Hydrothermal catalytic processing of saturated and unsaturated fatty acids to hydrocarbons with glycerol for in situ hydrogen production

Article abstract of DOI:10.1039/c3gc41798k

Lipids are a promising feedstock to produce renewable hydrocarbon fuels and H₂via catalytic hydrothermal processing. Upon exposure to hydrothermal media (e.g., 300 °C, 8-11 MPa), lipids rapidly hydrolyze to produce saturated and unsaturated free fatty acids in varying ratios, depending on the feedstock, as well as glycerol. This report demonstrates the potential of Pt-Re/C for the hydrothermal conversion of saturated and unsaturated fatty acids to hydrocarbons, using glycerol reforming for in situ H₂ production to meet process demands. Experiments showed that deoxygenation of stearic acid, a model saturated fatty acid, was significantly enhanced with Pt-Re/C under a reducing atmosphere compared to Pt/C. The coupled hydrogenation and deoxygenation (HYD-DOX) of oleic aid, a model unsaturated fatty acid, was also moderately enhanced under an inert atmosphere using glycerol for in situ H₂ production, with DOX as the rate-limiting step. Characterization of Pt-Re/C showed that Re had a significant effect on CO:H uptake ratio (2.2) compared to commercial Pt/C (1.3), with the metals dispersed as small crystallites (～3-4 nm) throughout carbon support. Experiments revealed that the initial system H₂ headspace loading <3.45 MPa greatly enhances fatty acid DOX kinetics via decarboxylation/decarbonylation without net H ₂ consumption. At higher initial H₂ loadings (≥3.45 MPa), fatty acid reduction was also observed as a minor DOX pathway. Experiments also showed that oleic acid HYD-DOX and glycerol reforming are affected by initial glycerol concentration and catalyst loading. Under optimized process conditions, complete HYD-DOX of oleic acid to heptadecane was achieved within 2 h with a net-zero H₂ consumption using a 1:3 glycerol-to-fatty acid ratio (i.e., the native ratio in triacylglycerides). X-ray photoelectron spectroscopy showed that H₂ in the reactor headspace results in lower oxidation states of Pt and Re, suggesting a possible mechanism for enhanced DOX kinetics. This approach holds promise for overcoming the high external H ₂ demands of conventional lipid hydrotreatment processes.

Full text of DOI:10.1039/c3gc41798k

Green Chemistry
View Article Online
View Journal | View Issue
PAPER
Hydrothermal catalytic processing of saturated
and unsaturated fatty acids to hydrocarbons with
glycerol for in situ hydrogen production
Derek R. Vardon,*a Brajendra K. Sharma,b Humberto Jaramillo,a Dongwook Kim,c
Jong Kwon Choe,a Peter N. Ciesielskid and Timothy J. Strathmanna
Cite this: Green Chem., 2014, 16,
1507
Lipids are a promising feedstock to produce renewable hydrocarbon fuels and H2 via catalytic hydrother-
mal processing. Upon exposure to hydrothermal media (e.g., 300 °C, 8–11 MPa), lipids rapidly hydrolyze
to produce saturated and unsaturated free fatty acids in varying ratios, depending on the feedstock, as
well as glycerol. This report demonstrates the potential of Pt–Re/C for the hydrothermal conversion of
saturated and unsaturated fatty acids to hydrocarbons, using glycerol reforming for in situ H2 production
to meet process demands. Experiments showed that deoxygenation of stearic acid, a model saturated
fatty acid, was significantly enhanced with Pt–Re/C under a reducing atmosphere compared to Pt/C. The
coupled hydrogenation and deoxygenation (HYD–DOX) of oleic aid, a model unsaturated fatty acid, was
also moderately enhanced under an inert atmosphere using glycerol for in situ H2 production, with DOX
as the rate-limiting step. Characterization of Pt–Re/C showed that Re had a significant effect on CO : H
uptake ratio (2.2) compared to commercial Pt/C (1.3), with the metals dispersed as small crystallites
(∼3–4 nm) throughout carbon support. Experiments revealed that the initial system H2 headspace loading
<3.45 MPa greatly enhances fatty acid DOX kinetics via decarboxylation/decarbonylation without net H2
consumption. At higher initial H2 loadings (≥3.45 MPa), fatty acid reduction was also observed as a minor
DOX pathway. Experiments also showed that oleic acid HYD–DOX and glycerol reforming are affected by
initial glycerol concentration and catalyst loading. Under optimized process conditions, complete HYD–
DOX of oleic acid to heptadecane was achieved within 2 h with a net-zero H2 consumption using a
1 : 3 glycerol-to-fatty acid ratio (i.e., the native ratio in triacylglycerides). X-ray photoelectron spectroscopy
showed that H2 in the reactor headspace results in lower oxidation states of Pt and Re, suggesting a poss-
ible mechanism for enhanced DOX kinetics. This approach holds promise for overcoming the high exter-
nal H2 demands of conventional lipid hydrotreatment processes.
Received 31st August 2013,
Accepted 19th December 2013
DOI: 10.1039/c3gc41798k
www.rsc.org/greenchem
fatty acid methyl esters (i.e., biodiesel), including their direct
compatibility with existing refinery operations, full integration
1. Introduction
With the effect of global climate change increasingly evident,
there is a pressing need to transition to renewable fuels and
petrochemical substitutes. Green diesel (e.g., renewable diesel-
grade hydrocarbons) produced from lipid feedstocks is a
potential alternative for partially displacing the 46 billion
gallons per year of petroleum diesel consumed in the United
States.1 Green diesel has several advantages over conventional
into the transportation infrastructure, and improved fuel
performance properties.2 Commercialization of green diesel by
catalytic hydrodeoxygenation (HDO) has been pursued by Neste
Oil’s NExBTL and UOP/Eni’s Ecofining processes. However, the
high moisture content of many lipid-rich feedstocks (e.g., micro-
algae, waste grease) can be problematic for downstream process-
ing operations, requiring energetic and costly pretreatment and
separation steps. In addition, the HDO process has high H2
process demands because oxygenated lipid functional groups
are primarily removed by catalytic reduction.3,4 For example, the
Ecofining process requires 1.5–3.8 wt% H2 of the incoming feed
stream for vegetable oil upgrading.2 The large H2 demand
necessitates colocation with hydrogen production facilities and
negatively impacts the process sustainability since H2 is primarily
derived from fossil fuels.3,4 As a result, alternative lipid process-
ing routes are being explored for green diesel production.
aDepartment of Civil and Environmental Engineering, University of Illinois at
Urbana-Champaign, 205 N. Mathews Ave., Urbana, IL 61801, USA.
E-mail: dvardon2@illinois.edu; Tel: +1(217) 766-3916
bIllinois Sustainable Technology Center, University of Illinois at Urbana-Champaign,
1 Hazelwood Dr., Champaign, IL 61820, USA
cDepartment of Chemistry, Korea Military Academy, Seoul 139-799, Republic of
Korea
dNational Renewable Energy Laboratory, 15013 Denver West Parkway, Golden,
CO 80401, USA
This journal is © The Royal Society of Chemistry 2014
Green Chem., 2014, 16, 1507–1520 | 1507

