Decomposition of glucose with in situ deoxygenation in a low H2 pressure environment – Pt. II: Bimetallic catalysts

Article abstract of DOI:10.1016/j.apcata.2019.03.013

Bimetallic catalysts, CoFe, NiFe, PdFe, and PtCo were studied for their ability to perform in situ deoxygenation of glucose decomposition production at 350 °C. Catalysts were prepared via co-impregnation and sequential impregnation methods on SiO₂

Full text of DOI:10.1016/j.apcata.2019.03.013

AppliedCatalysisA,General578(2019)10–19
Contents lists available at ScienceDirect
Applied Catalysis A, General
journal homepage: www.elsevier.com/locate/apcata
Decomposition of glucose with in situ deoxygenation in a low H2 pressure
environment – Pt. II: Bimetallic catalysts
⁎
Kyle A. Rogersa, Ying Zhenga,b,
a Department of Chemical Engineering, University of New Brunswick, 15 Dineen Drive, Fredericton, E3B 5A3, NB, Canada
b Department of Chemical and Biochemical Engineering, Western University, 1150 Richmond St, London, ON, N6A 3K7, Canada
A R T I C L E I N F O
A B S T R A C T
Keywords:
Bimetallic catalysts, CoFe, NiFe, PdFe, and PtCo were studied for their ability to perform in situ deoxygenation of
glucose decomposition production at 350 °C. Catalysts were prepared via co-impregnation and sequential im-
pregnation methods on SiO2 in combinations of 4 wt% and 0.5 wt% for the two metals. Easily identifiable furanic
compounds in the products were used to establish an understanding of deoxygenation activity. Co-impregnated
0.5%Co4%Fe/SiO2 and sequentially impregnated (Fe first) 0.5%Ni4%Fe/SiO2 catalysts provided the most fa-
vourable results. Thanks to the alloying of Co and Fe which stabilized Fe, the 0.5%Co4%Fe/SiO2 catalyst
eliminated furfural content and maintained a higher output of H2. The 0.5%Ni4%Fe/SiO2 catalyst required the
Ni and Fe phases to remain uninterrupted to perform their synergistic roles which boosted DMF content and
overall reduced solid residue.
Deoxygenation
Water-gas-shift
Glucose
Decomposition
Bimetallic catalysts
1. Introduction
reaction stage in which biomass is decomposed and the resulting pro-
ducts can immediately undergo deoxygenation in the presence of var-
Traditional technologies for producing petroleum-compatible bio-
fuels from lignocelluosic materials are heavily reliant on the use of
hydrogen. Thermochemical processes like pyrolysis produce oxygen-
rich bio-oil which must be upgraded via hydroprocessing [1–7]. Al-
though these processes claim to be renewable/sustainable they remain
reliant on fossil fuels as > 95% of the world’s H2 is produced via fossil
fuel resources [8].
The presence of CO2, CO, H2, H2O, and light hydrocarbons in the
outlets of lignocellulosic decomposition processes represents a missed
opportunity [2]. CO and H2O for example can potentially undergo
water-gas-shift reaction (WGSR) and produce H2 which can then be
used for in situ deoxygenation. The concept of producing H2 internally
and using it to produce renewable fuels in not entirely new. For ex-
ample, Dumesic’s research group from the University of Wisconsin de-
monstrated the aqueous phase reforming of sugars such as glucose to
produce hydrogen which was followed by upgrading with Pt catalyst
[10–12]. Aside from requiring a potentially expensive catalyst; such a
system would require initial processing of lignocellulosic material to
first separate lignin and then produce water-soluble sugars from cel-
lulose and hemicellulose. On the other hand, a thermochemical process
involving in situ deoxygenation may utilize raw lignocellulosic material
without the need of significant processing of the feedstock.
ious monometallic catalysts. Using glucose as a model compound for
cellulose, we reported the reduction of furfural into selectively deox-
ygenated products such as 2-methylfuran (2 M F) an 2,5-dimethylfuran
(DMF) using a 4%Co/SiO2. Although Ni catalysts were not as selective
towards 2 M F and DMF, they were however beneficial for reducing the
production of solid residue.
Bimetallic catalysts have received much attention for their activities
in the selective deoxygenation of phenolic and furanic compounds de-
rived from bio-oils [5]. Resasco and colleagues from the University of
Oklahoma studied the use of supported Ni-Fe bimetallic catalysts for the
selective deoxygenation of m-cresol and furfural and determined that
Fe’s oxophilic nature facilitated selective absorption via oxygen contain
functional groups while Ni facilitated H2 activation together permitting
hydrogenolysis [13,14]. Pd-Fe bimetallic catalysts have also received
much attention however Pd was reportedly responsible stabilizing Fe
and preventing permanent oxidation [15–18]. Other bimetallic cata-
lysts have included supported PtZn [19], supported NiCu [20,21], and
PtCo nanocrystal [22] catalysts.
Herein, Pt. 2 of the decomposition of glucose with in situ deox-
ygenation in a low H2 pressure environment focuses on the use of
supported bimetallic catalysts: NiFe, CoFe, PdFe, and PtCo. As with Part
1, this study used glucose as a model for cellulose. Since Ni/SiO2 and
Co/SiO2 were shown to yield the most ideal results in Part 1; within this
In Part 1 [9], we investigated the potential of performing a one-pot
⁎ Corresponding author at: Department of Chemical Engineering, University of New Brunswick, 15 Dineen Drive, Fredericton, E3B 5A3, NB, Canada.
E-mail address: ying.zheng@uwo.ca (Y. Zheng).
https://doi.org/10.1016/j.apcata.2019.03.013
Received 25 January 2019; Received in revised form 15 March 2019; Accepted 19 March 2019
Availableonline20March2019
0926-860X/©2019ElsevierB.V.Allrightsreserved.

K.A. Rogers and Y. Zheng
AppliedCatalysisA,General578(2019)10–19
study, NiFe and CoFe catalyst received much attention and were also
studied for the impact that impregnation method (co-impregnation
versus sequential/step-wise impregnation) has on the activity. Unlike
NiFe bimetallic catalysts, CoFe catalysts have previously not received
much or any attention within the literature for the use as a deox-
ygenation catalyst. They have however received some attention for
their potential use for WGSR which may also be useful in the present
system [23–25]. Although some of the aforementioned catalysts have
been used in WGSR studies; the in-situ production of H2 is not yet the
primary focus for the catalysts. As such, a small source of H2 is provided
(37.6 kPaa) in order to maintain activity of deoxygenation catalysts.
