In situ XAS and FTIR studies of a multi-component Ni/Fe/Cu catalyst for hydrogen production from ethanol

Article abstract of DOI:10.1016/j.apcata.2013.07.032

Multicomponent catalysts containing Ni, Fe, Cu active for ethanol reforming reactions, prepared by solution combustion synthesis are characterized by multiple techniques such as ex situ XRD, XPS, and in situ XAFS and FTIR. XRD results indicate copper to be present in the reduced state as CuNi bimetal while nickel and iron are observed to be partially in a spinel NiFe₂O ₄ structure. In situ XANES and XAFS analysis show a change in Ni, Fe and Cu oxidation states during reaction. Cu, which was fully reduced before reaction, became partly oxidized upon exposure to ethanol and oxygen. Ni is mostly (75%) reduced and does not seem to change its oxidation state during the reaction. Fe is not present in metallic form after reduction and during the reaction, but some change in the oxidation state from Fe(II) to Fe(III) occurred during the reaction. XPS and SEM images indicate the formation of carbon filament on the spent catalyst. XPS results also indicate the enrichment of surface by Fe and Cu during the reduction of the catalyst. Based on the activity and characterization results obtained, and literature review, the role of predominant phases during ethanol decomposition reaction is proposed.

Full text of DOI:10.1016/j.apcata.2013.07.032

Accepted Manuscript
Title: In-situ XAS and FTIR studies of a multi-component
Ni/Fe/Cu catalyst for hydrogen production from ethanol
Authors: A. Kumar J.T. Miller A.S. Mukasyan E.E. Wolf
PII:
S0926-860X(13)00430-4
DOI:
Reference:
http://dx.doi.org/doi:10.1016/j.apcata.2013.07.032
APCATA 14348
To appear in:
Applied Catalysis A: General
Received date:
Revised date:
Accepted date:
26-4-2013
10-7-2013
14-7-2013
Please cite this article as: A. Kumar, J.T. Miller, A.S. Mukasyan, E.E.
Wolf, In-situ XAS and FTIR studies of a multi-component Ni/Fe/Cu catalyst
for hydrogen production from ethanol, Applied Catalysis A, General (2013),
http://dx.doi.org/10.1016/j.apcata.2013.07.032
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.

1
2
3
4
5
6
An active catalyst for ethanol decomposition reaction, containing nano-composites of Ni/Fe/Cu
was synthesized using combustion based techniques. Different characterization results indicate
that the segregated phases were active at different temperature ranges resulting in gradual change
the product selectivity with temperature.
7
8
9
10
11
12
1
Page 1 of 41

12
13
14
15
16
17
18
19
20
21
ꢀ NiFeCu catalyst appears homogeneous at the mesoscale but highly heterogeneous at the
nanoscale.
ꢀ Cu oxidation state changes upon exposure to ethanol but Ni remains 75% reduced
without further changes in oxidation state during reaction.
ꢀ Iron is not present in metallic form under any of the conditions used.
ꢀ XPS results indicate the enrichment of surface by Fe and Cu during the reduction.
ꢀ IR results show changes in the amount and type of adsorbed species with temperature and
correlates with the active role of each metal at a given temperature.
2
Page 2 of 41

21
22
23
In-situ XAS and FTIR studies of a multi-component Ni/Fe/Cu catalyst for hydrogen
production from ethanol
A. Kumar1,3, J. T Miller2, A. S. Mukasyan1, E. E. Wolf1*
24
25
26
1Chemical and Biomolecular Engineering Dept. University of Notre Dame, Notre Dame, IN
46556; 2 Chemical Sciences Div. Argonne National Laboratory, Argonne, Ill, 3
27
Abstract
28
29
30
31
32
33
34
35
36
37
38
39
40
41
Multicomponent catalysts containing Ni, Fe, Cu active for ethanol reforming reactions,
prepared by solution combustion synthesis are characterized by multiple techniques such as ex-
situ XRD, XPS, and in-situ XAFS and FTIR. XRD results indicate copper to be present in the
reduced state as Cu-Ni bimetal while nickel and iron are observed to be partially in a spinel
NiFe2O4 structure. In-situ XANES and XAFS analysis show a change in Ni, Fe and Cu oxidation
states during reaction. Cu, which was fully reduced before reaction, became partly oxidized upon
exposure to ethanol and oxygen. Ni is mostly (75%) reduced and does not seem to change its
oxidation state during the reaction. Fe is not present in metallic form after reduction and during
the reaction, but some change in the oxidation state from Fe(II) to Fe(III) occurred during the
reaction. XPS and SEM images indicate the formation of carbon filament on the spent catalyst.
XPS results also indicate the enrichment of surface by Fe and Cu during the reduction of the
catalyst. Based on the activity and characterization results obtained, and literature review, the
role of predominant phases during ethanol decomposition reaction is proposed.
3
Current location: Department of Chemical Engineering, Qatar University, Doha, P. O. Box
2713, Qatar
3
Page 3 of 41