View Article Online
Paper
Green Chemistry
Catalytic hydrothermal processing is a promising method the conversion of saturated and unsaturated fatty acids to
for producing renewable hydrocarbons from lipids by utilizing hydrocarbons with integrated glycerol APR for in situ H2
water as the reaction medium.5–8 Under subcritical conditions production has not been investigated.
(200–374 °C, 5–20 MPa), water becomes a unique reaction
This study examines the use of Pt–Re/C for the hydro-
medium with a reduced dielectric constant, providing lipid thermal deoxygenation of stearic acid, a model saturated fatty
solubility similar to non-polar solvents.9 The interphase mass acid, as well as the coupled hydrogenation and deoxygenation
transfer and water self-dissociation constant increase, creating (HYD–DOX) of oleic acid, a model unsaturated fatty acid, with
a
highly reactive processing environment.9 Additionally, in situ H2 production from glycerol. Experiments were con-
water as a solvent is well suited for high-moisture waste lipid ducted to (i) evaluate the material properties of Pt–Re/C and
feedstocks such as fish and animal fat processing residues,10 its stability in hydrothermal media, (ii) determine the use of
sewage sludge,11 and algal biomass.12 Algae are a particularly Re as a promoter metal to enhance saturated fatty acid DOX
promising feedstock due to their high lipid productivity rates and generate H2 in situ from glycerol to facilitate complete
compared to terrestrial biomass, potential integration with HYD–DOX of unsaturated fatty acids, (iii) investigate the influ-
wastewater treatment, and ability to use non-arable land for ence of initial H2 gas addition on fatty acid HYD–DOX kinetics,
cultivation.12–14
reaction pathways, catalyst oxidation state, and net H2 con-
Triacylglycerides (TAGs), the predominant component of sumption, and (iv) explore the impact of glycerol and catalyst
neutral lipids, rapidly hydrolyze in hydrothermal media to loading on product yields. Results support a proposed reaction
produce saturated and unsaturated free fatty acids (FFAs) (e.g., scheme for one-pot hydrothermal catalytic conversion of TAGs
lauric, palmitic, palmitoleic, stearic, oleic, linoleic) in varying to liquid hydrocarbon fuels, involving hydrolysis and catalytic
ratios, depending on the feedstock, as well as glycerol. Recent reforming, hydrogenation, and deoxygenation reactions.
work has shown that saturated fatty acids (e.g., lauric, palmitic,
stearic) rapidly undergo catalytic deoxygenation (DOX) to
produce linear hydrocarbons and CO2 via decarboxylation/de-
carbonylation without external H2 addition over Pt/C and Pd/C
2. Results and discussion
2.1. Catalyst characterization
catalysts, with Pt/C displaying significantly greater DOX
activity.5,6 In contrast, oleic acid (mono-unsaturated C18 fatty
Commercial Pt/C and as prepared Pt–Re/C catalysts were
acid) does not readily undergo DOX under the same con-
ditions. Instead, only partial hydrogenation (HYD) is observed,
with stearic acid (saturated C18 fatty acid) being the primary
reaction product.5,6 Complete conversion of unsaturated fatty
acids to renewable hydrocarbons is not achieved, suggesting
that olefin HYD (a H2-consuming reaction) is necessary prior
to DOX with noble metals on activated carbon.
initially characterized to determine their material properties.
The metal loading of commercial Pt/C as received was 5 wt%
nominal. Pt–Re/C was prepared by aqueous deposition of Re
onto commercial Pt/C, followed by in situ reduction. This
method produced a catalyst with a Pt loading of 5.2 wt% and a
Re loading of 4.2 wt% (Pt–Re molar ratio of 1.2), as measured
by ICP.
In addition, other studies have examined hydrothermal
catalytic reforming of glycerol, a by-product of TAG hydrolysis,
to generate H2, commonly referred to as aqueous phase
reforming (APR).15–20 Typically, glycerol is assumed to be
derived as a by-product from biodiesel production. However,
utilization of glycerol APR for in situ H2 production may facili-
tate unsaturated fatty acid HYD, allowing for subsequent fatty
acid DOX (HYD–DOX). Furthermore, the addition of Re to Pt/C
has been shown to enhance glycerol APR by reducing the
affinity for CO,18,21 which may allow for improved turnover fre-
quency. Increased CO turnover frequency may enhance fatty
acid DOX via decarbonylation, since CO is a reaction by-
product;22,23 however, to our knowledge the use of Pt–Re/C for
Multipoint N2-physisorption provided additional infor-
mation regarding the catalyst support properties, as shown in
Table 1. BET analysis determined that commercial Pt/C had a
support surface area of 1075 m2 g−1, while BJH desorption ana-
lysis determined that the pore volume was 0.715 cm3 g−1, with
an average pore diameter of 10.23 Å. The addition of Re to
commercial Pt/C reduced the support surface area to 750 m2
g−1 and the pore volume to 0.607 cm3 g−1, while the average
pore diameter remained fairly constant 10.06 Å. The reduction
in surface area and pore volume is not atypical for high sec-
ondary metal loadings, which may result in pore blockage.24
Recent work by our group to prepare Ru–Sn/C from com-
mercial Ru/C by incipient wetness with SnCl2 also resulted in a
Table 1 Physisorption and chemisorption parameters for commercial Pt/C, Pt–Re/C as prepared, and Pt–Re/C following hydrothermal exposure
SBET
Pore volumea
Avg. pore
dia.a (Å)
CO uptake
H uptake
CO : H
uptake ratio
Dispersionc
(%)
Catalyst
(m2 g−1
)
(cm3 g−1
)
(μ mol g−1
)
(μ mol g−1
)
Pt/C commercial
1075
750
675
0.715
0.607
0.507
10.23
10.06
10.04
130
112
84
102
52
30
1.3
2.2
2.8
51
42
32
Pt–Re/C as prepared
Pt–Re/C HT exposureb
a Pore volume and average pore diameter determined by BJH desorption. b Hydrothermal exposure at 300 °C for 3 h. c Dispersion calculated
based on CO chemisorption.
1508 | Green Chem., 2014, 16, 1507–1520
This journal is © The Royal Society of Chemistry 2014

View Article Online
Green Chemistry
Paper
decrease in surface area by 268 m2 g−1, without in situ H uptake with Pt–Re/C was markedly reduced by 49% (52 μmol
reduction; however, for Pt–Re/C, the influence of hydrothermal g−1) compared to commercial Pt/C. The reduction of H uptake
in situ reduction may also be a factor.
by nearly two-fold (CO : H uptake ratio of 2.2) is similar to pre-
CO and H2 pulse chemisorption was then performed to vious findings by Simonetti et al. (CO : H uptake ratio of 2.3),25
evaluate catalyst metal site properties, as shown in Table 1. which confirms the significant influence of Re addition and
The measured value of irreversible CO uptake for commercial suggests Pt–Re alloy formation.26
Pt/C was 130 μmol g−1, slightly greater than the value for H
XRD analysis of unloaded activated carbon, commercial
uptake (112 μmol g−1). This resulted in a calculated metal dis- Pt/C, and Pt–Re/C as prepared provided additional information
persion of 51% for Pt/C, based on CO uptake and an assumed regarding long-range order and average metal crystallite size,
CO : M ratio of unity. The addition of Re moderately lowered as shown in Fig. 1. Distinct peaks were observed on the bare
the measured CO uptake by 13% (112 μmol g−1), with a calcu- carbon support, while the lack of sharp, pronounced peaks
lated metal dispersion of 42% based on Pt loading. This with Pt/C and Pt–Re/C suggests that the metals were highly dis-
suggests an increase in particle size, similar to findings from persed as small crystallites. In comparison to Pt/C, Pt–Re/C
XRD and SEM-EDS described below, or Re coverage of Pt sites displayed a broad XRD peak near 40° (2θ value), with a bulk
during secondary metal loading. In contrast to CO uptake, the metal crystallite size of ∼3.4 nm estimated by the Scherer
equation.
Pt–Re/C was further characterized to evaluate the nature of
bimetallic crystallites after aqueous adsorption and in situ
reduction. Scanning electron microscopy coupled to energy
dispersive X-ray spectroscopy (SEM-EDS) provided a general
survey of the Pt and Re distribution across several carbon
support particles, as shown in Fig. 2a–d. Although small crys-
tallites comprised predominantly of rhenium were evident,
overall both metals were fairly well dispersed throughout the
support.
Transmission electron microscopy (TEM) of Pt–Re/C
revealed disperse metal crystallites <5 nm in diameter
(Fig. 3a), consistent with crystallite size estimates from XRD.
The distribution of Pt and Re within individual metal crystal-
lites was also evaluated using scanning transmission electron
microscopy coupled to energy dispersive X-ray spectroscopy
(STEM-EDS), as shown in Fig. 3b and 3c. EDS analysis con-
firmed that Pt and Re were present throughout the crystallites,
but a fixed metallic ratio was not observed. Based on these
results, it was evident that Pt and Re were well dispersed and
Fig. 1 XRD spectra of unloaded activated carbon, commercial Pt/C as
largely co-localized throughout the support; however, a defini-
tive assessment of the atomic scale, bimetallic crystallite
received, Pt–Re/C as prepared, and Pt–Re/C after exposure to hydro-
thermal media.
Fig. 2 SEM images and EDS maps of Pt–Re/C catalyst particles as prepared (a–d) and post hydrothermal media exposure (e–h), with single-metal
crystallites indicated by white arrowheads.
This journal is © The Royal Society of Chemistry 2014
Green Chem., 2014, 16, 1507–1520 | 1509

