A Multiphase Protocol for Selective Hydrogenation and Reductive Amination of Levulinic Acid with Integrated Catalyst Recovery

Article abstract of DOI:10.1002/cssc.201900925

At 60–150 °C and 15–35 bar H₂, two model reactions of levulinic acid (LA), hydrogenation and reductive amination with cyclohexylamine, were explored in a multiphase system composed of an aqueous solution of reactants, a hydrocarbon, and commercial 5 % Ru/C as a heterogeneous catalyst. By tuning the relative volume of the immiscible water/hydrocarbon phases and the concentration of the aqueous solution, a quantitative conversion of LA was achieved with formation of γ-valerolactone or N-(cyclohexylmethyl)pyrrolidone in >95 and 88 % selectivity, respectively. Additionally, the catalyst could be segregated in the hydrocarbon phase and recycled in an effective semi-continuous protocol. Under such conditions, formic acid additive affected the reactivity of LA through a competitive adsorption on the catalyst surface. This effect was crucial to improve selectivity for the reductive amination process. The comparison of 5 % Ru/C with a series of carbon supports demonstrated that the segregation phenomenon in the hydrocarbon phase, never previously reported, was pH-dependent and effective for samples displaying a moderate surface acidity.

Full text of DOI:10.1002/cssc.201900925

Accepted Article
Title: A Multiphase Protocol for Selective Hydrogenation and Reductive
Amination of Levulinic Acid with Integrated Recovery of Carbon-
Supported Ru-based Catalysts
Authors: Maurizio Selva, Alvise Perosa, Giulia Fiorani, Alessandro
Bellè, Fabrizio Cavani, and Tommaso Tabanelli
This manuscript has been accepted after peer review and appears as an
Accepted Article online prior to editing, proofing, and formal publication
of the final Version of Record (VoR). This work is currently citable by
using the Digital Object Identifier (DOI) given below. The VoR will be
published online in Early View as soon as possible and may be different
to this Accepted Article as a result of editing. Readers should obtain
the VoR from the journal website shown below when it is published
to ensure accuracy of information. The authors are responsible for the
content of this Accepted Article.
To be cited as: ChemSusChem 10.1002/cssc.201900925
Link to VoR: http://dx.doi.org/10.1002/cssc.201900925
A Journal of
www.chemsuschem.org

ChemSusChem
10.1002/cssc.201900925
FULL PAPER
A Multiphase Protocol for Selective Hydrogenation and Reductive
Amination of Levulinic Acid with Integrated Recovery of Carbon-Supported
Ru-based Catalysts
a
b
a
a
Alessandro Bellè, Tommaso Tabanelli, Giulia Fiorani, Alvise Perosa,
b
a*
Fabrizio Cavani, and Maurizio Selva,
Abstract: At 60-150 °C and 15-35 bar of H
2
, two model reactions of
including enantioselective ones; oxidations of benzylic alcohols to
the corresponding carbonyls; epoxidations and cycloadditions of
levulinic acid as the hydrogenation and the reductive amination with
cyclohexylamine, were explored in a multiphase system composed of
an aqueous solution of reactants, a hydrocarbon, and a commercial
CO
2
; C−C and C−X bond-forming processes, etc.[3] Applications
have emerged also in the conversion of biomass and bio-sourced
compounds such as the synthesis of 5-hydroxymethylfurfural
5
% Ru/C as a heterogenous catalyst. By tuning the relative volume of
the immiscible water-hydrocarbon phases and the concentration of aq. (HMF) from fructose,[4] the hydrogenation/dehydration of levulinic
[5]
solutions, a quantitative conversion of LA was achieved with formation
acid to -valerolactone, the enzymatic hydrolysis of cellulose to
[6]
of -valerolactone or N-cyclohexylmethyl pyrrolidone in >95% and
glucose, and the production of 2-phenyl ethanol from yeast
cells.[ Other MP systems based on mutually insoluble solvents
have been applied to the acid catalyzed dehydration of mono-, di-,
7]
88% selectivity, respectively, while the catalyst was segregated in the
hydrocarbon phase where it could be recycled in an effective semi-
continuous protocol. Under such conditions, formic acid as an additive, and poly-saccharides in water/DMSO solutions in the presence of
affected the reactivity of LA through a competitive adsorption on the
catalyst surface. This effect was crucial to improve selectivity for the
reductive amination process. The comparison of 5% Ru/C to a series
of carbon supports demonstrated that the segregation phenomenon
in the hydrocarbon phase, never previously reported, was pH-
dependent and effective for samples displaying a moderate surface
acidity.
hydrophobic organic solvents such as methyl isobutyl ketone
(MIBK), 1-butanol, 2-butanol, 1-hexane, etc. acting as extractants
of products;[8] and to the hydrolytic breakdown of dimethyl-furan
for the synthesis of 2,5-hexanedione.[9]
Pertinent to this context are also MP systems combining ILs and
2
supercritical CO on the principle of univocal solubility which
2 2
means that ILs do not dissolve in scCO , but scCO is soluble in
ILs.[10,11] Such systems were successfully used to confine reacting
substrates and catalysts in the IL-phase, while products were
recovered by compressed CO
2
(Scheme 1).[
12]
Introduction
CO2
Multiphase systems (MPs) can be achieved by combinations of
multiple immiscible phases such as aqueous solutions, ionic
liquids (ILs), and non-polar solvents based on organic and
supercritical media; or thermoregulated phases comprised of
mixtures of organic and perfluorinated compounds.[1] MP systems
can be used to steer the conversion or selectivity of organic
reactions or to aid in the separation of products and catalysts. For
example, the confinement/stabilization of metal catalysts in ILs in
the form of nanoparticles or heterogeneous solids.[2] MP systems
based on ILs have thus been described for: hydrogenations
products
IL
cat
reagents
Scheme 1. Products recovery from ILs/CO biphase systems
2
Examples were described for the extraction of HMF obtained by
the dehydration of sugars,[13] the recovery of carvacrol derived
from the hydrogenation of carvone,[14] the purification of phenolics
from the fractionation of lignocellulosic biomass,[ and the
recovery of cellulose from [EMIm][DEP]/DMF solutions.[
15]
16]
The potential of MP systems for the upgrading of the platform
chemical levulinic acid (LA) has been further explored in this
paper. The aim was to integrate the catalytic synthesis of added-
value derivatives of LA with an efficient catalyst/products
separation procedure for the recycle of the catalyst. In particular,
the hydrogenation of LA to -valerolactone, and its reductive
[
a]
Dott. Alessandro Bellè, Dr. Giulia Fiorani, Prof. Alvise Perosa, Prof.
Maurizio Selva
Department of Molecular Sciences and Nanosystems
Ca’ Foscari University of Venezia
Via Torino, 155 – 30175, Venezia-Mestre, Italy
E-mail: selva@unive.it
Dr. Tommaso Tabanelli, Prof. Fabrizio Cavani
Department of Industrial Chemistry “Toso Montanari”
University of Bologna
[b]
amination
to
1-cyclohexyl-5-methylpyrrolidin-2-one
were
investigated in mutually immiscible aqueous/hydrocarbon phases,
both with and without selected ionic liquids, and in the presence
of C-supported Ru which is among the most versatile and active
heterogeneous catalysts for processing water-soluble bio-
Viale del Risorgimento, 4 – 40136, Bologna, Italy
Supporting information for this article is given via a link at the end of
the document.
This article is protected by copyright. All rights reserved.

