Eco-Friendly Natural Clay: Montmorillonite Modified with Nickel or Ruthenium as an Effective Catalyst in Gamma-Valerolactone Synthesis

Article abstract of DOI:10.1007/s10562-021-03740-3

Ni/Ru metals supported on cheap and available support montmorillonite K10 were used for the selective hydrogenation of levulinic acid to γ-valerolactone. Different loadings of the metals were applied by the impregnation method, and detailed characterization was performed (UV–VIS, XRD, TPR, TPD, particle size distribution, SEM, XRF). Metals’ homogeneous distribution on the surface was confirmed. The selectivity to the desired product was almost independent on the used material. A detailed study of the influence of solvents on the studied reaction was also performed—protic alcohol-based solvents caused the formation of levulinic and valeric acid esters in the reaction mixture. The selectivity was influenced mainly by the alcohol structure (the highest selectivity obtained using isopropyl alcohol and sec-butanol). Mainly the solvent’s donor number (except ethanol) influenced the reaction rate. The prepared catalysts are promising, available, and cheap materials for the studied reaction. Solvent may significantly influence the yield of γ-valerolactone. Graphic Abstract: [Figure not available: see fulltext.].

Full text of DOI:10.1007/s10562-021-03740-3

Catalysis Letters
https://doi.org/10.1007/s10562-021-03740-3
Eco‑Friendly Natural Clay: Montmorillonite Modifed with Nickel
or Ruthenium as an Efective Catalyst in Gamma‑Valerolactone
Synthesis
Eliška Vyskočilová1 · Eva Vrbková1 · Jiří Trejbal1 · Michaela Vaňková1 · Libor Červený1
Received: 5 May 2021 / Accepted: 12 July 2021
© The Author(s), under exclusive licence to Springer Science+Business Media, LLC, part of Springer Nature 2021
Abstract
Ni/Ru metals supported on cheap and available support montmorillonite K10 were used for the selective hydrogenation of
levulinic acid to γ-valerolactone. Diferent loadings of the metals were applied by the impregnation method, and detailed
characterization was performed (UV–VIS, XRD, TPR, TPD, particle size distribution, SEM, XRF). Metals’ homogeneous
distribution on the surface was confrmed. The selectivity to the desired product was almost independent on the used mate-
rial. A detailed study of the infuence of solvents on the studied reaction was also performed—protic alcohol-based solvents
caused the formation of levulinic and valeric acid esters in the reaction mixture. The selectivity was infuenced mainly by
the alcohol structure (the highest selectivity obtained using isopropyl alcohol and sec-butanol). Mainly the solvent’s donor
number (except ethanol) infuenced the reaction rate. The prepared catalysts are promising, available, and cheap materials
for the studied reaction. Solvent may signifcantly infuence the yield of γ-valerolactone.
Graphic Abstract
Keywords γ-Valerolactone · Levulinic acid · Montmorillonite K10 · Nickel · Ruthenium
1 Introduction
* Eliška Vyskočilová
eliska.vyskocilova@vscht.cz
Catalytic hydrogenation of levulinic acid (LA, from biomass
resources) giving γ-valerolactone was widely studied in past
decades due to the above-mentioned lactone’s signifcant
potential. γ-Valerolactone (GVL) is considered a nontoxic
and biodegradable molecule, which can be used as a safe
* Eva Vrbková
eva.vrbkova@vscht.cz
1
Department of Organic Technology, University
of Chemistry and Technology, Technická 5, 16628 Prague,
Czech Republic
Vol.:(0123456789)
1 3

E. Vyskočilová et al.
food additive, eco-friendly solvent, or a building block for
the synthesis of other organic substances [1, 2]. According to
published literature, this catalytic hydrogenation can be per-
formed as a batch or continuous process. A hydrogen source
can be molecular hydrogen (in the gaseous phase) or, e.g.,
formic acid [2]. Usage of formic acid can be advantageous,
as it originates during levulinic acid synthesis [2].
one catalyst was advantageous for levulinic acid hydrogena-
tion and was reported in several publications [10, 11, 20].
Nickel–copper materials were also used as catalysts in a con-
tinuous arrangement. Almost total conversion of levulinic
acid was obtained using Ni–Cu–Al hydrotalcite-like material
(275 °C, H2 fow 20 ml min−1 [10]), and 98% GVL yield
was obtained using Ni–Cu–SiO2 combined catalyst (99%
conversion of levulinic acid, 250 °C, H2 fow 30 ml min−1).
Cobalt–Al2O3 material was successfully used in GVL forma-
tion (total conversion of levulinic acid,>99% selectivity of
GVL formation, 180 °C, 5 MPa, 3 h [14]).
The heterogeneous catalyst used in GVL synthesis can
be based on precious metals (ruthenium, palladium, irid-
ium [3–8]) or transition metals (nickel, cobalt, copper, zinc
[9–15]). Ruthenium-based materials are very selective
toward GVL formation from levulinic acid, as ruthenium
is able to selectively hydrogenate the C=O group, leav-
ing other double bonds untouched [3, 16]. Ruthenium on
mesoporous carbon support was used to obtain GVL from
levulinic acid (99% yield of GVL, total conversion of lev-
ulinic acid, 1 h, 110 °C, 4 MPa [17]). The catalytic activ-
ity of ruthenium on graphene solid support in this reaction
was also reported—total conversion of levulinic acid and
96% GVL yield (6 h, 30 °C, 6 MPa [4]) or 97% GVL yield
(2 h, 80 °C, 6 MPa [18]). Zirconium or titanium oxides as
solid support for ruthenium-based materials were also used
(>80% levulinic acid conversion, 99.9% selectivity of GVL
formation, 130 °C, 4 MPa, 3 h [3, 5]).