View Article Online
Paper
Green Chemistry
Fig. 3 Pt–Re/C images of the catalyst as prepared by TEM (a) and STEM (b), with localized STEM-EDS spectra obtained from the region of an indi-
vidual crystallite (c). Note that Cu peaks present due to STEM sample holder.
composition (e.g., collocated mono-metallics, intermetallic reaction products from fatty acid decarbonylation (Rxn. 2),22,23
alloys, core-shell bimetallics, etc.) could not be concluded from terminal alkenes and CO, were not detected in significant
this data.
amounts.
2.2. Catalyst activity
2.2.1. Stearic acid DOX with Pt/C and Pt–Re/C. The influ-
ence of Re on the activity of Pt/C for stearic acid DOX was
initially investigated, as shown in Rxn. 1, since saturated fatty
acids can comprise a significant component of lipids.
ð2Þ
However, CO can be rapidly converted to CO2 by the WGS
reaction, and terminal alkenes can be readily hydrogenated
with a H2 headspace, as shown in Rxn. 3 and Rxn. 4,
respectively.
ð1Þ
ð3Þ
Experiments were performed with Pt/C and Pt–Re/C using a
reducing H2 headspace (<3.45 MPa/557 mmol H2), as shown
in Fig. 4. The conversion of stearic acid to heptadecane at
300 °C occurred with high selectively (>90% molar yield) to
heptadecane for both Pt and Pt–Re/C, with octadecane
ð4Þ
In contrast to hydrothermal media, organic solvents (e.g.,
observed in trace amounts (3–8% molar yield). The high dodecane) produce CO as the primary C-gas product during
selectivity to heptadecane suggests decarboxylation/decarbony- fatty acid DOX with a H2 headspace, since the DOX pathway
lation as the primary DOX pathway,3,4 similar to past findings sharply shifts from decarboxylation to decarbonylation.22,23
by Fu et al. using Pt/C with an inert N2 headspace.5 The main This may be problematic due to competitive adsorption of CO
Fig. 4 Timecourse for stearic acid DOX at 300 °C with an initial H2 reactor headspace using Pt/C and Pt–Re/C (a). Pseudo-first order kinetic para-
meter fits for heptadecane formation indicated by color solid lines for Pt–Re/C (green) and Pt/C (blue). Effect of temperature on stearic acid DOX
3 h heptadecane molar yields using Pt/C and Pt–Re/C with an initial H2 reactor headspace (b). Reaction conditions: 20 g stearic acid, 0.5 g catalyst,
80 g H2O, initial head-gas = 3.45 MPa H2.
1510 | Green Chem., 2014, 16, 1507–1520
This journal is © The Royal Society of Chemistry 2014

View Article Online
Green Chemistry
Paper
on active catalyst sites and the lack of H2 regeneration by the
WGS reaction in organic media.27
ð5Þ
The addition of Re to Pt/C had a significant effect on the
apparent stearic acid DOX kinetics at 300 °C, as shown in
Fig. 4a. The Pt–Re/C stearic acid pseudo-first order DOX rate
constant for the given liquid reactant loading, normalized to
catalyst loading (kapp Pt–Re/C DOX = 2.72 0.24 × 10−4 s−1 g−1),
ð6Þ
was nearly 3-fold greater compared to Pt/C (kapp
0.94
=
Pt/C DOX
0.05 × 10−4 s−1 g−1). Although multi-phase reaction
Pt/C and Pt–Re/C catalysts were tested with a reactor head-
space initially pressurized with N2 gas (0.35 MPa/56 mmol) at a
reaction temperature of 300 °C, as shown in Table 2. With Pt/C
and no glycerol addition, oleic acid was only partially hydro-
genated (31% molar yield) to stearic acid, similar to findings
by Fu et al.5 Based on the oleic acid loading (71 mmol),
22 mmol of H2 was required to account for hydrogenation of
the single olefin bond in oleic acid. It has been suggested that
water, the carbon support, or organic reactants may serve as
potential H2 sources in hydrothermal media.5,6,29 Hydrocarbon
products resulting from fatty acid DOX were not observed (e.g.,
heptadecane, octadecane, or lower hydrocarbons). Likewise,
analysis of the reactor headspace gas following the reaction
revealed no significant quantities (<1 mmol) of H2 remaining.
Therefore, it is presumed that the small amount of H2 gene-
rated in situ was consumed by HYD of oleic acid.
kinetics are significantly more complex than a pseudo-first
order assumption, this simplification allowed for generalized
activity comparisons. In addition, Pt–Re/C also displayed
improved DOX performance at low temperatures (i.e., <340 °C),
as indicated by the heptadecane 3 h molar yields shown in
Fig. 4b.
Due to the limitations of collecting accurate hydrothermal
kinetic data in a large batch reactor, detailed kinetic analysis
was not addressed. Multiple factors can potentially mask
intrinsic catalytic activity in batch, including partial de-
activation with time on stream, inhibition from reaction by-
products, and artifacts introduced during the time delay for
batch reactor heat-up.28 However, the effect of Re addition on
catalyst activity was pronounced and in-depth kinetic analysis
will be addressed in follow-up work.
The increased activity with Re addition may be due to a
number of factors, including enhanced CO turnover for decar-
bonylation as a DOX pathway, alteration to active site mor-
phology, or changes to active site electronic properties.
Previous work has shown that addition of Re to Pt/C can sig-
nificantly reduce CO adsorption affinity,18,21 which may allow
for greater active site turnover and availability for fatty acid
DOX and glycerol reforming reactions. As a control, the activity
of the Re precursor alone was tested for stearic acid DOX, with
no conversion to heptadecane observed. This suggests that Re
may only function as promoter metal with Pt, although further
investigation into the mechanism for stearic acid DOX in
hydrothermal media is needed.
In contrast, when glycerol (5 g/54 mmol) was added to the
reactor with Pt/C, complete HYD of oleic acid was observed
and partial DOX of the resulting stearic acid occurred (24%
molar yield), producing heptadecane with high selectivity
(>98%). However, heptadecane yields were significantly lower
compared to initial experiments with stearic acid in a reducing
headspace (Fig. 4), suggesting that the system H2 level greatly
affects catalyst DOX performance. Residual H2 was also
detected in the headspace following the conversion (48 mmol),
indicating significant in situ H2 production with glycerol
addition. It is estimated that 119 mmol of H2 was produced
in situ if the stoichiometric H2 requirement for oleic acid HYD
is added to the measured headspace residual.
2.2.2. Glycerol for facilitating oleic acid HYD–DOX. The
utilization of glycerol APR for in situ H2 production (Rxn. 5)
was then examined to facilitate HYD of oleic acid to stearic
acid (Rxn. 6), and subsequent fatty acid DOX (Rxn. 1). Ideally,
in situ H2 production from glycerol would alleviate the need
for external H2 consumption, which is problematic for conven-
tional lipid hydrotreatment processes.3,4
Similar to findings with a reducing headspace, the use of
Pt–Re/C enhanced stearic acid DOX (38% heptadecane
molar yield), although the extent of oleic acid HYD was lower
(92% molar yield). As with Pt/C, no significant levels of gly-
cerol remained in the aqueous phase after 9 h for Pt–Re/C. The
residual H2 measured in the headspace (50 mmol) and esti-
mated total in situ production (115 mmol; 30% of theoretical
Table 2 Glycerol addition to produce in situ H2 production and promote oleic acid HYD–DOX using Pt/C and Pt–Re/C with an initial inert gas
reactor headspacea
Fatty acid and hydrocarbon molar yields
Glycerol
H2 residual in
Oleic acid
Catalyst
loading (g)
headspace (mmol)
conversion (%)
Stearic acid (%)
Heptadecane (%)
Pt/C
Pt/C
Pt–Re/C
0
5
5
ND
48
50
31
100
92
31
75
53
ND
24
37
a Reaction conditions: 300 °C, initial head-gas = 0.35 MPa N2, 20 g oleic acid, 0.5 g catalyst, 80 g water, time at temperature 9 h. (Note: ND = not
detected.)
This journal is © The Royal Society of Chemistry 2014
Green Chem., 2014, 16, 1507–1520 | 1511