ChemSusChem
10.1002/cssc.201900925
FULL PAPER
sourced organic reactants.[17] Herein, a new IL-free multiphase
protocol is described whereby the desired derivatives of LA are
obtained in high yields and selectivity, accompanied by easy and
highly efficient recovery and recycle of the metal catalyst and
product isolation. Indeed, at the end of MP-reactions, the Ru/C
catalyst can be segregated in a hydrocarbon solvent (more often
isooctane) with no leaching in the aqueous phase where organic
products are dissolved.
The reliability of the method has been tested under several
different conditions by using either LA as such or in mixture with
formic acid, and at T and p up to 150 °C and 35 bar, respectively.
The study proved the robustness of the procedure and its
suitability to implementing the investigated processes in a semi-
continuous mode.
(further details on effects of p, time, and stirring speed are in the
SI section, Figures S1, S2, and S4).
As anticipated above, this reaction had been previously explored
by us in a multiphase system comprised of three immiscible
aqueous, hydrocarbon (typically, i-octane), and hydrophobic IL
(an onium salt) phases.[5] The reaction proceeded in the water
phase where both the LA and the product were soluble, yielding
GVL in a quantitative yield. Ru/C as a catalyst was perfectly
segregated in the IL medium due to the presence of the organic
phase that prompted phase-separation. This protocol however,
though highly efficient for catalyst/product separation, did not
allowed the recovery of the original catalyst embodied in the
viscous IL-layer and it suffered from IL degradation issue.
The present experiments described in table 1 highlighted that in
the absence of the IL, not only the reaction was feasible, but its
outcome was improved. At 80 °C, after 1 hour, conversion of LA
increased from 70% in the presence of the onium salt to 91% in
the water/isooctane system (entries 1 and 3). Although with a
smaller difference, a similar trend was appreciated also at 100 °C
where the corresponding conversions were 95 and >99% (entries
Besides practical advantages of this finding, the surface and
morphological characterization of Ru/C used in this study and its
comparison to other C-supports as such, offers a perspective on
the complex phenomena of confinement of the catalytic system
into the hydrocarbon medium.
2
and 4). The selectivity to GVL was quantitative only once the
Results and Discussion
process was complete; otherwise, in analogy to previous literature
observations,[ -hydroxyvaleric acid (HVA) was detected as an
intermediate (entries 1 and 3; cfr Scheme 2). In all cases, the
mass balance was validated by using an external standard (see
experimental) proving that products and unconverted reagent
were quantitatively confined in the aqueous phase.
5]
Hydrogenation/dehydration of levulinic acid: volumes and
concentration of immiscible phases. The conversion of levulinic
acid (LA) to -valerolactone (GVL) was used a model reaction to
begin the investigation (Scheme 2).
The most striking and unexpected behavior was that without any
IL, conditions of entries 3-4 allowed the selective confinement of
the catalyst in the hydrocarbon phase, meaning that the
catalyst/product separation was made possible only by adjusting
the relative amounts and concentration of water and i-octane.
Moreover, albeit acting out of the aqueous phase where the
reaction took place, the catalyst displayed an activity (TOF=234
O
O
O
H2
O
OH
OH
Ru/C
O
LA
OH
GVL
H O
2
Scheme 2. Catalytic hydrogenation/dehydration of LA to GVL
-1
h
at 80 °C) only slightly lower than that reported for the
hydrogenation of LA catalyzed by 5% Ru/C in conventional water
The used MP system was comprised by an aqueous solution of
LA, a hydrocarbon (i-octane), and a commercially available
catalyst, 5% Ru/C, sourced from Aldrich. Comparative
experiments were also run with the addition of an extra phase of
methyl trioctyl ammonium or phosphonium bistriflimide salts
solution (TOF=273 h , 70 °C, 30 bar of H
2
).[17f]
-
1
Table 1 shows photographs of the multiphase system at the end
of the process in the presence and in the absence of the IL,
respectively (last column, top and bottom). By contrast, before the
reaction, Ru/C appeared manifestly partitioned in the immiscible
phases.
(
[Q8881][Ntf2], Q= N, P; 360 mg each) as ionic liquids apt to
segregate the catalyst, based on our previous work.[
18]
A
The temperature and pressure did not appreciably affect the
catalyst separation in the IL-free system. It was noticed that both
the conversion of LA and the selectivity to GVL smoothly
increased up to quantitative values, in the range of 60 to 100 °C
and 10 to 35 bar, respectively (Figures S1 and S2). Regardless of
T and p however, Ru/C was neatly suspended in i-octane only.
The reaction could also be easily scaled-up. At 100 °C, a test
carried out by triplicating the volumes and amounts of solutions,
reactants, and the catalyst with respect to entry 4 of Table 1,
confirmed a quantitative reaction after 1 hour and a perfect
catalyst/product segregation in the two immiscible phases. GVL
was distilled from the aqueous solution, and isolated in an 88%
yield, thus further confirming the mass balance.
screening of the reaction was conducted by changing T and p,
and the relative amounts and concentration of the aqueous
solution, the hydrocarbon phase, and the catalyst. Operative
intervals for such variables were chosen based on exploratory
experiments and previous results reported by us.[5] Tests were
carried out in the range of 60-130 °C and 10-50 bar of H ,
2
respectively, with volumes of aqueous and i-octane phase
variable between 5 and 10 mL, concentrations of aqueous LA of
0.2-0.5 M, and 20-50 mg of Ru/C.
Conversion of LA and products selectivity were determined by GC,
while the structure of products was assigned by GC/MS, and by
comparison to an authentic sample of GVL. All tests were run in
duplicate to ensure reproducibility.
Table
1 summarizes the most representative results by
comparing the effect of the IL and the reaction temperature
This article is protected by copyright. All rights reserved.

ChemSusChem
10.1002/cssc.201900925
FULL PAPER
Table 1. Conversion of LA to GVL under MP-conditions with and w/o [N8881][Ntf
2
] as an IL.[a]
Entry
[P8881][Ntf
2
]
T
(°C)
t/p
(h/bar, [c])
LA Conv.
(%)
GVL Sel.
(%)
Observed
Separation
[b]
[d]
[d]
[e]
(
IL:Ru, wt/wt)
1
2
15
80
1/35
1/35
70
95
80[f]
15
100
95
3
4
none
none
80
1/35
1/35
91
82[f]
>99
100
>99
[
0
a] Reactions were carried out in a steel autoclave fitted with a glass liner charged with an aq. solution of LA (10 mL,
.36 M), i-octane (5 mL), 5% Ru/C (28 mg), and [P8881][Ntf ] (360 mg when indicated; entries 1-2). [b] Ionic liquid:Ru
weight ratio. [c] Pressure (bar) of H . [d] Conversion of LA and selectivity of GVL were determined by GC using
2
2
internal standards (see experimental section). [e] Visual appearance of the multiphase system at the end of the
experiments. [f] GVL and hydroxyvaleric acid (HVA) were the sole reaction products.
LA Conversion
Selectivity to GVL
To the best of our knowledge on MP reaction systems, these
80
findings had no precedents and interestingly, they could be
exploited not only to simplify the overall procedure but also to
improve the performance, recovery, and recycle of Ru/C. It should
be here noted that the separation of C-supported catalysts from
organic/aqueous solutions is a well-documented issue in industry:
even conventional techniques of filtration and centrifugation
become expensive, if not impracticable, for fine powdered
carbons with low particles size (often the most active supports),[
and alternative approaches based on the engineering of
magnetically separable carbon materials have been proposed. [
Catalyst recycling and leaching tests. The cost of catalysts in a
liquid-phase reaction may represent up to one third of the total
cost of the process, implying that its loss, e.g. by leaching, can be
critical and its recovery and reuse is imperative. [21,22] Accordingly,
the above-described IL-free protocol was explored to investigate
on the recyclability and stability of Ru/C. Conditions of entry 4 in
Table 1 (aq. LA, 0.36 M; i-octane, 5 mL; 5% Ru/C, 28 mg; 35 bar
70
60
50
40
30
19]
20]
20
10
0
1
2
3
4
run (n)
Figure 1. Recycle of Ru/C on 4 subsequent runs in the conversion of LA into
GVL carried out at 65 °C for 2 hours. Other conditions were those of Figure 1
(aq. LA, 0.36 M; i-octane, 5mL; 5% Ru/C, 28 mg; 35 bar of H ).
2
2
of H ) were used for the recycling tests, though, temperature and
time were set at 65 °C and 2 hours to moderate the final
conversion and to ensure a more accurate control of the reaction
outcome. Once the first run was complete, the lower aqueous
phase was removed from the vessel, the isooctane layer
containing the catalyst was washed with Milli-Q water (3 x 5 mL),
and a fresh aqueous solution of LA (0.36 M; 10 mL) was added.
The recycling procedure was repeated four times, and the whole
set of reactions was run twice to ensure reproducibility. The
results are illustrated in Figure 1.
Notably, -hydroxyvaleric acid (HVA) was observed in all cases,
thereby indicating that at 65 °C, the temperature was insufficient
to complete the dehydration step of the intermediate HVA to GVL
(Scheme 2; see also later, Figure 2). This was corroborated by
further experiments in which the effect of the reaction time was
explored for the hydrogenation/dehydration of LA carried out
according to the multiphase setup of Figure 1 (details are in the
SI section, Figures S3-S4 and Table S1). In the time interval from
1 to 4 hours, conversion and selectivity smoothly increased up to
95% and 75%, respectively; moreover, if the mixture was allowed
Tests demonstrated that conversion of LA and selectivity to GVL
remained constant in the range 62-70% and 65-75%, respectively, to stand, HVA slowly continued to dehydrate to GVL even at room
through the 4 runs (green and red bars).
temperature. Similar trends were noticed also in the reduction of
LA catalyzed by Ru/C, in methanol solutions. [23]
This article is protected by copyright. All rights reserved.

