Res Chem Intermed (2018) 44:3713–3722

https://doi.org/10.1007/s11164-018-3376-2

Catalytic deoxygenation of waste soybean oil

over hybrid catalyst for production of bio-jet fuel:
in situ supply of hydrogen by aqueous-phase reforming
(APR) of glycerol

Il-Ho Choi1 • Kyung-Ran Hwang1 • Hye-Young Choi1 •

Jin-Suk Lee1

Received: 5 September 2016 / Accepted: 24 November 2016 / Published online: 14 March 2018
 Springer Science+Business Media B.V., part of Springer Nature 2018

Abstract A one-pot reaction was performed to produce oxygen-free saturated

hydrocarbons via the catalytic deoxygenation and hydrogenation of waste soybean
oil over a hybrid catalyst (Pd/C and NiO/c-Al2O3). We utilized in situ hydrogen
generated from a reforming reaction of glycerol, a byproduct of triglyceride
hydrolysis, for the one-pot reaction to produce hydrocarbons. When NiO/c-Al2O3
(2 g) was used along with Pd/C (1 g), most of the unsaturated free fatty acids
(FFAs) were hydrogenated into saturated FFAs, and the percentage of desirable
hydrocarbons in the liquid product increased, in contrast to the case when only Pd/C
(1 g) was used. This result means that using a hybrid catalyst is better for promoting
the catalytic deoxygenation reaction than increasing the degree of loading of Pd/C,
and suggests that it should be possible to decrease the amount of precious metal
catalysts to be used for deoxygenation reaction.

Keywords Bio-jet fuel  Deoxygenation  Hydrogenation  One-pot

reaction  Hybrid catalyst  Waste soybean oil  APR of glycerol  In situ
hydrogen

Introduction

Nonedible oil is a potential inexpensive source for producing hydrocarbons via

catalytic processes in the range of diesel and jet fuel [1]. Biofuels such as green
diesel and bio-jet have several advantages. For example, they are directly
compatible with the present transportation infrastructure and exhibit improved fuel

& Kyung-Ran Hwang

[email protected]
1
Biomass and Waste Energy Laboratory, Korea Institute of Energy Research, 102, Gajeong-ro,
Yuseong-gu, Daejeon 34129, Republic of Korea

123
3714 I.-H. Choi et al.

properties, as they contain oxy-free hydrocarbons with carbon numbers in the proper
range [2].
Neste Oil’s NEXBTL and UOP/Eni’s Ecofining commercial processes [2–4]
produce green diesel through a hydrodeoxygenation (HDO) reaction. Further, bio-
jet fuel can be obtained through a series of additional catalytic reactions. However,
these processes have a few problems associated with them. For instance, they
involve costly pretreatments of the feedstocks, require a separation step, and use
large quantities of hydrogen; nearly 300–420 m3 of H2 is needed per unit volume
(m3) of vegetable oil [5, 6], which means that the price of the final product would be
increased.
In order to overcome the above drawbacks, a decarboxylation reaction which
removes the oxygen from free fatty acids (FFAs) in the form of CO2 can be used as
an alternate of HDO to produce hydrocarbons without hydrogen [7]. However, this
reaction causes a significant deactivation of the catalyst in the absence of hydrogen
[8, 9], and a small amount of hydrogen is still needed to delay the deactivation of the
catalyst. Here, we have focused on glycerol, the by-product of the hydrolysis of
triglyceride (TG). An amount of 1 mol of glycerol can be decomposed to 7 mol of
hydrogen and 3 mol of carbon dioxide by aqueous-phase reforming (APR) and be
utilized as a source of hydrogen.
Recently, the hydrothermal decarboxylation of hydrolyzed aqueous FFAs was
performed in a single-stage reactor without using a separation step; glycerol was
used as the source for the in situ production of hydrogen [10–12], in order to
simplify the process and to mitigate catalyst deactivation. Fu et al. reported the
hydrothermal decarboxylation of unsaturated FFAs in the absence of hydrogen [4].
However, low selectivity to oxy-free hydrocarbon was exhibited on a Pt[5]/C
catalyst, because, prior to catalytic deoxygenation to obtain saturated hydrocarbons,
the FFAs underwent hydrogenation of olefins. This means that extra hydrogen
addition was needed to obtain saturated hydrocarbons. Vardon et al. [11] examined
the coupled hydrothermal hydrogenation and deoxygenation of model compounds
using a Pt[5.2]–Re[4.2]/C catalyst while employing the hydrogen generated in situ
from glycerol. However, they found that the loading of hydrogen in the batch
reactor in the initial stage was necessary to complete the conversion of FFAs into
desirable hydrocarbons, as the catalytic deoxygenation rate of FFAs was a rate-
limiting step. Hollak et al. [12] examined the hydrothermal deoxygenation of TG
and FFAs over a Pd[5]/C catalyst while using hydrogen produced in situ from
glycerol. However, the main hydrocarbon products were long-chain olefins even
though the hydrogen produced in situ had a positive effect on the deoxygenation
activity. Furthermore, additional external sources such as glycerol and methanol
were needed to obtain saturated hydrocarbons, since the conversion rate of glycerol
using a Pd/C catalyst was relatively low in spite of the long reaction time (20 h).
Even though there have been many studies on one-pot reactions, the hydrothermal
catalytic processing of raw nonedible oil into hydrocarbons in a one-pot reactor has
not been investigated. Furthermore, this reaction requires the use of the precious
metal group catalysts because complex reactions, such as the hydrolysis of TG, the
reforming of glycerol into hydrogen, the hydrogenation of FFAs, and catalytic
deoxygenation reactions, are involved, as illustrated in Fig. 1.