The usage of copper-based materials is advantageous
as this metal is selective in C=O bond hydrogenolysis and
inactive during C–C bond hydrogenolysis. These materials
can also decompose formic acid, which can be used as a
hydrogen source in this reaction [12, 13]. The use of cop-
per–ZrO2 materials was reported in the work of Ishikawa
et al. [21] with the result of 90% conversion of levulinic
acid and>90% selectivity of GVL formation (1 h, 200 °C,
3.5 MPa). Copper supported on graphene provided total con-
version of levulinic acid and 99% yield of GVL (4 h, 140 °C,
4.2 MPa [22]).
In the presence of a nonalcoholic solvent, this reaction
can proceed via two ways (Fig. 1). The frst mechanism is
starting with levulinic acid (a) hydrogenation providing
intermediate 4-hydroxypentanoic acid (b), which is follow-
ingly dehydrogenated and cyclized to desired GVL (c). The
second mechanism starts with the dehydrogenation of lev-
ulinic acid (a), giving an intermediate angelica lactone (d),
which is subsequently hydrogenated to desired GVL (c).
In this work, we ofer an extensive discussion dealing
with the solvent infuence on the reaction course that, based
on our knowledge, was not presented before. The discus-
sion about the infuence of used solvent describe the efect
of chosen parameters on the reaction course. We would
like to present also the usage of montmorillonite (natural
cheap ecofriendly clay), as a support for active metal for
this reaction. We have compared the performance of two
Palladium-MOF structure was reported to be used in lev-
ulinic acid hydrogenation (total conversion of levulinic acid,
140 °C, 2 MPa, 2 h [6]). As an example of diferent precious
metal, iridium-based complexes may serve. Their catalytic
activity in GVL formation was described, e.g., in [7, 8].
Moving to transition metals, we cannot avoid nickel-
based catalysts. Nickel-based materials are known for
selective hydrogenation abilities in systems having multi-
ple hydrogenable groups (e.g., C=C and C=O bonds [19]).
Zhang et al. [9] reported the usage of nickel-modifed zeolite
HZSM-5 in the studied reaction (total conversion of lev-
ulinic acid, 95% yield of GVL formation, 170 °C, 3 MPa,
2 h, solvent- water). Ni–Al2O3 was used as a catalyst in the
work of Varkolu et al. [2]. The copper-nickel combination in
Fig. 1 Possible reaction
pathways from levulinic acid to
γ-valerolactone via two possible
intermediates: 4-hydroxypen-
tanoic acid (b) and angelica
lactone (d)
1 3

Eco‑Friendly Natural Clay: Montmorillonite Modifed with Nickel or Ruthenium as an Efective…
types of metal-modifed forms of this material—nickel, and
ruthenium. We have chosen these two types of metals based
on their previously discussed properties in levulinic acid
hydrogenation, giving the desired GVL but using diferent
support types.
2.3 Calculation of Reaction Rate
Initial reaction rate was calculated based on the fact that at
the beginning of the reaction the reaction rate may be taken
as of the frst order, meaning linear, and a tangent may be
constructed with only low error. Thus, the equation used
for the calculation of the initial reaction rate was following:
2 Experimental
Δc
rrinit
=
,
(1)
Δt × mcat
2.1 Catalyst Synthesis
where Δc is the diference in concentrations of LA at the
beginning of the reaction and at appropriate time in mmol,
Δt is a time diference in min, and mcat is amount of catalyst
(g).
Montmorillonite K 10 (Sigma Aldrich) was mixed with
solution of nickel(II) nitrate hexahydrate (Lachema, p.a.) or
ruthenium (III) chloride (Sigma Aldrich, 99%) in deminer-
alized water (e.g. in case of K10-Ni30 was used 42.46 g of
nitrate and 20 g of montmorillonite K10; in case of K10-Ru5
was used 2.75 g of chloride and 20 g of montmorillonite
K10). Suspension was stirred overnight at room tempera-
ture and then water was evaporated using rotary evaporator.
Prepared material was dried (air, 100 °C, 12 h) and calcined
(air, 450 °C, 12 h). Finally, materials were reduced using
hydrogen–nitrogen fow (H2:N2 ratio 1:9) at specifed tem-
perature—400 °C (ruthenium-based materials) and 500 °C
(nickel-based materials) for 3 h. We have chosen such no
typical process for preparation of nickel-based catalyst due
to the fact that the commercial montmorillonite is an avail-
able and cheap material (usually such high metal loadings
are prepared by mixed precipitation).
Activation energies were calculated from Arrhenius equa-
tions based on dependence of reaction rates (reaction con-
stants) on temperatures. Taken in to account the fact that
the dependence of ln(k) on T1 is linear, the linear coef-
cient of the obtained equation is proportional to the activa-
tion energy. R2 of linear equations were 0.9991 and 0.9944
for K10-Ni50 and K10-Ru2.5, resp. Ability of solvent to
dissolve hydrogen were calculated according to NRTL/
Redlich–Kwong equation of state with Henry’s law using
Aspen Plus 11 database.
2.4 Characterization Methods
Temperature programmed reduction (TPR) was carried out
using an AutoChem II Micromeritics Instrument 2920. Cata-
lyst sample (0.1 g) was placed in a quartz U-shaped tube and
during pretreatment the material was heated to 150 °C under
argon fow (30 ml min−1) and kept under this temperature
for 1 h to remove physisorbed water. Afterward, the linear
temperature program (15 °C min−1) started at a temperature
of 150 °C and the sample was heated up in hydrogen-argon
fow (10% of hydrogen in argon, 30 ml min−1) up to a tem-
perature of 700 °C. Temperature programmed desorption of
pyridine (TPD) was performed to monitor material’s acidity.
