Chemoselective Hydrodeoxygenation of Carboxylic Acids to Hydrocarbons over Nitrogen-Doped Carbon-Alumina Hybrid Supported Iron Catalysts

Article abstract of DOI:10.1021/acscatal.8b04967

The establishment of catalyst systems for the chemoselective hydrodeoxygenation (HDO) of carboxylic acids to hydrocarbons, such as the HDO of long-chain fatty acids to alkanes, is important for biomass to biofuel conversion. As the most abundant and probably the cheapest transition metal on the earth, iron is a promising non-noble-metal alternative to precious metals for large-scale conversion of biomass. However, it usually suffers from unsatisfactory activity. In this work, a nitrogen-doped carbon-alumina hybrid supported iron (Fe-N-C@Al₂O₃) catalyst is established for chemoselective HDO of carboxylic acids to hydrocarbons. By using stearic acid HDO as the model reaction, n-octadecane and n-heptadecane are produced with yields of 91.9% and 6.0%, respectively. Triglycerides can also be converted into liquid alkanes with a total molar yield of >92%. In addition, the iron catalyst can chemoselectively catalyze the HDO of the carboxylic acid group in the presence of other functional groups such as an aromatic ring. This chemoselectivity has rarely been seen before because the aromatic ring is usually more easily hydrogenated in comparison to HDO of the carboxylic acid group. The characterization results showed that both the formation of a nitrogen-doped carbon-alumina hybrid and the iron loading are important for the Lewis basicity of these catalysts, in order to adsorb the acid substrates. The addition of melamine as the nitrogen precursor during pyrolysis eliminates undesired reactions between the iron precursor and alumina support to form an inactive hercynite phase, leading to the formation of an Fe₃C active phase for the hydrogenation of -COOH to -CH₂OH and the hybrid of N-C and alumina for the HDO of -CH₂OH to -CH₃.

Full text of DOI:10.1021/acscatal.8b04967

Subscriber access provided by UNIV OF LOUISIANA
Article
Chemoselective Hydrodeoxygenation of Carboxylic Acids to Hydrocarbons
over Nitrogen-doped Carbon-Alumina Hybrid-Supported Iron Catalysts
Jiang Li, Junjie Zhang, Shuai Wang, Guangyue Xu, Hao Wang, and Dionisios G. Vlachos
ACS Catal., Just Accepted Manuscript • DOI: 10.1021/acscatal.8b04967 • Publication Date (Web): 04 Jan 2019
Downloaded from http://pubs.acs.org on January 5, 2019
Just Accepted
“Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted
online prior to technical editing, formatting for publication and author proofing. The American Chemical
Society provides “Just Accepted” as a service to the research community to expedite the dissemination
of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in
full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully
peer reviewed, but should not be considered the official version of record. They are citable by the
Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore,
the “Just Accepted” Web site may not include all articles that will be published in the journal. After
a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web
site and published as an ASAP article. Note that technical editing may introduce minor changes
to the manuscript text and/or graphics which could affect content, and all legal disclaimers and
ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or
consequences arising from the use of information contained in these “Just Accepted” manuscripts.
is published by the American Chemical Society. 1155 Sixteenth Street N.W.,
Washington, DC 20036
Published by American Chemical Society. Copyright © American Chemical Society.
However, no copyright claim is made to original U.S. Government works, or works
produced by employees of any Commonwealth realm Crown government in the course
of their duties.

Page 1 of 45
ACS Catalysis
1
2
3
4
5
6
7
8
9
Chemoselective Hydrodeoxygenation of Carboxylic
Acids to Hydrocarbons over Nitrogen-Doped
Carbon-Alumina Hybrid-Supported Iron Catalysts
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
*
,†
†
†
‡
‡
Jiang Li, Junjie Zhang, Shuai Wang, Guangyue Xu, Hao Wang, and Dionisios G. Vlachos,
*
,§
†State Key Laboratory of Heavy Oil Processing, College of New Energy and Materials, China
University of Petroleum (Beijing), Beijing 102249, China
‡Anhui Province Key Laboratory of Biomass Clean Energy, Department of Chemistry,
University of Science and Technology of China, Hefei 230026, China
§Department of Chemical and Biomolecular Engineering and Catalysis Center for Energy
Innovation, University of Delaware, 221 Academy St., Newark, Delaware19716, United States
ABSTRACT: The establishment of catalyst systems for chemoselective hydrodeoxygenation
(HDO) of carboxylic acids to hydrocarbons, such as the HDO of long-chain fatty acids to alkanes,
is important for biomass-to-biofuel conversion. As the most abundant and probably the cheapest
transition-metal on the earth, iron is a promising non-noble alternative to precious metals for
large-scale conversion of biomass. However, it usually suffers from unsatisfactory activity. In
this work, nitrogen-doped carbon-alumina hybrid-supported iron (Fe-N-C@Al O ) catalyst is
2
3
ACS Paragon Plus Environment
1

ACS Catalysis
Page 2 of 45
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
established for chemoselective HDO of carboxylic acids to hydrocarbons. By using stearic acid
HDO as the model reaction, n-octadecane and n-hepadecane are produced with yields of 91.9%
and 6.0%, respectively. Triglycerides can also be converted into liquid alkanes with total molar
yield of >92%. In addition, the iron catalyst can chemoselectively catalyze the HDO of
carboxylic acid group in the presence of other functional groups such as an aromatic ring. This
chemoselectivity is rarely seen before because the aromatic ring is usually easier to be
hydrogenated than carrying out HDO of carboxylic acid group. The characterization results
showed that both the formation of nitrogen-doped carbon-alumina hybrid and the iron-loading
are important for the Lewis basicity of these catalysts, in order to adsorb the acid substrates. The
addition of melamine as nitrogen precursor during pyrolysis eliminates undesired reactions
between iron precursor and alumina support to form inactive hercynite phase, leading to the
formation of Fe C active phase for the hydrogenation of –COOH to –CH OH and the hybrid of
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
2
N-C and alumina for the HDO of -CH OH to –CH .
2
3
KEYWORDS: Biomass conversion, Hydrodeoxygenation, Carboxylic Acids, Iron, Nitrogen-
doping, Heterogeneous catalysis
1
. INTRODUCTION
Rapid depletion of fossil reserves, increasing energy demand and global warming concerns
prompt the utilization of sustainable sources such as biomass to produce chemicals and fuels.1,2
Triglycerides, which are esters of long-chain fatty acids and glycerol, are one of the most
important sustainable feedstocks for biofuel production. The first generation biodiesel, fatty acid
methyl ester (FAME), is produced by transesterification of triglycerides with methanol with an
ACS Paragon Plus Environment
2