Future research may further investigate in-situ hydrogen production.
Furanic compounds in the products, which have distinct structures and
deoxygenation pathways, were used to establish an understanding of
deoxygenation activity.
from the reactor, the solid contents were washed with acetone and
dried in an oven at 80 °C overnight. Liquid outputs were analyzed using
GC–MS (Shimadzu GCMS QP5000) and GC-FID (Varian GC450). The
hydrogen to carbon ratio of the solid residue was determined via CHNS
analysis (Leco CHNS-932).
TGA analyses were performed using a TA Instruments TGA Q500 to
determine the direct impact of selected catalysts on the decomposition/
weight loss of glucose. The catalysts and glucose were loaded into the
TGA at a ratio of 1:5. The TGA was heated from room temperature to
950 °C at a rate of 5 °C/min under flowing He (50 nm L/min).
2.3. Catalyst characterization
Catalysts selected for characterization underwent TPR, XRD, and
(S)TEM analyses with the objective of investigating the bimetallic cat-
alyst structures and how the catalysts differ based on preparation
method. Temperature programmed reduction (TPR) analyses were
performed using an Autosorb 1-C with a RGA200 mass spectrometer as
the detector. 150 mg of catalyst was loaded in a quartz u-cell and va-
cuum dried at 300 °C for 20 min under flowing helium (50 nm L/min).
TPR analyses was then performed using 1% H2 in He gas mixture at a
flow rate of 50 nm L/min. The Quartz cell was heated from room tem-
perature to 900 °C at a rate of 10 °C/min. X-ray diffraction (XRD) of
reduced catalysts was performed at the UNB Geochemical and
Spectrographic Facility using a Bruker AXS D8 XRD. XRD spectrums
were gathered within the 2Theta range of 10-90° at a rate of 0.02°/
second.
(S)TEM analyses were performed at the UNB Microscopy and
Microanalysis Facility with a JEOL JEM-2010 (S)TEM. Images were
collected with a Gatan Ultrascan camera using Digital Micrograph.
Metallic sites were identified on the supporting material (SiO2) using
both HRTEM (High Resolution TEM) and high angle annular darkfield
STEM (HAADF-STEM) mode. With both techniques, a high contrast was
observed between catalyst sites and the supporting SiO2 material.
Verification of the metallic sites was determined by collecting EDS
spectra with an EDAX Genesis 4000 Energy Dispersive X-ray (EDS)
analyser. In addition, mapping of the metallic sites was performed using
the EDS detector whilst in STEM mode to further determine the ex-
istence and distribution of both metals across an area of material. An
accelerating voltage of 200 kV was used for the imaging and analysis.
Images were processed using an open source image processing plat-
form, Fiji [32].
2. Experimental
2.1. Catalyst preparation
Various bimetallic catalysts were prepared using incipient wetness
impregnation onto a SiO2 supporting material (Aerosil 200). Both co-
impregnation and step-wise impregnation methods were used. The
catalysts that were prepared consisted of NiFe, CoFe, PtCo, and PdFe
catalysts. Catalysts were prepared with one metal at 0.5 wt% and the
other at 4 wt%. Catalysts were prepared using the following salts from
Sigma Aldrich: nickel (II) nitrate hexahydrate, cobalt (II) nitrate hex-
ahydrate, iron (III) nitrate nonahydrate, tetraamineplatinum (II) ni-
trate, and palladium (II) nitrate dihydrate. Before impregnation, the
silica support was dried at 120 °C for 6 h and calcined at 550 °C for 4 h.
After impregnation, catalysts were dried at 120 °C for 6 h and calcined
in air at 550 °C for 6 h with heating rates of 5 °C/min.
Step-wise impregnation was performed by impregnating the first
metal onto the support then performing calcination. Upon cooling of
the material; the second impregnation was performed followed by an-
other calcination process. Catalysts that were prepared via step wise
impregnation were denoted as “SWM” where M represents the metal
that was impregnated first (ie. 4%Ni0.5%Fe/SiO2-SWFe – Fe was im-
pregnated onto the support first followed by the 4%Ni).
For performance testing, catalysts were pelletized and crush to a
size range of 40–60 mesh. Catalyst were reduced in flowing H2 at
50 nm L/min at desired reduction temperatures with a heating rate of
10 °C/min. Temperatures used for reduction were 450 °C for NiFe/SiO2,
550 °C for CoFe/SiO2, 450 °C for PdFe/SiO2, and 450 °C for PtCo/SiO2
[13,14,18,26].
3. Results and discussion
3.1. Solid products
2.2. Activity tests
A high level of reduction of the solid content is desired to attain a
liquid product and provide easy separation from the catalyst. Reduction
of the solid content results from the conversion of glucose itself and the
prevention of the formation of solid products such are char and heavy
compounds and polymers attributed as tars. Catalysts may prevent this
by supporting tar cracking/reforming reactions to eliminate heavy
compounds in favour of liquid and gaseous products [33–39]. As de-
monstrated in Fig. 1, most bimetallic catalysts were shown to aid in the
reduction of solid residue over tests with just glucose. This is also de-
monstrated in Fig. 2 where the H:C ratios of the solid residue were
higher upon addition of the catalysts representing the possibility of less
char formation.
As evidenced in Fig. 2, NiFe catalysts were the most capable at re-
ducing output solid residue. Based on what was previously observed for
monometallic catalysts in Part 1 [9], it is likely that the presence of Ni
had a significant role. 0.5%Ni4%Fe-SWFe for example provided a si-
milar reduction in solid residue to 0.5%Ni/SiO2 as well as a very similar
H:C ratio. Meanwhile, the co-impregnated and the SWNi variants of the
0.5%Ni4%Fe catalyst had comparatively lower levels of reduction in
Activity tests were performed in accordance with Part 1 of the study
[9]. Activity/performance tests were performed in a horizontal, stirred
55 mL stainless steel reactor for 1 h at 350 °C with 800 mg of glucose.