Detailed description for this measurement were performed
in our previous works [23–26]. The particle size distribu-
tions of all samples were determined by laser light scattering
(Malvern Mastersizer 3000 system equipped with Hydro EV
wet sampling unit, Malvern Instruments Ltd.). The materials
were characterized using a wet dispersion method. For each
sample, particle size distributions were recorded for at least
10 determinations at an obscuration range of 8–14%. The
particle size distributions of all materials were determined
using demineralized water (conductivity<1 μS cm−1) with
surfactant Triton X-100 (Aldrich) as a dispersion medium.
The reference refractive index for materials was 1.555.
ARL 9400 XP sequential WD-XRF spectrometer was used
2.2 Levulinic Acid Hydrogenation
Hydrogenation was performed in an autoclave (volume
160 ml, Parr). Autoclave was flled with an appropriate
amount of catalyst (10 wt% in case of Ru catalysts, 20 wt%
in case of Ni catalysts, based on levulinic acid), levulinic
acid (10 g, Merck, 98%), internal standard—mesitylene
(0.8 g, Acros Organics, 99%) and 100 ml of solvent—tolu-
ene, 1,4-dioxane, heptane, cyclohexane, methanol, propan-
2-ol, propan-1-ol, butan-1-ol, ethyl acetate (all Penta, p.a.),
ethanol (Penta, 96.5%) or butan-2-ol (Lachema, 99%). Sam-
ples taken during the reaction were centrifuged and ana-
lyzed using gas chromatograph coupled with mass detector
(Shimadzu 2010 Plus, GCMS-QP2010 Ultra) equipped with
nonpolar column DB-5. Injector temperature was 250 °C,
the programmed temperature program ranged from 80 °C to
250 °C using rate 10 °C min−1, detector voltage was 0.8 kV
and ion source had temperature 220 °C. Experiments were
repeated two times and the error between these two measure-
ments was not higher than 8%. Reuse experiment: reaction
mixture was centrifuged, catalyst was separated and washed
3 times with toluene (10 ml). Wet catalyst was mixed with
fresh reactants and the reaction was performed as usual.
1 3

E. Vyskočilová et al.
to perform X-Ray fuorescence analysis. Powders for SEM
analysis were applied on carbon adhesive tape and then gold-
ened by 5 nm of gold via Quorum Q150 ES (Quorum Tech-
nologies Ltd., Laughton, UK). Footage were made by scan-
ning electron microscope TESCAN VEGA 3 LMU (Tescan
Brno, Czech Republic) in regime low vacuum (UniVac)
at pressure in working chamber 1 Pa, accelerating voltage
20 kV and mod BSE. For elemental analysis, EDS analyzer
OXFORD instrument X-max 20 mm2 with software Aztec
(Oxford Instruments, Abingdon, UK) was used. Tescan
Lyra3GMU/EDS/EBSD/STEM/TOFSIMS (Tescan Brno,
Czech Republic) was used for STEM image using STEM
grid. UV–VIS–NIR absorption spectra of the prepared
samples were recorded using spectrophotometer (Shimadzu
UV2600i) with integration sphere. X-ray difraction powder
data were measured at RT on θ–θ powder difractometer
X’Pert PRO in Bragg–Brentano parafocusing geometry with
CoKα radiation (λ=1.7903 Å, U=35 kV, I=40 mA). The
specifc surface area was determined using nitrogen adsorp-
tion (3Flex volumetric analyzer, Micromeritics, USA). The
specifc surface area of materials was calculated via the BET
equation and t-plot method.
3 Results and Discussion
3.1 Material Characterization
The temperature-programmed reduction of calcined mate-
rials was performed (Fig. 2) to characterize metal-sup-
port interactions of material. In the case of Ni-containing
materials, two reduction peaks are visible–the frst around
430–450 °C. This can be assigned to Ni species encapsulated
into the mesopores of the montmorillonite layer framework.
The second reduction peak around 550–580 °C is indicating
that the cationic exchange of Ni2+ occurred during impreg-
nation. This peak is assigned to Ni species located in mont-
morillonite interlayers [27, 28]. In the case of ruthenium-
based materials, the reduction peak in the range 170–250 °C
was observed. This can be attributed to the reduction of
RuO2 [29].
X-ray fuorescence (XRF) results confrmed successful
metal deposition and metal content similar to desired val-
ues (Table 1). Temperature programmed desorption (TPD)
of pyridine revealed that material acidity decreased after
modifcation with metals. That can be caused by the fact
that the metal particles partially covered the montmorillonite
surface. This could induce some acid sites’ unavailability
in the montmorillonite structure for pyridine during TPD
measurement. All materials possessed by one broad peak
with the maximum in the range of 280–300 °C (Online
Resource, Fig. S1) reveal that more types of acid sites were
present on the material. A lower decrease of material acidity
in Ni 30% modifed montmorillonite was probably caused
by the fact that Ni is weak Lewis acid and may possess by
some amount of weak acid sites. On the other side, the low-
est amount of acid sites was detected in the case of K10
modifed by 50% of Ni. In the case of a higher amount of
nickel on the support, the blockage of interlayers of mont-
morillonite (that are responsible for the material acidity)
probably occurred. In the case of Ru modifed materials, the
diference was negligible.