Page 3 of 45
ACS Catalysis
1
2
3
4
5
6
7
8
9
3
annual manufacture of more than 20 million tons. However, its poor low-temperature flow
property and unsatisfactory thermal and oxidation stability limit its application as a high-grade
4
fuel in engines. The direct conversion of triglycerides into diesel-range alkanes (C15-C18), on
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
the other hand, has attracted significant attention because the as-prepared alkane biodiesel has no
oxygen, and its better low-temperature flow properties make it compatible with current engines.
To date, there are two approaches for the production of alkanes from triglycerides via
hydrodeoxygenation (HDO). The first involves the use of conventional sulfide CoMo and NiMo
catalysts,5-7 but suffers from contamination of the alkane products by unavoidable sulfur catalyst
leaching. The second employs sulfur-free supported metal catalysts. Lercher and coworkers
8
9
reported two impressive studies about using Ni/HBeta and Ni/ZrO catalysts to catalyze the
2
HDO of stearic acid to octadecane and n-heptadecane, respectively via two different reaction
1
0
11
12,13
14
pathways. Since then, noble metals such as Pd, Pt, Ru,
and Ir-ReOx, and non-noble
metals such as Ni,15-19 Co,20,21 W, Mo,
22
23,24
and Fe have been tested. In view of economic
25
considerations and abundance, non-noble metal catalysts are attractive. Among them, Fe is very
promising due to the anticipated large scale processing of biomass, and the facts that Fe is 3,000-
3
0,000 times more abundant and 20-150 times cheaper than other metals (figure 1). Currently,
the only study using Fe as metal center for the HDO of fatty acids was reported by Slowing et
al.25 Iron nanoparticles supported on mesoporous silica nanoparticles (Fe-MSN) afforded n-
octadecane and n-hepadecane with yields of 83% and 10%, respectively from oleic acid.
However, the Fe-MSN catalyst was ineffective in catalyzing the conversion of triglycerides such
as microalgal oil to liquid alkanes. 16% alcohols, 33% unsaturated hydrocarbons and 18%
saturated hydrocarbons are obtained at 67% conversion. Thus, it is still desirable to develop an
effective Fe-based catalyst system with improved HDO activity.
ACS Paragon Plus Environment
3

ACS Catalysis
Page 4 of 45
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 1. Abundance and price of some representative non-noble transition metals. The former
2
6
is reported in Ref. The prices of Fe, Co, Ni, Mo are quoted in London Metal Exchange (LME)
on May 22, 2018. The price of W is its export price in China on May 22, 2018.
Recently, nitrogen-doping has been reported to improve the catalytic activity of iron
catalyst. For example, numerous Fe-N-C catalysts have exhibited catalytic performance
comparable to commercial Pt/C catalyst towards electrochemical reactions such as the oxygen
27
reduction reaction. Notably, Beller et al. reported some pioneer work using Fe-phen catalysts,
o
which were prepared by the pyrolysis of iron-phenanthroline complexes on supports at 800 C,
towards organic synthesis such as nitroarenes reduction, oxidative dehydrogenation of N-
heterocycles, and hydrogen-free reductive aminations.28-30 For biomass conversion, we recently
found that a Fe-N-C catalyst is capable of carrying out catalytic transfer hydrogenation of
furfural, the HDO of 5-hydroxymethylfurfural, and the reductive cleavage of lignin-derived ether
linkage.31-33 Even though the exact nature of active sites of Fe-N-C catalyst is still not clear, Fe
28
NPs encapsulated in a nitrogen-doped carbon shell is usually observed. It is well known that
nanocarbon materials are ideal supports due to their unique physical and chemical properties,34-36
and the nitrogen dopant could introduce additional unpaired electrons to the conjugation of
37
graphitic  systems to facilitate electron transfer and substrate adsorption. Catalyst activity can
ACS Paragon Plus Environment
4

Page 5 of 45
ACS Catalysis
1
2
3
4
5
6
7
8
9
38
further be boosted by Fe/Fe C species. It should be noted that the activity of nitrogen-doped
3
iron catalysts greatly depends on the selection of precursors and the heat treatment process.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Scheme 1. Hydrodeoxygenation of carboxylic acids.
The graphitic carbon nitride (g-C N ) polymer, which can be easily prepared by simple
3
4
o
thermal polymerization of melamine, cyanamide, dicyanamide, urea, and thiourea at 550 C, has
recently drawn great attention because of its appealing electronic band structure, peculiar
3
9
physicochemical stability, and “earth-abundant” nature. The high N content (ca. 60%) and the
as-formed “nitrogen pots” with six nitrogen lone-pair electrons create ideal sites for metal
inclusion (such as Fe), forming active sites for chemical conversions. In this work, nitrogen-
doped carbon-alumina hybrid-supported iron (Fe-N-C@Al O ) catalysts, prepared by
2
3
simultaneous pyrolysis of iron acetylacetone (Fe(acac) ) and melamine onto alumina at
3
o
temperatures ranged from 300 to 1100 C, are demonstrated to be active towards selective HDO
of carboxylic acids to hydrocarbons (Scheme 1). By using stearic acid HDO as a model reaction,
the effect of pyrolysis temperature, support, nitrogen-doping, and iron loading are explored in
detail. Under optimal reaction conditions, n-octadecane and n-hepadecane are produced with
yields of 91.9% and 6.0%, respectively at quantitative conversion of stearic acid over Fe-N-
ACS Paragon Plus Environment
5

ACS Catalysis
Page 6 of 45
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
C@Al O -900 catalyst. In addition, triglycerides such as cyperus esculentus oil and jatropha oil
2
3
can also be converted into liquid alkanes with total weight yields (molar yields) of 78.8 wt%
(92.1%) and 81.8 wt% (95.2%), respectively. The iron catalyst can chemoselectively catalyze the
HDO of carboxylic acid group in the presence of other functional groups, such as an aromatic
ring. This chemoselectivity is rarely seen because hydrogenation of the aromatic ring is usually
easier than the HDO of carboxylic acid group.40-42 To our knowledge, the systematic study of
chemoselective HDO of arylcarboxylic acid has not been reported yet. Only a few studies on the
conversion of benzoic acid to toluene have been reported. Cu/Al O and LiAlH with Ti catalysts
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
3
4
afford toluene with yields of 40% and 61%, respectively.43,44 The ZnO catalyst generated toluene
as by-product during the conversion of benzoic acid to benzaldehyde, suffering from high
o
45
reaction temperature (>400 C) and energy-consuming gasification of benzoic acid. Thus, our
work provides the first comprehensive investigation of chemoselective conversion of carboxylic
acid groups to methyl groups in the presence of other functional groups such as aromatic rings.
The catalyst is characterized by various techniques. The addition of melamine as N precursor
during pyrolysis is important for the improved HDO activity of iron catalysts. It eliminates
undesired reactions between iron precursor and alumina support to form inactive hercynite
phase, leading to the formation of Fe C active phase for the HDO reaction especially the
3
hydrogenation of –COOH to –CH OH. The Lewis basicity of the catalyst favors adsorption of
2
the acid substrates to form carboxylate intermediates, which is influenced by the formation of
nitrogen-doped carbon-alumina hybrid and iron-loading. The formation of hybrid also increased
the ratio of Bronsted acid/Lewis acid sites of Al O to enhance its activity towards HDO of –
2
3
CH OH to –CH .
2
3
2. EXPERIMENTAL SECTION
ACS Paragon Plus Environment
6