ChemSusChem
10.1002/cssc.201900925
FULL PAPER
Recycling tests proved that the stirring speed was a key
parameter for an effective reuse of the catalyst. Figure 1 refers to
reactions carried out in MP mixtures magnetically stirred at 1300
rpm. However, if the same experiments were repeated at a lower
speed, the reaction outcome was remarkably different: for
example, by halving the agitation rate at 650 rpm, both conversion
and selectivity progressively decreased to reach values of 23 and
1
(
8%, respectively, after four subsequent reuses of the catalyst
details are in the SI section, Figure S5). It could be anticipated
that mass transfer limitations of solid-catalyzed gas-liquid
[24]
processes like hydrogenations, were even more pronounced in
the investigated MP-arrangement than those observed in
conventional single-liquid phase systems,[22,25] because the
catalyst was in a hydrocarbon phase and an aq./org interfacial
barrier had to be overcome for the reaction to occur. An inspection
of this aspect was carried out by closely studying the effect of the
stirring speed on the reaction: under the conditions of Figure 1,
once the catalyst was suspended in i-octane, LA conversion
increased from 32 up to 63% in the range from 300 to 1000 rpm.
Thereafter, it (conversion) did not appreciably improve by a more
effective mixing at 1300 rpm (Figure S6). Although a detailed
kinetic study was beyond the scope of the paper, results indicated
that external mass transfer limitations were no longer significant
for the MP-process at stirring rates ≥1000 rpm.
Scheme 3. Semi-continuous catalytic hydrogenation of LA under IL-free
multiphase conditions
Multiphase hydrogenation of mixtures of levulinic and formic acid.
Bio-based levulinic acid is currently synthesized from saccharides
through an acid-catalyzed sequence yielding 5-(hydroxymethyl)
furfural (HMF), which in turn, generates an equimolar mixture of
LA and formic acid (FA).[
27]
2 2
Since FA can be catalytically decomposed to CO and H over
Ru/C, [28] it (FA) may act as an in-situ transfer hydrogenation agent.
This approach has been successfully applied for the direct
conversion of the aqueous stream of LA and FA into GVL, without
The complete segregation of Ru/C in the i-octane phase (entries
3
-4 of Table 1) was confirmed by both indirect experiments based
external H
2
supply. [29] However, limitations exist due to the high
[26]
on the Sheldon test, and direct ICP analyses of the aqueous
phase. In the Sheldon test, once the first and the second runs of
Figure 1 were complete, the corresponding aqueous solutions
were recovered and set to react separately, as such for 2 hours
temperature (≥ 190 °C) required for the dissociation of formic acid,
and the need of excess FA (up to 4 molar equivalents) for the
reaction to proceed.[ This behavior was confirmed by the results
obtained in the IL-free multiphase system. Under the conditions
of Figure 1, at 190 °C, the reaction of an aqueous equimolar
solution of LA and FA (0.36 M in each acid, 10 mL) provided a
negligible formation of GVL (1%) even after 12 hours. The extent
of FA decomposition did not exceed 14% (for further details, see
Figure S7).
30]
2
at 65 °C and 35 bar of H , under magnetic stirring. Both tests were
negative: GC/MS analysis showed that no further hydrogenation
of LA took place (conversion, 62-65%, did not change with respect
to Figure 1), thereby indicating that no active catalyst was present
in the aq. solutions. If any homogeneous water-soluble Ru-
species formed, these did not contribute to the reaction. The
absence of metal leaching in the aqueous phase was finally
demonstrated by ICP/MS analyses proving that the dissolved Ru
was <0.01 wt% with respect to the metal amount in the catalyst
used for the reactivity tests (details of ICP/MS measures are in
the SI section, Table S2).
Multiphase reactions, however, were successful when performed
under a H pressure. The most representative result is illustrated
2
in Figure 2 reporting a substantially quantitative conversion of LA
and selectivity > 95% to GVL, for an experiment carried out for 12
2
h at 130 °C and 35 bar of H [other conditions: LA and FA in a 1:1
molar ratio (0.36 M in each acid, 10 mL), i-octane (5 mL), and 5%
Ru/C (28 mg)]. Compared to Figure S3 showing that aq. LA alone
was fully converted at only 65 °C and 4 hours, Figure 2 proved a
remarkable detrimental role of FA on the kinetics. This was
ascribed to the competitive adsorption of FA over the Ru/C
surface which, as described in the literature, disfavors the
hydrogenation of LA. [29,30] However, the excellent selectivity to
GVL (red profile) highlighted a promoting effect of the high
temperature on the dehydration of the intermediate HVA (see
Figure 1). It should be noted that the catalyst always appeared
confined in the i-octane phase where it was perfectly separated
from the product. By contrast, our previous studies indicated the
hydrogenation of LA in the presence of FA was not practicable
when IL-mediated multiphase systems were used: not only
Overall, Ru/C was stable and reusable without loss of
performance in the hydrocarbon phase. The MP-system was
perfectly suited to the design of a semi-continuous hydrogenation
of LA to GVL, as illustrated in Scheme 3. The experiments
involved simple removal the aqueous GVL-containing phase once
the first reaction was complete, and replacement with fresh
aqueous solution of LA. Then, the mixture could be once more
subjected to the reaction conditions for the first recycle, and so on.
If necessary, i-octane could be removed by rotary evaporation,
thus yielding a quantitative recovery of the original catalyst as a
dry powder.
The MP-protocol was then explored to extend its applicability for
the synthesis of GVL from the reaction of LA in the presence of
formic acid.
2
harsher conditions were required (150 °C, 35 bar of H , 16-32 h),
but only moderate conversions, not exceeding 21%, were
observed. [
5]
This article is protected by copyright. All rights reserved.

ChemSusChem
10.1002/cssc.201900925
FULL PAPER
O
the reaction.[39] Unfortunately, initial tests showed that the catalyst
remained partitioned between the two immiscible phases of the
system, and most importantly, the major reaction was the
competitive hydrogenation of LA to GVL and not the desired
formation of 1-cyclohexyl-5-methylpyrrolidin-2-one (CyMP). For
O
H
OH
O
aq. FA (0.36 M)
OH
O
1
5
30 °C, H =35 bar
% Ru/C, i-octane
2
O
aq. LA (0.36 M)
up to 99%
100
2
example, at 130 °C and 15 bar of H , when an equimolar aq.
90
80
70
60
50
40
30
20
10
0
solution of LA and CyNH (3.6 mmol each; total volume = 5 mL)
2
was set to react in the presence of 5% Ru/C (28 mg), and i-octane
(5 mL), levulinic acid was quantitatively converted after 16 h,
yielding a mixture of GVL (83%), CyMP (9%), and other
derivatives (8% in total) whose structures were assigned by
GC/MS and comparison of literature data (Scheme 5, further
details are in the SI section, Figure S9). [31,33,35,38]
LA Conversion
Selectivity to GVL
O
O
O
O
O
R
R
R
R
O
N
N
N
N
H
O
GVL
CyMP
m/z=197
m/z=179
0
100
200
300
400
500
600
700
800
time (min)
R=Cy
Scheme 5. Products observed during MP-reaction of LA and CyNH
Figure 2. Conversion of LA and selectivity to GVL for the multiphase reaction
performed using an equimolar aq. solution of LA and FA (0.36 M in each acid,
2
10 mL). Other conditions: 130 °C, 35 bar of H
2
, i-octane (5 mL), and 5% Ru/C
(28 mg).
To disfavor the formation of GVL during the MP-process, we
envisaged adding formic acid (FA), known to selectively inhibit the
adsorption of LA over Ru/C (see previous paragraph, [29,30]).
Further screening studies were therefore carried out with an
equimolar mixture of LA and FA (3.6 mmol each), by adjusting the
Recycle tests proved that Ru/C could be reused without loss of
activity, acting as a suspension in the i-octane phase. Three
subsequent reactions repeated on the same catalyst showed that
under the conditions of Figure 2, after 4 hours, conversion of LA
and selectivity to GVL were steady in the range of 70-75% and
quantity of CyNH
varying T and p (130 and 150 °C, and 15 and 35 bar H
respectively). Other conditions were not modified [16 h, 5% Ru/C
28 mg), and i-octane (5 mL)], implying that the total reaction
2
(3.6-7.2 mmol) and water (0.5-5 mL), and by
>95%, respectively (details of recycle tests are in the SI section,
2
Figure S8).
Although the transfer hydrogenation from FA to LA was not
(
possible, the proposed MP-method worked satisfactorily under H
2
volume changed from 5.5 to 10 mL according to the variations of
the aqueous phase. All tests were run in duplicate to ensure
reproducibility. The most representative results are reported in
Table 2 which contains also the experiments carried out without
FA, for comparison.
At 130 °C and in the absence of FA, the role of both water and
2
CyNH was highlighted. Selectivity towards the formation of
CyMP improved progressively from 9 to 27% by decreasing the
volume of water from 5 to 0. 5 mL (entries 1-3); below this limit
pressure, and it could in principle, be used for the synthesis of
GVL starting from mixtures of LA/FA as received from the
decomposition of sugars.
Multiphase reductive amination of levulinic acid. The reductive
amination of LA provides access to N-alkyl pyrrolidones of
renewable origin, relevant as solvents, surfactants, complexing
agents, and intermediates for inks and fiber dyes (Scheme 4).[31]
(
2
0.5 mL), the partitioning of reactants (particularly CyNH ) and
products in i-octane inhibited segregation of the catalyst and
favored the formation of a mixture of unidentified by-products.[
The volume of water was thereafter maintained between 0.5 and
40]
Scheme 4. N-alkylpyrrolidones from levulinic acid
2
mL. Doubling of the reactant amine was also beneficial,
imparting a 10% increase of the selectivity of the reductive
amination (entries 2 and 4). Yet, the main reaction path remained
the hydrogenation/dehydration of LA.
At 130 °C, the best results in terms of selectivity were achieved
by the addition of FA, that reversed the products distribution. The
combined effects of FA, excess amine, and reduced water volume
promoted the formation of CyMP in up to 80% and concurrently
limited GVL to almost negligible amounts (3-5%) (entries 5-7).
Homogeneous and heterogenous catalysts (i.e. Au/ZrO
MoO /TiO , C-coated Ni nanoparticles, Ir/SiO -SO H, and Ru-, Ir-,
and In-complexes) prove active for the reaction, using either
2
, Pt-
X
2
2
3
2
formic acid as a transfer hydrogenation or H or even hydrosilanes
as reductants.[
31,32,33,34,35,36,37,38]
The catalyst/product separation
and catalyst reuse, however, are serious drawbacks, especially
for large-scale applications.[31] To cope with these problems, the
present multiphase approach was explored for the model reaction
of LA with cyclohexylamine. An important issue was the need to
minimize the presence of water because of its adverse effects on
This article is protected by copyright. All rights reserved.