123
Catalytic deoxygenation of waste soybean oil over hybrid… 3715

Fig. 1 The reactions associated with the one-pot catalytic process of nonedible oil

Here, we report the catalytic deoxygenation and hydrogenation of raw nonedible

oil (wasted soy bean oil; SBO) over a hybrid catalyst in a one-pot reactor, with the
aim of examining the possibility of decreasing the amount of precious metal catalyst
to be used for the deoxygenation reaction and of simplifying existing processes.

Experimental

Materials

The soybean oil purchased from Sajo was heated several days with air stirring to
obtain waste SBO. Distilled water (DW) was obtained using a lPure water
purification system. N,O-Bis(trimethylsilyl) trifluoroacetamide (C 98.0%, BSTFA;
Acros Organics) and pyridine (C 99.0%,;Samchun Pure Chemical) were used for
the silylation of the gas chromatography (GC) samples. In order to dilute the
samples for the GC/mass spectroscopy (MS) analysis, we used ethyl acetate
(C 99.5%), which was purchased from Samchun Pure Chemical.

Catalyst Preparation and Characterization

Palladium (5 wt%) loaded on activated charcoal was purchased from Aldrich, and
was used for the catalytic deoxygenation reaction [13]. The catalyst NiO/c-Al2O3
was employed for APR of glycerol because it exhibited good performance during
the direct conversion of glycerol generated from hydrolysis of TG into H2 and CO2
[14]. Nickel oxide (25 wt%) supported on c-Al2O3 (99.97%; Alfa Aesar) was
prepared by the incipient wetness impregnation method. Ni(NO3)26H2O, which
was used as a precursor, was obtained from Junsei Chemical. The synthesized NiO/
c-Al2O3 catalyst was dried at 105 C for 5 h and calcined at 500 C for 3 h in air.
The Brunauer–Emmett–Teller surface areas (SBET) and pore volumes of the
catalysts were measured by N2-adsorption at - 196 C using Micromeritics ASAP
2020. Prior to the analysis, catalysts were outgassed at 200 C for 4 h under
vacuum. Pore volume and pore diameter were calculated using the Barrett–Joyner–

123
3716 I.-H. Choi et al.

Halenda desorption. In order to determine the metal dispersion of Pd/C and NiO/c-
Al2O3, CO and H2 chemisorption were preformed, respectively, using the
Micromeritics ASAP 2020 system.

One-pot reaction

The one-pot reaction was carried out in a bench-top microreactor (100 mL;
Parr4590) made by Parr Instrument. The reaction conditions were: 0.5–1.5 g of Pd/
C, 0–3 g of NiO/c-Al2O3, 15 g of DW as medium, and 10 g of SBO. Pd/C was
reduced at 200 C for 2 h with 17.8% H2/N2, and nickel in NiO/c-Al2O3 was
employed as the form of metal oxide based on Ref. [14]. The catalysts, SBO and
DW were placed into the vessel, and the reactor was mounted in a heating jacket.
Before the start of the reaction, the reactor was purged with N2 (99.999%; Special
Gas, Republic of Korea) for 10 min, and the pressure within was charged to 2 bar
(N2). The reactions were performed at 300 C for 3–6 h. The stirrer (150 rpm) and
the timer were started when the heater was turned on. After the reaction was
completed, the reactor was allowed to cool to room temperature in air. Then, the
liquid and gas products were collected and analyzed.