View Article Online
Paper
Green Chemistry
maximum APR yield) were also similar to yields with Pt/C. As a decarbonylation and APR of glycerol, and is a known inhibitor
control, glycerol APR with Pt–Re/C was tested without oleic of noble metal catalyst activity.31
acid, which produced a comparable level of H2 (142 mmol;
2.2.3. Pt–Re/C activity with an inert and reducing head-
37% of theoretical), indicating no significant inhibition at the space. The reactivity of Pt–Re/C for oleic acid HYD–DOX with
time scale investigated. Past work examining Pt–Re/C solely for in situ H2 production from glycerol was further explored using
glycerol APR found enhanced H2 productivity (88.7% glycerol a reactor headspace initially pressurized with either an inert
conversion, 24.5% H2 selectivity) compared to Pt/C (5.3% con- gas (0.35 MPa/56 mmol N2) or reducing gas (2.59 MPa/
version, 56.5% H2 selectivity);18 however, a significantly lower 417 mmol H2) (Fig. 5). With an initial headspace of N2
temperature was investigated (225 °C).
(Fig. 5a), oleic acid HYD occurred to a significant extent (91%)
Analysis of the aqueous phase following the conversion after 1.5 h at temperature, followed by a slow progression until
revealed that little residual glycerol remained (<1 wt% of completion at 12 h. The H2 level in the headspace remained
initial loading), suggesting complete decomposition at the fairly constant throughout the reaction (44 4 mmol), and no
elevated temperature (300 °C) and extended reaction time significant quantities of glycerol were detected in the aqueous
(9 h). The estimated in situ production of H2 represented 31% phase, indicating that decomposition was complete during
of the theoretical maximum H2 yield for glycerol APR (7 H2 per reactor heatup to the setpoint temperature. The estimated
glycerol). Direct comparisons with previous results for glycerol total in situ production of H2 (measured headspace + estimated
APR are complicated due to differing temperature regimes, stoichiometric demand for oleic acid HYD) was 115 mmol,
reactor configurations (batch vs. continuous), active metals, consistent with findings from catalyst screening experiments.
and catalyst supports. For example, initial investigations with As the reaction progressed, DOX molar yields increased signifi-
Pt/Al2O3 for glycerol APR at 255 °C demonstrated 100% C-con- cantly from 7% after 3 h to 38% after 9 h, before leveling off at
version to the gas phase with 51% selectivity to H2. The lower 41% after 12 h with heptadecane as the primary DOX product
H2 yield from glycerol observed in this study may be due to (>98% selectivity). The addition of glycerol at high levels (5 g)
undesirable C–O bond cleavage pathways that generate H2 con- appeared to have an inhibitory effect on the rate of stearic acid
suming side-products and prevent CO formation.18 High temp- DOX, so pseudo-first order rate constants were not deter-
eratures have also been shown to decrease H2 selectivity mined. The DOX process progressed much slower compared
during glycerol APR, despite increased conversion.19 Likewise, to HYD of oleic acid and APR of glycerol, suggesting that
higher temperatures can impair the water gas shift (WGS) reac- additional work should target increasing catalytic activity for
tion,30 lowering H2 yields. However, fatty acid DOX kinetics the former.
increase with temperature, following Arrhenius parameters,5
suggesting a tradeoff between optimum conditions for glycerol 417 mmol), a dramatic enhancement was observed in the appar-
APR and fatty acid DOX. ent rate of oleic acid HYD–DOX with glycerol, as shown in
With the initial headspace gas switched to H2 (2.586 MPa/
Based on these results, Pt–Re/C was then chosen for further Fig. 5b. Analysis of the reaction products after 1 h at temperature
study due to enhanced acid DOX kinetics compared to Pt/C revealed complete HYD, as well as a small amount of DOX (13%).
(Fig. 4 and Table 2), together with past reports of enhanced As the reaction progressed, the elevated initial H2 level greatly
glycerol APR18,20 and reduced affinity for binding CO enhanced fatty acid DOX kinetics, resulting in complete conver-
(Table 1).18 As noted, CO is produced during fatty acid sion within 7 h and high selectivity for heptadecane (>98%).
Fig. 5 Hydrothermal catalytic HYD–DOX kinetics of oleic acid using glycerol as an in situ H2 source and an initial reactor headspace containing
N2 (a) and H2 (b). Reaction conditions: 300 °C, 20 g oleic acid, 5 g glycerol, 0.5 g Pt–Re/C, 80 g H2O, 0.35 MPa N2 or 2.59 MPa H2.
1512 | Green Chem., 2014, 16, 1507–1520
This journal is © The Royal Society of Chemistry 2014

View Article Online
Green Chemistry
Paper
Enhancement of oleic acid HYD–DOX with the addition of solvent media (e.g., dodecane), where selectivity is highly sensi-
H2 to the headspace was not surprising since oleic acid HYD tive to H2 loading. In organic solvents, reaction products are
appears to be a necessary step prior to DOX in hydrothermal observed from all three fatty acid DOX pathways (e.g., decar-
media with noble metal catalysts on activated carbon.5,6,29 boxylation, decarbonylation, reduction) depending on the H2
However, it was surprising that headspace analysis revealed a loading, with a sharp shift towards decarbonylation as the H2
surplus of H2 (479 9 mmol) in the reactor compared to the partial pressure is increased.23,32,33
initial external loading (417 mmol H2). After accounting for
Headspace analysis also revealed that the net process H2
the extent of oleic acid HYD and final headspace composition, balance was affected by the initial external H2 loading, as
it was estimated that 133 mmol of H2 was produced in situ. shown in Fig. 6b. At lower initial H2 loadings (<3.45 MPa/
Thus, while increased H2 pressure provides a significant 557 mmol), in situ H2 production from glycerol provided a net
kinetic enhancement, in situ H2 production from glycerol APR surplus of system H2 (net increase of 45 12 mmol) despite
exceeded the H2 consumption requirement for oleic acid HYD the consumption of H2 by oleic acid HYD (estimated require-
and yielded a net positive H2 balance for the overall catalytic ment of 71 mmol H2). However, at high initial loadings of H2
process. It follows that after initial pressurization with H2, no (≥3.45 MPa/557 mmol), a net reduction in system H2 was
additional external inputs of H2 would be required for the cata- observed after 3 h. The net reduction was attributed to
lytic hydrothermal conversion of lipids if H2 can be recycled in additional H2 consumption by fatty acid reduction, which con-
the process.
sumes 3 moles of H2 to fully reduce the carboxylate group. In
2.2.4. Influence of initial hydrogen loading. Due to the addition, high H2 pressures can suppress the WGS reaction,34
dramatic influence of H2 on DOX kinetics, the effect of varying lowering H2 production.
initial H2 loading (0–5.17 MPa/0–833 mmol) was investigated
Based on these results, the system H2 pressure, either
further by examining the 3 h DOX yields and resulting head- provided from external sources or produced by in situ APR of
space gas composition. As the initial H2 reactor loading glycerol, was determined to be a key parameter influencing both
increased, the 3 h yield of DOX products increased from 7% fatty acid DOX kinetics and conversion product selectivity. It is
(0 Pa initial H2) to 94% (5.17 MPa initial H2) (Fig. 6a). Hepta- also worth noting that the shift in DOX pathways at elevated H2
decane was observed as the only major DOX product at initial pressure suggests that process H2 levels will be self-limiting, since
H2 loadings <3.45 MPa. At higher initial H2 loadings, fatty acid increasing H2 pressure will shift conversion towards H2-
reduction products (e.g., octadecane, stearic alcohol) were consuming reduction processes, preventing further accumulation.
observed (11.3%), indicating a minor shift towards reduction
of the carboxylate group as a secondary DOX pathway (Rxn. 7).
2.2.5. Effect of glycerol and catalyst loading. Additional
experiments were then performed to determine the effect of
varying glycerol concentrations (Fig. 7a) and catalyst loadings
(Fig. 7b) on the 3 h oleic acid HYD–DOX molar yields with
in situ H2 production using an initial reducing headspace
(2.59 MPa/417 mmol H2). When no glycerol was added to the
reactor, fatty acid HYD was complete and the DOX hepta-
decane molar yield was still significant (45%) due to the initial
ð7Þ
The high selectivity for heptadecane observed in hydrother-
mal media is in contrast to observations reported for organic
Fig. 6 Influence of initial headspace H2 loading on oleic acid HYD–DOX (a) and in situ H2 production by glycerol APR (b). Reaction conditions: t =
3 h, 300 °C, 20 g oleic acid, 5 g glycerol, 0.5 g Pt–Re/C, 80 g H2O. H2 consumed by HYD is estimated based on the observed saturated compounds
derived from oleic acid.
This journal is © The Royal Society of Chemistry 2014
Green Chem., 2014, 16, 1507–1520 | 1513