Page 7 of 45
ACS Catalysis
1
2
3
4
5
6
7
8
9
2
.1 List of chemicals
Iron(III) acetyl acetone (98%), metal oxide supports such as α-Al O (30 nm, hydrophilic,
2
3
99.9%), TiO (99%) and SiO (99.9%), dodecane (98%), decane (98%), myristic acid (98%),
2
2
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
tetradecane (99%), n-octane (99.5%) and salicylic acid (99%) were purchased from Aladdin
Reagent Co. Ltd. Iron(III) nitrate nonahydrate (98.5%), ethanol (99.5%) and benzoic acid (99%)
were purchased from Sinopharm Chemical Reagent Co. Ltd. Urea (99%), toluene (99.5%), o-
xylene (98.5%), p-xylene (98%) were purchased from Guangfu Chemical Reagent Co. Ltd.
Dicyandiamide (98%), 1,10-phenanthroline monohydrate (99%), melamine (98%), activated
carbon (99%), octadecane (98%), 1-octadecene (90%), octadecanol (98%), octadecanal (95%),
phenol (99.5%), octoic acid (98%), p-phthalic acid (99%), o-phthalic acid (99%), p-
chlorobenzoic acid (99%), suberic acid (99%), α,α,α-trifluoro-p-toluic acid (98%), 4-tert-
butylbenzoic acid (99%), p-hydroxybenzoic acid (99%), p-anisic acid (99%), p-fluorobenzoic
acid (98%), 3-phenylpropanoic acid (98%), 4-chlorotoluene (98%), 4-fluorotoluene (99%), 4-
methylbenzotrifluoride (98%), 4-tert-butyltoluene (95%), 4-methoxytoluene (98%), p-cresol
(
99%) and o-cresol (99%) were purchased from TCI.
.2 Catalyst preparation
The iron catalysts were prepared by simultaneous pyrolysis of iron and nitrogen precursors
2
onto various supports under Ar at 300-1100 °C. The iron loading is 20 wt% unless otherwise
specified. Iron precursors such as iron(III) acetylacetonate and iron(III) nitrate, nitrogen
precursors such as melamine, dicyandiamide, urea, and 1,10-phenanthroline, supports such as
Al O , TiO and SiO and activated carbon were examined in this work. Typical preparation is as
2
3
2
2
follows: a certain amount of iron(III) acetylacetonate was firstly added to 100 mL ethanol at
room temperature. After the precursor was completely dissolved, melamine was added into the
ACS Paragon Plus Environment
7

ACS Catalysis
Page 8 of 45
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
solution. Then 1 g Al O was added into the mixture and the solution was stirred at 60 °C for 16
2
3
h followed by evaporation at 40 °C to remove the ethanol solvent. The obtained solid was dried
at 60°C overnight. Then the mixture was grounded into fine powder and pyrolyzed in a tubular
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
-1
furnace with an argon flow rate of 100 mL min . To prepare the best iron catalyst, the following
-1
furnace temperature program was used: 20 °C hold for 60 min, ramp 10 °C min to 550 °C and
-1
hold for 3 h, then ramp 5 °C min to 900 °C and hold for 2 h. To prepare Fe-N-C@Al O -900
2
3
catalyst with different iron loadings, the amount of melamine was changed accordingly to keep
the Fe/N ratio. The iron catalysts prepared from different batches exhibited the same catalytic
performance, demonstrating that our preparation method is highly reproducible.
2.3 Catalyst characterization
The morphology of the catalysts was characterized by Hitachi SU8010 scanning electron
microscopy (SEM, Japan) at 200 kV. TEM, HAADF-STEM, and the energy dispersive
spectrometer (EDS) mapping were performed on a JEOL JEM-2100F high-resolution
transmission electron microscope operated at 200 kV. N adsorption measurements were
2
conducted using an ASAP2020M adsorption analyzer which reports the adsorption isotherm,
specific surface area, and pore volume automatically. The surface areas of catalysts were
calculated using the BET method in the range of relative pressures between 0.05-0.20. The pore
size and distribution were calculated from the adsorption branch of the isotherms using the BJH
method. XRD patterns were recorded on an X-ray diffractometer (TTR-III, Rigaku Corp., Japan)
with Cu Kα radiation (λ=1.54056 Å). The data were recorded over 2θ ranges of 10-70°. XPS was
obtained using an X-ray photoelectron spectroscopy (ESCALAB 250Xi, Thermo-VG Scientific,
USA) with monochromatized Al Kα radiation (1486.92 eV). The nitrogen content was detected
by elemental analysis (Eurovector EA 3000). The iron content was detected by ICP-AES
ACS Paragon Plus Environment
8

Page 9 of 45
ACS Catalysis
1
2
3
4
5
6
7
8
9
(
Optima 7000DV, PerkinElmer Inc.). For NH and CO -TPD tests, approximately 100 mg
3
2
sample was loaded in a quartz reactor and then heated at 500 °C under argon flow for 2 h. Then
o
the adsorption of NH or CO was carried out at 40 C for 1 h. Subsequently, the catalysts were
3
2
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
flushed with argon at 40 °C for 1 h, and then heated to 700 or 800 °C with a heating ramp rate of
o
-1
1
0 C min . The desorbed NH or CO was measured by a gas chromatograph equipped with a
3 2
thermal conductivity detector (TCD). Raman spectra were measured on a Raman spectrometer
Renishaw, λ= 532 nm). FTIR spectra were recorded on a Nicolet 8700 FTIR spectrometer in the
(
-1
wavenumber range from 400 to 4000 cm . To confirm the chemical adsorption of stearic acid on
Fe-N-C@Al O -900 catalyst surface, the treated catalyst was prepared as in our previous
2
3
4
6
studies. 0.1 g of Fe-N-C@Al O -900 catalyst was added into 20 mL of a 0.05 M n-hexane
2
3
solution of stearic acid, and the mixture was stirred at room temperature for 12 h. Then the solid
was separated by centrifugation by washing it 15 times with n-hexane. The residue was dried at
80 °C overnight under N . Pyridine adsorption IR (Py-IR) is used to explore the acid types. A 10
2
mg tableting sample was first pretreated at 400 °C under vacuum for 2 h. After cooling down to
room temperature, pyridine was adsorbed for 10 min, and the temperature was increased to
−
3
2
00 °C to desorb for 0.5 h (1×10 Pa) and then increased to 350 °C to detect the medium and
strong acid sites.
2
.4 Experimental procedure
The catalytic HDO of carboxylic acids was carried out in a 50mL stainless reactor
purchased from Anhui Kemi Machinery Technology Co., Ltd. For a typical reaction, 0.5 mmol
stearic acid, 100 mg heterogeneous catalyst, and 20 mL dodecane solvent were loaded into a
quartz lining in the reactor. The reactor was purged by hydrogen for 3 times, and then purged
with 4 MPa H at room temperature. The reaction was carried out at 320 °C for 8 h with a
2
ACS Paragon Plus Environment
9