ChemSusChem
10.1002/cssc.201900925
FULL PAPER
2
Table 2. Multiphase reductive amination of LA with CyNH both in the presence and in the absence of FA.
FA
(mol:mol)
T
(°C)
p
(bar)
Conv.
(%)
Selecitivity (%)
LA:CyNH
2
[
2
H O
(mL)
[
b]
[c]
Entry
a]
(mol:mol)
CyMP
GVL
Others [d]
1
2
3
4
5
6
7
8
9
1
1
1
2
1
2
2
2
2
2
5
none
130
130
130
130
130
130
130
150
150
150
15
15
15
15
15
15
15
15
35
35
>99
>99
99
9
83
65
68
62
20
5
8
2
none
20
27
32
65
80
80
79
82
88
15
5
0.5
2
none
none
95
6
2
Yes (1)
Yes (1)
Yes (1)
Yes (1)
Yes (1)
Yes (1)
94
15
15
17
14
12
7
2
>99
>99
>99
>99
96
0.5
2
3
7
2
6
10
0.5
5
All reactions were carried out for 16 h, in the presence of 5% Ru/C (28 mg), and i-octane (5 mL). The total reaction
volume changed from 5.5 to 10 mL according to the variations of the aqueous phase. [a] LA:CyNH molar ratio. [b]
2
LA:FA molar ratio. [c] Conversion of LA (determined by GC). [d] Compounds of Scheme 5 other than GVL and CyMP.
Finally, by performing reactions at 150 °C and 35 bar, the quantity
literature substantially exclude any role of noble metals as
catalysts in the formation of imines and amides. For example,
without any catalyst and solvent, the reaction of n-octylamine and
LA provided the corresponding imine [4-(octylimino)pentanoic
acid] in a 14% yield after 10 min at room temperature.[31] Moreover,
at 120 °C (solventless conditions), a quantitative yield of the imine
derived from ethyl levulinate and aniline was obtained in the
2
presence of Pt/TiO catalyst, but it was demonstrated that the
of products termed as “others”, was reduced allowing to obtain
the desired pyrrolidone with 88% selectivity at almost complete
conversion (entry 10). No further optimization was explored.
Therefore, we observed that by tuning of the reaction conditions
and parameters, the selectivity of the multiphase reductive
amination could be improved by a factor of 10. Such a dramatic
change was ascribed to concurrent effects inferred by examining
the accepted mechanisms for catalytic reductive amination,
reaction did not require noble metal catalysis; the process was
[
31,33,34,38]
and comparing the structures of co-products designated
4+
strongly promoted by the Lewis acidity associated to Ti sites on
as “others” in Table 2. First, FA acted as an inhibitor for the
adsorption of LA on Ru/C, thereby favoring the reaction of LA with
CyNH rather than its hydrogenation to GVL and. Reductive
2
support. [41,42] On the other hand, also amides of LA could be
TiO
2
[33]
obtained in the absence of catalysts and solvents.
Specifically,
N-cyclohexyl-4-oxopentanamide, compound B of scheme 6, was
observed in the uncatalyzed reaction of LA and CyNH at 120 °C
amination plausibly followed pathways illustrated in Scheme 6.
2
for 12 h. [38] To speculate on these aspects, additional experiments
were performed also in aqueous solution. According to the
conditions of entry 10 in Table 2, a mixture of LA (3.6 mmol) and
2
CyNH in a 1:2 molar ratio in water (0.5 mL) was set to react at
different temperatures, from ambient up to 100 °C. Neither the
imine A nor the amide B (or other products) were observed,
thereby confirming the adverse effect of water on the formation of
such compounds, and consequently on the overall process of
reductive amination. [31,32,43] However, the presence of amide B
among by-products of the reaction carried out at 150 °C (see
scheme 5) not only provided evidence for its intermediacy in the
synthesis of the pyrrolidone product, but it indirectly supported the
same role for imine A. The latter compound, being highly reactive,
could not be observed. Indeed, both the catalytic hydrogenation
of the imine C=N bond to produce the amine C and the
subsequent dehydration/cyclization of C to CyMP were reported
as easy and fast transformations over noble metals (Pt, Ru, Pd)
Scheme 6. Mechanism for multiphase reductive amination of LA with
cyclohexylamine.
The initial condensation of LA and cyclohexylamine may produce
either the imine A and the amide B. Investigations already
reported in the
(
Scheme 6, top and right), [33] On the other hand, GC/MS analyses
This article is protected by copyright. All rights reserved.