View Article Online
Paper
Green Chemistry
Fig. 7 Influence of glycerol loading (a) and Pt–Re/C loading (b) on 3 h oleic acid HYD–DOX yields and H2 headspace concentrations. Reaction
conditions: 300 °C, 20 g oleic acid, 80 g H2O, initial H2 loading = 2.59 MPa, 0.5 g Pt–Re/C loading (for panel a), 5 g glycerol (for panel b).
loading of H2 in the system; however, the final measured H2
level (345 mmol) decreased proportionally due to oleic acid
HYD. With a glycerol loading (2.2 g) native to the TAG triolein
(1 mol of glycerol for every 3 mol of oleic acid), the DOX molar
yield increased modestly, resulting in a 3 h molar conversion
to heptadecane of 55%. The final headspace H2 level
(394 mmol) dropped below the initial loading (417 mmol);
but, to a lesser extent compared to the reaction with no
added glycerol (345 mmol). After accounting for H2 consump-
tion estimated for oleic acid HYD, a total of 48 mmol of H2
was produced in situ. Lastly, a glycerol loading of 5 g
(54 mmol) resulted in surplus H2 in the system (479 9 mmol)
compared to the initial loading, with a DOX heptadecane
molar yield (55%) comparable to the native triolein loading
(2.2 g/24 mmol). The estimated in situ H2 yield was ∼30% of
the theoretical maximum yield for glycerol APR. After account-
ing for oleic acid HYD, 133 mmol of H2 was produced in situ.
This suggests that further enhancements in glycerol APR
activity may be needed to facilitate net zero or net positive
system H2 levels for lipid feedstocks with greater degrees of
unsaturation compared to triolein.
Fig. 7b shows the effect of catalyst loading on the 3 h oleic
acid HYD–DOX molar yields and in situ H2 production from
glycerol. As discussed earlier, the standard catalyst loading
used throughout this study (0.5 g) resulted in complete HYD
and partial DOX (55%) of oleic acid after 3 h, with a net
surplus of H2 in the headspace (478 mmol 51 mmol). At
higher catalyst loadings, the increased number of active sites
resulted in a greater degree of fatty acid conversion, as
expected. With 1 g of catalyst, complete HYD–DOX of oleic
acid to heptadecane occurred with high selectivity (>98%). The
yield of in situ H2 remained the same as that observed with
0.5 g catalyst loading, consistent with the observation that gly-
cerol APR has gone to completion for the current sampling
time scale. When the catalyst loading was increased to 2 g
(1 : 10 fatty acid to catalyst ratio by weight), a modest increase
Fig. 8 Timecourse yields for oleic acid HYD–DOX using a glycerol
loading native to triolein (1 : 3 glycerol-to-oleic acid molar ratio) and an
elevated loading of Pt–Re/C (2.0 g). Other reaction conditions: 300 °C,
20 g oleic acid, 2.2 g glycerol, 80 g H2O, initial H2 headspace =
2.59 MPa. Pseudo-first order kinetic parameter fits for stearic acid DOX
(blue) to produce heptadecane (green) are shown with solid color lines.
was also observed for in situ H2 production (556 mmol), which
resulted in an observed H2 yield of 47% of the theoretical
maximum. This suggests that the glycerol-to-fatty acid ratio
for a given catalyst loading may influence in situ H2 yields,
although further study is needed.
2.2.6. Native triolein loading. As a demonstration, Fig. 8
shows a final time series experiment conducted for oleic acid
using a glycerol loading representative of triolein, a TAG with
oleic acid side chains (1 : 3 glycerol-to-oleic acid molar ratio),
and an increased catalyst loading (2 g Pt–Re/C) due to
enhanced in situ H2 production from glycerol APR (Fig. 7b).
Analysis of the reaction products revealed that oleic acid HYD
to stearic acid was complete within 15 min at 300 °C (time at
which first sample was collected), with a significant extent
likely occurring during heat-up from ambient temperature.
Despite H2 consumption from oleic acid HYD, the headspace
H2 level remained fairly constant near the initial loading
1514 | Green Chem., 2014, 16, 1507–1520
This journal is © The Royal Society of Chemistry 2014

View Article Online
Green Chemistry
Paper
(417 mmol), suggesting rapid APR of glycerol and improved H2 hydrothermal media at lower temperature (225 °C) using
yields compared to earlier results with 0.5 g of catalyst. No sig- in situ XPS.20
nificant quantities of glycerol were detected in the aqueous
In contrast to the above, XPS indicated that Pt and Re were
phase after 15 min at temperature. After 2 h at temperature, in more reduced states following exposure to hydrothermal
fatty acid DOX reached completion, significantly faster than media with H2 headspace (2.59 MPa/417 mmol), as shown in
the 7 h required with 0.5 g of catalyst despite a higher glycerol Fig. 9c and 9d. The BE of the 4f7/2 peak for Pt (71.9 eV) was
loading (5 g glycerol; Fig. 5b). Analysis of the headspace at 2 h consistent with a highly reduced state, whereas the measured
confirmed that the overall process was net zero in H2 BE for Re (41.4 eV) was closest to +I reference standard pre-
consumption (421
pseudo-first order DOX rate constant for the given liquid reac- surface atomic ratio under a reducing atmosphere (1.0) was
25 mmol). The Pt–Re/C stearic acid viously collected by our group (41.9 eV).35 Analysis of the Pt–Re
tant loading, normalized to catalyst loading (kapp Pt–Re/C DOX
=
identical to the inert headspace above, suggesting no signifi-
2.83 0.38 × 10−4 s−1 g−1) was comparable to previous results cant changes in crystallite elemental distribution. To our
obtained with 0.5 g of catalyst under a reducing headspace knowledge, this is the first study to examine the effect of H2 on
without glycerol (kapp Pt–Re/C DOX = 2.72 0.24 × 10−4 s−1 g−1
;
the oxidation state of Pt–Re/C in hydrothermal media. The
Fig. 4). The rapid apparent DOX kinetics and lower glycerol lower oxidation states observed with a H2 headspace may
loading may the mask inhibitory effects of glycerol and its account for the increased activity of the catalyst for DOX pro-
degradation products noted previously. These results demon- cesses that are H2 neutral. However, other factors such as coke
strate an initial proof of concept for the catalytic hydrothermal minimization, saturation of reaction by-products,32 alteration
processing of unsaturated TAGs with H2 consumption require- of carbon support functional groups,29,37 or promotion of dec-
ments being met through APR of glycerol, a byproduct of TAG arbonylation as a DOX pathway22,38,39 may also contribute to
hydrolysis.
the enhanced fatty acid DOX kinetics observed with H2 reactor
headspace.
It is also worth noting that the standard catalyst pre-
reduction step used for all reactions in this study (catalyst
exposed to 1.38 MPa H2 for 2 h at 200 °C prior to flushing the
2.3. Influence of headspace composition on Pt–Re oxidation
state
The influence headspace gas composition on the Pt–Re oxi- headspace and introducing reactants) had a significant effect
dation state during exposure to hydrothermal media was then on the Pt–Re/C fatty acid DOX activity, even when the head-
examined due to the pronounced effect on fatty acid DOX kine- space was initially pressurized with H2 (2.59 MPa). With cata-
tics (Fig. 5 and 6). X-ray photoelectron spectroscopy (XPS) was lyst pre-reduction, the 3 h DOX molar yield of heptadecane
used to characterize the oxidation states of Pt and Re after from oleic acid was 34 3%, compared to 7% without pre-
exposure to hydrothermal media (300 °C, 3 h) with either an reduction. This contrasts with past studies that reported no
inert headspace (0.35 MPa/56 mmol N2) or a reducing head- enhancement for DOX of saturated fatty acids when Pt/C was
space (2.59 MPa/417 mmol H2). Individual oxidation states of pre-reduced (Fu et al., 2010). The difference in DOX perform-
Pt and Re exhibit characteristic doublet pairs due to spin–orbit ance with Pt–Re/C may be due to rapid re-oxidation of the cata-
coupling effects. Metal oxidation states were assessed by com- lyst after exposure to hydrothermal media with an inert
paring binding energy (BE) of the 4f7/2 peak against reference headspace20 or changes in the metal crystallite composition
standards collected previously,35 as shown in Fig. 9.
distribution (i.e., alloy formation), as indicated by H2 chemi-
Following exposure to hydrothermal media with an inert N2 sorption results.
headspace, Pt–Re/C was characterized by metals with a high
BE and corresponding oxidation state, as shown in Fig. 9a and
2.4. Pt–Re/C hydrothermal stability
9b, respectively. The Pt–Re surface atomic ratio (1.0) deter- The stability of Pt–Re/C in hydrothermal media (300 °C, 3 h)
mined by XPS was comparable to the bulk ratio measured by with a reducing headspace (2.59 MPa/417 mmol H2) was then
ICP (1.2), suggesting a relatively uniform surface and bulk evaluated to determine changes in catalyst composition or
metal ratio throughout the metal crystallites, as opposed to a morphology. ICP analysis of a representative Pt–Re/C catalyst
core–shell orientation. The BE of the 4f7/2 peak for Pt was fit sample after hydrothermal exposure showed no significant
with two peaks (71.7 eV, 73.9 eV), indicating a mixed oxidation changes in metal loading. Subsequent analysis of the aqueous
state based on the range of values previously reported in litera- phase also confirmed that metal leaching was not prominent,
ture for the NIST XPS database (Pt0 = 70.7–71.3 eV; Pt–Re alloy = although trace levels (<1% of the total metal loading) of Pt and
71.6–71.8 eV; PtII = 72.4–74.6 eV;).36 The BE of the 4f7/2 peak Re initially loaded were observed (Pt = 0.18 ppm; Re =
for Re was fit with a single peak (45.6 eV), consistent with a 4.16 ppm).
ReVI-VII oxidation state based on ReVI NIST database values
Analysis of the carbon support by multipoint N2 physisorp-
(44.3–46.8 eV) and ReVII reference standards previously col- tion revealed pronounced changes in the surface area, as
lected by our group (45.4–46.7 eV),35 despite in situ pre- shown in Table 1. Following hydrothermal exposure, the
reduction with H2 at 1.38 MPa for 2 h prior switching the support surface area of Pt–Re/C decreased by 11% (675 m2
headspace gas over to N2. This finding agrees with previous g−1), similar to past reports for activated carbon noble metal
work analyzing the oxidation state of Pt–Re/C exposed to catalysts in hydrothermal media.5,24 Although this decrease is
This journal is © The Royal Society of Chemistry 2014
Green Chem., 2014, 16, 1507–1520 | 1515