ACS Catalysis
Page 10 of 45
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
stirring speed of 800 rpm. After reaction, the gaseous phase was analyzed by gas
chromatography (GC). A ShinCarbon ST 80/100 packed column (Restek) and a thermal
conductivity detector (TCD) were used to determine the yields of H , CO, CO and CH . A Plot
2
2
4
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Q column and a flame ionization detector (FID) were used to determine the yields of gaseous
hydrocarbons such as CH , C H , and C H . The liquid products were collected, and 0.5 mmol
4
2
6
3
8
eicosane was added into the solution as internal standard. The products were analyzed using gas
chromatography (GC) and gas chromatography-mass spectrometry (GC-MS). GC-MS analysis
was conducted by an Agilent 7890B Gas Chromatograph equipped with a HP-5MS 30 m × 0.25
mm × 0.25 μm capillary column (Agilent). The GC was directly interfaced to an Agilent 5977
mass selective detector (EI, 70 eV). Some typical GC oven temperature programs were listed as
-1
follows: (1) For stearic acid, 210 °C hold for 2 min, ramp 20 °C min to 300 °C and hold for 2
-1
min;(2) For other substrates, 40 °C hold for 2 min, ramp 10 °C min to 150 °C, then ramp 20 °C
-1
min to 300 °C and hold for 2 min. Representative GC spectra are shown in supporting
information (figure S1 and S2).
Some important experiments were performed at least twice to ensure reproducibility. The
carbon loss can be attributed to the formation of undetected products in GC or coke formation.
The conversion and yield for the HDO of carboxylic acids were calculated by mol% as follows:
푚표푙푎푟 푎푚표푢푛푡 표푓 푐푎푟푏표푥푦푙푖푐 푎푐푖푑푠 푎푓푡푒푟 푟푒푎푐푡푖표푛
Conversion = 1 ―
× 100%
(
푚표푙푎푟 푎푚표푢푛푡 표푓 푐푎푟푏표푥푦푙푖푐 푎푐푖푑푠 before 푟푒푎푐푡푖표푛)
푚표푙푎푟 푎푚표푢푛푡 표푓 푒푎푐ℎ 푝푟표푑푢푐푡 푎푓푡푒푟 푟푒푎푐푡푖표푛
Yield =
× 100%
푚표푙푎푟 푎푚표푢푛푡 표푓 푐푎푟푏표푥푦푙푖푐 푎푐푖푑푠 before 푟푒푎푐푡푖표푛
Some carboxylic acids are insoluble in dodecane, so that additional procedures are required
to determine the conversions in some cases in table 2. (1) For stearic acid, benzoic acid, 4-tert-
butylbenzoic acid, and 3-phenylpropanoic acid (entries 1, 5, 9, and 13 in table 2), no additional
ACS Paragon Plus Environment
1
0

Page 11 of 45
ACS Catalysis
1
2
3
4
5
6
7
8
9
procedures are required; (2) For p-fluorobenzoic acid, p-chlorobenzoic acid, and o-phthalic acid
(entries 6, 7, and 14 in table 2), the production solution was diluted with THF to 100 mL to
dissolve all the acid substrates; (3) For α,α,α-trifluoro-p-toluic acid, p-hydroxybenzoic acid, p-
anisic acid, and p-phthalic acid (entries 8, 10, 11, and 15 in table 2), the acid substrates are
extracted from product solution by N,N-dimethylformamide (DMF) after the determination of
yields of alkane products, and phenylethyl alcohol is used as internal standard to determine the
conversions; (4) For myristic acid (entry 2 in table 2), decane is used as the solvent instead of
dodecane to avoid the interference of the determination of yields of alkane products by the
solvent; (5) For octoic acid and suberic acid (entries 3 and 4 in table 2), a control experiment
without addition of substrate is performed to determine the yields of alkanes derived from the
cracking of the solvent; (6) For salicylic acid (entry 12 in table 2), the conversion is not
determined because the retention time of substrate is overlapped with the solvent on a HP-5MS
column. It cannot be detected on a DB-WAX column because of high boiling point and high
polarity.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
The weight yield and total molar yields of liquid alkanes products for the HDO of plant oil
were calculated as follows:
푤푒푖푔ℎ푡 표푓 푒푎푐ℎ 푝푟표푑푢푐푡 푎푓푡푒푟 푟푒푎푐푡푖표푛
Yield(wt%) =
∗ 100%
푤푒푖푔ℎ푡 표푓 푐푦푝푒푟푢푠 푒푠푐푢푙푒푛푡푢푠 표푖푙 before 푟푒푎푐푡푖표푛
푡표푡푎푙 푚표푙푎푟 푎푚표푢푛푡푠 표푓 푙푖푞푢푖푑 푎푙푘푎푛푒푠 푎푓푡푒푟 푟푒푎푐푡푖표푛
Total yield(mol%) =
∗ 100%
푚표푙푎푟 푎푚표푢푛푡푠 표푓 푐푎푟푏표푥푦푙푖푐 푎푐푖푑 푔푟표푢푝푠 before 푟푒푎푐푡푖표푛
3
. RESULTS AND DISCUSSION
3
.1 Catalyst characterization
ACS Paragon Plus Environment
1
1

ACS Catalysis
Page 12 of 45
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 2. TEM image (a), HAADF-STEM (b), Fe, N, C, O, Al, and combined elemental maps
(c-h) of Fe-N-C@Al O -900 catalyst.
2
3
TEM images of various Fe-N-C@Al O catalysts are shown in the Supporting Information
2
3
(
figure S4-S8). For Fe-N-C@Al O -300 catalyst, Fe atoms are homogeneously dispersed on the
2 3
alumina flake without the formation of nanoparticles (NPs), as confirmed by the EDX
o
spectroscopy (figure S4). Raising the pyrolysis temperature to 550 C led to the formation of a
structure where Fe NPs were encapsulated by a carbon shell. The Fe NPs are confirmed by EDX
spectroscopy (figure S5), and the average particle size is 28.63 nm. When the pyrolysis
o
temperature was further raised to 700 C, cavities appeared and the average particle size
increased to 52.03 nm (figure S6). Surprisingly, it seems that only alumina flakes and amorphous
carbon can be observed in Fe-N-C@Al O -900 and Fe-N-C@Al O -1100 catalysts (figure S7
2
3
2
3
and S8). It is hard to distinguish Fe NPs, alumina, and carbon flakes in the darker areas in the
TEM images.
To get a clearer realization of the morphology of Fe-N-C@Al O -900 catalyst, EDX
2
3
ACS Paragon Plus Environment
1
2