41
42
1. Introduction
43
44
45
1.1 Combustion synthesis:
Combustion synthesis (CS) is an attractive technique for materials synthesis on account of
being simple, economical, fast and energy efficient. This simplicity and flexibility to synthesize a
wide range of materials has led to an increased research interest as indicated by the increase in
the publications on CS [1, 2]. The conventional solid-solid CS can be broadly classified into two
types based on the way combustion reaction takes place. The combustion reaction can proceed as
a self-sustained propagating reaction front after local ignition of the reactive solid form, or it can
be combusted simultaneously all over the volume by using an external uniform heating source.
The former method is known as Self-Propagating High-Temperature Synthesis (SHS) and the
later as Volume Combustion Synthesis (VCS). Recent innovations have led to the use of CS in
fluid phases such as solution combustion synthesis or SCS [3].
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
Among the above-mentioned methods, solution combustion synthesis, due to its ability to
produce nano-materials, has recently gained attention and it is being applied to diverse areas of
materials synthesis such as pigments, catalysis, electronic and magnetic materials, drug delivery
etc [1, 2, 4]. SCS is considered as a redox reaction consisting of oxidizing agents (e.g. metal
nitrates) and reducing agents (e.g. glycine, urea, hydrazine etc), also referred as fuel. The
exothermic reaction between metal nitrates and the fuel provides the energy required for
sustaining the combustion synthesis reaction without adding external energy. Furthermore, the
energy released is high enough to evaporate volatile compounds and calcine the products formed
leading to the formation of crystalline phases. Because the combustion reaction rate is very high,
the crystallites formed do not have enough time to sinter, leading to the formation of nano-
powders with higher surface area than obtained from conventional synthesis. Thus, SCS can
4
Page 4 of 41

66
67
68
69
70
71
72
73
74
yield highly pure, crystalline materials with high surface area in a single step synthesis without
requiring additional thermal treatments. In our previous work with methanol partial oxidation
[5], we compared the co-precipitation vs CS preparation and found the later more active. Cu/Zn
catalyst prepared by co-precipitation is reported to have a surface area of 17 m2/g [5], whereas
the catalysts prepared by combustion methods have area as high as 31 m2/g [6]. Therefore in this
work we focused mainly on the CS preparation. Work is underway comparing various methods
of preparation on supported catalysts for ethanol decomposition.
The CS reaction between metal nitrates and glycine, used as fuel, can be represented by the
following widely accepted scheme:
5
5
M v (NO3 )v + ( vϕ)CH2 NH2COOH + v (ϕ −1)O2
9
4
75
(1)
10
→ MvOv (s) + ( vϕ)CO2 (g) + vϕH2O(g) + v(
18
25
5ϕ + 9
18
)N2 (g)
9
2
76
77
78
79
80
81
82
83
84
85
86
87
where Mv is a ν-valence metal. The parameter φ, the fuel to oxidizer ratio, is defined such that φ
= 1 corresponds to a stoichiometric oxygen concentration, meaning that the initial mixture does
not require atmospheric oxygen for complete oxidation of the fuel, while φ > 1 (<1) implies fuel-
rich (or lean) conditions.
According to the above scheme, (Eq. 1), SCS can be used for the synthesis of metal
oxides. Recent publications from our group [7-10] demonstrated that controlling φ during SCS
one can synthesize reduced metals nano-powders as well (e.g. Ni, Cu rather than NiO, CuO). A
reaction pathway was proposed to describe the controlled synthesis of different phases [9, 10].
The active solution containing a metal nitrate and glycine can be impregnated on a thin media
(e.g. cellulose paper, carbon nano-tubes or graphite sheet etc.) before combustion synthesis to
facilitate the cooling of the products obtained after CS [11-13]. This fast reaction and cooling,
and the unique microstructure of the products, results in finer particles with higher surface area
5
Page 5 of 41

88
89
than calcined oxides, which has potential for catalytic applications, as shown by previous results
and model studies [12, 13].
90
This study complements our previous work on the synthesis and activity studies of
multicomponent Ni/Fe/Cu based catalyst for hydrogen production from ethanol using SCS [14]
which showed, Ni1Fe0.5Cu1 to be the optimum catalyst composition among the several catalysts
investigated for hydrogen production from ethanol partial oxidation and decomposition
reactions. The focus of this work was to understand the role of individual metals in such
complex multicomponent system rather than following an activity–selectivity optimization
scheme by combinatorial compositional changes of the various components as previously
reported. Ethanol’s decomposition products were measured in detail on individual metals/oxides
(Ni, Cu and Fe) as well as on the most active and selective multicomponent catalyst, Ni1Fe0.5Cu1,
previously studied [14]. This paper attempted to correlate the activity of each component with its
oxidation state, and in the differences in such results under in-situ conditions, without reaction,
and under operando conditions i.e. in the presence of the reaction.
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
As stated earlier, SCS has numerous advantages over other catalyst preparation methods,
such as co-precipitation that requires separation of the products after precipitation followed by
calcination, which leads to sintering. As indicated by equation (1), except for the metal oxide
product, all other products are gas phase products, which can be controlled by varying the
parameter φ. The gases generated during CS form micro-channels on the solid products as they
are released during reaction, thus contributing towards the porosity of the material synthesized,
potentially leading to high surface area.
1.2
Ethanol reforming:
6
Page 6 of 41

110
111
112
113
114
As the research interest on the fuel cells technology rises, the need for a pure hydrogen
supply is also rising. Hydrogen can be abstracted from many hydrogen-containing precursors,
and among them light alcohols (e.g. methanol, ethanol), are attractive renewable sources as they
can be produced from corn stover [15] and other biomass byproducts [16]. The main routes for
hydrogen generation from ethanol and the corresponding heat of reaction are as follows:
115
116
117
118
Ethanol Dehydrogenation and Decomposition:
−ΔH2098 = −68KJmol−1
−ΔH2098 = −49KJmol−1
C2H5OH → CH3CHO + H2
C2H5OH → CH4 + CO + H2
Ethanol Steam Reforming:
(2)
(3)
119
120
121
122
−ΔH2098 = −174KJmol−1
C2H5OH + 3H2O → 2CO2 + 6H2
(4)
(5)
123
124
125
Ethanol Partial Oxidation:
3
−ΔH2098 = 509KJmol−1
126
C2H5OH + O2 → 2CO2 +3H2
2
127
128
129
130
131
132
133
134
Ethanol steam reforming (ESR) is a highly endothermic reaction and requires steam to
sustain the reaction. Ethanol decomposition (ED) is less endothermic than ethanol steam
reforming, whereas ethanol partial oxidation (EPOx) is exothermic and releases energy during
reaction. We previously studied the ED and EPOx reactions on Ni/Fe/Cu based catalysts on
account of lesser energy requirements for hydrogen production. A review of the previously used
catalysts for ethanol reforming is described in our previous paper [14]. On the basis of literature
results, we selected Ni, Fe and Cu for the synthesis of multicomponent catalysts using
combustion synthesis. Activity results from a family of such catalysts using a high throughput
7
Page 7 of 41