Product analysis

Three types of samples were collected. The first type was a liquid-phase product
containing hydrocarbons and fatty acids; this was the top layer of the formed
product. Next was a water solvent containing dissolved glycerol as a by-product
derived from the conversion of TG. The last type was a gas-phase product. The top
of the liquid-phase product was analyzed using GC [Shimadzu GC2010 Plus,
equipped with a flame ionization detector and a Restek RTX-5 capillary column
(60 m 9 0.32 mm 9 0.25 lm)]–MS (Shimadzu GCMS-QP2010 SE). In order to
convert the carboxylic acid in the liquid product into a compound with a low
evaporation temperature, the sample was silylated with pyridine and BSTFA before
the GC and GC/MS analysis. The GC oven temperature was programmed as
follows: 1 min hold at 100 C, 3.5 C/min ramp to 200 C, 3 min hold at 200 C,
3.5 C/min ramp to 250 C, 3 min hold at 250 C, 5 C/min ramp to 280 C, and
1 min hold at 280 C. The detector temperature was maintained at 280 C, and the
samples were injected into the column with a 20:1 split ratio. In order to identify the
liquid-phase products, a GC/MS system equipped with the same column as that in
the GC system was used. For use in the MS analysis, the sample employed for the
GC analysis was 10 times diluted with ethyl acetate. The FFA content of the liquid-
phase produce was evaluated by quantifying the acid number (ASTM D974). To
quantify the acid number and the concentration of the FFAs, a known amount of the
sample was added to a mixture of ethanol and benzene (ethanol/benzene ratio of
1:1). This solution mixture was then titrated with a solution of KOH (0.5 N) using
phenolphthalein as a color indicator. As soon as the color of the sample changed to
purple, the titration process was halted, and the concentration of the FFAs was
calculated from the volume of the KOH solution added. The equation used was:

123
Catalytic deoxygenation of waste soybean oil over hybrid… 3717

NV
Acid number ¼ 56:1 
M
Acid number
FFAð%Þ ¼
2
where N is 0.5 (N), V is the volume of KOH added (mL), and M is the mass of the
sample (g). The water-soluble solvent obtained after the reaction was analyzed
using a high-performance liquid chromatograph (Shimadzu) with a refractive index
detector and an Aminex HPX-87H column; this was done to determine the con-
centration of glycerol in the solvent. A GC system [7890A, Agilent Technologies
equipped with a thermal conductivity detector and a Supelco Carboxen-1004
stainless steel column (2 m 9 1/16 in. 9 0.75 mm)] was us ed to analyze the gas-
phase product.

Results and discussion

One-pot reaction with Pd/C catalyst

The physiochemical properties of the prepared catalysts, Pd/C for the decarboxy-
lation reaction of the nonedible oils and NiO/c-Al2O3 for the in situ hydrogen
production reaction, are listed in Table 1. The purchased Pd/C catalyst had a large
average pore size and a low metal dispersion. The NiO/c-Al2O3 catalyst had an
extremely low total surface area as well as a low metal dispersion, owing to the fact
the active material was impregnated in it to a high degree.
The One-pot catalytic reaction of SBO with catalysts was carried out in order to
observe the progression of the following reactions: the hydrolysis of TG to FFAs,
the reforming of glycerol into hydrogen, the ohydrogenation of FFAs, and catalytic
deoxygenation. The distribution of the liquid product and the glycerol concentration
in the water-soluble liquid after the one-pot reaction of SBO are shown in Fig. 2; the
reaction conditions were: 300 C and 3 h in the presence of Pd/C and water. In
general, SBO is composed of linoleic acid (53.6 mol%), oleic acid (25.2 mol%),
palmitic acid (10.5 mol%), linolenic acid (7.7 mol%), and stearic acid (1.4 mol%).
When we tested the reaction without any catalyst, the FFA% of liquid product was
approximately 95% (wasted soybean oil’s FFA% & 1). Additionally, 84.4% of

Table 1 The physiochemical property of the prepared catalysts

Catalyst SBET (m2/g) Pore volume (cm3/g) Mean pore Metal
size (nm) dispersion
Total Micro Meso Macro Total Micro Meso Macro (%)

Pd/C 795.8 562.8 207.0 3.5 0.654 0.232 0.376 0.043 6.4 28.5
NiO/c- 50.2 9.9 36.0 3.6 0.273 0.004 0.222 0.048 24.2 1.8
Al2O3

123
3718 I.-H. Choi et al.