Page 13 of 45
ACS Catalysis
1
2
3
4
5
6
7
8
9
elemental mapping was further performed (figure 2). The brightest particles in HAADF-STEM
image are Fe NPs as confirmed by Fe elemental map. The carbon and alumina flake can easily be
distinguished by C and Al elemental maps. In contrast, it is hard to distinguish them in the TEM
images. N atoms are found in the carbon and alumina flake, confirming the formation of
nitrogen-doped carbon and alumina. The Fe Nps are supported onto both carbon and alumina
flakes.
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
ACS Paragon Plus Environment
1
3

ACS Catalysis
Page 14 of 45
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 3. XRD patterns of various iron catalysts. (a) Fe-N-C@Al O catalysts prepared at
2
3
different pyrolysis temperatures and Fe-Al O -900 catalyst. (b) Iron catalysts prepared on
2
3
different supports with and without nitrogen-doping. (c) Fe-N-C@Al O catalysts with different
2
3
iron loadings.
The XRD patterns of iron catalysts prepared at different pyrolysis temperatures are shown
in figure 3a. Only characteristic peaks of alumina support (corumdum, syn, PDF#10-0173) are
found in the XRD pattern of Fe-N-C@Al O -300 catalyst. The metallic iron phase can be
2
3
o
detected at a pyrolysis temperature of 550 C. Its intensity significantly increases when the
o
pyrolysis temperature is raised to 700 C. Besides the metallic iron phase, weak peaks of Fe C
3
phase (PDF#89-7271) are detected in the XRD pattern of Fe-N-C@Al O -900 catalyst (see
2
3
o
figure S10 for a clearer observation of XRD pattern in the 2 range of 40-50 ). The absence of
peaks for FeNx species is probably due to the quite low contents of N (0.19 wt% vs 11.34 wt%
of C, table S1) and homogeneous dispersion without the formation of crystalline phase. The
o
peaks of Fe C phase disappear when the pyrolysis temperature is further raised to 1100 C. The
3
XRD pattern of Fe-Al O -900 catalyst show that the iron precursor reacts with the alumina
2
3
support to form hercynite phase in the absence of nitrogen precursor during the pyrolysis.
ACS Paragon Plus Environment
1
4

Page 15 of 45
ACS Catalysis
1
2
3
4
5
6
7
8
9
The XRD patterns of iron catalysts prepared using different supports with and without
nitrogen-doping are shown in figure 3b. The iron precursor reacts with metal oxide supports
such Al O , TiO and SiO to form hercynite, FeTiO , and fayalite phases, respectively. The
2
3
2
2
3
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
formation of these phases can be avoided by the addition of nitrogen precursor such as melamine
during pyrolysis. Intense peaks of metallic iron phase can be found in the Fe-N-C@TiO and Fe-
2
N-C@SiO catalysts. For the Fe-N-C@Al O -900 catalyst, peaks of Fe C phase are also
2
2
3
3
observed besides metallic iron phase. When activated carbon was used as support, intense peaks
of metallic phase are always observed irrespective of whether the nitrogen precursor is added. In
addition, extremely weak peaks of Fe O species are formed without nitrogen precursor, and
3
4
extremely weak peaks Fe C phase are detected in the presence of nitrogen precursor.
3
The XRD patterns of Fe-N-C@Al O catalysts with different Fe loadings are shown in
2
3
figure 3c. Hercynite is the only iron species at 3 and 5wt% Fe loading, and the color of the
catalysts is light grey (figure S11). Metallic iron can be found in the XRD patterns at iron
loadings above 10 wt%. The peaks of hercynite phase completely disappeared at iron loading of
2
0 wt%, and weak peaks of Fe C phase can be detected. The color of the iron catalysts turn
3
darker with increasing iron loading (figure S11).
ACS Paragon Plus Environment
1
5

ACS Catalysis
Page 16 of 45
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
Figure 4. FT-IR spectra of (Fe)-N-C@Al O -900 catalysts with and without treating with stearic
2
3
acid.
o
It is well known that the pyrolysis of melamine at 550 C leads to the formation of graphitic
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
carbon nitride (g-C N ). Since characteristic bands of g-C N are found in the IR spectra of Fe-
3
4
3
4
N-C@Al O catalysts (figure S12), g-C N was probably formed during the catalyst
2
3
3
4
4
7
preparation. In addition, the interaction between (Fe)-N-C@Al O -900 catalysts and stearic
2
3
acid is further revealed using FT-IR (figure 4). The characteristic peaks at 2954, 2918, 2850, and
467 cm−1 correspond to asymmetric stretching vibration of C−H in −CH , asymmetric
1
3
stretching vibration of C−H in −CH −, symmetric stretching vibration of C−H in −CH −, and
2
2
deformation vibration of C−H of alkyl chain in stearic acid. These peaks also exist in the FT-IR
spectra of (Fe)-N-C@Al O -900 catalysts upon stearic acid adsorption, confirming its adsorption
2
3
onto the support and the iron catalyst. This adsorption is chemsorption rather than physisorption
−
1
because the characteristic peak of carboxyl group in stearic acid at 1705 cm is missing in the
−1
FT-IR spectra of (Fe)-N-C@Al O -900 catalysts, and instead, a weak new peak at 1563 cm ,
2
3
1
3
which could be assigned to the formation of carboxylate, is observed. The carboxylate is a key
intermediate in the hydrogenation of carboxylic acid to aldehyde during stearic acid HDO.9
ACS Paragon Plus Environment
1
6

Page 17 of 45
ACS Catalysis
1
2
3
4
5
6
7
8
9
Figure 5. CO -TPD profiles of Al O , N-C@Al O -900, Fe-Al O -900, and Fe-N-C@Al O -900
2
2
3
2
3
2
3
2
3
catalysts.
The Lewis acidity and basicity of Fe-N-C@Al O -900 catalyst were explored using NH -
2
3
3
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
and CO -TPD, respectively. The NH -TPD profile (figure S13) indicates that the Fe-N-
2
3
C@Al O -900 catalyst has no Lewis acidity. By comparison, the two intense peaks at 100-300
2
3
o
and 550-800 C in the CO -TPD profile (figure 5) suggest that the catalyst possesses both weak
2
and strong Lewis basicity. This basicity facilitates chemisorption of carboxylic acid, as
o
corroborated by FI-IR in figure 4. A much weaker peak at 100-300 C observed in the CO -TPD
2
profile of Fe-Al O -900 catalyst, and the shift of the peak for strong Lewis basicity to lower
2
3
temperature, indicate that the formation of nitrogen-doped carbon-alumina hybrid leads to
o
stronger Lewis basicity. The broad peak at 160-610 C in the CO -TPD profile of N-C@Al O -
2
2
3
9
00 catalyst indicates that it has medium Lewis basicity. Thus, the Lewis basicity of Fe-N-
C@Al O -900 catalyst is controlled by both the formation of the nitrogen-doped carbon-alumina
2
3
hybrid and the iron-loading.
Figure 6. N 1s XPS spectra of Fe-N-C@Al O catalysts prepared at different pyrolysis
2
3
temperatures.
ACS Paragon Plus Environment
1
7