ChemSusChem
10.1002/cssc.201900925
FULL PAPER
showed evidence for two isomeric unsaturated cyclic precursors
of CyMP (m/z=179; Scheme 5, E and F) as minor products.
Alternative pathways can be then formulated considering either
the equilibrium of imine A with enamine D followed by dehydration
the hydrocarbon phase, in all cases. iv) All the other C-samples
(1, 3, 5 and 6) appeared dispersed in water and i-octane,
regardless of time and pH of aq. phase.
Experiments confirmed the separation of samples
0 (the
(
(
Scheme 6, mid), or the cyclization/dehydration of amide B
scheme 6, left) to get both compounds E and F. CyMP is finally
reference 5% Ru/C), 2, and 4 of Table 3 (in the case of sample 4,
details are shown by photographs in Figure S10). The behaviour
of samples 0, 2 and 4 was general: similar results to those of
Table 3 were noticed when i-octane was replaced with other
hydrocarbons (hexane, cyclohexane; 5 mL) or water-insoluble
solvents such as diethyl carbonate (5 mL). Overall, the observed
segregation behaviour was ascribed to the surface composition of
the carbons, usually comprised by different oxides (carboxylic-,
phenolic-, lactonic-, and ether- groups) whose presence affects
the hydrophobicity of such materials.[44] Nonetheless, solids
morphology and even the presence of inorganic impurities could
not be ruled out. [21b, 45]
produced by a catalytic C=C bond hydrogenation. It should be
noted that similar cyclic intermediates have been proposed in two
recent studies of reductive amination of LA, [33,38] but they have
been synthesized only as derivatives of benzylamine. [33] Our
attempts to isolate the corresponding products from CyNH
been unsuccessful.
2
have
The investigation proved the concept that the multiphase protocol
could be extended to the reductive amination of LA. The water
sensitivity of the reaction implied the use of minimal volumes of
the aqueous phase whose presence, however, was necessary for
the catalyst/product segregation. [40] The latter (separation) was
achieved at the end of the process, by adjusting both the water/i-
octane proportions in a 2:1 volumetric ratio, and pH at 2.5 with
few drops of aq. HCl. It should be noted here that: i) no apparent
correlation could be inferred between the selectivity and the
catalyst separation. The presence of FA was crucial to improve
the formation of CyNMP, but it was not compulsory to induce the
segregation of Ru/C in the organic phase (see also next
paragraph); ii) even after the reaction at 150 °C and 35 bar, the
recovery of the suspension of Ru/C in i-octane (5 mL) was
quantitative. Moreover, the hydrocarbon proved stable with
unaltered composition and purity; iii) once i-octane was rotary
evaporated, the catalyst was tested for a single recycle under the
conditions of entry 10 (Table 2). No appreciable changes of the
catalytic performance were noticed with respect to the fresh
catalyst.
2
An array of characterization techniques such as N adsorption,
temperature-programmed desorption (TPD), Raman and infrared
spectroscopy methods (DRIFTS), SEM coupled with energy
dispersive X-Ray analysis (EDX), and chemical titration methods,
[45,46,47,48,49]
was therefore used to further compare the reference
catalyst to carbons 3 (ENGELHARD, 44853) and 4 (NORIT SX
1G) which were selected as model for one inactive (3) and one
active (4) sample towards separation in biphase systems,
respectively. They were finely ground in a mortar prior to
characterization.
Physical and chemical characterization of Ru/C and samples 3
and 4. It should be first noted that most of the analysis of the
2
textural properties carried out by N adsorption isotherms and
SEM, and the composition/properties of the surfaces performed
by spectroscopic and TPD techniques, did not highlight specific
features or differences between the Ru/C catalyst, carbon 4, and
the inactive sample 3, ascribable to the ability of the material to
segregate in the hydrocarbon phase.
The selective segregation of the Ru/C on hydrocarbon phases:
choice of model C-supports. In the investigated MP-conditions, a
hypothesis for the unusual selective partitioning of the catalyst in
the hydrocarbon phase was based on the role of surface and bulk
properties of the C-support. Initial experiments were then aimed
to identify other C-supports with an affinity for liquid hydrocarbons
similar to that of the investigated Ru/C. Six commercial carbons
Some representative results are summarized in Table S3, Figures
S11-S13, and Figure 3. (Details on the preparation of samples,
instruments and executions of experiments are in the
experimental section).
Carbons 3 and 4 showed similar surface areas and pore volumes
(Table S3), despite they had a remarkably different behavior in
the biphase system.
Morphological analogies were also consistent with SEM analysis.
The comparison of SEM images extended to all samples of Table
3 (the reference catalysts and carbons 1-6) did not allow
appreciable differentiations among the specimens. (Full details of
SEM analyses are reported in SI section, Figure S11).
DRIFT spectra recorded for carbons 3 and 4 by using pyridine as
a probe molecule, showed almost superimposable profiles,
meaning that Lewis and Brønsted acid sites available at the
surface of the carbon materials and identified by adsorption bands
(
1-6) were selected for the purpose. If not already provided as
powders, specimens were finely grinded before tests. Each
carbon sample (30 mg) was introduced in a biphase system
comprised of water (milli Q, 10 mL) or an acidic aqueous solution
(
10 mL), and i-octane (5 mL). Acid solutions were obtained by
adding levulinic acid (0.36 M) or HCl (3x10- M) so that the
resulting pH (2-2.5) reproduced conditions of starting mixtures in
the studied multiphase reactions. Then, at rt, the system was kept
under vigorous magnetic stirring.
3
Results are reported in Table 3.
Four facts were evident from visual inspection. i) Segregation
occurred only for a limited number of samples (the reference 5%
Ru/C, and 2 and 4: entries 1, 2, and 5). ii) When present, complete
partitioning of the sample in i-octane took place after 60-90 min
only with an acid aqueous solution, independently from the acid
-1
-1
-1
-1 [50,51]
at 1445 cm , 1465 cm , 1574 cm and 1530-1550 cm ,
were hardly distinguishable either by strength or relative amounts
(Figure S12). Similar results were obtained from Raman
spectroscopy (see SI section, Figure S13).
(
LA or HCl) and at constant pH (2.0-2.5). In pure milli-Q water,
Starting from the accepted patterns of thermal decomposition of
carboxylic acids and anhydrides, lactones, phenols, and
carbonyl/quinones, [45,47,52]
samples were uniformly suspended in the biphase system. iii)
Both 5% Ru/C and C-supports (2 and 4) were neatly confined in
This article is protected by copyright. All rights reserved.

ChemSusChem
10.1002/cssc.201900925
FULL PAPER
Figure 3 describes the results of TPD analyses by grouping the
relative % amount of the different classes of oxygen-bearing
surface groups determined for 5% Ru/C and carbons 3 and 4
Table 3. Comparison of 5% Ru/C and commercial carbon samples in a
biphase system
Entry
Sample
#
Source
MP-conditions[a]
Observed
separation
(
details including calibration curves, desorption profiles and the
corresponding deconvolutions, and the evaluation of
deconvoluted quantities, are fully reported in the SI section,
Figures S14-S18). The (moderate) fluctuations observed in the
distribution of the surface groups did not point to any trend by
which specimens could be distinguished for the behavior shown
under multiphase conditions of Table 3.
[
b]
1
2
3
4
5
6
7
0,
H
2
O/i-octane
Acid sol./ i-octane
O/i-octane
Acid sol./ i-octane
O/i-octane
Acid sol./ i-octane
O/i-octane
Acid sol./ i-octane
O/i-octane
Acid sol./ i-octane
O/i-octane
Acid sol./ i-octane
O/i-octane
Acid sol./ i-octane
None
Yes[c]
None
None
None
Yes[c]
None
None
None
Yes[c]
None
None
None
None
Aldrich, Lot
MKBW5890V
#
5%
Ru/C
1
H
2
BASF 288954,
Lot # 13601
3
Ru/C
6
1%
5
2%
ENGELHARD S-
2
3
4
5
6
H
2
45502; Lot
#
12775, Moist.
Cont. 43.3%
2
8%
ENGELHARD,
44853; Lot #
3
%
23%
H
2
1
3%
1
2%
9%
4
% 3%
1%
12823,
Moist.
4
Cont. 56.9%
4
Carboxylic acids
Anhydrides
Lactones
H
2
NORIT SX 1G
Lot # A-10536
Quinones
Phenols
2
9%
H
2
4%
NORIT
Extra
RX1
Lot
Lot
#
1%
17%
660658
Figure 3. Distribution of the different classes of oxygen-bearing surface groups
determined by TPD tests, in the reference catalyst, and samples 3 and 4. TPD
tests were performed on 218 mg of each sample which were first degassed
under a flow of helium, and then, heated at a rate of 5 °C min-1 up to 910 °C
H
2
NORIT
Extra
RX3
#
(details are in both the experimental and the SI sections).
580092
[
[
a] pH of both milli-Q water and aq. acid solutions did not vary during tests.
b] Segregation of the catalyst or the carbon in i-octane. [c] Segregation was
On the other hand, some appreciable differences emerged by
EDX analyses and further surface chemical characterizations
carried out Boehm titration and point of zero charge (pHpzc).
EDX mapping other elements than C, showed that both the
reference catalyst and carbon 4 active for segregation, contained
quantities of Na in the range of 0.1-0.2% (entries 1 and 3, Table
observed after 60-90 min at rt.
respectively (entries 1 and 3). Likewise, the lowest pHpzc (2.2) was
measured for 3, thereby confirming its higher surface acidity
among the investigated samples. The two methods not only
provided consistent results, but they suggested a correlation for
the behaviour in the biphase systems: the higher the surface
acidity (for 3), the higher the hydrophilicity due to H-bonding, the
poorer the separation in the hydrocarbon phase.
4
2
), while the same metal was virtually absent in carbon 3 (entry
). This held true also from the comparison of carbon 2 (active for
separation in a biphase system; Na=0.2%) with carbons 1, 5 and
(inactive) in which the Na amount was even below the detection
6
limit (see Table S4 in the SI section).
Overall, the physico-chemical characterization studies lead to
hypothesise multiple reasons for the segregation described in
Table 3, among which a role is played by properties that are
apparently common to the reference catalyst and samples 2 and
4 active for separation in i-octane: the presence of non-negligible
quantities (0.1-0.2%) of Na-based impurities and a relatively low
surface acidity. The phenomenon, however, is far from being
understood since the nature of these properties and their relative
contributions/effects to the separation in the multiphase systems
still remain to be clarified. Moreover, the incongruity between the
results offered by Boehm titration and pHpzc analyses from one
side, and from the other, those obtained by the
TPD/spectroscopic measures might be due to the different
environments under which the techniques operate. Titration-
based methods are carried out in aqueous solutions, while TPD
and DRIFT/Raman spectra are acquired on a partially dehydrated
sample where solvation effects on surface functional groups are
mitigated, if occurring at all.
Boehm titration was carried out according to a standardized
protocol by which the investigated sample (0.50 g) was titrated
using three aq. solutions diversified by their basic strengths,
3 2 3
containing NaHCO , Na CO
and NaOH, respectively.[49] The
point of zero charge (pHpzc), i.e. pH at which the charge of the
carbon surface is zero, was determined by equilibrating selected
amounts of both the reference catalyst and carbons 3 and 4 in
standard acid aqueous solutions through procedures described
[49,53]
elsewhere.
(Details of the used methods are in the
experimental section).
Results are reported in Table 5.
The Boehm titration showed that the total acidity due to oxygen
surface groups did not vary significantly between the samples,
oscillating in the range of 120-140 µeq/g. Though, strong acid
functions as carboxylic ones accounted for more than 90% of the
acidity of sample 3 (entry 2); while, the contribution of the same
(
carboxylic) groups was 13% and 36% for Ru/C and carbon 4,
This article is protected by copyright. All rights reserved.