135
136
approach showed that a Ni1Fe0.5Cu1 catalyst gave the highest activity and selectivity among a
limited series of catalyst studied.
137
138
2. Experimental
2.1
Catalyst synthesis
Aqueous solutions of metal nitrates (Alfa Aesar) Mv(NO3)v ·xH2O (where Me = Ni, Cu,
139
140
141
142
143
144
145
146
147
148
149
150
151
152
Fe) and glycine (C2H5NO2, as fuel) were used to synthesize catalysts in volume combustion
synthesis (VCS) mode by forming a sol-gel of the precursors and fuel. The amounts of precursor
materials were calculated based on the production of 3g of catalyst. These precursors were
dissolved in 75ml of deionized water in a 400ml beaker. In the sol-gel VCS method the beaker
containing the homogeneous solution of metal nitrates and fuel is heated over a hot plate until the
sol-gel solution reaches the ignition temperature after evaporation of water. The fuel/oxidizer
ratio φ, has been maintained at 1.75, which is the optimized value obtained from our previous
studies [9] for the production of pure metals using combustion synthesis. Once ignited, the
reaction proceeds very fast, which can lead to either an explosive combustion mode or to the
self-propagating mode (SHS) mode inside the beaker itself. After synthesis, the resulting
powders were milled using a planetary ball mill (Retsch PM 100) at 650 rpm for two minutes to
get a mixture with uniform particle size. The ratio of catalyst sample and milling balls was kept
at 1:2 in all the cases.
153
2.2
Catalyst characterization
154
155
156
Multiple techniques were used to characterize the Ni/Fe/Cu catalyst. XPS and XRD methods
were used to determine elemental surface concentrations and identification of the bulk phases
respectively. XANES and EXAFS experiments were conducted to determine oxidation states of
8
Page 8 of 41

157
158
159
160
161
Ni, Fe and Cu under different reaction conditions. In-situ DRIFT studies were performed to get
information about the adsorbed intermediates on the surface of the catalyst. The catalytic activity
and selectivity data were published in our earlier paper [14]. The hydrogen selectivity reported is
defined as the amount of hydrogen produced to the maximum amount of hydrogen that could
have been produced with the reacted ethanol.
162
163
164
2.2.1 XRD, SEM and TEM measurements:
165
166
167
168
169
170
171
XRD measurements were carried out in air at room temperature in an X-ray
diffractometer (Scintag Inc) using Cu-Kα radiation of 1.54056 Ǻ wavelength. The powdered
samples were loaded on an amorphous glass sample stage to avoid signals from the sample
holder itself. Powder microstructures were imaged using a field emission SEM (Magellan 400
FEI) at Notre Dame Integrated Imaging Facility (NDIIF). Transmission electron microscopy
images were collected using Titan 80-300, which is capable of providing atomic level imaging
and is equipped with nano-scale resolution EDS (energy dispersive spectra) facility.
172
173
174
175
176
177
178
2.2.2 X-ray photoelectron spectroscopy.
X-ray photoelectron spectra measurements were obtained using a Kratos XSAM-800
system using Mg-Kα radiation at 1253.6 eV and an electron takeoff angle at 90° (sample
normal). Powder samples were mounted on sample stubs with double sided conductive carbon
tapes. The peak areas were analyzed using the CasaXPS software package with relative
sensitivity factors obtained from Kratos XSAM library. These results were used to estimate the
surface concentration of different elements and their oxidation states under vacuum using carbon
9
Page 9 of 41

179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
as standard for calibrating the peak locations. Samples were reduced ex-situ at 300ºC for 1 hour
under continuous hydrogen flow. The reduced samples were transferred to the XPS sample
transfer chamber using a glove bag. A sealed reactor containing the reduced catalyst was placed
inside a glove bag, which was mounted on the door of transfer chamber and purged multiple
times with Ar to eliminate residual air that could oxidize the sample before opening the reactor
and transferring the sample. The same procedure could not be used with the used samples that
were exposed to air after removal from the reactor.
2.2.3 TPR experiments:
The TPR experiments were conducted on a catalyst characterization unit of our own
design equipped with a flow reactor connected to a 6-port sample valve for injecting pulses, a
thermal conductivity detector, and automated flow and temperature controllers. Prior to a TPR
run, each catalyst (approx 100 mg) was oxidized using 20% O2 in Ar, flowing at 50cc/min while
being heated from room temperature to 300ºC in 30min then kept at this temperature for an hour.
After that, the flow was switched to pure He/Ar to flush the system while it cools down to room
temperature. For TPR measurements a carrier gas containing 5% H2 in Ar flows through one side
of the TC cell before entering the flow reactor containing the catalyst sample. The reactor
effluent is connected to a water trap to adsorb or condense the water formed during reduction.
The water-free effluent gas flows over the sample side of the TC cell allowing it to detect any
loss in hydrogen that may occur due to reduction of the catalyst.
2.2.4 X-ray absorption studies:
The XANES and EXAFS experiments were carried out at the Advanced Photon Source
(APS) facility at the Argonne National Laboratory (ANL). The spectra were obtained in
transmission mode at the sector 10 MRCAT (Material Research Collaborative Access Team)
10
Page 10 of 41