ACS Catalysis
Page 18 of 45
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
The N 1s XPS spectra of Fe-N-C@Al O catalysts are shown in figure 6 and S14. The N 1s
2
3
XPS spectrum of Fe-N-C@Al O -300 was displayed separately (figure S14) because its N
2
3
content (23.26 at%, detected by XPS) is much higher than that of other catalysts (no more than
.2 at%), leading to much more intense N peaks (8.5-28 times higher than that of Fe-N-
C@Al O -550 and Fe-N-C@Al O -1100 catalysts, respectively). Other N 1s spectra will be
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
5
2
3
2
3
hardly distinguished if they are co-displayed with the spectrum of Fe-N-C@Al O -300 catalyst.
2
3
The N 1s spectra of N-C materials are usually fitted into four species: pyridinic N at 398.3-
98.6 eV, pyrrolic N at 400.5 eV, graphitic N at 401.2 eV, and oxidized N at 402-405 eV.48,49 In
3
addition, a peak at 399.7 eV can also be observed in Fe-N-C@Al O catalysts prepared under
2
3
o
550 C, which is assigned to bridge N in g-C N (tertiary N bonded to carbon atoms in the form
3
4
47
of N–(C) or H–N–(C) ). The possible N species and corresponding contents are shown in table
3
2
o
S2. The increase of pyrolysis temperature from 300 to 900 C led to a significant decrease of the
pyridinic N and bridge N and an increase in graphitic N. Oxidized N appeared when the
o
pyrolysis temperature reached 1100 C, but the signal to noise ratio is quite low because of
extremely low N content (entry 8, table 1).
The C 1s XPS spectra are shown in figure S16. Graphitic C is observed as the only C
o
species for Fe-N-C@Al O catalysts prepared at pyrolysis temperature above 550 C. For Fe-N-
2
3
2
47
C@Al O -300 catalyst, the C peak at 288.17 eV was identified as sp -bonded carbon (N–C=N),
2
3
suggesting that graphitic carbon nitride is formed during the pyrolysis, consistent with FTIR
results in figure S12.
ACS Paragon Plus Environment
1
8

Page 19 of 45
ACS Catalysis
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 7. Raman spectra of Fe-N-C-900 and some alumina-based catalysts.
The Raman spectra of Fe-N-C-900 and some alumina-based catalysts indicated the
presence of a nitrogen-doped carbon-alumina hybrid (figure 7). The two peaks at ca. 1360 and
-1
[50]
1
390 cm are characteristic of -Al O . The Raman spectra remained unchanged after single
2
3
nitrogen-doping or Fe loading. By contrast, the simultaneous loading of Fe and N-C showed
typical peaks of D and G bands of carbon. In addition, the characteristic peaks of -Al O
2
3
significantly changed because they overlap with the D band of carbon.
Table 1. The nitrogen contents and N adsorption/desorption data of various catalysts.
2
Entry Catalysts
N
BET surface area
m²/g]
Pore Volume
[cm³/g]
Pore Size
[nm]
a
content
[
[
-
-
wt%]
1
2
3
4
5
Al O
61.3
6.15
8.39
14.0
56.7
0.46
0.02
0.03
0.06
0.21
29.67
14.68
13.17
3.66
2
3
Fe-Al O -900
2
3
N-C@Al O -900
2.93
7.20
0.80
2
3
Fe-N-C@Al O -300
2
3
Fe-N-C@Al O -550
3.77
2
3
ACS Paragon Plus Environment
1
9

ACS Catalysis
Page 20 of 45
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
6
7
8
9
1
1
Fe-N-C@Al O -700
0.35
0.19
0.08
0.50
2.10
0.73
54.6
46.9
35.9
704
0.22
0.19
0.17
0.63
0.08
0.73
3.74
3.76
3.69
4.77
9.61
25.49
2
3
Fe-N-C@Al O -900
2
3
Fe-N-C@Al O -1100
2
3
Fe-N-C-900
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
0
1
Fe-N-C@TiO -900
31.7
114
2
Fe-N-C@SiO -900
2
aDetermined by elemental analysis.
The nitrogen content and N adsorption/desorption data of various catalysts are shown in
2
table 1. The N content of Fe-N-C@Al O catalysts continually decreased with increasing
2
3
2
−1
pyrolysis temperature. The alumina support has a BET surface area of 61.3 m g , a pore
3
−1
volume of 0.46 cm g , and a pore width of 29.67 nm (entry 1). Loading of Fe or N-C leads to
significant decrease in BET surface area, pore volume, and pore size (entries 2 and 3). By
o
comparison, the simultaneous loading of Fe and N-C at 900 C afforded much higher BET
surface area and pore volume (entry 7). In addition, the pore size was further decreased to 3.76
nm. The N sorption isotherm of Fe-N-C@Al O -900 catalyst (figure S17) can be classified as a
2
2
3
type IV isotherm with a nitrogen uptake at low relative pressures and an H3 hysteresis loop
around a relative pressure of 0.4–1.0 P/P , suggesting that the catalyst possesses well-defined
0
51
mesopores as confirmed by the pore distribution. The hysteresis loop could also be observed in
the N sorption isotherm of Fe-N-C@Al O catalysts prepared at other pyrolysis temperatures
2
2
3
(figure S17). The BET surface area of iron catalyst prepared on carbon support was much higher
than those prepared on metal oxides.
3
.2 Catalyst screening
ACS Paragon Plus Environment
2
0

Page 21 of 45
ACS Catalysis
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 8. HDO of stearic acid over various iron catalysts. Reaction conditions: 0.5 mmol stearic
o
acid, 20 mL solvent, 0.1 g catalyst, 4 MPa H , T=320 C, and t=8 h. n-C17=heptadecane.
2
C18=octadecene + octadecane. C18-OH= 1-octadecanol. C17-CHO=1-octadecanal. Trace yields
of octadecene are observed in most cases except for Fe-N-C@Al O -300 and 550 catalysts.
2
3
The iron catalysts are first tested for the HDO of stearic acid to liquid alkanes as a model
reaction (figure 8). A glass vial was used to exclude influence from the reactor body. The blank
test showed that the HDO reaction is unlikely to occur without any catalyst. The Fe-Al O -900
2
3
catalyst exhibited poor HDO activity, affording only 1.9% yield of heptadecane (n-C17) as the
major liquid alkane product with 12.3% yield of 1-octadecanal as the major by-product. This
poor HDO activity indicated that the hercynite species, the only iron species in the XRD pattern
of Fe-Al O -900 catalyst, formed by the reaction of iron precursor with the alumina support in
2
3
ACS Paragon Plus Environment
2
1