Reduced catalyst catalysts were used in the reactor at a 1:5 catalyst to
glucose ratio. 1-methylnaphthalene (1 MN) (Sigma-Aldrich) was used
as a solvent (20:1 solvent to glucose ratio) to promote heat and mass
transfer. 1 MN was once again selected as the solvent due to its high
boiling point temperature and its stability at reaction conditions even
within the presence of a catalyst showing no discernable changes in H2
content and 1 MN related by-products within the liquid phase. 350 °C
was selected as a reaction temperature as it represents a suitable tem-
perature for both cellulose decomposition and WGSR [27–31]. Prior to
reactions, the filled reactors were vacuum purged with He then filled
with 37.6 kPaa of H2 then topped off with He to a total pressure of 310
kPaa to prevent evaporation of the solvent.
Gases within the reactor were tested using a RGA200 mass spec-
trometer at room temperature after undergoing separation via a HP-
PLOT/U column held at 30 °C. Upon separation from the liquid contents
11

K.A. Rogers and Y. Zheng
AppliedCatalysisA,General578(2019)10–19
produce a more uniform NiFe alloy phase rather than a strictly separate
Ni-phase which is more active towards either the initial decomposition
of glucose or tar cracking (see catalyst characterization in Section 3.4).
When 0.5%Fe is layered on top of the 4%Ni, it may be possible that
both Ni and Fe phases are exposed. The Fe phase may not completely
cover the Ni phase possibly enabling a synergistic affect between the
two metals in which each metal favours a different role in the reaction.
This would explain why the SWNi performed slightly better than the
SWFe layout which would have had a more completely exposed Ni
phase.
CoFe catalysts appear to behave in the opposite way and favour co-
impregnation over the step-wise impregnation methods. Co-im-
pregnated 4%Co0.5%Fe reduced the organic solid content by ˜77% –
slightly higher than the 4%Co/SiO2 catalyst from the previous study
with 74%. Even more subtle differences are witnessed for the two se-
quentially impregnated 4%Co0.5%Fe catalysts. It is noteworthy too
that as monometallic catalysts, 0.5%Fe/SiO2 and 4%Co/SiO2 resulted
in similar reductions in organic solid residue. Combing the two into a
bimetallic catalyst apparently offered no significant improvement.
Similar observations can be made for the 0.5%Co4%Fe catalysts.
For PdFe and PtCo, an odd observation was made where exposing
the noble metals on top of the base metals provided slightly lower re-
duction in solid residue. However, in general, using bimetallic config-
urations of NiFe, CoFe, PdFe, and PtCo provides little advantage over
their monometallic counterparts in terms of solid residue reduction.
Especially in the cases of PdFe and PtCo catalysts, the additional costs
of these catalysts outweigh the benefits.
Fig. 1. Percent removal of dry organic solids with bimetallic catalysts with data
for 0.5%Ni/SiO2 from Part 1 [9].
To further understand the impact the catalysts, have on the de-
composition of glucose, TGA was performed using 0.5%Ni4%Fe and
0.5%Co4%Fe catalysts. As demonstrated in Fig. 3, the catalysts, had a
significant effect on the decomposition of glucose by reducing the
temperatures at which substantial decomposition begins. Without a
catalyst additive, the decomposition of glucose has two main periods of
decomposition one at 215 °C and again at 293 °C where glucose un-
dergoes a faster rate of decomposition. With the catalyst additives
though, the first stage of decomposition had the higher rate of de-
composition however it was not as significantly fast. As seen in Table 1,
the results for the two chosen catalysts and their three variants did not
have appreciable differences. Iron played an important role in influ-
encing the decomposition of glucose as the results are mostly compar-
able to those of 4%Fe/SiO2 from Part 1. In some cases, though, there are
discernable differences such as for co-impregnated 0.5%Ni4%Fe and
0.5%Co4%Fe catalysts (as shown in Fig. 3) where the catalyst con-
taining nickel achieved an overall lower residue content at the end of
Fig. 2. H:C Molar Ratio in solid products.
solid residue. For these two catalysts, the Ni phase may not have been
directly exposed like the SWFe catalyst and thus may have suffered
from an interference with Fe. However, the second highest reduction in
solid residue occurred on the 4%Ni0.5%Fe-SWNi catalyst which may
not have had Ni-phase that was exposed as the 4%Ni0.5%Fe-SWFe
variant which had slightly more solid residue. Meanwhile, the co-im-
pregnated 4%Ni0.5%Fe catalyst suffered a loss in solid reduction
compared to the 4%Ni/SiO2 catalyst in Part 1. Co-impregnation may
Fig. 3. TGA results of glucose decomposition with and without 0.5%Ni4%Fe/SiO2 and 0.5%Co4%Fe/SiO2 catalyst. Catalysts added at 1:5 ratio with glucose.
12

K.A. Rogers and Y. Zheng
AppliedCatalysisA,General578(2019)10–19
Table 1
TGA results summary for glucose decomposition with catalyst additives. Temperature (T) program: 25 °C–900 °C at 5 °C/min. Y = percent of solid remaining (does
not include catalyst weight).
Catalyst
Initial Major Weight Loss
Valley
Second Major Weight Loss
Ta(°C) Ya(wt%) Tb(°C) Yb(wt%) (-dY/dT)b(%/K) Tc(°C) Yc(wt%) (-dY/dT)c(%/K) Td(°C) Yd(wt%) (-dY/dT)d(%/K) Te(°C) Ye(wt%)
No Catalyst
0.5%Ni4%Fe/SiO2
0.5%Ni4%Fe/SiO2 SWFe 170
0.5%Ni4%Fe/SiO2 SWNi 171
0.5%Co4%Fe/SiO2
178
173
99.6%
99.5%
99.6%
99.8%
99.6%
99.6%
99.7%
215
211
194
200
195
200
199
90.0%
81.6%
89.5%
87.2%
87.9%
87.0%
86.6%
0.49
0.63
0.67
0.61
0.61
0.64
0.68
242
244
230
234
227
236
234
79.4%
60.3%
66.7%
66.8%
67.5%
64.4%
64.2%
0.30
0.46
0.40
0.42
0.43
0.44
0.43
293
253
242
246
245
248
245
49.1%
54.6%
60.8%
60.5%
57.9%
58.0%
58.2%
0.92
0.56
0.44
0.49
0.48
0.51
0.47
417
435
431
433
423
434
435
16.4%
12.2%
24.9%
21.2%
23.3%
18.8%
20.7%
169
0.5%Co4%Fe/SiO2 SWFe 172
0.5%Co4%Fe/SiO2 SWCo 171
a
Point at which mass loss rate begins to exceed 0.05%/K.