202
203
204
205
206
207
208
beam line 10-BM, equipped with a bending magnet (BM). This beamline has a double crystal
Si(111) monochromator to select the incident x-ray photon energy. A detuning of 50% was used
to accomplish the rejection of higher order harmonics in the beam. Self-supported catalyst wafers
were prepared by pressing the sample, diluted by silica, into a cylindrical holder with multiple
channels that can contain up to six catalysts at a time. The dilution of a sample with silica and the
amount of diluted sample to make wafers were adjusted to get the optimized absorption edge
height χ(E) to about 1 unit. In Cu, Ni, Fe samples, the dilution ratio varied between 1:2 to 1:7
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
(wt sample: wt silica) and the amount required for wafer preparation varied from 7mg-15mg of
the diluted sample. The 6-channel multiple sample holder was placed in the center of a quartz
tube of 0.75” ID and 18” length. The tube has fitting arrangements at both ends with an inlet and
outlet for the reactant gases to flow through the reactor, and an inlet for a K-type thermocouple
to measure the sample temperature. The reactor tube end fittings have openings at both ends that
were covered with polyimide film providing a window to allow transmission of the X-ray beam
and to seal the tube. An electrical furnace with a temperature control heated the catalysts to set
the pretreatment and reaction temperatures. This set up allowed us to obtain X-ray absorption
(XAS) spectroscopy measurements during in-situ and under reaction conditions i.e. in operando
mode under the following conditions:
A: In air at room temperature; B: In-situ reduced sample at room temperature; C: Operando at
270 ºC during reaction (ethanol decomposition and ethanol partial oxidation reaction).
The sample was first reduced at 300 ºC for 1 hr in a 5% H2 (balance He) stream flowing at 50
cc/min in a hood located outside the x-ray beam station to optimize utilization of the beam line.
After reduction, the sample was cooled down in the gas flow to room temperature, then the flow
was stopped and the inlet and outlets valves were closed to isolate the reactor from the
11
Page 11 of 41

225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
atmosphere and avoid air contact. The gas lines were disconnected and the reactor was
transferred to the beam line and reactant gas lines were reconnected. The fitting tubes and valves
were purged before flowing reactant gases inside the reactor. XAS spectra were first obtained at
room temperature on the reduced samples then the reactor temperature was raised from room
temperature to 270 ºC under the continuous flow of reactants. Operando measurements were
taken at this temperature at the same conditions used for ethanol decomposition and ethanol
partial oxidation reaction. Standards for the reduced metals were obtained from metal foils (for
Cu-Cu, Ni-Ni and Fe-Fe phase and amplitude parameters) and for the oxides using standard
CuO, Cu2O (for Cu-O parameters), NiO (for Ni-O parameters), FeO, Fe2O3 (for Fe-O
parameters). The experimental data obtained is fitted to standard scattering equation to get the
coordination numbers; inter-atomic distances etc. using standard WINXAS97 and ATHENA
software. The Debye-Waller (DW) factor was obtained by fitting the spectra of the reduced
sample at the 270 oC and at room temperature using reference spectra. The presence of an oxide
fraction makes it difficult to calculate the nanoparticle size directly from the coordination
number. Therefore, estimates were made by using the metallic fraction present in the XANES
spectra of the nanoparticle. For example a XANES spectra at room temperature in air shows 75%
metallic copper and 25% Cu(I) present in the catalyst nanoparticle. The actual Cu-Cu
coordination used for size estimation was 11.5 (8.6/0.75). The % of each phase in the NiFe05Cu
catalyst was estimated from the XANES spectra using a linear combination of the peak height
over its value past the edge energy as obtained from the standards used [17].
245
2.2.5 Diffuse Reflectance Infrared Fourier Transform Spectroscopy (DRIFTS):
246
247
DRIFTS measurements were carried out in a Bruker FTIR spectrometer (Equinox 55)
using a DRIFTS reaction cell (Harrick, HVC) with a flat ZnSe windows and a praying mantis
12
Page 12 of 41

248
249
250
251
252
253
254
optical assembly (Harricks, 3-3S). The distance between the ZnSe window and the catalyst
sample was kept at minimum to reduce the gas phase signals. The catalyst samples were diluted
with fumed SiO2 in a ratio of 1:2 to 1:4 to reduce IR absorption by the dark samples and enhance
reflectance. Each sample was first reduced in-situ at 300ºC for 1 hour in a continuous flow of 30
cc/min containing 20% H2 (balance He) gas. Thereafter, the system was purged with He (25
cc/min) at 300ºC for half an hour. Background spectra were taken first at different temperatures
in continuous He flow.
255
256
257
258
259
260
Ethanol was introduced by directing the He through an ethanol bubbler kept at room
temperature for 10 min during desorption studies. The system was purged with continuous He
flow, while desorption spectra were taken at regular intervals. The ethanol decomposition
reaction was studied under continuous flow of ethanol while increasing the temperature. Spectra
were obtained using a MTC detector in the frequency range of 370– 4000 cm-1 with a resolution
of 4 cm-1 by collecting 128 scans at a scan velocity of 3 K-Hz using a MTC detector.
261
3.
Results and discussion
262
263
264
265
266
267
268
269
As described earlier all the catalysts in reference [14] were synthesized by using SCS method.
The BET area of the synthesized catalysts varied from 11 m2/g to 30 m2/g. The BET area for
Ni1Fe0.5Cu1catalyst used for detailed characterization is 16 m2/g.
3.1 XRD analysis:
XRD results (ex-situ in air at RT) in figure 1 shows diffraction peaks obtained after CS of the
single metal nitrate-glycine system (Ni), the bi (NiCu) and the tri-metal nitrate-glycine
(Ni1Fe0.5Cu1) catalysts by adding Cu(NO3)2 and then Fe(NO3)3 to the Ni(NO3)2-glycine mixture
and maintaining the same fuel to oxidizer ratio (ϕ = 1.75). The ratio, ϕ = 1.75, was optimized to
13
Page 13 of 41