View Article Online
Paper
Green Chemistry
Fig. 9 Influence of an initial inert headspace (0.35 MPa N2) on the oxidation state of Pt (a) and Re (b), as well as the influence of an initial reducing
headspace (2.59 MPa H2) with Pt (c) and Re (d) after exposure to hydrothermal media for 3 h at 300 °C.
not insignificant, the degree of degradation is much smaller in Fig. 2g and h. These results indicate the need for additional
compared to oxide supports which can degrade by over 90% investigation into the hydrothermal stability and activity of Pt–
hydrothermal media.40 The support pore volume also Re/C for fatty acid deoxygenation with time on stream.
decreased by 16% (0.507 cm3 g−1), while the average pore dia- Although time on stream testing was beyond the scope of this
meter remained fairly constant (10.04 Å).
initial study, evaluation of Pt–Re supported on activated
CO and H2 chemisorption following hydrothermal exposure carbon pellets for the hydrodeoxygenation of phenol in hydro-
revealed significant changes to the exposed metal area, as thermal media demonstrated high stability and activity for
shown in Table 1. Irreversible CO uptake decreased by 25% over 140 h on stream in a packed bed flow reactor,24 highlight-
(CO uptake 84 µmol g−1), resulting in a calculated dispersion ing the potential of Pt–Re/C for facilitating hydrothermal
of 31.5%. This indicates that high temperature hydrothermal chemistry.
exposure likely results in some sintering of the immobilized
metal crystallites, consistent with findings by XRD and
2.5. Integrated reaction scheme
SEM-EDS described below. In addition, irreversible H uptake
decreased by 42% (H uptake 30 µmol g−1), resulting in a
CO : H uptake ratio of 2.8. The high CO : H uptake ratio indi-
cates that Re continues to influence the chemisorption behav-
ior of Pt, although significant morphological changes have
occurred compared to as prepared Pt–Re/C.
XRD analysis of Pt–Re/C after hydrothermal media exposure
showed support and crystallite peaks similar to the as pre-
pared catalyst (Fig. 1). Estimation of the metal crystallite size
by XRD after hydrothermal exposure (3.8 nm compared with
3.4 nm for fresh catalyst) also suggests that sintering occurred.
SEM-EDS elemental mapping of Pt and Re revealed that the
majority of metallic sites remained co-dispersed, but single-
element crystallites were present to a greater degree, as shown
Experimental results support an integrated catalytic hydrother-
mal reaction scheme for the conversion of lipids to hydro-
carbons, with in situ H2 production from glycerol (Fig. 10). As
mentioned, in hydrothermal media triacylglycerides rapidly
hydrolyze to produce free fatty acids and glycerol. This process
is utilized at commercial scale, commonly referred to as fat-
splitting.41–43 The utilization of hydrolyzed lipids facilitates the
use of low-quality lipid feedstocks which are typically high in
free fatty acids and problematic for conventional biodiesel pro-
cessing.44 Polar lipids (e.g., phosphatidylcholine) are also
known to rapidly hydrolyze to produce free fatty acids, glycerol,
and P and N containing moieties,45 potentially expanding the
range of viable lipid feedstocks for hydrothermal conversion to
hydrocarbons.
1516 | Green Chem., 2014, 16, 1507–1520
This journal is © The Royal Society of Chemistry 2014

View Article Online
Green Chemistry
Paper
the CO produced from the initial decarbonylation yields
additional H2 via the WGS reaction (Rxn. 3). In contrast to
H2-neutral decarboxylation and decarbonylation pathways, fatty
acid reduction pathways observed at higher initial H2 loadings
(Fig. 3a) consume H2. Complete fatty acid reduction to the
corresponding hydrocarbon consumes 3 moles of H2 (Rxn. 7).
3. Conclusion
In summary, catalytic hydrothermal processing is a promising
method for converting saturated and unsaturated fatty acids
into hydrocarbon fuels while using glycerol APR to meet
process H2 consumption needs. To our knowledge, this is the
first report demonstrating complete HYD–DOX of unsaturated
fatty acids with net zero H2 consumption using glycerol.
Experiments revealed that fatty acid DOX was influenced by
the addition of Re to Pt/C, system H2 level, and glycerol and
catalyst loading. Fatty acid decarboxylation/decarbonylation
was observed as the primary DOX pathway, with reduction
being observed as a minor pathway at high initial H2 loadings
(≥3.45 MPa/557 mmol). Although several mechanisms may
account for enhanced DOX yields with H2 initially loaded into
the system, the effect on catalyst activity and oxidation state
was pronounced. Characterization of Pt–Re/C determined that
Re had a significant effect on CO : H uptake ratio (2.2) com-
pared to commercial Pt/C (1.3), with metals dispersed as small
crystallites (∼3–4 nm) throughout carbon support. Optimized
conditions were applied with a 1 : 3 glycerol-to-oleic acid
molar ratio (i.e., native ratio formed upon hydrolysis of the
TAG triolein), which resulted in complete DOX within 2 h at
temperature and net-zero H2 consumption. Catalyst characteri-
zation after hydrothermal exposure revealed that moderate
sintering had occurred, suggesting additional work is needed
to evaluate Pt–Re/C stability with time on stream. A H2-neutral
(or positive) scheme for hydrothermal catalytic conversion of
TAGs is proposed, involving TAG hydrolysis, catalytic reform-
ing of glycerol, and fatty acid hydrogenation and deoxygena-
tion processes resulting in saturated hydrocarbon products.
Results from this study suggest a promising strategy for over-
coming some of the major limitations of conventional lipid-to-
biofuel pathways.
Fig. 10 Proposed reaction scheme and product yields for hydrothermal
processing of the triacylglyceride triolein into hydrocarbons and in situ
H2.
Following hydrolysis, liberated glycerol can undergo cata-
lytic decomposition reactions to generate H2 and CO, with CO
undergoing the WGS reaction to produce additional H2 over
the appropriate catalyst (Rxn. 5).30 Collectively, aqueous phase
reforming can theoretically produce 7 moles of H2 from
1 mole of glycerol.
As noted, free fatty acids generated from lipid hydrolysis
can contain varying ratios of saturated, unsaturated, and poly-
unsaturated fatty acids, depending on the feedstock origin. For
example, waste grease feedstocks (e.g., yellow grease, brown
grease, tallow) may contain saturated fatty acids ranging from
39–52% of the total fatty acid profile. Algal lipid feedstocks, in
contrast, may contain as low as 10% saturated fatty acids,3 and
over 30% polyunsaturated fatty acids depending on the
strain.46 Fatty acid unsaturation can impart a H2 consumption
demand since HYD to the corresponding saturated fatty acids
(Rxn. 6) is necessary prior to DOX of the carboxylic acid group
(Rxn. 1 and Rxn. 2). The continuous H2 demand can poten-
tially be met by APR of glycerol released upon lipid hydrolysis,
yielding a net neutral or net positive system H2 balance. Lipid
feedstocks can be blended prior to processing to maintain an
optimal degree of unsaturation during continuous processing.
Additional H2 can initially be introduced to accelerate kinetics;
however, unlike conventional hydrotreatment processes, con-
tinuous H2 input may not be required.
4. Materials and methods
4.1. Materials
Stearic acid (98%) was obtained from TCI Chemicals and oleic
acid (90%) was obtained from Alfa Aesar. Glycerol (99.5%),
activated carbon (Darco 100 mesh), powdered Pt/C (5 wt%)
catalyst, and ammonium perrhenate salt (NH4ReO4) were pur-
chased from Sigma-Aldrich. Hydrocarbon standards (e.g., pen-
tadecane, hexadecane, heptadecane, octadecane) were
obtained from Chem Service Inc. Analytical-grade dichloro-
methane (DCM) and methanol were obtained from Fisher
Scientific. For a typical experiment, Pt–Re/C was prepared via
Saturated fatty acids undergo DOX during catalytic hydro-
thermal processing to produce linear hydrocarbons with high
selectivity. The DOX pathway may progress via decarboxylation
(Rxn. 1), or decarbonylation (Rxn. 2) rapidly followed by HYD
of the terminal alkene (Rxn. 4).
Although terminal alkene HYD includes consumption of
H2, the overall decarbonylation process is H2 neutral because
This journal is © The Royal Society of Chemistry 2014
Green Chem., 2014, 16, 1507–1520 | 1517