First peak point on DTG curve.
Valley point on DTG curve between first and second peak.
Second peak point on DTG curve.
b
c
d
e
Point at which mass loss rate declines below 0.05%/K.
the analysis.
Given that 4%Fe likely played an important role in glucose de-
previous study, 0.5%Co and 4%Fe catalysts produced relatively low
amounts of H2. Combining the two however appears to have a greater
impact due to either a synergistic effect or combined activity. Seeing as
how no activity combination effect is witnessed for the 4%Co0.5%Fe
catalyst, it is possible there is a synergistic role between the two metals
which is most present when they are co-impregnated. A similar trend is
also noted for CO2 production highlighting the likelihood that WGSR
occurred. The fact that the co-impregnation method provided the most
H2 suggests that there is a stronger relationship between the two metals
when they are formed together perhaps as an alloy. It is suggested that
this is due to the similar oxide structures of the two metals producing
cobalt iron oxides Co3-xFexO4.
composition, it is yet to be explained why the 0.5%Ni4%Fe-SWFe cat-
alyst offered the highest reduction in solid residue during the activity
tests. It is most likely that the catalyst reduced the formation of solids
from heavy hydrocarbons or char. In Part 1, 0.5%Ni/SiO2 was shown to
reduce solid residue due to its tar cracking abilities and its overall
catalyst site dispersion. Having been impregnated separately from the
Fe phase; the 0.5%Ni phase on the 0.5%Ni4%Fe-SWFe may have been
separated and uninfluenced by the 4%Fe phase allowing the 0.5%Ni
phase to act on its own in tar cracking.
Having the more noble metal in lower concentration than oxophillic
Fe also appeared to benefit the NiFe catalysts in terms of H2 output
alone. The 0.5%Ni4%Fe configurations of NiFe had a higher output of
H2 than the 4%Ni0.5%Fe catalysts. Although 0.5%Ni produced more H2
as a monometal than 4%Fe in Part 1, co-impregnation or impregnation
of the Fe last is more beneficial for H2 output signifying an importance
in Fe exposure. Like CoFe, a similar trend is also noted for CO2 pro-
duction over the NiFe catalysts suggesting WGSR. According to
Watanabe, et al. [42], NiFe catalysts behave synergistically for WGSR
where water oxidizes the Fe and H2 is produced meanwhile, Ni facil-
itates better CO absorption which leads to the oxygenation of CO to CO2
via the oxidized Fe site. It may be possible that CoFe may also behave in
a similar fashion where Co replaces the role of Ni.
3.2. Gaseous products
One major purpose for the use of bimetallics in the proposed system
is for the in-situ production of hydrogen. Bimetallic catalysts such as
supported NiFe and PdFe catalysts have been studied for their activity
in WGSR reactions [31,40–42]. Pt 1 demonstrated that monometallic
catalysts can increase H2 output during glucose decomposition. It was
therefore anticipated that the synergy between two metals together
would further improve the H2 production over the monometallic cata-
lysts.
Of all catalysts that were tested, 0.5%Co4%Fe catalysts had the
highest output of H2 exceeding double the initial H2 and the H2 pro-
duction that was witnessed over monometallic catalysts in previous
study (see Table 2). Overall, CoFe catalysts produced the most H2 with
co-impregnated 0.5%Co4%Fe offering the most H2. Interestingly, in the
For 4%Ni0.5%Fe it is noted that stepwise impregnation beginning
with 4%Ni is most beneficial to produce CO2 with no discernable
Table 2
Outputs of H2, CO2, and CO.
0.5%(M1)
4%(M2)
0.5%(M1) 4%(M2) -SW
(M2)
0.5%(M1) 4%(M2) -SW
(M1)
4%(M1)
0.5%(M2)
4%(M1) 0.5%(M2) -SW
(M2)
4%(M1) 0.5%(M2) -SW
(M1)
H2 (kPaa)
Initial Input 38
No Catalyst 64
NiFe/SiO2
CoFe/SiO2
PdFe/SiO2
PtCo/SiO2
48
78
35
70
14
28
51
64
14
30
21
66
20
58
22
61
CO2 (kPaa) No Catalyst 99
NiFe/SiO2
CoFe/SiO2
PdFe/SiO2
PtCo/SiO2
163
153
138
129
103
109
118
114
122
108
100
145
122
121
136
130
CO (kPaa)
No Catalyst 62
NiFe/SiO2
CoFe/SiO2
PdFe/SiO2
PtCo/SiO2
52
53
67
47
37
44
44
44
42
39
44
56
55
52
57
44
13

K.A. Rogers and Y. Zheng
AppliedCatalysisA,General578(2019)10–19
differences in H2 production among all three impregnation methods
used for 4%Ni0.5%Fe. One possibility to consider is that internal H2 is
being produced along with CO2 via WGSR on the 4%Ni0.5%Fe catalysts
however the H2 may be rapidly consumed by other reactions. Seeing as
how the H2 output dips below the initial amount that was added –
unlike what was seen on monometallic 4%Ni and 0.5%Fe catalysts in
Part 1 – significant H2 consumption is highly likely. Therefore, it is
possible that the SWNi variant of the 4%Ni0.5%Fe catalyst produced
more internal H2 than the other two variants as signified by the higher
CO2 output.
exceptions such as 4%Ni0.5%Fe/SiO2 versus 4%Co0.5%Fe/SiO2. With
no surprise, 0.5%Ni4%Fe/SiO2-SWFe had the highest yield of liquid
products among all bimetallic catalysts, as this catalyst was active to-
wards tar cracking. However, the liquid yield remained lower than that
of 0.5%Ni/SiO2 from Pt.1 due to its higher yields of CO and CO2. Co-
impregnated 0.5%Co4%Fe/SiO2 yielded far less products in the liquid
phase as it favoured the production of H2 and CO2. Along with the drop
in CO production, it is suggested that the reduction in liquid phase
products is due to a reduction in water content as water would have
been consumed for water-gas-shift.