270
produce pure metal (Ni) and a (CuNi) alloy [13]. Maintaining the same ratio ϕ = 1.75 in
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
presence of Cu(NO3)2 and Fe(NO3)3 gives CuNi (the CuNi[111] peak at 2θ=43.94, 51.19, 75.31) ,
pure Ni[111] (peak at 2θ=44.5, 51.84, 76.37) and NiFe2O4 spinel phases indicating that in the
bulk of these catalysts, Cu is completely reduced, Ni partially reduced and Fe to be in oxide form
(Fe2O3). There are many factors that affect the oxidation state of a synthesized catalyst. Copper
exhibits three oxidation states during the CS process which gradually changes from Cu(II) to
Cu(I) and then to Cu(0) as we move from φ value 0 to 3. Ni exhibits only two oxidation states
and changes from Ni(II) to Ni(0) when φ is changed from 0 to 1.75. Both Ni and Cu have
comparable affinity for oxygen (Cu-O and Ni-O comparable enthalpy of dissociation). The
detailed mechanism that leads to reduction of these metals instead of the oxide phase predicted
by Eq (1), is not known in detail and is under investigation, but it involves several intermediate
decomposition reactions of the nitrates and glycine that yields reducing gases that lead to the
metal phase. It appears that extra hydrogen is required to reduce Cu(I) to Cu(0) and thus a higher
value of φ. As shown in the reference [9] using φ =1.75 yields Cu(I) and Cu(0) oxidation states
and no Cu(II).
14
Page 14 of 41

287
Figure 1: XRD results on the products from Ni(NO3)2-Glycine, Ni(NO3)2-Cu(NO3)2-
Glycine and Ni(NO3)2-Fe(NO3)3-Cu(NO3)2-Glycine mixtures combustion
288
289
290
291
292
293
294
295
296
3.2
TPR analysis:
Figure 2 shows the TPR analysis conducted on Fe2O3, CuO, NiO and Ni1Fe0.5Cu1
catalysts prepared by CS. Fe2O3 is the most difficult metal oxide to reduce, giving a peak for
hydrogen uptake at 453ºC, most likely due to its partial reduction from Fe2O3 to FeO, with
another peak appearing above 650ºC. CuO shows a peak for maximum hydrogen uptake at
250ºC while NiO reduces at a lower temperature than CuO at 237ºC. This could be due to the
effect of the fuel to oxidizer ratio used which was optimized to =1.75 to produce pure Ni metal
ϕ
297
298
as compared to CuO and Fe O where the same value of =1.75 results in metal-oxide phases
ϕ
2 3
for Cu. A higher value of ϕ =3.0 is required to produce pure Cu using combustion synthesis [10],
15
Page 15 of 41

299
300
whereas Fe is more difficult to produce and may require the combustion synthesis to take place
in an inert atmosphere, in addition of using a higher value. Ni1Fe0.5Cu1 gives a first peak at
ϕ
301
225ºC and another broader peak between 385ºC - 500ºC.
302
Figure 2: TPR analysis on Fe, Cu, Ni, and Ni1Fe0.5Cu1 catalysts prepared by combustion
synthesis
303
304
It is clear that the presence of Ni, Fe and Cu together increases the reducibility of Ni/Cu
mixtures and decreases the reduction temperature to a lower value of 225ºC in comparison to
237ºC and 250ºC, as observed in pure NiO and CuO phases respectively. Fe2O3 also seems to
reduce earlier in Ni1Fe0.5Cu1 as the TPR profile becomes flat after 500ºC and there is no
indication of a third peak around 650ºC as in the single Fe2O3 TPR profile. So the broader peak
between 385ºC-500ºC could be a combination of two close peaks of Fe2O3. From the TPR
analysis it can be concluded that the three metallic/oxide phases together makes the catalyst more
reducible in comparison to each individual metal oxide phase. The metals and metal oxides
present together not only effect the reducibility of the multicomponent catalyst, but these phases
305
306
307
308
309
310
311
312
16
Page 16 of 41

313
314
315
316
also rearrange themselves during different treatments of the catalyst such as reduction and
exposure to ethanol etc. The actual surface concentrations of individual metals under vacuum
were determined by XPS as described below.
3.3
X-ray photoelectron spectroscopy
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
XPS results of the Ni1Fe0.5Cu1 catalyst were determined after exposure to three different
states: in air, reduced state and after the ethanol decomposition reaction. Since the sample was
exposed to air after reaction, the XPS spectra of the used catalyst essentially matches the one of
the sample in air. Figure 3 shows the XPS spectra at energies corresponding to Cu 2p (fig 3a), Ni
2p (fig 3b) and Fe 2p (fig 3c) photoemission from the catalyst treated in air, reduced, and after
reaction. Apart from these peaks, carbon and oxygen peaks were also observed. Copper was
expected to be in the reduced state as the XRD results indicated. Surface oxidation from residual
atmospheric oxygen could not be avoided, however, as indicated by the shift in the Cu 2p peaks
towards a higher binding energy on the samples in air and after the reaction as compared to the
reduced sample, indicating the presence of Cu+2 oxidation state on the surface. Ni 2p peaks in air
and used samples are also shifted toward higher oxidation state than the reduced ones showing
that the surface oxidation differ from the bulk sample, which is mostly reduced, according to
XRD. The binding energy of the Fe 2p peaks does not change after reduction and is present in
the Fe+3 oxidation state. The separation between the 2p peaks of Ni, Cu and Fe is due to spin
orbital splitting.
17
Page 17 of 41

