy Fuels,
2010, 24, 2747–2757.
36 N. Mimura, I. Takahara, M. Saito, T. Hattori, K. Ohkuma
and M. Ando, Catal. Today, 1998, 45, 61–64.
37 T. S. Hendren and K. M. Dooley, Catal. Today, 2003, 85,
333–351.
39 A. V. Ignatchenko, J. Phys. Chem. C, 2011, 115, 16012–16018.
40 K. M. Dooley, A. K. Bhat, C. P. Plaisance and A. D. Roy, Appl.
Catal., A, 2007, 320, 122–133.
41 D. Martin and D. Duprez, J. Phys. Chem., 1996, 100,
9429–9438.
42 T. Yamaguchi, Catal. Today, 1994, 20, 199–217.
43 F. Fally, V. Perrichon, H. Vidal, J. Kaspar, G. Blanco, J.
Pintado, S. Bernal, G. Colon, M. Daturi and J. Lavalley, Catal.
Today, 2000, 59, 373–386.
38 J. Kuriacose and S. Jewur, J. Catal., 1977, 50, 330–341.
This journal is © The Royal Society of Chemistry 2021
Catal. Sci. Technol., 2021, 11, 2762–2769 | 2769