When observing the results for the output of CO it is noted hat the
output of CO does not correlate with that of CO2 as expected. For a
system purely based on WGSR, it is expected that the output of CO
would decrease as the output of CO2 increases. However, as seen in the
results of some catalysts such as 4%Ni0.5%Fe, it is noted the trend of
CO2 increasing as per the impregnation method does not correspond
with a decrease in CO. What is likely occurring is that the production
and subsequent consumption of internal hydrogen leads to the pro-
duction of more CO and H2O. Therefore, it is not fair to say that a
higher output of CO alone represents low WGSR activity here as WGSR
could have already occurred producing H2 which enabled the produc-
tion of additional CO.
Similarly, for 0.5Pd4Fe, impregnating 4%Fe after 0.5%Pd appears
to produce more CO2 than the alternative step-wise impregnation
method as well as monometallic 4Fe. Under the assumption that the 4%
Fe covers the 0.5%Pd, this suggests that there is some synergistic re-
lationship whereby Pd causes electron delocalization which stabilizes
the Fe surface and preventing it from becoming permanently oxidized
[15]. Alternatively, if Pd is directly exposed, it made become con-
taminated or easily coked. Similarly, to monometallic Pd, both con-
figurations of 0.5%Pd4%Fe had an overall reduction of H2 from initial
input to output.
In the previous study for monometallic catalysts, it was determined
that the Co catalysts were the most effective at performing in situ
deoxygenation with the highest selectivities to 2 M F and DMF over
furanic intermediates such as furfural. For 4%Co/SiO2 in particular;
˜87.2% of the furanic compounds were 2 M F and DMF. As seen in
Fig. 5, none of the bimetallic catalysts quite achieved the same se-
lectivity, however various 4%Fe-containing catalysts yielded competi-
tive results as high as ˜82%. For comparison, in the previous study, the
fraction of 2 M F and DMF totalled 65.6%.
Combining 4%Fe and 0.5%Ni made a more competitive catalyst
than either of the metals on their own. Overall, 0.5%Ni4%Fe catalysts
out performed 4%Ni0.5%Fe which appeared to offer little improvement
over monometallic 4%Ni in terms of 2 M F and DMF selectivity. This is
likely related to higher production of internal H2 with the 0.5%Ni4%Fe
catalysts which had higher outputs of both H2 and CO2 suggesting
higher degree of WGSR. Stepwise impregnation benefits this catalyst.
When the 4%Fe phase was impregnated first, the result was a total
elimination of the 5-methylfurfural intermediate and a higher se-
lectivity towards DMF than the co-impregnated catalyst. Of the three
variants, the 0.5%Ni4%Fe-SWFe catalyst had the lowest output of H2
which as detailed here may have been due to an elevated rate of con-
sumption via deoxygenation.
Combining Pt and Co together as a bimetallic catalyst negatively
impacted the output of H2 as both Pt and Co produced more as mono
metals however, together, they appear to consume H2. Both stepwise
configurations of 0.5%Pt4%Co catalysts produced the same amount of
CO2 however combining Co with Pt appears to offer no advantage over
Fe. Therefore, the oxophilic nature of Fe serves an important role in the
CoFe catalysts by as suggested earlier, serving the same role as Fe in
NiFe catalysts – facilitating the oxygenation of CO to CO2.
The selective deoxygenation function of NiFe catalysts have re-
ceived much attention. Both metals have been suggested to serve im-
portant roles where Fe facilitates adsorption preferentially via the
oxygen containing functional groups while Ni activates hydrogen for
hydrogenolysis [13,14,26]. Therefore, it makes sense that both metals
would have to be exposed. Impregnating 4%Fe after 0.5%Ni may have
had a negative impact on Ni’s ability to activate hydrogen and thus the
selectivities of 2 M F and DMF are lower on the SWNi catalyst than the
SWFe. The co-impregnated catalyst on the other hand may have suf-
fered from a more intimate interaction between the two metals which
may have formed a NiFe mixed or alloy phase.
3.3. Liquids products
Similar to the NiFe catalysts; the 0.5%Co4%Fe catalysts appeared to
outperform the 4%Co0.5%Fe catalysts. The co-impregnated and SWFe
variants of 0.5%Co4%Fe eliminated the output of the furfural. Unlike
the NiFe catalysts, the co-impregnation method was very suitable for
CoFe in general boosting competitively high 2 M F and DMF selectiv-
ities for both the 0.5%Co4%Fe and 4%Co0.5%Fe catalysts. This pro-
vides additional evidence that the CoFe catalysts benefit from a strong
interaction between the two metals perhaps as an alloy. 4%Co0.5%Fe-
SWFe produced interesting results in which the furnaic compounds
consisted of a large portion (28.2%) of 2-acetylfuran - an unusual
furanic compound that would signify either a major change in glucose
As one may elude from the results reported for solid and gaseous
products, the catalysts had an impact on the yield of liquid products. As
seen in Fig. 4, the relative trend between the catalysts, for the most
parts follows that of the organic solid conversion from Fig. 1 with a few
decomposition or possibly
a side reaction involving furfural.
Trimethylfuran was noted as being present in the products likely due
trans-alkylation reactions however in low amounts (< 5% of total
furanics). Interestingly although the fraction of 2 M F achieved here was
not as significant as 4%Co/SiO2 in the previous study, 0.5%Co4%Fe/
SiO2 had a lower furfural fraction at 0%.
As a monometallic catalyst; 0.5%Pd was not very effective at re-
ducing the amount of furfural among all furanic compounds. However,
0.5%Pd and 4%Fe paired together significantly reduced the content of
furfural, completely eliminating it when the 0.5%Pd phase was im-
pregnated onto the support first. Like NiFe, PdFe catalysts have re-
ceived a lot of attention in the selective deoxygenation of furanic and
Fig. 4. Percent yield of liquid products determine via mass balance where %
Yield = (total mass of liquid products)/(mass of glucose feed).