332
333
334
Fig 3: XPS spectra of Ni1Fe0.5Cu1 catalyst in air, after reduction and after reaction (a) Cu2p
(b) Ni2p and (c) Fe2p
Table 1: Atomic ratio of Ni, Fe and Cu on the surface of Ni1Fe0.5Cu1 catalyst
Total atomic ratio
Ni1Fe0.5Cu1 , in air
Fe/Ni
1.3
Cu/Ni
2
Ni1Fe0.5Cu1 , reduced
2.4
2.6
Ni1Fe0.5Cu1 , after reaction
1.2
2.1
335
18
Page 18 of 41

336
337
338
339
340
341
342
343
344
345
346
347
348
Table 1 gives the atomic ratio of Ni, Cu and Fe relative to Ni near the surface region,
(referred hereafter as surface) of the Ni1Fe0.5Cu1 catalyst respectively. The relative
concentrations of different elements are calculated based on the area of the XPS and the
sensitivity factor for each element. Although the bulk catalyst was synthesized based on an
atomic ratio of 1:0.5:1 of Ni:Fe:Cu, the surface composition of the catalysts exposed to air is
different and contains Ni:Fe:Cu in 1:1.3:2 ratios, most of which are the oxides of the different
metals. After reduction, Fe/Ni and Cu/Ni atomic ratios increase to 2.4 and 2.6 respectively.
Surface reconstruction and composition are determined by the surface’s Gibbs free energy. The
oxide phases of Ni, Fe and Cu have significantly lower surface energies than their pure metallic
phases and for this reason their metallic surfaces tend to form oxides when exposed to oxygen
[18, 19]. The spinels (MN2O4, M3O4) have lower surface energy than the metals (M), rocksalt
oxides (MO) and trivalent oxides (M2O3) [19]. This could be the reason why only pure Ni is
obtained using combustion synthesis at an optimized fuel to oxidizer ratio of =1.75, whereas
ϕ
349
350
351
352
353
354
355
356
357
358
the same fuel to oxidizer ratio in the presence of Fe and Cu gives spinel (NiFe2O4) phases (see
XRD pattern, fig 1). The surface goes back to its initial state after the reaction and exposure to
air.
The XPS spectrum of the C1s photoelectrons (shown in supplementary figure S1)
observed at 284.5 eV shows a large peak on the spent Ni1Fe0.5Cu1 catalyst after a 40 hour time on
stream at 400ºC. Ni and Fe, known for their activity towards C―C bond scission, are likely to be
responsible for carbon formation. As previously reported [14], deactivation which reduces the
catalyst hydrogen selectivity but not the conversion as a function of time on stream (TOS), could
be attributed to the formation of carbon species altering the distribution of active sites. It is
anticipated that at longer TOS the conversion will also decline. Activity experiments conducted
19
Page 19 of 41

359
360
361
362
363
364
365
on individual metals [14], indicate no carbon formation on Cu only catalyst. Carbon formation
has been reported on Ni, Fe, Co based catalysts [20] and it grows forming filament-like
structures as shown in the SEM micrograph shown in figure 4a. A mechanism for the carbon
filament formation and growth was proposed by Baker et. al [20] and later refined by others [21-
23] wherein the metal particle grows at the top of the filament, which is consistent with the
sustained conversion shown with TOS but the significant effect on the H2 selectivity in favor of
total oxidation.
366
367
Figure 4: (a) SEM micrographs of carbon filaments on the surface of spent Ni1Fe0.5Cu1
catalyst. (b) TEM image and (c) phase distribution using HRTEM on a Ni1Fe0.5Cu1 catalyst.
368
369
370
371
372
373
XPS studies provide information about the surface concentrations under ultra-high
vacuum condition, which is different than under reaction conditions, nonetheless the results
indicate that on the reduced catalyst surface the various phases are in a different ratio than in the
synthesized bulk catalyst. The bulk catalysts oxidation states, were studied under in-situ and
operando conditions which are described later in the XAS section. A TEM image along with
EDS is presented next to describe the phase distribution at nano-scale.
20
Page 20 of 41

374
3.4
Transmission Electron Microscopy:
375
376
377
378
379
380
381
382
383
384
385
386
387
Figure 4b shows the results obtained on high resolution TEM. The catalysts particles are
agglomerates of heterogeneous nanoparticles ranging from 10-100 nm size. The structure
appears to be very non-uniform contrary to our expectations. As the solution combustion
synthesis involves molecular level mixing of different metal nitrates and glycine, a homogeneous
product with uniform properties is expected. Different phases are segregated around the bulk of
the catalyst (fig 4c). This could be partly due to the difference in the solubility limits and
precipitation of different components in a solution containing multiple salts. Although
combustion synthesis is a fast process and mainly dependent on the ignition of the decomposition
products of metal nitrates and glycine (ammonia and nitric acid), it is difficult to rule out any sort
of precipitation for the fact that individual metal nitrates have different decomposition
temperature and it is highly probable that solution reaches solubility limit just before its ignition.
This heterogeneity is restricted to nanoscale only, as the EDS data obtained using LEO SEM (not
shown), which gives information from ~1 μ m depth, shows uniform distribution of all the
388
389
390
391
phases. This heterogeneity at the nanoscale clearly explains the results obtained during ethanol
reforming reaction [14]. Different segregated phases seem to dominate the product distribution at
different temperature ranges resulting in a gradual shift in the product selectivity with
temperature.
392
3.5
In-situ and operando XANES and EXAFS studies
393
394
395
The Ni edge XANES spectra in air, reduced, and under reaction, of the Ni1Fe0.5Cu1
catalyst, are shown in figure 5. The trends in the Cu edge look similar to Ni edge, and the Fe
edge spectra did not show much change in oxidation states. Ni edge results (figure 5) in
21
Page 21 of 41