ACS Catalysis
Page 22 of 45
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
the absence of nitrogen precursor during pyrolysis, are inactive towards the HDO reaction. In
addition, the poor catalytic performance of N-C@Al O -900 and Fe-C@Al O -900 catalysts
2
3
2
3
indicated that both iron and nitrogen are indispensable to reach satisfactory HDO activity. A
prominent peak of metallic Fe can be found in the XRD pattern of Fe-C@Al O -900 catalyst,
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
2
3
and the crystalline structure of alumina support significantly changed (figure S18). Specifically,
strong peaks of alumina support (corundum, syn, PDF#10-0173) became much weaker, and new
weak peaks of alumina (alumina, PDF#11-0517) appeared.
The HDO activity of the iron catalysts is significantly improved upon nitrogen-doping. 1-
octadecanol was the major product over Fe-N-C@Al O -300 catalyst (30.1% yield). Meanwhile,
2
3
octadecane (n-C18) and octadecene (C18 ) were the major hydrocarbon products. The presence
=
of octadecene suggested that n-C18 was probably generated by the dehydration of 1-octadecanol
to octadecene followed by hydrogenation. In addition, the Fe-N-C@Al O -300 catalyst is
2
3
unstable during reaction, leading to opaque production solution and unsatisfactory carbon
balance (see figure S19 in the supporting information). When the pyrolysis temperature was
o
raised to 550 C, the yield of 1-octadecanol further increased to 60.8%, suggesting that the
hydrogenation activity of the iron catalysts increased significantly.
When the pyrolysis temperature was further raised, the yields of alkanes increased
dramatically, and the highest selectivity of alkanes was achieved for the catalyst with a pyrolysis
o
temperature of 900 C. 91.9% yield of n-C18 and 6.0% yield of n-C17 are obtained with only
1.9% yield of cracking alkane products. No oxygenated products were detected (see the GC
spectrum in figure S19). The alkane yields significantly decreased when the pyrolysis
o
temperature was raised to 1100 C, and the major by-product was the 1-octadecanol with a yield
of 56.9% , suggesting that the HDO activity of the iron catalyst significantly decreased.
ACS Paragon Plus Environment
2
2

Page 23 of 45
ACS Catalysis
1
2
3
4
5
6
7
8
9
Iron catalyst with 1,10-phenanthroline as N precursor (Fe-phen-C@Al O -900) was also
2
3
examined because 1,10-phenanthroline has been known as excellent N precursor for nitrogen-
doped iron catalysts.28-33 However, it showed inferior HDO activity, giving 23.2% yield of 1-
octadecanol as major by-product. By contrast, use of other g-C N precursors such as
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
3
4
dicyandiamide and urea as N precursor led to similar HDO activity as melamine. The nitrogen
content of 1,10-phenanthroline is much lower than g-C N precursors such as melamine (17.3
3
4
wt% vs 66.7 wt%), so a much higher dosage of 1,10-phenanthroline was used to keep the same
Fe/N ratios in the precursors. Correspondingly, the C/Fe ratios in the precursors of Fe-phen-
C@Al O -900 catalyst are much higher than that of Fe-N-C@Al O -900 catalyst. This caused
2
3
2
3
significant changes in the crystalline structure of alumina support, which is similar to Fe-
C@Al O -900 catalyst (figure S18). The replacement of iron(III) acetylacetonate with ferric
2
3
nitrate led to the formation of hercynite phase as the predominant iron phase (see XRD pattern in
figure S20), thus resulting in poor HDO activity. The color of the catalyst also turned into light
grey (figure S21). The undesired redox reaction between nitrate ion and N precursor melamine
during pyrolysis may hamper the reaction between Fe and melamine to form FeNx and Fe/Fe C
3
species, so that hercynite phase is formed as the major iron species. Much inferior HDO catalytic
performance was also observed if other supports such as activated carbon, TiO , and SiO were
2
2
used.
As intense peaks of metallic iron phase were detected in the XRD pattern of Fe-N-
o
C@Al O catalysts prepared above 550 C, one may presume that the metallic iron phase is the
2
3
active center to catalyze stearic acid HDO. To test this hypothesis, Fe-C-900 catalyst, in which
metallic iron is the most predominant iron species, was used for the HDO reaction. However,
9.3% yield of n-C17 was obtained as the major product, and the yield of 1-octadecanol was only
ACS Paragon Plus Environment
2
3

ACS Catalysis
Page 24 of 45
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
.2%. The poor yield of hydrogenated product and the different product distribution suggested
that the metallic iron is not the key active center for the hydrogenation reaction. By contrast, the
Fe-N-C-900 afforded 45.1% yield of 1-octadecanol as the major product. As weak Fe C phases
3
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
are observed in the XRD pattern, the presence of Fe C phase is probably indispensible for the
3
enhanced hydrogenation activity. The XRD pattern of Fe-N-C@Al O -900 catalyst, which is the
2
3
most effective iron catalyst towards the HDO reaction, also has similarly weak peaks of Fe C
3
species. Similar synergy has recently been reported, where Fe/Fe C nanoparticles were found to
3
3
8
boost the activity of a Fe-N-C catalyst. The role of Fe C phase during hydrogenation of –
3
COOH group is further elucidated in the stability test section.
The iron loading plays also a key role on the HDO activity of the Fe-N-C@Al O -900
2
3
catalysts (figure 9). The Fe-N-C@Al O catalysts with Fe loading less than 10 wt% afforded
2
3
poor yields of HDO products, and 1-octadecanal was one of the major products. As stated above,
the XRD patterns showed that hercynite is the only iron species in these iron catalysts, further
confirming that hercynite is inactive towards the HDO reaction. When the iron loading increased
from 5 wt% to 10 wt%, metallic iron species can be found in the XRD pattern, and the yield of 1-
octadecanol increased from 14% to 31.6%. The optimal iron loading is 20 wt%. Further increase
of the iron loading to 30 wt% leads to more n-C17 and cracking products.
ACS Paragon Plus Environment
2
4