14

K.A. Rogers and Y. Zheng
AppliedCatalysisA,General578(2019)10–19
Fig. 5. Distribution of total furanic products.
phenolic compounds. The literature states that Pd alone was more ac-
tive towards ring interaction however when modified with Fe, the
catalyst favours selective deoxygenation of constituents attached to the
aromatic ring [15–18]. Hensley et al. [15], described the roles of the
two metals as Fe facilitating oxygenate absorption while Pd imposes an
electron delocalization effect effectively stabilizing the Fe. As it out-
performed the SWFe variant; it is suspected that the SWPd variant of
0.5%Pd4%Fe provided a stronger interaction between the two metals
and/or was able to prevent Pd’s interaction with the aromatic ring more
effectively. Like the PtCo catalysts though which also had competitive
results among all the catalysts, the PdFe catalysts appear to offer no
significant advantages over the NiFe and CoFe catalysts which would
justify the additional costs associated with using noble metal catalysts
Pd and Pt.
3.4. Characterization
Fig. 6. TPR Profiles for 0.5%Ni4%Fe/SiO2 and 0.5%Co4%Fe/SiO2 catalysts.
0.5%Ni4%Fe and 0.5%Co4%Fe were selected for further tests and
characterizations due to the overall value that they provided. Despite
having lower concentrations of expensive, active materials (Ni and Co),
these catalysts provided very competitive results in terms of both gas
and liquid products along with the potential to help reduce solid re-
sidue.
Due to the low concentrations of Ni and Co, TPR profiles (as shown
in Fig. 6) were faint as iron itself generally has a faint profile. Inter-
estingly though, the broad peaks which generally form at 450 °C and
600 °C for Fe were shifted and stretched when a second metal was used.
For example – the 0.5%Ni4%Fe/SiO2-SWFe catalyst had and an initial
peak ˜425 °C followed by a very broad peak ˜700 °C. Clearly, there was
an interaction between the two metals. The CoFe catalysts had very
undiscernible peaks especially when prepared via stepwise impregna-
tion.
Fig. 7) revealed that there was not an oxide phase present. Due to their
low concentration; Ni and Co phases were not detectable on their own.
XRD spectra for CoFe and NiFe alloys coincide with spectra for Fe
therefore making it difficult to distinguish if the peaks at ˜45° corre-
spond only to Fe or if they indicate the presence of an alloy phase. For
the two co-impregnated catalysts, the peak appears to be a double peak
which could be described by the presence of both Fe and an alloy phase
however the scale and resolution of these peaks are not strong enough
to allow for verification.
HRTEM and STEM-HAADF analyses with were performed with EDS
to determine the presence of Co/Ni and Fe in the catalysts and to es-
tablish if there were separated or together as alloy phases. HRTEM
images – as shown in Fig. 8 – show no distinguishable deference
XRD analyses of the 0.5%Co4%Fe and 0.5%Ni4%Fe catalysts (see
15

K.A. Rogers and Y. Zheng
AppliedCatalysisA,General578(2019)10–19
(Fig. 9a–c) clearly show the dispersion of the catalyst sites and their
relative densities compared to the SiO2 support. Corresponding EDS
mappings (Fig. 9d–f) demonstrated distinct differences between co-
impregnated and step-wise impregnated 0.5%Ni4%Fe. Fig. 9d shows
that the Ni sites easily coincide with the Fe sites without separation,
strongly suggesting that the two metals are together as an alloy. Fig. 9e
and f on the other hand show a clear separation of the Ni and Fe sites for
the 0.5%Ni4%Fe catalysts prepared via step-wise impregnation. This
matches earlier suggestions that the co-impregnated catalysts experi-
enced alloying while the step-wise impregnated catalysts consisted of
separate phases. EDS mapping of the CoFe catalysts was not possible as
the spectra for Fe can interfere with that of Co.
Given the earlier observations, it can be concluded that 0.5%Co4%
Fe/SiO2 benefits most from alloying. This co-impregnated catalyst
boosted competitive levels of deoxygenation and the highest output of
H2. Based on findings of other bimetallic catalysts, it is hypothesized
that incorporating Co into Fe effectively helps stabilize Fe which fa-
cilitates absorption via oxygen containing functional group [15]. For
0.5%Ni4%Fe/SiO2 alloying Ni and Fe via co-impregnation, although
competitive, was not as favourable as step-wise impregnation. Step-
wise impregnation (SWFe) kept some of the 0.5%Ni phase separate
from the Fe allowing for greater reduction in solid residue and achieved
better deoxygenation. The SWFe catalyst however had lower output of
H2 perhaps due to higher consumption. Ni and Fe may need to be kept
separate for them to perform separate roles for both solids decom-
position and deoxygenation.
Fig. 7. XRD Spectra for 0.5%Ni4%Fe/SiO2 and 0.5%Co4%Fe/SiO2 catalysts.
between catalyst particles of different metals or the combination
thereof. EDS proved to be the only method of indicating the presence of
either metal or possibly both at one location. Dispersion of the metals
appeared to be most significant on the SWFe catalysts as catalyst sites
tended to be small and separated. The other two preparation methods
(especially the SWNi/SWCo variants) yielded catalyst sites that were
quite clustered perhaps. This effect was mostly independent on the
selection of the most active metal (Ni vs. Co). For the co-impregnated
samples, it was a regular occurrence to find sites that were composed of
both metals. Meanwhile, for the step-wise impregnated samples, me-
tallic sites consisted of both monometallic and bimetallic phases. Even
the metals of lower concentrations (Co and Ni) were found separated.
STEM-HAADF images and EDS mappings were performed to further
investigate the metallic site compositions. STEM-HAADF images
3.5. Magnetite tests
Of all the bimetallic catalysts that were tested, the two catalysts that
proved to most interesting were the 0.5%Ni4%Fe/SiO2-SWFe and
0.5%Co4%Fe/SiO2 catalysts. The former provided the best reduction in
organic solid residue at 83.7% reduction while the latter provided the
highest output of H2 at 78.2 kPaa. Both catalysts were competitive
Fig. 8. HRTEM Images of: (a) 0.5%Ni4%Fe/SiO2 (b) 0.5%Ni4%Fe/SiO2-SWFe (c) 0.5%Ni4%Fe/SiO2-SWNi (d) 0.5%Co4%Fe/SiO2 (d) 0.5%Co4%Fe/SiO2-SWFe and
(e) 0.5%Co4%Fe/SiO2-SWCo.
16

K.A. Rogers and Y. Zheng
AppliedCatalysisA,General578(2019)10–19
Fig. 9. STEM-HAADF images and EDS mappings of 0.5%Ni4%Fe/SiO2 catalysts. (a)-(c): STEM-HAADF images of NiFe, NiFe-SWFe, and NiFe-SWNi respectively. (d)-
(f): EDS mappings of NiFe, NiFe-SWFe, and NiFe-SWNi respectively.
towards producing deoxygenated furanic compounds 2 M F and DMF.