100 120
Pentadecane
Distribution of liquid product (area%)
Heptadecane

Conc. of glycerol in solvent (g/L)

100 Heptadecene
80 Palmitic acid
Linoleic acid
Oleic acid
80 Stearic acid
60 etc.
Glycerol
60

40
40

20
20

0 0
0/0 0/1 0/2 0.5 / 0 1/0 1.5 / 0

Amount of catalyst ( Pd/C / NiO/γ-Al2O3 )

Fig. 2 The distribution of liquid product and the glycerol concentration in the water-soluble liquid after
the one-pot reaction of SBO over Pd/C or NiO/c-Al2O3

unsaturated FFA (linoleic and oleic acid) and 15.6% of saturated FFA (palmitic and
stearic acid) were identified by GC analysis as shown in the 1st column of the table,
showing that SBO was mostly hydrolyzed to FFA at our reaction condition. In order
to confirm whether or not FFA can be deoxygenated by NiO/c-Al2O3, we tested
with using only NiO/c-Al2O3 and the results are depicted in the 2nd and, 3rd
column. While 1 g of NiO catalyst showed negligible change of the distribution of
the liquid product compared with that of the non-catalyst test, 2 g of NiO catalyst
made unsaturated FFA partially saturated with decreasing linoleic acid and
increased the other FFAs. Nevertheless, no hydrocarbon was detected in either
samples, revealing that NiO/c-Al2O3 has no catalytic activity for the deoxygenation
of SBO in this reaction. When 0.5 g of only the Pd/C catalyst was used, the FFAs
were more saturated. The FFAs produced were in the order shown: linoleic acid
(C18:2) & oleic acid (C18:1) [ palmitic acid (C16:0) [ stearic acid (C18:0). This
indicates that a few of the FFAs with double bonds in their molecule structures, such
as linolenic, linoleic, and oleic acids, were hydrogenated to oleic or stearic acids;
however, no hydrocarbons were observed, meaning that catalytic deoxygenation did
not occur during the one-pot reaction using 0.5 g of Pd/C. When an increase in the
amount of Pd/C used (1 g), linoleic acid, which was the most abundant FFA,
disappeared and a large amount of oleic acid was observed instead. This means that
the hydrogenation reaction was promoted by the hydrogen produced in situ with the
increase in the amount of Pd/C catalyst used. When the amount of the Pd/C catalyst
loaded was increased to 1.5 g, the primary FFA, oleic acid, disappeared and
deoxygenated products, namely, heptadecane, heptadecene, and pentadecane, were
observed. The catalytic deoxygenated products accounted for approximately 60% of
the liquid product. The results of the analysis of the gas product showed that H2 was

123
Catalytic deoxygenation of waste soybean oil over hybrid… 3719

not present in the case of 0.5 g Pd/C; however, small amounts of CO, CO2 and CH4
were detected. On the other hand, the amounts of H2 and CO2 produced increased
with an increase in the amount of catalyst loaded. With respect to glycerol, which
was a by-product of the hydrolysis of TG during the one-pot reaction, most of it was
consumed during the in situ production of hydrogen when 1.5 g of Pd/C was used.
This indicated that Pd/C acted as a catalyst both for the in situ production of
hydrogen and for the deoxygenation reaction [12]. However, a relatively high
amount of the Pd/C catalyst was needed for the two reactions. Furthermore, the
other oxy-free compounds such as hydrocarbons including the benzene ring (carbon
number greater than C17) were detected in significant amounts in the liquid product
with an increase in the amount of the Pd/C catalyst loaded. There was no significant
change in the component distribution of the liquid product; however, the amounts of
the above-mentioned hydrocarbons with benzene rings increased slightly when the
reaction time was increased to 6 h for 1.5 g of Pd/C (not shown here).
The component distribution of the liquid product after the one-pot reaction of
SBO with Pd/C is shown in Fig. 3; the reaction conditions were 300 C and 3 h, and
the reaction was performed both with and without DW as the medium. When water
was added during the one-pot reaction, the area percentage of the desirable
hydrocarbons reached 60%, with heptadecene making up only 1.4% of the liquid
product. This was because the hydrolysis of TG into FFAs and glycerol, the aqueous
reforming of the glycerol, and the hydrogenation of the FFAs occurred in sequence,
as illustrated in Fig. 1. On the other hand, the desirable hydrocarbons constituted
only 43% of the liquid product, with heptadecene forming a significant portion in
the absence of water. Moreover, the concentration of other compounds such as
alkyl-substituted aromatics and cyclohexanol increased to approximately 25%. Fu
et al. [10] reported that the catalytic deoxygenation rates appeared to be independent
of the carbon number; however, the number of double bonds present in the molecule