ChemSusChem
10.1002/cssc.201900925
FULL PAPER
Table 4. EDX-analyses of the reference catalyst and samples 3 and 4. The values reported are calculated as average of
five different elemental analyses (EDX wide range mapping of 350x300µm).
Entry
1
Sample
#
Elements (wt, %)
C
O
Na
Mg
Ca
S
Cl
Al
K
Ru
0, 5%
86.15
8.49
0.17
0.04
0.13
0.14
4.88
Ru/C
2
3
3
4
93.17
93.43
8.14
5.90
0.02
0.08
0.02
0.04
0.06
0.21
0.16
0.08
0.39
0.08
0.02
hydrocarbon medium, while products are recovered from water
solution, free of any cross-contamination. The method is reliable
under a variety of conditions for the reactions of levulinic acid as
such or in mixture with formic acid (FA). FA acts as a competitor
of levulinic acid for the catalyst active sites, thereby always
disfavoring the kinetics of the investigated processes; its
presence however, is crucial to improve the selectivity of MP-
reductive amination. Moreover, the procedure has been
successfully scaled up for the preparation of -valerolactone,
demonstrating its applicability to overcome issues posed by
filtration/centrifugation for the separation of finely powdered C-
supported catalysts.
Table 5. Boehm titration and evaluation of pHPZC for the reference catalyst
and carbons 3 and 4
Entry
Sample
#
pHpzc
Acidity (µeq/g)
Total
140
Carboxylic
Lactones
46
Phenol
41
1
0,
5%
3.6
50
Ru/C
2
3
3
4
2.2
2.6
140
120
130
15
10
85
20
The investigation of factors affecting switching and confinement
of Ru/C in the hydrocarbon medium, has provided evidence of a
complex phenomenon which shows a dependence from the pH of
the aqueous phase and most plausibly, it occurs when the C-
support exhibits a moderate surface acidity and contains minor
amounts of alkali-metal impurities. However, reasons and the
nature of the effects responsible of this intriguing behaviour are
far from being understood and will be the object of future
investigations.
Finally, although the present study has been focused on levulinic
acid, the versatility of the multiphase protocol paves the way for
its extension to the upgrading of many hydrosoluble derivatives of
renewable origin.
Conclusions
The present results highlight the potential of a multiphase system
comprised of two immiscible aqueous and hydrocarbon phases,
and a heterogenous Ru-based catalyst supported on C, for the
synthesis of valuable derivatives of levulinic acid as -
valerolactone and N-cyclohexyl methyl pyrrolidone, through
reactions of hydrogenation and reductive amination, respectively.
Compared to other MP-systems based on ionic liquids, the
procedure not only allows an enhanced reactivity, but if offers an
original solution to separate reaction products from the catalyst in
a semi-continuous mode. By tuning the relative volumes of water
and hydrocarbon and the concentration of reactants in the
aqueous solution, Ru/C can be completely segregated, without
aqueous leaching, and recycled as
a suspension in the
Experimental Section
General
film = 0.25 mm) and an Elite-624 capillary column (L = 30 m, Ø = 0.32 mm,
Levulinic acid, formic acid, 5% Ru/C, i-octane, cyclohexane, diethyl
carbonate, cyclohexylamine, diethylenglycol dimethylether [diglyme,
MeO(CH CH O) Me] were commercially available compounds sourced by
2 2 2
Aldrich. If not otherwise specified, they were employed without further
purification. Ionic liquids as methyl trioctyl ammonium and phosphonium
film =1.8 mm), respectively.
1H NMR spectra were recorded at 400 or 300 MHz, and 13C NMR spectra
were recorded at 100 MHz; chemical shifts are reported downfield from
tetramethylsilane (TMS), and CDCl was used as the solvent.
3
bistriflimide salts were prepared according to
a
method described
Reaction Procedures
[
18]
2 2
elsewhere. Water was milli-Q grade. H and N gases were purchased
from SIAD, Italy.
Hydrogenation of Levulinic Acid. In a typical hydrogenation experiment, a
5-mL tubular reactor of borosilicate glass (Pyrex) was charged with a 0.36
M aqueous solution of levulinic acid (LA, 3.6 mmol; total volume=10 mL),
% Ru/C (30 mg, 0,015 mmol of Ru) as catalyst and isooctane (5 mL). The
2
GC–MS (EI, 70 eV) and GC (flame ionisation detector; CG/FID) analyses
were performed with an HP5-MS capillary column (L = 30 m, Ø = 0.32 mm,
5
This article is protected by copyright. All rights reserved.