Like previous tests with 0.5%Ni/SiO2 and 0.5%Re/SiO2 in Part 1; these
two catalysts were tested again without an initial charge of H2 but in-
stead with Fe3O4 fed at the same ratio as the catalysts. The intended
purpose for the Fe3O4 was to produce internal H2 (via WGSR), which
the catalysts would be dependent upon.
The addition of the deoxygenation catalysts appeared to have only a
small added affect on the outcome of the process as seen in Table 3.
Adding the catalysts managed in slightly reduce the amount of organic
solids from a 82.0% reduction to 85.2% upon addition of the NiFe
catalyst and 86.3% for the CoFe catalyst. The NiFe catalyst seemingly
increased the output of gases, especially CO2 perhaps attributed to
additional WGSR and reforming activity. However, where the CoFe
catalyst mostly maintained the selectivity of furanic compounds. the
NiFe catalyst seemed to reduce the selectivity of 2 M F and DMF in fa-
vour of other furanic compounds such as 2-acetylfuran.
can be produced on the surface and readily consumed for deoxygena-
tion eliminating many mass transfer barriers.
Combining WGSR and deoxygenation on a single catalyst surface
would mitigate the need for H2 (from WGSR) or even water (from de-
hydration) to be desorbed from the catalyst surface – only CO2. The
mechanism would ideally proceed similarly to the WGSR mechanism
suggested for NiFe/ CeO2-ZrO2 by Watanabe et al. [42], however, in-
stead of the active material (NiFe) being oxidized by water, it is oxi-
dized by the oxygenates undergoing hydrogenolysis. Catalysts would
require an initial amount of hydrogen to provide activation to start
hydrogenolysis. The oxygen on the surface may then transfer to an
appropriate supporting material containing oxygen vacancies thereby
temporarily stabilizing the active metal. CO is then required to reduce
the catalyst producing CO2. Surface CO may be derived either via the
actual absorption of carbon monoxide or possibly via decarbonylation.
In summary, water which is typically produced during hydrogenolysis
is not immediately associated and desorbed from the surface, and es-
sentially oxygen is instead removed as CO2. A summary of this theo-
retical process is demonstrated in Fig. 10 based on the mechanisms
described by Watanabe et al. [42], for WGSR and Sitthisa et al. [14], for
deoxygenation.
Overall, the catalysts do not appear to add any significant bonus on
top of the Fe3O4. One issue that has been considered is that the hy-
drogen produced on the magnetite is not readily available to the added
catalysts to provide activation. Ideally, WGS and deoxygenation reac-
tions should occur on the same catalyst on nearby sites so that hydrogen
Table 3
Summary of results for tests with magnetite. NiFe catalyst = 0.5%Ni4%Fe/SiO2-SWFe. CoFe catalyst = 0.5%Co4%Fe/SiO2.
Glucose
Glucose + Fe3O4
Glucose + Fe3O4 + NiFe Catalyst
Glucose + Fe3O4 + CoFe Catalyst
Gaseous Products:
PH2 (kPaa)
PCO2 (kPaa)
PCO (kPaa)
Solid Residue:
%OSC
3.8
96.5
52.0
9.1
181.7
47.5
10.0
200.3
53.4
10.5
170.6
51.3
75.8%
0.76
82.0%
0.96
85.2%
0.97
86.3%
0.90
H:C
Fraction of Total Furanic Products:
2 M F
DMF
Furfural
5-methylfurfural
Other Furanics
28.4%
45.8%
37.8%
0.9%
2.8%
12.7%
41.9%
32.5%
3.3%
4.0%
18.2%
45.5%
39.4%
2.0%
0.0%
13.2%
25.9%
24.9%
14.2%
6.6%
17

K.A. Rogers and Y. Zheng
AppliedCatalysisA,General578(2019)10–19
Fig. 10. Proposed combination of WGSR and Deoxygenation (furfuryl alcohol shown as example) on the surface of an active metal with an active support with
oxygen vacancies. (A) Adsorption of furfuryl alcohol; adsorption and dissociation of H2. (B) Furfuryl alcohol is adsorbed via the oxygen of the hydroxymethyl group
(likely on Fe of NiFe catalyst). (C) Hydrogenolysis of furfuryl alcohol. (D) 2 M F desorbed from the surface; oxygen is transferred to the active support’s oxygen
vacancy (such as CeO2-ZrO2). (E) CO is adsorbed onto catalyst surface (likely via Ni of NiFe); hydrogen atoms at any time may associate. (F) Oxygen back-transfer
from support to active metal (Fe of NiFe); desorption of H2. (G) Oxidation of CO to produce CO2. (H) Final desorption of CO2.
4. Conclusions
[3] S. De, B. Saha, R. Luque, Bioresour. Technol. 178 (2015) 108–118.
[4] P.M. Mortensen, J.D. Grunwaldt, P.A. Jensen, K.G. Knudsen, A.D. Jensen, Appl.
Catal. A 407 (2011) 1–19.
Of all the catalysts that were studied herein, the superior catalysts
were selected as 0.5%Co4%Fe/SiO2 and 0.5%Ni4%Fe/SiO2-SWFe. Due
to alloying of the two metals, 0.5%Co4%Fe/SiO2 offered competitive
deoxygenation properties, eliminating furfural content all while
boosting in-situ H2 production. As an alloy, Co stabilizes Fe, preventing
permanent oxide formation. On the other hand, sequential impregna-
tion of 0.5%Ni4%Fe/SiO2 catalyst was favourable as it provided a
distinct separation of the Ni and Fe phases allowing them to perform
their individual roles. As a result, the 0.5%Ni4%Fe/SiO2-SWFe catalyst
provided competitive deoxygenation properties with elevated DMF
fractions and was superior at reducing solid residue content. Aside from
their ability to help reduce solid residue, catalysts with 4 wt% Co/Ni
0.5% provided underwhelming results. Pd4%Fe/SiO2-SWPd also
eliminated furfural content however H2 consumption was a concern.