View Article Online
Paper
Green Chemistry
aqueous adsorption of NH4ReO4 onto 0.5 g of Pt/C catalyst sus- (1253.6 eV). The pass energy employed for high resolution
pended in 50 g of deionized (DI) water. The contents were Pt and Re XPS spectra was 35.8 eV. Pt–Re/C catalysts for XPS
loaded into a Parr 4575 500 mL high temperature/pressure analysis were exposed to hydrothermal media (300 °C, 3 h)
reactor. Pt/C and Pt–Re/C catalysts were pre-reduced in situ using the Parr 4575 reactor loaded initially with either an inert
prior to use by purging and venting the reactor with 1.38 MPa (0.35 MPa/56 mmol N2) or reducing (2.59 MPa/417 mmol H2)
of H2 over three cycles, followed by heating to 200 °C with con- headspace. After exposure, catalysts were sampled in situ using
tinuous mixing for 2 h at temperature. The reactor was then a Parr 4532-D collection attachment. The isolation valve on the
cooled to ambient temperature and maintained under positive reactor was briefly released to charge the sampling accessory,
pressure with either an inert (N2) or reducing (H2) gas.
which was then cooled to ambient pressure using a water-
cooling sleeve. To prevent air exposure, catalyst samples
were dried inside an anoxic glove box and loaded onto conduc-
4.2. Catalyst characterization
Catalyst metal loadings were measured using a Spector Arcos tive copper tape located inside an anoxic XPS-sample
Inductively Couples Plasma Atomic Emission Spectrometer holder chamber, as described previously.35 High-resolution
(ICP-AES) after digestion in concentrated acid. Support surface (i.e., 0.1 eV resolution) XPS spectra were collected from 38.0 eV
area, pore volume, and average pore diameter were determined to 58.0 eV for Re and 68.0 eV to 88.0 eV for Pt. Spectra were
by multipoint N2 physisorption using a QuandraSorb SI analy- energy-normalized relative to the dominant C 1s peak from the
zer. Catalyst specific surface area was calculated by BET, and activated carbon support (284.5 eV). Spectra were then normal-
pore volume and pore diameter were calculated by BJH deso- ized and fit with a convoluted Gaussian and Lorentzian func-
rption. Prior to analysis, catalysts were outgassed at 250 °C for tion after constraining the separation distance and peak area
18 h. Physisorption measurements were performed at the ratio of contributing doublet peaks with CasaXPS version 2.3
temperature of liquid N2.
(Casa Software Ltd).
Catalyst metal dispersion and the influence of Re on Pt was
determined by CO and H2 chemisorption using a Micro-
4.3. Catalytic hydrothermal processing
meritics Autochem II 2920 pulse analyzer. Prior to analysis, Hydrothermal conversions were conducted in the Parr 4575
catalysts were degassed at 40 °C for 0.2 h under Ar, dried at reactor. The reactor containing pre-reduced catalyst was loaded
100 °C, and reduced at 280 °C in flowing 10% H2/Ar (50 mL with the 20 g of oleic acid, and 5 g of glycerol dissolved in 30 g
min−1) for 1 h. Catalysts were then purged at 280 °C for 0.5 h of DI (80 g DI total including water from catalyst pre-
with either He (for CO chemisorption) or Ar (for H2 chemi- reduction). The reactor vessel was pressurized to 1.38 MPa
sorption). Chemisorption measurements were performed at with the desired headspace gas (N2 or H2) and purged for
45 °C and metal dispersion was calculated based on CO three cycles under constant stirring. The headspace pressure
uptake with a CO : M ratio of unity.
was then increased to the desired initial value corrected to
Long-range catalyst material order was analyzed by XRD 25 °C. Reactor temperature was raised to 300 °C at a rate of
using a Rigaku Ultima IV X-ray diffractometer. The X-ray ∼10 °C min−1 (heat-up time ∼1 h) with constant stirring at
voltage was set to 40 kV and 44 mA, with a sampling width of high speed (>1000 rpm) applied for the desired reaction time.
0.02° and scan speed of 5° min−1. Metal crystallite size para- Once the reaction time elapsed, the reactor was rapidly cooled
meters were estimated using the Scherer equation with PDXL to ambient temperature by initiating water flow through
version 1.6.0.1 (Rigaku Corporation).
internal cooling coils. For time series studies, independent
Imaging and elemental of mapping of multiple catalyst par- batch reactions were performed for each time point. Where
ticles was performed by SEM-EDS using a FEI Quanta 400 FEG indicated, triplicate reactions were performed independently
scanning electron microscope equipped with an EDAX X-ray to determine experimental variability, with error bars repre-
detector. Samples were mounted onto aluminum stubs using a senting standard deviations.
conductive carbon tape to adhere the particles to the stub and
provide electrical conductivity. Elemental mapping was per-
4.4. Reaction product analysis
formed using an accelerated voltage of 30 kV. The signal was The total reactor contents were collected, fractionated, and
based on net X-ray intensity using a Z absorption florescence analyzed gravimetrically to ensure mass balance closure of the
(ZAF) correction over a dwell time of 300 ms per pixel.
aqueous and organic product phases. After confirming mass
High-resolution TEM images of individual catalyst particles balance closure, the recovered reactor contents were heated
were collected using a JEOL 2010LAB6 TEM instrument. while stirring to produce a bi-phase system for sampling. The
Samples were dispersed in acetone and mounted onto holey top organic layer containing fatty acids and hydrocarbons was
carbon Cu grids prior to analysis. STEM-EDS elemental sampled, dissolved in DCM (1 wt% sample), and filtered
mapping of metallic crystallites was also performed using a (0.45-µm PTFE) prior to analysis to remove catalyst particles.
JEOL 2010F field emission TEM instrument, coupled to an Dissolved glycerol in the bottom aqueous phase was sampled,
Oxford INCA EDS detector with a 1 nm probe size.
dissolved in methanol (1 wt% sample) and filtered (0.45-µm
Catalyst surface oxidation states were examined using a PTFE) prior to analysis. As a control, each complete bilayer was
Physical Electronics PHI 5400 X-ray photoelectron spectro- also dissolved in excess solvent and analyzed to ensure repre-
meter equipped with a monochromatized Mg Kα source sentative results from sampling.
1518 | Green Chem., 2014, 16, 1507–1520
This journal is © The Royal Society of Chemistry 2014