Fig. 3 The distribution of 100

Pentadecane
liquid product according to
Heptadecane
solvent after the one-pot
Heptadecene
reaction of SBO with Pd/C
Distribution of liquid product (area%)

80 Palmitic acid
(1.5 g)
Linoleic acid
Oleic acid
Stearic acid
60 etc.

40

20

0
DW None

123
3720 I.-H. Choi et al.

structure could be affecting the decarboxylation rate, potentially resulting in the

formation of high-molecular weight products. Thus, water is probably an essential
medium for the one-pot reactor, as it facilitates complex reactions, acting as a
medium during the production of hydrogen. However, even though it is possible to
produce hydrogen in situ during the one-pot reaction while using water as the
medium, a relatively high amount of the Pd/C catalyst is needed, as shown in Fig. 2.
Thus, the NiO/c-Al2O3 catalyst was further added in the one-pot reaction to promote
in situ hydrogen production by the aqueous reforming of glycerol [14], in
expectation of a reduction in the quantity of Pd/C participating in various reactions.

One-pot reaction with hybrid catalysts

As mentioned above, the one-pot reaction of SBO with a hybrid catalyst (Pd/C and
NiO/Al2O3 catalysts) was carried out at 300 C for 3 h in water, and the results are
shown in Fig. 4. When Pd/C (0.5 g) and NiO/Al2O3 (0.5 g) were used together, the
amount of linoleic acid present in the liquid product declined significantly, while
those of oleic acid and stearic acid increased. This results were markedly different
from that obtained when only 0.5 g of Pd/C was used but similar to that obtained
when 1 g of Pd/C was used, as shown in Fig. 2. Linoleic acid disappeared
completely when 1 g of NiO/c-Al2O3 was used; however, the amount of the
desirable deoxygenated hydrocarbons was not significantly increased with an
increase in the amount of NiO/c-Al2O3 loaded. This might be owing to the fact that
the amount of Pd/C used (0.5 g) was too small compared to the amount of the
reactant, and because the NiO/c-Al2O3 catalyst did not exhibit catalytic activity on

100 120
Pentadecane
Heptadecane
100 Heptadecene
Distribution of liquid product (ara%)

80 Palmitic acid
Conc. of glycerol in solvent (g/L)

Linoleic acid
Oleic acid
80 Stearic acid
60 etc.
Glycerol
60

40
40

20
20

0 0
0.0 0.5 1.0 1.5 2.0
Ni/γ-Al2O3 loading (g)

Fig. 4 The distribution of liquid product and glycerol concentration in water-soluble liquid after the one-
pot reaction of SBO with the hybrid catalyst (Pd/C 0.5 g and NiO/c-Al2O3 v g)

123
Catalytic deoxygenation of waste soybean oil over hybrid… 3721

the deoxygenation reaction. Moreover, the distribution of FFAs in the liquid-phase

product did not change much, while the decomposition rate of glycerol decreased.
The one-pot reaction to obtain deoxygenated hydrocarbons was performed using
higher amounts of the catalyst and glycerol, and the results are shown in Fig. 5. In
contrast to the case when only Pd/C (1 g) was used, the addition of the NiO/c-Al2O3
catalyst (2 g) raised the area percentage of the hydrocarbons to approximately 35%.
This means that the rate of the catalytic deoxygenation reaction was higher when the
hybrid catalysts were used than when greater amounts of Pd/C were used. In
general, theoretically, the available hydrogen is 1.53 mol to hydrogenate 1 mol of
SBO and the maximum hydrogen available from aqueous reforming of glycerol is
7 mol, along with the TG hydrolysis. Thus, it is sufficient to supply in situ hydrogen
for the hydrogenation of the double bonds present in FFAs. As shown in Fig. 5,
most of the FFAs were saturated hydrocarbons, with stearic acid and palmitic acid
also present in high concentrations. When glycerol was further added, the
concentration of the desirable hydrocarbons increased slightly, while the hep-
tadecene disappeared.