Future work should focus on improving catalyst site dispersion on
supporting materials and the selection of appropriate materials that
promote in-situ hydrogen production. A bifunctional catalyst that pro-
motes both deoxygenation and WGSR instead of two separate catalysts
would likely eliminate barriers associated with mass transfer of H2.
[5] K.A. Rogers, Y. Zheng, ChemSusChem (2016).
[6] A. Sanna, Bioenergy Res. 7 (2014) 36–47.
[7] H. Wang, J. Male, Y. Wang, ACS Catal. 3 (2013) 1047–1070.
[8] B.C.R. Ewan, R.W.K. Allen, Int. J. Hydrogen Energy 30 (2005) 809–819.
[9] K.A. Rogers, Y. Zheng, Appl. Catal. A (2019) 75–85.
[10] G.W. Huber, R.D. Cortright, J.A. Dumesic, Angew. Chem. Int. Ed. 43 (2004)
1549–1551.
[11] R.D. Cortright, R.R. Davda, J.A. Dumesic, Nature 418 (2002) 964–967.
[12] G.W. Huber, J.A. Dumesic, Catal. Today 111 (2006) 119–132.
[13] L. Nie, P.M. De Souza, F.B. Noronha, W. An, T. Sooknoi, D.E. Resasco, J. Mol. Catal.
A Chem. 388-389 (2014) 47–55.
[14] S. Sitthisa, W. An, D.E. Resasco, J. Catal. 284 (2011) 90–101.
[15] A.J.R. Hensley, Y. Hong, R. Zhang, H. Zhang, J. Sun, Y. Wang, J.S. McEwen, ACS
Catal. 4 (2014) 3381–3392.
[16] A.J.R. Hensley, Y. Wang, J.S. McEwen, ACS Catal. 5 (2015) 523–536.
[17] Y. Hong, H. Zhang, J. Sun, K.M. Ayman, A.J.R. Hensley, M. Gu, M.H. Engelhard,
J.S. McEwen, Y. Wang, ACS Catal. 4 (2014) 3335–3345.
[18] J. Sun, A.M. Karim, H. Zhang, L. Kovarik, X.S. Li, A.J. Hensley, J.S. McEwen,
Y. Wang, J. Catal. 306 (2013) 47–57.
[19] D. Shi, J.M. Vohs, ACS Catal. 5 (2015) 2177–2183.
[20] Q. Guo, M. Wu, K. Wang, L. Zhang, X. Xu, Ind. Eng. Chem. Res. 54 (2015) 890–899.
[21] X. Zhang, T. Wang, L. Ma, Q. Zhang, Y. Yu, Q. Liu, Catal. Commun. 33 (2013)
15–19.
[22] G.H. Wang, J. Hilgert, F.H. Richter, F. Wang, H.J. Bongard, B. Spliethoff,
C. Weidenthaler, F. Schüth, Nat. Mater. 13 (2014) 293–300.
[23] R. Watanabe, S. Watanabe, N. Hirata, C. Fukuhara, Catal. Lett. 146 (2016)
2478–2484.
Acknowledgements
[24] F. Meshkani, M. Rezaei, Int. J. Hydrogen Energy 39 (2014) 16318–16328.
[25] K. Sonal, K.K. Kondamudi, S. Pant, Upadhyayula, Ind. Eng. Chem. Res. 56 (2017)
4659–4671.
[26] W. Yu, K. Xiong, N. Ji, M.D. Porosoff, J.G. Chen, J. Catal. 317 (2014) 253–262.
[27] M. Widyawati, T.L. Church, N.H. Florin, A.T. Harris, Int. J. Hydrogen Energy 36
(2011) 4800–4813.
The authors thank the Natural Science and Engineering Research
Council of Canada (NSERC) and the Canada Foundation for Innovation
(CFI) for their financial support.
[28] R. Vinu, L.J. Broadbelt, Energy Environ. Sci. 5 (2012) 9808–9826.
[29] D.K. Shen, S. Gu, Bioresour. Technol. 100 (2009) 6496–6504.
[30] N.H. Florin, A.T. Harris, Int. J. Hydrogen Energy 32 (2007) 4119–4134.
[31] C. Ratnasamy, J.P. Wagner, Catal. Rev. 51 (2009) 325–440.
[32] J. Schindelin, I. Arganda-Carreras, E. Frise, V. Kaynig, M. Longair, T. Pietzsch,
S. Preibisch, C. Rueden, S. Saalfeld, B. Schmid, J.-Y. Tinevez, D.J. White,
References
[1] M. Asadieraghi, W.M. Ashri Wan Daud, H.F. Abbas, RSC Adv. 5 (2015)
22234–22255.
[2] F.X. Collard, J. Blin, Renew. Sustain. Energy Rev. 38 (2014) 594–608.
18

K.A. Rogers and Y. Zheng
AppliedCatalysisA,General578(2019)10–19
V. Hartenstein, K. Eliceiri, P. Tomancak, A. Cardona, Nat. Methods 9 (2012) 676.
[33] R.N. Widyaningrum, T.L. Church, M. Zhao, A.T. Harris, Int. J. Hydrogen Energy 37
(2012) 9590–9601.
[37] M. Zhao, T.L. Church, A.T. Harris, Appl. Catal. B: Environ. 101 (2011) 522–530.
[38] M. Zhao, N.H. Florin, A.T. Harris, Appl. Catal. B: Environ. 92 (2009) 185–193.
[39] M. Zhao, N.H. Florin, A.T. Harris, Appl. Catal. B: Environ. 97 (2010) 142–150.
[40] P. Tepamatr, N. Laosiripojana, S. Charojrochkul, Appl. Catal. A 523 (2016)
255–262.
[34] J. Grams, N. Potrzebowska, J. Goscianska, B. Michalkiewicz, A.M. Ruppert, Int. J.
Hydrogen Energy 41 (2016) 8656–8667.
[35] A.M. Ruppert, M. Niewiadomski, J. Grams, W. Kwapiński, Appl. Catal. B: Environ.
145 (2014) 85–90.
[36] R.N. Widyaningrum, T.L. Church, A.T. Harris, Catal. Commun. 35 (2013) 45–50.
[41] X. Wang, R.J. Gorte, Appl. Catal. A 247 (2003) 157–162.
[42] K. Watanabe, T. Miyao, K. Higashiyama, H. Yamashita, M. Watanabe, Catal.
Commun. 12 (2011) 976–979.
19