A decrease of specifc surface area was observed after
modifcation by both metals. In the case of Ru modifed
materials, this decrease of specifc surface area was only
Fig. 2 Reduction peaks of material modifed with ruthenium and
nickel (TCD signal)
Table 1 Results of X-ray
Material
XRF
Calculated
Ni (%)
TPD
SBET
m2 g−1
fuorescence, temperature
programmed desorption of
pyridine and specifc surface
area
NiO (%)
RuO4 (%)
Ru (%)
npyr des
−1
(µmol gcat
)
K10
–
–
–
–
4.6
9.6
–
–
–
–
2.8
5.9
87
57
15
34
41
233
142
97
220
203
K10-Ni30
K10-Ni50
K10-Ru2.5
K10-Ru5
32.9
56.8
–
25.8
44.6
–
–
–
1 3

Eco‑Friendly Natural Clay: Montmorillonite Modifed with Nickel or Ruthenium as an Efective…
slight (from 233 to 220 resp 203 m2 g−1) and corresponded
to the loading of Ru. Ni modifed materials showed a more
signifcant decrease was (from 233 to 142 resp. 97 m2 g−1).
A signifcantly higher modifcator amount caused this con-
siderable decrease of specifc surface area in the case of
Ni-modifed materials.
Table 2 Characteristics of materials obtained using laser difraction
measurement
Material
Span value
Particle size (µm)
Dv(10)
Dv(50)
Dv(90)
K10
3.7
3.6
3.3
3.3
4.9
3.96
3.48
3.68
2.90
0.20
17.7
12.8
13.3
13.8
69.7
49.1
47.4
49.2
44.6
K10-Ru2.5
K10-Ru5
K10-Ni30
K10-Ni50
Scanning Electron Microscopy with Energy Dispersive
Spectroscopy (SEM-EDS) confirmed the homogeneous
distribution of desired metals (nickel, ruthenium) on cat-
alyst surface (Fig. 3). STEM image of K10-Ni50 (Online
resource, Fig. S10) showed the particle size of Ni in this
material.
9.06
In all cases after the modifcation of montmorillonite
with metal, particle sizes Dv(50) and Dv(90) decreased
compared to non-modifed material (Table 2). Materials
prepared by modifcation using ruthenium possess very
similar Dv(50) and Dv(90) values (12.8 µm and 13.3 µm,
and 49.1 µm and 47.4 µm respectively). This can be caused
by the fact, that the absolute ruthenium amount on material
is not so signifcant as in case of nickel (2.5–5% compared
to 30–50%). A signifcant decrease of particle size with
increasing nickel amount (Dv(10) value 0.2 µm for Ni50
material compared to 3.96 µm for nonmodifed montmo-
rilonite) can be caused by fact, that calcination of nickel
Fig. 3 SEM-EDS images
of nickel-modifed material
(K10-Ni30, up) and ruthenium-
modifed material (K10-Ru2.5,
down), magnifcation 1000x
1 3

E. Vyskočilová et al.
precursor can led to origination of NiO particle, which size
is smaller than common montmorillonite particles [30].
Particle size distribution (Fig. 4) for non-modifed mont-
morillonite is similar to already published results [31, 32].
Origination of smaller particles in the case of nickel modi-
fed materials as mentioned above is clearly visible on the
graph obtained from laser difraction measurement.
All samples were analyzed through the difuse refectance
arrangement in the range of 220–1400 nm (Fig. 5). In the
near-infrared region, no signifcant absorption was observed.
A spectrum obtained for pure montmorillonite was similar
to the literature [33]—the only band was observed with a
maximum around 250 nm. A slight shift of this maximum
to c.a. 230 nm was observed in the case of modifcation
with ruthenium. On the other hand, nickel-modifed materi-
als possessed a signifcant band around 320 nm, which was
already observed at Ni-modifed montmorillonite [30].
X-ray difraction (XRD) of prepared calcined materials
showed that montmorillonite was in the amorphous form.
Still, both metals (Ru and Ni) were successfully reduced
and were present in the form of crystallites. The difracto-
grams are shown in the Online resource (Fig. S2). To see
the changes in the size of interlayers was impossible since
the used commercial montmorillonite K10 was amorphous.
Nevertheless, other characterization methods showed the
uniform distribution of metals on the surface.
Fig. 5 UV–VIS spectra of (non)modifed montmorillonite
acid the choice of the solvent. Alcohols in the presence
of an acidic catalyst (in this case, the support montmoril-
lonite) may react with levulinic acid by means of esteri-
fcation. In the reaction mixture, including alcohols and
levulinic acid, esters of the starting material are present
together with esters of valeric acid. The explanation of
valeric acid formation is two: (i) the frst ofered in the lit-
erature [34] is based on the supposition that the intermedi-
ate is pent-3-enoic acid in the following step hydrogenated
to valeric acid. Pent-3-enoic acid is formed in the reaction
mixture from GVL in the presence of an acid catalyst.
Ester of valeric acid is then formed by esterifcation with
the present alcohol (Fig. 6). (ii) The second possibility of
valeric acid formation is based on direct hydrogenolysis
of GVL in the presence of hydrogenation catalyst. In this
3.2 Reaction Results
3.2.1 Infuence of the Solvent
From the point of view of the product, variability is the
most exciting parameter in the hydrogenation of levulinic
Fig. 4 Partile size distribution obtained using laser difraction measurement
1 3

Eco‑Friendly Natural Clay: Montmorillonite Modifed with Nickel or Ruthenium as an Efective…
Fig. 6 Mechanism of valeric acid formation from GVL [34]
Fig. 7 Possible hydrogenation course of LA to GVL with subsequent valeric acid formation
case, valeric acid is directly formed from GVL. And again,
ester of valeric acid is produced by esterifcation (Fig. 7).
Both ways may participate in the formation of valeric
acid. It cannot be decided which one is the more probable
as the rate of pent-3-enoic acid hydrogenation would be
very fast under used hydrogenation conditions. This inter-
mediate would not be detected in the reaction mixture.