Page 25 of 45
ACS Catalysis
1
2
3
4
5
6
7
8
9
Figure 9. HDO of stearic acid over iron catalysts with different iron loadings. Reaction
o
conditions: 0.5 mmol stearic acid, 20 mL solvent, 0.1 g catalyst, 4 MPa H , T=320 C, and t=8 h.
2
3.3 Effect of support
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
The HDO activities of various iron catalysts with and without nitrogen-doping are further
examined to explore the effect of support (figure 10). The Fe-N-C@Al O -900 and Fe-N-C-900
2
3
catalysts exhibit much higher HDO products than the Fe-Al O -900 and Fe-C-900 catalysts,
2
3
respectively, which possess Fe/Fe C species in their XRD patterns. By contrast, the effect of
3
nitrogen-doping is not so remarkable for Fe-N-TiO -900 and Fe-N-SiO -900 catalysts, for which
2
2
predominant peaks of metallic Fe are observed in the XRD patterns.
Figure 10. HDO of stearic acid over iron catalysts prepared using different supports with and
without nitrogen-doping. Reaction conditions: 0.5 mmol stearic acid, 20 mL solvent, 0.1 g
o
catalyst, 4 MPa H , T=320 C, and t=8 h.
2
ACS Paragon Plus Environment
2
5

ACS Catalysis
Page 26 of 45
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 11. HDO of 1-octadecanol over various supports with and without nitrogen-doping.
Reaction conditions: 0.5 mmol 1-octadecanol, 20 mL solvent, 0.1 g catalyst, 4 MPa H , T=320
2
oC, and t=8 h.
Alumina is well-known to be an efficient catalyst for alcohol dehydration since 1797.[52] As
stated above and discussed later in the reaction pathway section (Section 3.4), the dehydration
reaction could be implicated in the HDO of stearic acid to C18. Thus, one may presume that the
alumina support plays an important role towards the dehydration of 1-octadecanol. To test this
hypothesis, the HDO of 1-octadecanol over various supports is examined (figure 11). On
alumina support, the yields are 2.5% to C17, 9.1% to C18, 24.6% to C18 , 5.5% to cracking
=
products, and 1.5% to C17-CHO at a conversion of 77.3%. Surprisingly, TiO exhibited much
2
better catalytic performance, while SiO and C gave significantly inferior results. Nitrogen-
2
doping led to a positive effect on alumina, towards both dehydration of –OH and hydrogenation
of C=C, but a negative effect on TiO . Such a promotion effect by hybrid-formation was also
2
reported by Su et al.[53] In addition, the addition of Fe significantly improves the catalytic
performance of alumina-based catalyst, but leaves that of TiO -based catalyst unchanged.
2
ACS Paragon Plus Environment
2
6

Page 27 of 45
ACS Catalysis
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 12. IR spectra of pyridine adsorbed on Al O (a) and TiO (b).
2
3
2
The acidity of Al O and TiO was explored using Py-IR to elucidate their different activity
2
3
2
towards HDO of 1-octadecanol (figure 12). The amounts of Brønsted (B) and Lewis (L) acid
sites are listed in table S4. Lewis acid sites are predominant on both supports, and the ratio of
B/L acids sites is higher on TiO (0.24) than on Al O (0.05). Although the peaks for Brønsted
2
2
3
-1
acid sites (ca. 1540 and 1635 cm ) are hard to distinguish on both supports, the peaks for B+L
acid sites are more intense on TiO . It is well known that Bronsted acid sites are important for
2
dehydration, rationalizing in part why TiO exhibited a better catalytic performance than Al O
3
2
2
towards HDO of 1-octadecanol. After nitrogen-doping, the ratios of B/L sites greatly increased
ACS Paragon Plus Environment
2
7

ACS Catalysis
Page 28 of 45
1
2
3
4
5
6
7
8
9
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
from 0.05 to 0.17 on Al O , leading to better catalytic performance. By contrast, nitrogen-doping
2
3
led to a slightly lower ratio of B/L sites and inferior activity on TiO2.
3.4 Reaction pathway
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
Figure 13. Time courses for the HDO of stearic acid (a) and 1-octadecanol (b). Reaction
conditions: 0.5 mmol substrate, 20 mL solvent, 0.1 g Fe-N-C@Al O -900 catalyst, 4 MPa H ,
2
3
2
o
T=320 C.
The time course for stearic acid conversion was recorded (figure 13a). 88.6% yield of 1-
octadecanol was obtained at 0.5 h, showing that 1-octadecanol is an important intermediate
during the HDO of stearic acid. After 6 h, 1-octadecanol was completely consumed, and the
ACS Paragon Plus Environment
2
8

Page 29 of 45
ACS Catalysis
1
2
3
4
5
6
7
8
9
yield of n-C18 increased from 2.4% at 0.5 h to 92.2% at 6 h. No changes in the product
distribution were observed when the reaction was prolonged to 8 h, confirming that the produced
alkanes would not further react over the Fe-N-C@Al O catalyst.
2
3
1
1
1
1
1
1
1
1
1
1
2
2
2
2
2
2
2
2
2
2
3
3
3
3
3
3
3
3
3
3
4
4
4
4
4
4
4
4
4
4
5
5
5
5
5
5
5
5
5
5
6
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
2
3
4
5
6
7
8
9
0
1
-Octadecanal was detected initially (~1 h) with yields less than 1%, so that the minor
product heptadecane was probably generated by the decarbonylation of 1-octadecanal. The time
course for the HDO of 1-octadecanol (figure 13b) also showed the generation of trace yields of
1
-octadecanal during the first 2 h, suggesting reversible transformation between 1-octadecanol
and 1-octadecanal, which is well known for transition-metal catalyzed HDO of stearic acid.
Trace octadecene (C ) was obtained during the first 4 h, suggesting that octadecane was
18=
probably produced by the dehydration of 1-octadecanol followed by hydrogenation. HDO of
oleic acid confirms that the iron catalysts are capable of catalyzing the hydrogenation of C=C
bond, for which stearic acid is observed as a major intermediate. Another possible mechanism
entails the reverse Mars-Van Krevelen mechanism during the HDO of 1-octadecanol to
octadecane. This mechanism is implicated by the reactivity of benzoic acid over the Fe-N-
C@Al O catalyst. Although there is no α-H next to the carboxylic acid group in benzoic acid, it
2
3
is still converted to toluene with 97.8% yield (entry 5, table 2). As benzyl alcohol is one of the
key intermediate molecules, its HDO should proceed via a reverse Mars-Van Krevelen kind of
mechanism rather than dehydration followed by hydrogenation because of the lack of α-H. The
reverse Mars–van Krevelen mechanism, which is usually involved in HDO over reducible metal-
oxides.25,54 can directly remove the O atom in the oxygenate substrate by C-O bond scission
over an oxygen vacancy and H back donation from the OH, leaving the O in the oxygen vacancy.
The aromatic ring does not participate in the reaction during the HDO of benzyl alcohol,
similarly to a recent report, by one of us, on the highly selective HDO of m-cresol to toluene
ACS Paragon Plus Environment
2
9