396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
Ni1Fe0.5Cu1, indicates that Ni is partially oxidized at room temperature with an estimated 58%
NiO and only 42% Ni phase. The amount of Ni phase increases to 75% after reduction though it
is still not completely reduced. Adding ethanol does not seem to affect the oxidation state of Ni
as no changes in Ni phase is observed in presence of ethanol at 270°C, while the ethanol and
oxygen together increase the Ni oxidation state as indicated in the supplementary table S1 where
the fraction of pure nickel decreases to 70% from the initial value of 75% after reduction. It
should be noted that these changes are time dependent and they initially take place in a few
minutes but then they level off and remain nearly constant up to 40 mins. Limitations on beam
line use prevented us to study changes at longer time on stream.
Results from the Cu edge are as follows: Cu was found to be partly in Cu(I) oxidation
state in the Ni1Fe0.5Cu1 catalyst at room temperature, in air, having an estimated 25%Cu(I) and
75%Cu(0). This partial oxidation of Cu could be a result of long-term exposure to air during
sample preparation for XAS analysis. After reduction, the edge matches quite well with metallic
Cu reference foil edge (Fig 5b), indicating complete reduction to metallic Cu(0) oxidation state.
In the presence of ethanol at 270°C, a small change in oxidation state takes place and 15% of
Cu(0) changes to Cu(I), whereas the oxidation state in the presence of both ethanol and oxygen,
appears more pronounced giving 30% Cu(I), but in all the cases, a large fraction of Cu remains
in the reduced state. Results from the Fe edge in air at room temperature indicate that the sample
contains mainly Fe(+3) and some Fe(+2), giving an overall oxidation state of the entire sample to
be 2.3 (Fig 5c). After reduction, some fraction reduces to Fe(+2), but the edge height is much
higher than the reference Fe metal foil (Fig 5c), showing the absence of metallic Fe. At 270ºC, in
presence of ethanol as well as in presence of ethanol and oxygen, only a minor change was
22
Page 22 of 41

418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
observed in the edge shape and height indicating that only small changes in the oxidation state of
Fe occurs during exposure to the reaction mixture.
In addition to the XANES spectra, EXAFS spectra were taken at all the three, Cu, Ni and
Fe edges, to study the change in oxidation states, and estimate the size of nano-particles (NP) and
the coordination state of each element in Ni1Fe0.5Cu1. The phase and amplitude files of the
constituent phases in R (Fourier Transform) space from the EXAFS results can be fitted with the
phase and amplitude files obtained from references (e.g. metal-metal and metal-oxygen) to
obtain nearest neighbor and scattering distances. The Debye-Waller factors and energy
corrections (E0) used for fittings are listed in the tables provided in the supplementary file (Table
S1-S3).
Table S1 summarizes the results obtained on the Ni edge for Ni1Fe0.5Cu1 in different
conditions. Ni1Fe0.5Cu1 in air contains 42% Ni in reduced Ni(0) state which increases to about
75% Ni(0) metal after reduction, and in presence of ethanol at 270°C. In EtOH + O2 at 270ºC, a
slightly more oxidized NiO, by about 5%, is estimated from EXAFS fitting. Pre-treatment with
EtOH (either with or without O2) reduces the size of Ni nanoparticles to 2-3 nm whereas just
after reduction in H2 (at 300ºC) the nanoparticles were about 4-5 nm. It is interesting to note that
the metal-metal bond distance is 2.53 Å, which is an indicative of Cu-Ni bimetallic particle. Ni-
Ni scattering distances in Ni only catalyst is 2.49 Å, while Cu-Cu in Cu only catalyst is 2.55 Å.
An alloy of Ni-Cu would have an intermediate bond distance.
437
438
439
440
Fourier Transform of the real (solid lines) and imaginary part (dotted lines) of the fitted
k2·Chi(k) function of the EXAFS spectra of Ni-only catalyst reduced at 300ºC and the Ni foil are
shown in the supplemental information, Figs S2a and S2b. The smaller size of the magnitude of
FT indicates that the Ni nanoparticles are small and the imaginary part of the first shell and
23
Page 23 of 41

441
442
Figure 5: XAS spectra on Ni1Fe0.5Cu1 catalyst; in air, reduced and during reaction, (a)
Ni edge XANES, (b) Cu edge XANES and (c) Fe edge XANES
24
Page 24 of 41

443
444
445
446
447
448
449
450
higher shells are identical to that of Ni foil, indicating identical bond distances of the Ni only
catalyst w.r.t. the Ni foil (Fig. S2a). A comparison between the FT of k2·Chi(k) of Ni K-edge of
Ni-only catalyst and Ni1Fe0.5Cu1 catalyst after reduction at 300ºC, shows a large shift in the
imaginary part of the FT for the Ni1Fe0.5Cu1 catalyst indicating a different bond distance of 2.52
Å (Fig. S2b) than the reduced Ni-only catalyst. Since this value is in between the Cu reduced
bond distance (2.55 Å) and Ni (2.49 Å) it suggests that there is some intermediate degree of
phase distribution or alloying between Cu and Ni and the presence of an alloy is expected rather
than pure Cu or Ni phases.
451
452
453
454
455
456
Figure 6a shows the FT of k2·Chi(k) of the Ni edge of the Ni1Fe0.5Cu1 catalyst after
reduction and during the ethanol decomposition reaction at 270ºC. It is clear that these two peaks
fit very well indicating similar percentage of reduction in both conditions. During ethanol
decomposition, the reducibility of Ni nanoparticles (NP) does not change and remain nearly
constant at 75%. However the smaller magnitude of the FT indicates the presence of smaller NP
during ethanol decomposition reaction.
457
458
459
460
461
462
463
464
465
Figure 6b shows the magnitude of the Fourier Transform of the k2·Chi(k) function of the
EXAFS spectra of the Cu K-edge on the Ni1Fe0.5Cu1 catalyst after reduction and during ethanol
decomposition at 270ºC, along with results from a Cu foil used as reference. Cu in the
Ni1Fe0.5Cu1-air catalyst is 25% Cu(+1) and 75% Cu(0) metallic (Table S2). The smaller size of
the magnitude of k2·Chi(k) in presence of ethanol indicates a smaller particle size than after
reduction in H2. Upon reduction, Cu is fully reduced and the nanoparticles are larger than 9 nm,
as determined by an empirical relationship obtained by fitting the particle size and the
coordination number [24-26]. These estimates, however, may not be accurate at these large
sizes, i.e., at N greater than about 11. It is clear from fig 6b that most of the Cu in ethanol
25
Page 25 of 41