We tested the linear alcohols as solvents in levulinic
acid hydrogenation in the row from methanol to n-butanol
and to see the infuence of alcohol branching, isopropyl
alcohol, and sec-butanol. pKa values, donor numbers, die-
lectric constants of all used solvents and solvents’ ability
to dissolve hydrogen are given in Table 3.
Table 3 Properties of used solvents (dissociation constant—pKa,
donor number—DN and relative permittivity—ε)
Solvent
pKa [35–37] DN
ε [39] Solubility of
) H2 (ppm w/w)
(kJ mol−1
[38]
Toluene
1,4-dioxane
Heptane
Cyclohexane
Ethyl acetate
Methanol
Ethanol
n-Propanol
n-Butanol
41
− 3.9
n/a
n/a
25
15.5
15.5
16.1
16.1
0.1
14.8
n/a
2.38 585
2.25 578
1.92 848
2.02 630
6.02 537
n/a
17.1
19.0
19.2
19.8
29
21.1
19.5
32.7
24.5
18.3
17.8
17.9
17.3
572
516
462
416
529
485
As we expected, the correspoding levulinates and valer-
ates were detected in the reaction mixtures where alco-
hols were used as solvents. We have calculated the initial
reaction rates (Table 4) for diferent alcohols (from the
decrease of concentration of levulinic acid in the reac-
tion mixture) and tried to correlate them with the solvent
properties. The overall reaction mixture’s complexity did
not enable us to evaluate the exact dependence of reaction
rate on two of chosen properties, dissociation constant and
relative permittivity. In the case of donor number, its infu-
ence on the initial reaction rate was observable (Fig. 8).
The observed trend could follow the quadratic regression.
The value of the coefcient of determination R2 for the
Eq. (2) was 0.8818.
Isopropylalcohol 17.2
sec-Butanol 17.6
rrinit = −0.0049 × (DN)2 + 0.2361 × (DN) − 2.5753
(2)
Nevertheless, we detected one exception—ethanol, where
the reaction rate was the fastest (and the donor number does
not difer a lot). The reaction rate was the slowest, using
n-butanol as a solvent that may be connected with a rela-
tively high donor number that may cause the blockage of
acid sites for catalyst necessary for the reaction to proceed.
1 3

E. Vyskočilová et al.
Table 4 Initial reaction rate calculated for the reactions in diferent
alcohols as solvents (catalyst K10-Ru5, 10 g LA, 10% of catalyst,
100 ml of solvent, 120 °C, 4 MPa)
the sterical hindrance. The highest selectivity to GVL was
obtained using isopropyl alcohol together with sec-butanol,
both secondary alcohols in our choice (Table 5). The difer-
ence between these two alcohols was negligible. The lowest
selectivity to GVL was observed using methanol, the alco-
hol with the shortest chain and the most suitable alcohol
for performing alcoholysis (methanolysis) and the esterif-
cation. That’s why the highest concentration of levulinate
together with valerate was detected in the reaction mixture.
From the obtained data (Table 5), the levulinate behavior as
reaction intermediate was apparent (not only in the case of
methanol as a solvent, Fig. S4). Using methanol, the main
product of the levulinic acid hydrogenation seemed to be
methyl valerate. The reaction was relatively slow, and total
conversion was not achieved in 5 h. Thus, we only suppose
that GVL concentration would decrease, showing that this
product was subject to methanolysis. Similar results were
already described [41]. In the row ethanol—n-propanol—
n-butanol, the highest selectivity to GVL was achieved using
ethanol. The alcoholysis of GVL represented by selectivity
to the corresponding valerate was comparable for all these
used alcohols. Thus, they difered in the extent of esterifca-
tion of starting levulinic acid. The amount of levulinate in
the reaction mixture increased with the increasing alcohol
chain length (Table 5). It was a little bit surprising because
the esterifcation typically proceeded faster in alcohols with
shorter chains [42], but some data, same as in our case, can
be found [43, 44]. However, in the latter mentioned cases,
the authors explained that behavior by the physical proper-
ties of used alcohols. Our system was more complicated.
Based on chosen solvent properties in Table 3, the higher
basicity of the solvent and lower polarity boosted the esteri-
fcation. However, we can conclude that in the case of valeric
acid esters’ desired production (e.g., for biofuel production),
methanol would be an advisable solvent [41]. When GVL
is the desired product, the use of alcohol should be avoided.
Based on the previous paragraph, we tested other solvents
(toluene, 1,4-dioxane, ethyl acetate, heptane, and cyclohex-
ane) that were aprotic to avoid the reaction between sol-
vent and products in the reaction mixture. The selectivity
in all these solvents was 100% to GVL; no other products
or intermediates were detected in the reaction mixture. For
Solvent
rrinit
−1
(mmol min−1 gcat
)
Methanol
Ethanol
n-Propanol
n-Butanol
Isopropylalcohol
sec-Butanol
0.14
0.55
0.20
0.18
0.23
0.19
Fig. 8 The dependence of initial reaction rate on donor number (cata-
lyst K10-Ru5, 10 g LA, 10% of catalyst, 100 ml of solvent, 120 °C,
4 MPa)
We have to mention that all used alcohols independently on
their structure were included in this evaluation.