View Article Online
Green Chemistry
Paper
Fatty acids, alkanes, alcohols, (e.g., stearyl alcohol), glycerol views of the EPA. This material is also based upon work sup-
and glycol derivatives (e.g., ethylene glycol, propylene glycol) ported by the National Science Foundation Graduate Research
were quantified using a HP 5890 Series II FID gas chromato- Fellowship (DGE-1144245). Any opinion, findings, and con-
graph outfitted with a Restek Stabilwax-DA column (30 m × clusions or recommendations expressed in this material are
0.25-mm id, 0.25-µm film). Helium (6 mL min−1 column flow) those of the authors and do not necessarily reflect the views of
was used as the carrier gas with the injector split flow set to the National Science Foundation. In addition, financial
60 ml min−1. The oven temperature was increased from 40 °C support was provided by the Department of Civil and Environ-
to 250 °C at 10 °C min−1. The injector volume was set to 1 µL mental Engineering at the University of Illinois and the
and the inlet temperature set at 250 °C. The detector tempera- National Science Foundation Division of Chemical, Bioengi-
ture was set to 250 °C, H2 gas set to 32 ml min−1, and airflow neering,
set to 400 ml min−1. Compound retention times and FID-cali- (CBET-0746453).
bration response curves were conducted using fatty acid, gly- We thank Rick Haasch for assistance with XPS analysis at
Environmental,
and
Transport
Systems
cerol, and alkane standards. Reaction products were also the Center of Microanalysis of Materials (UIUC), which is sup-
confirmed by GC-MS analysis using a Varian 3800 gas chro- ported by the U.S. Department of Energy under grant DEFG02-
matograph with a Varian 2000 mass spectrometer and Restek 91-ER45439. Catalyst characterization was also supported in
RTX-5MS column and by 1H NMR using a Varian Unity part by the Center for Direct Catalytic Conversion of Biomass
400 MHz spectrometer with conditions described previously.47
to Biofuels (C3Bio), an Energy Frontier Research Center
Reactor headspace gas samples were collected after the funded by the U.S. DOE, Office of Science, Office of Basic
reaction using 0.5-L Restek Tedlar sample bags, and analyzed Energy Sciences, Award Number DE-SC0000997. Additionally,
for H2, CH4, CO, and CO2 using a HP 5890 Series II TCD gas we would like to thank Martin Menart at the Colorado School
chromatograph outfitted with
a Carboxen-1010 Supelco of Mines for assistance with catalyst chemisorption
column. For the carrier gas, N2 (1 mL min−1) was used to measurements.
analyze H2, while He was used to analyze carbon-based gases
(1 mL min−1). A split flow of 10 mL min−1 was used with an
inlet temperature of 225 °C. An isothermal oven temperature
profile was used to detect H2 (35 °C) and carbon-based gases
References
(125 °C). The TCD detector was set to 225 °C at high sensitivity
with a reference gas flow of 15 mL min−1 and auxiliary gas
1 U.S. EIA, 2011.
flow of 7 mL min−1
.
2 T. N. Kalnes, K. P. Koers, T. Marker and D. R. Shonnard,
Environ. Prog. Sustainable Energy, 2009, 28, 111–120.
3 A. Berenblyum, V. Danyushevsky, E. Katsman,
T. Podoplelova and V. Flid, Pet. Chem., 2010, 50, 305–311.
4 E. Santillan-Jimenez and M. Crocker, J. Chem. Technol. Bio-
technol., 2012, 87, 1041–1050.
5 J. Fu, X. Lu and P. E. Savage, Energy Environ. Sci., 2010, 3,
311–317.
6 J. Fu, X. Lu and P. E. Savage, ChemSusChem, 2011, 4, 481–
486.
7 P. Biller, R. Riley and A. B. Ross, Bioresour. Technol., 2011,
102, 4841–4848.
8 T. M. Yeh, J. G. Dickinson, A. Franck, S. Linic,
L. T. Thompson and P. E. Savage, J. Chem. Technol. Biotech-
nol., 2013, 88, 13–24.
9 A. A. Peterson, F. Vogel, R. P. Lachance, M. Fröling,
M. J. Antal, Jr. and J. W. Tester, Energy Environ. Sci., 2008,
1, 32–65.
10 P. Jayasinghe and K. Hawboldt, Renewable Sustainable
Energy Rev., 2012, 16, 798–821.
Hydrocarbon product molar yields were calculated by divid-
ing the moles of recovered product by the moles of fatty acid
initially loaded into the reactor. Hydrocarbon selectivities were
calculated by dividing the moles of hydrocarbon product by
the moles of converted fatty acid. In situ H2 production yields
were determined using the ideal gas law based on the final
headspace gas pressure and compositional analysis deter-
mined by GC-TCD. Due to the low solubility of H2 in water at
ambient temperature compared to the gas loadings examined
in this study, dissolved H2 was assumed to be negligible. For
this work, H2 selectivity from glycerol APR is defined as the
ratio of H2 produced, divided by the theoretical amount of H2
produced if the converted glycerol was completely reformed
(e.g., decomposition and WGS) to generate H2 and CO2. Appar-
ent reaction rate constants for stearic acid deoxygenation were
estimated assuming pseudo-first order kinetics using non-
linear regression with GraphPad Prism version 6.00 (GraphPad
Software).
11 M. N. Siddiquee and S. Rohani, Renewable Sustainable
Energy Rev., 2011, 15, 1067–1072.
12 L. Brennan and P. Owende, Renewable Sustainable Energy
Rev., 2010, 14, 557–577.
Acknowledgements
The research described in this paper has been funded in part
by the United States Environmental Protection Agency (EPA) 13 J. K. Pittman, A. P. Dean and O. Osundeko, Bioresour.
under the Science to Achieve Results (STAR) Graduate Fellow- Technol., 2010, 102, 17–25.
ship Program. The EPA has not officially endorsed this publi- 14 J. A. Lercher, C. Zhao and T. Brueck, Green Chem., 2013, 15,
cation and the views expressed herein may not reflect the
1720–1739.
This journal is © The Royal Society of Chemistry 2014
Green Chem., 2014, 16, 1507–1520 | 1519

View Article Online
Paper
Green Chemistry
15 R. D. Cortright, R. R. Davda and J. A. Dumesic, Nature, 30 R. R. Davda, J. W. Shabaker, G. W. Huber,
2002, 418, 964–967.
16 J. Shabaker, G. Huber and J. Dumesic, J. Catal., 2004, 222,
180–191.
17 N. Luo, X. Fu, F. Cao, T. Xiao and P. P. Edwards, Fuel, 2008,
87, 3483–3489.
18 D. L. King, L. Zhang, G. Xia, A. M. Karim, D. J. Heldebrant,
R. D. Cortright and J. A. Dumesic, Appl. Catal., B, 2005,
56, 171–186.
31 R. R. Soares, D. A. Simonetti and J. A. Dumesic, Angew.
Chem., Int. Ed., 2006, 118, 4086–4089.
32 J. G. Immer and H. H. Lamb, Energy Fuels, 2010, 24, 5291–
5299.
X. Wang, T. Peterson and Y. Wang, Appl. Catal., B, 2010, 99, 33 J. G. Immer, M. J. Kelly and H. H. Lamb, Appl. Catal., A,
206–213. 2010, 375, 134–139.
19 D. Ö. Özgür and B. Z. Uysal, Biomass Bioenergy, 2011, 35, 34 J. W. Shabaker and J. A. Dumesic, Ind. Eng. Chem. Res.,
822–826. 2004, 43, 3105–3112.
20 L. Zhang, A. M. Karim, M. H. Engelhard, Z. Wei, D. L. King 35 J. K. Choe, J. R. Shapley, T. J. Strathmann and C. J. Werth,
and Y. Wang, J. Catal., 2012, 287, 37–43. Environ. Sci. Technol., 2010, 44, 4716–4721.
21 E. L. Kunkes, D. A. Simonetti, J. A. Dumesic, W. D. Pyrz, 36 NIST, X-Ray Photoelectron Spectroscopy Database, 2012.
L. E. Murillo, J. G. Chen and D. J. Buttrey, J. Catal., 2008, 37 H. P. Boehm, Carbon, 1994, 32, 759–769.
260, 164–177.
38 J. Lu, S. Behtash and A. Heyden, J. Phys. Chem. C, 2012,
116, 14328–14341.
22 A. S. Berenblyum, T. A. Podoplelova, R. S. Shamsiev,
E. A. Katsman and V. Y. Danyushevsky, Pet. Chem., 2011, 39 J. Lu, S. Behtash, M. Faheem and A. Heyden, J. Catal.,
51, 336–341. 2013, 305, 56–66.
23 B. Rozmysłowicz, P. Mäki-Arvela, A. Tokarev, A.-R. Leino, 40 H. N. Pham, A. E. Anderson, R. L. Johnson, K. Schmidt-
K. Eränen and D. Y. Murzin, Ind. Eng. Chem. Res., 2012, 51,
8922–8927.
Rohr and A. K. Datye, Angew. Chem., Int. Ed., 2012, 51,
13163–13167.
24 Y. T. Kim, J. A. Dumesic and G. W. Huber, J. Catal., 2013, 41 L. Lascaray, Ind. Eng. Chem., 1949, 41, 786–790.
304, 72–85.
42 T. A. Patil, D. N. Butala, T. S. Raghunathan and
H. S. Shankar, Ind. Eng. Chem. Res., 1988, 27, 727–735.
43 R. L. Holliday, J. W. King and G. R. List, Ind. Eng. Chem.
Res., 1997, 36, 932–935.
25 D. A. Simonetti, E. L. Kunkes and J. A. Dumesic, J. Catal.,
2007, 247, 298–306.
26 B. H. Isaacs and E. E. Petersen, J. Catal., 1984, 85, 1–7.
27 B. Rozmysłowicz, P. Mäki-Arvela and D. Murzin, in Biomass 44 F. Ma and M. A. Hanna, Bioresour. Technol., 1999, 70, 1–15.
Conversion, ed. C. Baskar, S. Baskar and R. S. Dhillon, 45 S. Changi, A. J. Matzger and P. E. Savage, Green Chem.,
Springer, Berlin Heidelberg, 2012, pp. 199–220.
2012, 14, 2856.
28 M. Behrens and A. K. Datye, Catalysis for the Conversion of 46 M. J. Griffiths and S. T. L. Harrison, J. Appl. Physiol., 2009,
Biomass and Its Derivatives, Max Planck Research Library
for the History and Development of Knowledge, 2013.
29 J. Fu, F. Shi, L. T. Thompson, X. Lu and P. E. Savage, ACS
Catal., 2011, 1, 227–231.
21, 493–507.
47 D. R. Vardon, B. K. Sharma, J. Scott, G. Yu, Z. Wang,
L. Schideman, Y. Zhang and T. J. Strathmann, Bioresour.
Technol., 2011, 102, 8295–8303.
1520 | Green Chem., 2014, 16, 1507–1520
This journal is © The Royal Society of Chemistry 2014