100
Pentadecane
Heptadecane
Heptadecene
Distribution of liquid product (area%)

80 Palmitic acid
Linoleic acid
Oleic acid
Stearic acid
60 Methyl-C16
Methyl-C18
etc.
40

20

0
0.5 / 0 / 0 0.5 / 1 / 0 1/0/0 1/2/0 1 / 2 / 2.5
Catalyst and additives loading (Pd/C / NiO/γ-Al2O3 /glycerol, g)

Fig. 5 The distribution of liquid product after the one-pot reaction of SBO with hybrid catalyst and
additional supply of glycerol

123
3722 I.-H. Choi et al.

Conclusions

A one-pot reaction was performed to produce oxy-free saturated hydrocarbons via

the catalytic deoxygenation and hydrogenation of SBO using a hybrid catalyst, Pd/C
and NiO/c-Al2O3. In the one-pot reaction, water was found to be an essential
ingredient for facilitating the complex reactions involved. When NiO/c-Al2O3 (2 g)
was used along with Pd/C (1 g), most of the unsaturated FFAs were hydrogenated
into saturated FFAs, and the concentration of the desirable hydrocarbons in the
liquid-phase product increased, in contrast to the case when only Pd/C (1 g) was
used. This means that in situ hydrogen supplied by APR of glycerol induced a
positive effect on the hydrogenation and deoxygenation reaction to produce
saturated compounds, and that the one-pot reaction is better promoted by using a
hybrid catalyst than by increasing the degree of Pd/C loading. This also suggests
that it should be possible to decrease the amounts of precious metals used as
catalysts for the deoxygenation reaction.

Acknowledgements This work was conducted under the framework of Research and Development
Program of the Korea Institute of Energy Research (KIER) (B8-2432-03).

References
1. E. Sari, M. Kim, S.O. Salley, K.Y.S. Ng, Appl. Catal. A 467, 261 (2013)
2. T.N. Kalnes, K.P. Koers, T. Marker, D.R. Shonnard, Environ. Prog. Sustain. Energy 28, 111 (2009)
3. J. Jakkula, V. Niemi, V.M. Purola, US Patent 20,040,230,085 (2004)
4. Neste Oil Corpration, Finland, Neste Oil (2012)
5. S. Lennox, L. Kristof, T. Adam, A. Bereczky, M. Mbarawa, A. Penninger, A. Kolesnikov, Fuel 107,
409 (2013)
6. D. Kubička, Collect. Czech. Chem. Commun. 73, 1015 (2008)
7. L. Hermida, A.Z. Abdullah, A.R. Mohamed, Renew. Sustain. Energy Rev. 42, 1223 (2015)
8. E.W. Pinga, J. Piersonb, R. Wallaceb, J.T. Millerc, T.F. Fuller, C.W. Jonesa, Appl. Catal. A 396, 85
(2011)
9. M. Ibáñez, B. Valle, J. Bilbao, A.G. Gayubo, P. Castãno, Catal. Today 195, 106 (2012)
10. J. Fu, X. Lu, P.E. Savage, Chemsuschem 4, 481 (2011)
11. D.R. Vardon, B.K. Sharma, H. Jaramillo, D. Kim, J.K. Choe, P.N. Ciesielski, T.J. Strathmann, Green
Chem. 16, 1507 (2014)
12. S.A. Hollak, M.A. Ariens, K.P. de Jong, D.S. van Es, Chemsuschem 7, 1057 (2014)
13. M. Snåre, I. Kubickova, P. Mäki-Arvela, K. Eränen, D.Y. Murzin, Ind. Eng. Chem. Res. 5, 5708
(2006)
14. G.C. Diaz, N.C.O. Tapanes, L.D.T. Camara, D.A.G. Aranda, Renew. Energy 64, 113 (2014)

123