As it is often discussed in the literature, the solvent may
signifcantly infuence the selectivity (e.g. [40]). In levulinic
acid hydrogenation using alcohol as a solvent, the infuence
is evident, as was mentioned above. Alcohols do no serve
only as solvents but also as reactants producing esters in
the reaction mixture. The most important property of the
solvents infuencing the selectivity was, without questions,
Table 5 Selectivities to chosen
Solvent
Selectivity at 40% conv. LA (%)
Selectivity at 60% conv. LA (%)
compounds in the reaction
mixture obtained at conversion
of levulinic acid 40 and 60%
(catalyst K10-Ru5, 10 g LA,
10% of catalyst, 100 ml of
solvent, 120 °C, 4 MPa)
GVL
Levulinate
Valerate
GVL
Levulinate
Valerate
Methanol
Ethanol
n-Propanol
n-Butanol
Isopropylalcohol
sec-Butanol
14.9
63.4
54.7
32.1
75.6
73.8
57.5
18.1
31.5
37.1
8.6
27.6
18.5
8.4
12.3
8.3
31.4
68.2
51.2
41.6
80.4
76.5
22.7
13.1
26.0
34.9
5.7
43.7
17.8
17.1
16.8
11.2
13.5
13.4
11.5
8.8
1 3

Eco‑Friendly Natural Clay: Montmorillonite Modifed with Nickel or Ruthenium as an Efective…
comparison, the reaction was also carried out without cata-
lyst—in this case no conversion of LA was observed. In hep-
tane and cyclohexane (under applied reaction conditions),
the initial reaction rate could not be calculated because, in
these solvents, the reaction proceeded almost immediately,
which is in a good correlation with values of hydrogen sol-
ubility in the solvents. These two solvents posess by the
highest hydrogen solubility among studied aprotic solvents
(Table 3, 848 and 630 ppm). The solubility of hydrogen in
other solvents did not difer a lot, thus their diferent perfor-
mance should be explained by other properties. In heptane,
total conversion of LA was achieved in 30 min, in cyclohex-
ane during one hour of the reaction. Both of these solvents
are nonpolar and do not possess any acido-basic properties.
It seems very convenient for levulinic acid hydrogenation to
Fig. 10 Infuence of the conversion of LA in 5 h on chosen solvent
GVL as the main product. The dependence of GVL concen-
tration in all solvents on reaction time is depicted in Fig. 9.
Dealing with the other three solvents (toluene, 1,4-dioxane,
and ethyl acetate), we found some interesting facts. From the
parameters (catalyst K10-Ru5, 10 g LA, 10% of catalyst, 100 ml of
solvent, 120 °C, 4 MPa)
point of view of initial reaction rate, we have seen almost
showing the infuence of pKa are exciting. The polar sol-
vents interact with the catalytic surface, blocking the active
site and decreasing the rate of the catalyzed reaction.
−1
no difference (toluene rrinit = 0.190 mmol min−1 gcat
;
1,4-dioxane rrinit =0.118 mmol min−1 gcat−1 and ethyl acetate
rrinit =0.120 mmol min−1 gcat−1). On the other side, checking
the results at the end of the reaction procedure (Fig. 9), the
achieved results difered more. We evaluate the infuence of
diferent parameters on the achieved conversion after 5 h of
the reaction (Fig. 10). We can conclude that the achieved
conversion did not show any trend based on polarity and
donor number in chosen low polar aprotic solvents. Still,
with increasing pKa (thus decreasing acidity), the achieved
conversion increased. Nevertheless, for the production of
GVL, the solvents without acido-basic properties would be
more convenient, as was mentioned above. Still, the results
3.2.2 Infuence of Reaction Conditions
Based on the results obtained by testing the solvent efects,
toluene was chosen as a solvent to study the infuence of
catalyst type and the reaction conditions. We have to say
that no by-products were detected in the reaction mixture;
thus, the selectivity to GVL was in all cases 100%. Also, no
intermediates were detectable during the reaction procedure.
Comparing metals used for hydrogenation on montmoril-
lonite support only confrmed the well-known fact that Ru
is more active than Ni. Under 110 °C and 6 MPa, the reac-
tion rate using K10-Ru2.5 was higher comparing K10-Ni50
(Tables 6 and 7). The acidity of materials, as same as parti-
cle size distribution, did not play a role in the reaction rate,
probably because they did not difer a lot. Thus the main was
the activity of the metal itself. The specifc surface area was
not the parameter infuencing the reaction rate because with
higher metal loading the specifc surface decreased and the
reaction rate increased.
The amount of the metal on the support in both cases
infuenced the reaction rate positively- with a higher amount
of the metal, the reaction rate increased. The values of reac-
tion rate using K10-Ni50 and K10-Ni30 (130 °C, 8 MPa)
−1
were 0.441 mmol min−1 gcat−1and 0.163 mmol min−1 gcat
resp., and for K10-Ru5 and K10-Ru2.5 (120 °C, 6 MPa)
−1
were 0.201 mmol min−1 gcat−1 and 0.134 mmol min−1 gcat
resp. (Tables 6 and 7). The increase corresponded to an
increase in metal loading in the material.
Fig. 9 Concentration of GVL in the reaction mixture over time (cata-
lyst K10-Ru5, 10 g LA, 10% of catalyst, 100 ml of solvent, 120 °C,
4 MPa)
The infuence of the pressure on the reaction rate showed
the expected trend—with increasing pressure the reaction
1 3