ChemSusChem
10.1002/cssc.201900925
FULL PAPER
vessel was placed in a jacketed steel autoclave equipped with
a
Characterization of 5% Ru/C and carbon samples 3 and 4
manometer and two needle valves by which, at rt, H
2
was admitted at the
Nitrogen adsorption–desorption isotherms were obtained at the liquid
nitrogen temperature with a Micromeritics ASAP 2010 system. Each
sample was degassed at 130 8Covernight before the measurement of the
N physisorption isotherm. From the data, the BET equation was used to
2
calculate the specific surface areas.
desired pressure. The autoclave was then heated by oil circulation at the
desired temperature (65-100 °C), while the mixture was kept under
magnetic stirring at a rate of 1300 rpm. After 4 hours, the autoclave was
cooled to rt, and purged. Thereafter, an aliquot (0.5 mL) of the water
solution was collected, mixed with an aq. solution of diethylenglycol
dimethylether as external standard [diglyme, MeO(CH
2 2 2
CH O) Me; 0.01 M,
The TPD analysis of carbon has been performed with a POROTEC
Chemisorption TPD/R/O 1100 automated system, promoting the controlled
decomposition of the sample in a temperature ramp under a flow of He (30
mL/min). The evolution of gaseous compounds was monitored with an on-
line MS (Cirrus 2, MKS Instrument). TPD tests were performed on 218 mg
0.5 mL], and analysed by GC to determine the conversion of LA and the
selectivity towards the observed products (-valerolactone and -
hydroxyvaleric acid).
Hydrogenation of Levulinic acid in presence of Formic Acid.
of each sample which were first degassed under a flow of helium, and then,
heated at a rate of 5 °C min-1 up to 910 °C. The determination of peak
areas corresponding to the release of CO and CO, deconvolutions, and
2
the evaluation of deconvoluted quantities were calculated by using Origin
software after calibration.[44]
Procedure A. Tests were carried out under the same conditions described
for the hydrogenation of LA as such (see above), except for the following
differences: i) an aqueous solution containing an equimolar mixture of
levulinic acid and formic acid (0.36 M each; total volume=10 mL) was used;
ii) the temperature was set at 130 °C; iii) the reaction time was up to 12 h.
Procedure B (Formic acid as a hydrogen donor). Tests were conducted as
2
Calibration curves for CO and CO . The calibration curves were prepared
2
for procedure B, except for: i) no gaseous H was used; ii) the temperature
was set at 190 °C. In this case, at the end of the reaction, aliquots (0.05
using a gas syringe. The gases used were 99,99% CO and a 10% CO
He. Results are reported in Figure S11-S12 (SI section).
2
in
mL) of the aqueous solution were taken, diluted with deuterated water
TPD results are reported in Figures S13-S15 (SI section). The distribution
percentage of the single species was calculated on the μmol/g value
calculated from the calibration curve.
1
(
0.5 mL) and analysed by H and 13C NMR to evaluate the decomposition
of formic acid (see Figure S5).
For both procedure A and B, GC-analyses of final mixtures and recycle
tests (only for procedure A) were carried out as described in the case of
LA as such. Recycle tests were run for 4 h.
In-Situ DRIFT spectra were acquired with a Bruker Vertex 70 instrument
equipped with a Pike DiffusIR cell attachment and recorded using an MCT
detector after 128 scans and with a 4 cm−1 resolution in the region 4000-
Reductive amination of levulinic acid with cyclohexylamine. Experiments
were carried out in a 25-mL tubular reactor of borosilicate glass (Pyrex)
which was charged with: i) with an aqueous solution containing an
equimolar mixture of levulinic acid and formic acid (3.6 mmol each). The
total volume was varied in the range of 0.5-4 mL; ii) cyclohexylamine (3.6-
4
50 cm−1. The equipment was coupled with a mass spectrometer EcoSys-
P from European Spectrometry 4 Systems. In a typical experiment, a
sample of carbon was loaded and pre-treated at 200 °C under a flow of He
(
10 mL/min) for 40 min in order to remove adsorbed molecules. Then,
using KBr as background, spectra were acquired at different temperatures
200-100-50-25 °C) while cooling the sample down to 25 °C. Then, a pulse
7.2 mmol); ii) 5% Ru/C (30 mg; 0.015 mmol of Ru) as catalyst, and
(
isooctane (5 mL). The glass vessel was then placed in a jacketed steel
autoclave and tests were then performed according to the procedure
above-described for the hydrogenation of LA as such. The temperature,
H pressure, and time were set at 130-150 °C, 35 bar, and 16 h,
2
respectively.
of pyridine (2 μL) was introduced. IR spectra were then acquired at 1 min
time intervals in order to follow the adsorption process. For the desorption
process, the spectra were acquired at 25 °C at 1 min time intervals for 10
minutes. Drifts spectra are reported in Figure 6.
Raman analyses were carried out using a Renishaw Raman System
RM1000 instrument, equipped with a Leica DLML confocal microscope,
with 5x, 20x, and 50x objectives, video camera, CCD detector, and laser
source Argon ion (514 nm) with power 25 mW. The maximum spatial
GC-analyses of final aqueous mixtures were carried out as described in
the case of LA as such.
Once the reaction was complete, N-cyclohexyl methyl pyrrolidone (CyMP)
was extracted from aqueous solution with acetyl acetate (3x10 mL).
Combined organic extracts were washed with diluted hydrochloric acid (3
mM, 10 mL) and then aq. NaOH (3 mM, 10 mL) to remove traces of the
unreacted (excess) amine and acid, respectively. The final solution of
-1
resolution is 0.5 µm, and the spectral resolution is 1 cm . For each sample,
at least five spectra were collected by changing the laser spot on the
surface. The parameters of spectrum acquisition are generally selected as
follows: 5 accumulations, 10 s, 25% of laser power to prevent sample
damage, and 50x objective. Raman spectra are reported in Figure S10 (SI
section).
acetyl acetate was dried over Na
2 4
SO , filtered, and rotary-evaporated. The
brown oily residue (from different runs: 0.58-0.62 g, Y=91-94%, purity 85-
8
9%; residual cyclohexylformamide was present) was analysed by GC/MS
1
and upon dissolution in CDCl
3
, further characterised by H and 13C NMR.
Scanning electron microscopy/energy dispersive spectroscopy
Both mass and NMR spectra were consistent to those reported in the
literature for CyMP.[38]
(SEM/EDX) analyses were performed using an EP EVO 50 Series
Instrument (EVO ZEISS) equipped with an INCA X-act Penta
FET®Precision EDS microanalysis and INCA Microanalysis Suite
Software to provide images of the spatial variation of elements in a sample
Attempts to isolate isomer intermediates detected in the final reaction
mixture (compounds E and F: m/z=179; see Schemes 5 and 6) by either
liquid-liquid extractions or column chromatography were unsuccessful.
Also, the synthesis of such products was attempted starting from LA and
cyclohexylmine, according to the procedure reported for the reaction of LA
with benzylamine.[33] All tests however, failed yielding mostly the
cyclohexylamide of LA.
(Oxford Instruments Analytical). An accelerating voltage of 20 kV was
applied with a spectra collection time of 60 s, and point measurements
were performed in 10-15 regions of interest. Secondary electron images
were collected. SEM images and results of EDX are reported in Figure S9
(SI section) and Table 5, respectively.
All hydrogenation and reductive amination tests were duplicated to check
for reproducibility. In the repeated tests performed under the same
conditions, the values for conversion and selectivity differed by less than
Bohem Titration. The procedure was carried out starting from the reference
catalyst (5% Ru/C) and carbons 3 and 4 of Table 3. A method reported in
the literature was followed.[48] Three identical amounts (0.50 g each) of the
investigated sample were introduced in three different 250-mL Erlenmeyer
5% from one reaction to another.
This article is protected by copyright. All rights reserved.