466
467
468
469
decomposition is in the reduced form but with formation of significantly smaller Cu
nanoparticles, ~ 3 nm. These results suggest that during oxidation-reduction–reaction
environments there is a significant structural transformation of the Cu-containing phases and a
phase segregation of Cu2O phase is likely to happen from NiCu bimetallic phases
470
471
Figure 6: (a) Ni K-edge Fourier transform of the Ni1Fe0.5Cu1 catalyst after reduction at 300ºC and
during ethanol decomposition at 270ºC. (b) Cu K-edge magnitude of Fourier transform of
Ni1Fe0.5Cu1 catalyst reduced at 300 ºC and in EtOH at 270ºC (k2: Δk = 2.7 – 10.7 Å-1)
471
472
473
474
None of the Fe containing catalysts show any metallic Fe in any of the conditions used, be
in air, reduced, or under reaction. Direct fitting of Fe XANES edge was not generally successful
compared with any of the reference samples and we lacked a NiFe2O4 reference that could have
provided a better fit.
475
476
477
XANES spectra of Fe oxide standards (FeO, Fe2O3 and Fe3O4) showed that the edge
energy shifts linearly toward higher values with increasing oxidation state of Fe (Fig. S3a). The
edge energy is linearly dependent on the oxidation state of the iron oxide samples as determined
26
Page 26 of 41

478
479
480
481
by a sharp peak in first derivative of the edge, with oxidation state (figure S3b). This
information was used to determine the average oxidation state of Fe in the samples (Table S3),
which changes slightly upon reduction with H2 at 300ºC, however, Fe is not completely reduced
to FeO or metallic Fe in any of the samples and conditions used.
482
483
484
485
486
487
488
It should be noted that the particles size suggested by XAFS are smaller than what is
displayed in the TEM image in figure 4. In Fig 4 a large particle was selected to obtain a
compositional map displaying the inhomogeneity at nanoscale and its size is not a statistically
average. The information obtained by XAFS is representative of a larger volume average value
from a sample volume determined from the beam diameter of 3 mm averaged at the mesophase
scale. Furthermore, since the particle is composed of multiple oxides, the XAFS particle size is
only an approximate estimation based on the metal fraction present in nanoparticles.
489
The XAS results discussed so far for Ni1Fe0.5Cu1 catalyst can be summarized as follows:
490
491
492
493
1) Ni was observed to be present giving 25% Ni(II) and 75% Ni(0) after reduction. Ni oxidation
state does not change during the ethanol decomposition reaction, however, it gets oxidized to
30% Ni(II) and 70% Ni(0) during ethanol partial oxidation reaction. Formation of mixed Cu-
Ni phases is also possible.
494
495
496
497
498
2) Cu in Ni1Fe0.5Cu1 catalyst in air is 25% Cu(I) and 75% Cu(0). During reduction in H2 at
300oC, it gets completely reduced to metallic Cu(0). During reaction, apart from pure Cu
metal nanoparticles, there is 15% Cu(I) during ethanol decomposition, which increases to
30% Cu(I) during ethanol partial oxidation. Particle size decreases from large sizes after
reduction to 3 nm during ethanol decomposition and ethanol partial oxidation reactions.
27
Page 27 of 41

499
500
501
3) Direct XANES fitting of Fe edge was not generally successful, however qualitatively no
metallic Fe was observed in the spectra and the oxidation state did not seem to be affected by
the reaction conditions.
502
503
504
505
506
507
508
509
510
511
512
513
514
515
The XPS and XAFS results were helpful in analyzing the surface and bulk concentration
of each element respectively as well as their oxidation states in different environmental
conditions. These techniques are effective in providing insights involving the reorganization of
the catalytic particles such as the rearrangement on the surface to minimize surface energy, and
change in oxidation state and size under reaction conditions. Because the technique operate in
different environments (UHV RT vs operando) the results are not the same but rather
complementary. Furthermore, the volume sampled in each case is different (surface vs bulk)
which will give similar results only if the catalyst are highly dispersed on a support and the XAS
signal are coming mainly from surface atoms. On the other hand the presence of the support
presents additional challenges to get a good signal/noise ratio.
3.5 Fourier Transform Infrared Spectroscopy:
FTIR studies were conducted to obtain information about the adsorbed intermediates on
the catalyst surface, and to be able to correlate them to the various oxidation states observed by
XAS studies.
28
Page 28 of 41

516
517
518
519
520
521
522
523
524
525
526
Fig. 7. In situ DRIFTS spectra of (a) Ni1Fe0.5Cu1 and (b) Cu only catalysts during ethanol
desorption at room temperature and various times.
Figure 7-a, 7-b show the results obtained during ethanol desorption studies on
Ni1Fe0.5Cu1 catalyst as well as on Cu only catalyst. To adsorb ethanol on the catalysts, a helium
stream saturated with ethanol was fed into the IR cell over the diluted catalysts for about 10 min
at room temperature. Then the cell was purged with pure He flow while recording the spectra at
various time intervals. The cell was modified to minimize the space between the window and
sample (~ 1mm) and reduce the signal from gaseous ethanol and ensure that most of the signals
are due to diffused reflectance from the catalyst’s surface and not from the gas phase. As
displayed in fig 7a, the intensity of the band corresponding to C―H stretching frequency (2800-
29
Page 29 of 41