E. Vyskočilová et al.
Table 6 The infuence of
reaction conditions on intitial
reaction rate catalyzed by Ni
modifed montmorillonite (10 g
of levulinic acid, 20 wt% of
catalyst, toluene)
K10-Ni50
K10-Ni30
Tempera- Pressure (MPa) rrinit
Temperature (°C) Pressure rrinit
−1
−1
ture (°C)
(mmol min−1 gcat
)
(MPa)
(mmol min−1 gcat
)
130
6
8
6
0.338
0.441
0.07
130
1
3
5
7
8
0.030
0.080
0.140
0.149
0.163
110
90
0.01
Table 7 The infuence of
reaction conditions on intitial
reaction rate catalyzed by Ru
modifed montmorillonite (10 g
of levulinic acid, 10 wt% of
catalyst, toluene)
K10-Ru5
K10-Ru2.5
Tempera- Pressure (MPa) rrinit
Tempera- Pressure (MPa) rrinit
−1
−1
ture (°C)
(mmol min−1 gcat
)
ture (°C)
(mmol min−1 gcat
)
120
4
6
8
0.160
0.201
0.251
90
100
110
120
6
0.041
0.059
0.100
0.134
rate also increased using both types of catalyst. For Ni modi-
fed montmorillonite, the pressure efect was measured using
K10-Ni30 (Table 6) under 130 °C, which was the temper-
ature convenient for performing the reaction. In the case
of Ru modifed support the infuence of the pressure was
measured using the materials containing 5% of Ru (Table 7)
under 120 °C, which was the highest tested temperature in
this type of catalyst.
times without any loss of activity (Online resource Fig. S6).
Reaction rates for these three experiments were very similar:
0.098–0.100 mmol min−1 gcat−1. In this case, the preserva-
tion of the material structure after reuse was monitored and
confrmed using XRD (Online resource Fig. S2).
4 Conclusion
We calculated the apparent activation energy from the
temperature influence (Tables 6 and 7) on the reaction
rate for both types of metals. For the reaction catalyzed by
K10-Ru2.5, the calculated activation energy of levulinic acid
hydrogenation was 48 kJ mol−1. This value corresponds to
the expected value for hydrogenation reactions. For the reac-
tion catalyzed by K10-Ni50, the calculated activation energy
reached the value 107 kJ mol−1 that is a relatively high value
but still acceptable for hydrogenations. Such high activation
energy may show that Ni’s activity in hydrogenation reac-
tions is signifcantly dependent on the temperature, and it is
necessary to use temperatures higher than 130 °C (even bet-
ter are temperatures higher than 150 °C) using this type of
catalyst. From Table 6, it is visible from the values of initial
reaction rates that at low temperatures (90 °C and 110 °C),
the desired hydrogenation of levulinic acid proceeded very
slow.
Nickel and ruthenium-based catalysts on montmorillonite
K10 support were used in the hydrogenation of levulinic acid
to γ-valerolactone. The detailed characterization showed the
appropriate reduction temperatures for each metal (TPR,
XRD), ability to prepare materials with the desired loading
of the metal on support (XRF), homogeneous coverage of
the montmorillonite by metals (SEM), relatively homogene-
ous particle size distribution (laser light scattering) and low
and similar acidity of all materials (TPD). Both metals on
chosen support are possible catalysts for the studied reac-
tion showing high selectivity to γ -valerolactone and, under
specifc conditions, with relatively high reaction rates. The
selectivity was infuenced neither by the type of metal nor
metal loading nor any other conditions, and achieved values
were up to 100%. We also ofered deep insight into the infu-
ence of solvent on the hydrogenation of levulinic acid. The
alcohol-based solvents in the reaction mixture supported the
formation of levulinic and valeric esters of corresponding
alcohol. The selectivity to esters was the highest in metha-
nol as a solvent—the smallest, non-branched alcohol. The
In Table 6, the results of the reproducibility test is also
shown. The diference in the case of K10-Ni30 (130 °C,
7 MPa) was 1%. In the case of and K10-Ni50, 130 °C,
6 MPa, the diference was little bit higher. We can say that
the obtained results are reproducible.
selectivity to γ -valerolactone was the highest ( ̴80%) using
Possibility of catalyst reuse was also monitored using
catalyst K10-Ru2.5. This material was sucessfully used three
branched alcohols (isopropyl alcohol and sec-butanol). In
the row ethanol—n-propyl alcohol—n-butyl alcohol, the
1 3

Eco‑Friendly Natural Clay: Montmorillonite Modifed with Nickel or Ruthenium as an Efective…
12. Zhang B, Chen Y, Li J, Pippel E, Yang H, Gao Z, Qin Y (2015) ACS
selectivity to γ-valerolactone decreased, and selectivity
to levulinate increased. The rate based on the conversion
of levulinic acid was dependent on the value of the donor
number of the used solvent. Using aprotic solvents (tolu-
ene, 1,4-dioxane, ethyl acetate, heptane, and cyclohexane),
the selectivity was not dependent on the used solvent, and
no by-products were detected in the reaction mixture. The
reaction proceeded with a high reaction rate in heptane
and cyclohexane—the solvents possessing no acido-basic
properties. In the case of other solvents from this group, the
comparison of achieved conversion dependence on chosen
solvent parameters showed a positive trend with the solvent’s
increasing pKa.
Catal 5:5567
13. Lomate S, Sultana A, Fujitani T (2017) Catal Sci Technol 7:3073
14. Long X, Sun P, Li Z, Lang R, Xia C, Li F (2015) Cuihua Xuebao
36:1512
15. Murugesan K, Alshammari AS, Sohail M, Jagadeesh RV (2019)
ACS Sustain Chem Eng 7:14756
16. Vrbková E, Skýpala T, Vyskočilová E, Červený L (2015) Res Chem
Intermed 41:9195
17. Liu X, Lan G, Su P, Qian L, Reina TR, Wang L, Li Y, Liu J (2020)
Catal Today 351:75
18. Tao H, Ding J, Xie C, Gao Y, Song J, Sun Z (2018) Nanotechnology
97:075708
19. Vrbková E, Vyskočilová E, Rott M, Zapletal M, Červený L (2017)
Res Chem Intermed 43:2603
20. Yoshida R, Sun D, Yamada Y, Sato S, Hutchings GJ (2017) Catal
Commun 97:79
In this work, we (i) have added some interesting infor-
mation and knowledge about the solvent infuence in the
hydrogenation of levulinic acid to γ-valerolactone, and we
(ii) proved that cheap and available montmorillonite K10
might serve as efcient support for Ni and Ru as active met-
als in the reaction mentioned above. Using these types of
catalysts leads to the high yield of the desired product. Pos-
sibility of catalyst reuse was confrmed.
21. Ishikawa S, Jones DR, Iqbal S, Reece C, Morgan DJ, Willock DJ,
Miedziak PJ, Bartley JK, Edwards JK, Murayama T, Ueda W,
Hutchings GJ (2017) Green Chem 19:225
22. Fu SY, Li YZ, Chu W, Li C, Tong DG (2015) Catal Sci Technol
5:1638
23. Vrbková E, Prejza T, Lhotka M, Vyskočilová E, Červený L (2021)
Catal Lett 151:1993
24. Vrbková E, Štefová B, Vyskočilová E, Červený L (2020) React
Kinet Mech Catal 131:213
25. Vrbková E, Štefová B, Zapletal M, Vyskočilová E, Červený L
(2020) Res Chem Intermed 46:4047
Supplementary Information The online version contains supplemen-
26. Vrbková E, Štefová B, Sekerová L, Krupka J, Vyskočilová E,
Červený L (2020) React Kinet Mech Catal 129:645
27. Jiang B, Zhang C, Wang K, Dou B, Song Y, Chen H, Xu Y (2016)
Appl Therm Eng 109:99
tary material available at https://doi.org/10.1007/s10562-021-03740-3.
Acknowledgements Authors would like to thank to Miloslav Lhotka
for measuring nitrogen physisorption and Martin Veselý for STEM
analysis.
28. Zhao YH, Hao QQ, Song YH, Fan WB, Liu ZT, Liu ZW (2013)
Energy Fuel 27:6362
29. Shi J, Hui F, Yuan J, Yu Q, Mei S, Zhang Q, Li J, Wang W, Yang J,
Lu J (2019) Catalysts 9:108
Declarations
30. Karthik K, Kalai Selvan G, Kanagaraj M, Arumugam S, Victor Jaya
N (2011) J Alloys Compd 509:181
Conflict of interest All authors certify that they have no afliations
with or involvement in any organization or entity with any fnancial
interest or non-fnancial interest in the subject matter or materials dis-
cussed in this manuscript.
31. Ma X, Fan Y, Dong X, Chen R, Li H, Sun D, Yao S (2018) Minerals
8:400
32. Sekerová L, Černá H, Vyskočilová E, Vrbková E, Červený L (2021).
Catal Lett. https://doi.org/10.1007/s10562-020-03514-3
33. Rao GR, Mishra BG (2005) Mater Chem Phys 89:110
34. Derle SN, Parikh PA (2014) Biomass Convers Biorefnery 4:293
35. https://www.engineeringtoolbox.com/parafnic-benzoic-hydroxy-
dioic-acids-structure-pkacarboxylic-dissociation-constant-alcoh
ol-phenol-d_1948.html. Accessed 5 Mar 2021
References
1. Feng J, Li M, Zhong Y, Xu Y, Meng X, Zhao Z, Feng C (2020)
Microporous Mesoporous Mater 294:109858
2. Varkolu M, Velpula V, Burri DR, Kamaraju SRR (2016) New J
Chem 40:3261
36. https://owl.oit.umass.edu/departments/OrganicChemistry/appendix/
pKaTable.html. Accessed 5 Mar 2021
37. https://go.drugbank.com/drugs/DB03316. Accessed 5 Mar 2021
38. Gutmann V (1978) The donor-acceptor approach to molecular inter-
actions. Plenum Press, New York
3. Cao W, Luo W, Ge H, Su Y, Wang A, Zhang T (2017) Green Chem
19:2201
4. Wu L, Song J, Zhou B, Wu T, Jiang T, Han B (2016) Chem—Asian
J 11:2792
39. Reichardt C (2003) Solvents and solvent efects in organic chemistry.
Wiley-VCH, Weinheim
5. Tan J, Cui J, Deng T, Cui X, Ding G, Zhu Y, Li Y (2015) Chem-
CatChem 7:508
40. Dyson PJ, Jessop PG (2016) Catal Sci Technol 6:3302
41. Yan L, Yao Q, Fu Y (2017) Green Chem 19:5527
42. Cardoso AL, Neves SCG, da Silva MJ (2009) Energy Fuels 23:1718
43. Braga VS, Barros ICL, Garcia FAC, Dias SCL, Dias JA (2008) Catal
Today 133:106
6. Cao S, Monnier JR, Regalbuto JR (2017) J Catal 347:72
7. Wang S, Dorcet V, Roisnel T, Bruneau C, Fischmeister C (2017)
Organometallics 36:708
8. Wang S, Huang H, Dorcet V, Roisnel T, Bruneau C, Fischmeister C
(2017) Organometallics 36:3152
44. Kastratović V, Bigović M (2018) Chem Ind Chem Eng Q 24:283
9. Zhang D, Zhao YP, Fan X, Liu ZQ, Wang RY, Wei XY (2018) Catal
Surv Asia 22:129
Publisher’s Note Springer Nature remains neutral with regard to
jurisdictional claims in published maps and institutional afliations.
10. Gundeboina R, Velisoju VK, Gutta N, Medak S, Aytam HP (2019)
React Kinet Mech Catal 127:601
11. Zhang R, Ma Y, You F, Peng T, He Z, Li K (2017) Int J Hydrog
Energy 42:25185
1 3