ChemSusChem
10.1002/cssc.201900925
FULL PAPER
flasks (A, B and C). A 0.05 M aq. solution of NaHCO
to flask A, a 0.05 M aq. solution Na CO (50 mL) was added to flask B,
2 3
3
(50 mL) was added
[5]
[6]
[7]
M. Selva, M. Gottardo, A. Perosa ACS Sustainable Chem. Eng. 2013, 1,
180−189
and finally, a 0.05 M aq. solution of NaOH (50 mL) was added to flask C.
The so-prepared suspensions were allowed to equilibrate for 72 h. Each
basic solution was filtered on a paper and stored in a closed vessel. Then,
a retro-titration method was used to evaluate the final pH: a 10-mL aliquot
of each of the recovered basic solutions was added with a 0.05 M aq.
solution of HCl and stirred for 10 min. The solutions as prepared were
titrated with a 0.05 M solution of NaOH and the equilibrium point was
evaluated using phenolphtalein as an indicator. Each titration was
K. Shill, S. Padmanabhan, Q. Xin, J. M. Prausnitz, D. S. Clark, H. W.
Blanch Biotechnol. Bioeng. 2011, 108, 511–520
M. Sendovski, N. Nir, A. Fishman J. Agric. Food Chem., 2010, 58, 2260–
2265
[8]
[9]
B. Saha, M. M. Abu-Omar Green Chem., 2014, 16, 24–38.
Y. Li, G. Lv, Y. Wang, T. Deng, Y. Wang, X. Hou, Y. Yang Chemistry
Select 2016, 6, 1252–1255.
[10] (a) L. A. Blanchard, J. F. Brennecke Ind. Eng. Chem. Res. 2001, 40, 287-
292; (b) S. Keskin, D. Kayrak-Talay, U. Akman, O. Hortacsu, J.
Supercritical Fluids 2007, 43, 150–180.
2 3
performed three times. In the case of aq. Na CO , 20 mL of 0.05 M HCl
were used. The quantities of the different surface groups (phenolic,
lactonic, and carboxyl groups) are calculated based on their different
[11] J.-M. Andanson, F. Jutz, A. Baiker J. Phys. Chem. B 2009, 113, 10249–
10254.
reactivity with bases: aq. NaOH neutralises all surface groups, aq. Na
reacts with both carboxyl and lactonic groups, and finally, NaHCO reacts
only with carboxylic groups. Therefore, the amount (µeq/g) of carboxylic
groups is determined directly from the reacted NaHCO , while the amount
of carboxyl and lactonic groups is calculated by the difference of reacted
Na CO and NaHCO , and the number of surface phenols is measured by
2 3
CO
3
[12] S. V. Dzyuba, R. A. Bartsch Angew. Chem. Int. Ed. 2003, 42(2), 148-150
[13] (a) X. Sun, Z. Liu, Z. Xue, Y. Zhanga, T. Mu Green Chem., 2015, 17,
2719–2722; (b) C. Shi, J. Xin, X. Liu, X.-G. Lu, S. Zhang ACS Sustainable
Chem. Eng., 2016, 4, 557–563.
3
2
3
3
[14] C. I. Melo, R. Bogel-Łukasik, E. Bogel-Łukasik J. Supercritical Fluids
2012, 61, 191–198.
the difference between reacted NaOH and Na
titration are reported in Table 6.
2 3
CO . Results of Bohem
[15] A. M. da Costa Lopes, M. Brenner, P. Falé, L. B. Roseiro, R. Bogel-
Łukasik ACS Sustainable Chem. Eng., 2016, 4, 3357–3367.
[
16] D. L. Minnick, A. M. Scurto Chem Commun (Camb). 2015, 51, 12649–
The point of zero charge (pHpzc) evaluation. The procedure was carried out
starting from the reference catalyst (5% Ru/C) and carbons 3 and 4 of
Table 3. A method reported in the literature was followed.52 A sample
12652.
[17]
(a) C. Michel, P. Gallezot ACS Catal., 2015, 5, 4130–4132; (b) M.
Pagliaro and M. Rossi, In The Future of Glycerol, 2nd Ed. Royal Society
of Chemistry, 2010; (c) C. Michel, J. Zaffran, A. M. Ruppert, J. Matras-
Michalska, M. Jędrzejczyk, J. Grams, P. Sautet Chem. Commun. 2014,
(0.020 g each) was introduced in a graduated cylinder to which milli-Q
water (10 mL) was added. A pH-meter coupled with a DLS analyser were
used to measure the pH and the potential of the suspension. The latter
50, 12450-12453; (d) J. Tan, J. Cui, G. Ding, T. Deng, Y. Zhu, Y-W. Li
(suspension) was then gradually acidified by an automatic device with
Catal. Sci. Technol. 2016, 6, 1469-1475; (e) M. G. Al-Shaal, W. R. H.
Wright, R. Palkovits Green Chem. 2012, 14, 1260-1263; (f) A. M. Raspolli
Galletti, C. Antonetti, V. De Luise, M. Martinelli Green Chem., 2012, 14,
which two different solutions, a 0.05 M of HCl and a 0.01 M of HCl, were
added dropwise. The potential was then measured as a function of pH
(Figure S16 exemplifies the results for carbons 3 and 4).
6
88–694; (g) J. Tan, J. Cui, T. Deng, X. Cui, G. Ding, Y. Zhu, Y. Li
ChemCatChem 2015, 7, 508–512
18] M. Fabris, V. Lucchini, M. Noè, A. Perosa, M. Selva Chem. Eur. J. 2009,
5, 12273–12282
[
[
Acknowledgements
1
19] (a) D. S. Cameron, S. J. Cooper, I. L. Dodgson, B. Harrison, J. W.
Jenkins Catal. Today 1990, 7, 113-137; (b) E. Auer, A. Freund, J. Pietsch,
T. Tacke Appl. Catal. A: General, 1998, 173, 259-271
Mrs F. Visin and Mrs M. Marchiori from DAIS and DSMN of
Ca‘ Foscari University are kindly acknowledged for their support
[
20] (a) G. M. Whitesides, C. L. Hill, J.-C. Brunie Ind. Eng. Chem., Proc. Des.
Dev., 1976, 15, 226-227; (b) A.-H. Lu, W. Schmidt, N. Matoussevitch, H.
Bonnemann, B. Spliethoff, B. Tesche, E. Bill, W. Kiefer, F. Schuth Angew.
Chem. Int. Ed. 2004, 43, 4303 –4306.
2
with ICP analysis and characterization by N adsorption isotherms,
respectively.
Keywords: multiphase systems • levulinic acid • hydrogenation •
[21] F. K. Kazi, A. D. Patel, J. C. Serrano-Ruiz, J. A. Dumesic and R. P. Anex,
Chem. Eng. J., 2011, 169, 329–338
reductive amination • catalyst recovery
[
22] I. Sádaba, M. López Granados, A. Riisager, E. Taarning Green Chem.,
2
015, 17, 4133–4145.
[
1]
(a) P. Tundo, A. Perosa, Chem. Soc. Rev., 2007, 36, 532– 550; (b)
Multiphase Homogeneous Catalysis, (Eds.: B. Cornils, W. A. Herrmann,
I. T. Horváth, W. Leitner, S. Mecking , H. Olivier-Bourbigou, D. Vogt),
Wiley-VCH, Weinheim, 2005; (c) Regulated Systems for Multiphase
Catalysis (Eds.: W. Leitner, M. Hölscher), Topics in Organometallic
Chemistry, Springer, 2008.
[
[
[
23] Z.-P. Yan, L. Lin, S. Liu Energy & Fuels 2009, 23, 3853–3858.
24] U. K. Singh, M. A. Vannice Appl. Catal. A: General 2001, 213, 1–24
25] J. M. Tukacs, B. Fridrich, G. Dibó, E. Székely, L. T. Mika Green Chem.,
2015, 17, 5189–5195.
[
[
26] (a) P. Puthiaraj, P. Suresh, K. Pitchumani Green Chem., 2014, 16, 2865–
2875; (b) D. J. Macquarrie, B. Gotov, S.Toma Platinum Metal Rev., 2001,
[2]
(a) V. I. Parvulescu, C. Hardacre Chem. Rev. 2007,107,2615−2665; (b)
Nanocatalysis in Ionic Liquids (Ed.: M. H. G. Prechtl), Wiley-VCH,
Weinheim, 2016; (c) M. Selva, A. Perosa, P. Canton Curr. Org. Chem.,
45, 102-110.
27] (a) L. Qi, Y. Fung Mui, S. Wing Lo, M. Y. Lui, G. R. Akien, I. T. Horvath
ACS Catal. 2014, 4, 1470-1477; (b) F. Salak Asghari, H. Yoshida Ind.
Eng. Chem. Res. 2007, 46, 7703-7710.
2017, 21, 2445-2454.
[
3]
4]
Catalysis in Ionic Liquids: From Catalyst Synthesis to Application (Eds.:
V. I. Parvulescu, C. Hardacre), Catalysis Series n° 15, the Royal Society
of Chemistry, 2014.
[
28] C. Fellay, P. J. Dyson, G. Laurenczy Angew. Chem., Int. Ed. 2008, 120,
4
030 –4032.
29] (a) V. Fabos, L. T. Mika, I. T. Horvath Organometallics 2014, 33, 181-
87; (b) Deng, L.; Li, J.; Lai, D. M.; Fu, Y.; Guo, Q. X. Angew. Chem., Int.
[
[
(a) S. Hu, Z. Zhang, Y. Zhou, B. Han, H. Fan, W. Li, J. Song, Y. Xie Green
Chem., 2008, 10, 1280–1283 (b) J. Y. G. Chan, Y. Zhang,
ChemSusChem 2009, 2, 731–734; (c) T. Stahlberg, W. Fu, J. M.
Woodley, and A. Riisager, A. ChemSusChem 2011, 4, 451–458.
1
Ed. 2009, 48, 6529-6532
[
30] A. M. Ruppert, M. Jędrzejczyk, O. Sneka-Płatek, N. Keller, A. S. Dumon,
C. Michel, P. Sautet, J. Grams Green Chem., 2016, 18, 2014–2028
This article is protected by copyright. All rights reserved.

ChemSusChem
10.1002/cssc.201900925
FULL PAPER
[
[
[
[
[
[
[
[
[
[
31] A. S. Touchy, S. M. A. H. Siddiki, K. Kon, K.-i. Shimizu ACS Catal. 2014,
[48] J. L. Figueiredo, M. F. R. Pereira, M. M. A. Freitas, J. J. M. Orfao, Ind.
Eng. Chem. Res. 2007, 46, 4110-4115
4, 3045-3050.
32] X.-L. Du, L. He, S. Zhao, Y.-M. Liu, Y. Cao, H.-Y. He, K.-N. Fan Angew.
Chem. Int. Ed. 2011, 50, 7815 –7819
[49] S. L. Goertzen, K. D. Theriault, A. M. Oickle, A. C. Tarasuk, H. A.
Andreas Carbon 2010, 48, 1252–12
33] G. Gao, P. Sun, Y. Li, F. Wang, Z. Zhao, Yong Qin, F. Li ACS Catal.
[50] G. Busca, Phys. Chem. Chem. Phys. 1999, 1, 723–736
[51] M. C. Ventura, Y. Akacem, and E. Kassab. J. Phys. Chem. C 2008, 112,
19045-19054
2
017, 7, 4927-4935.
34] Y. Ogiwara, T. Uchiyama, N. Sakai Angew. Chem. Int. Ed. 2016, 55,
864 –1867.
35] S. Wang, H. Huang, C. Bruneau, C. Fischmeister ChemSusChem 2017,
0, 4150 – 4154.
1
[52] C. Pevida, M. G. Plaza, B. Arias, J. Fermoso, F. Rubiera, J. J. Pis Appl.
Surf. Sci. 2008, 254, 7165–7172
1
[53] D. Prahas, Y. Kartika, N. Indraswati, S. Ismadji Chem. Eng. J. 2008, 140,
32–42.
36] J. J. Martínez, L. Silva, H. A. Rojas, G. P. Romanelli, L. A. Santos, T. C.
Ramalho, M. H. Brijaldo, F. B. Passos Catal. Today 2017, 296, 118–126
37] C. Wu, X. Luo, H. Zhang, X. Liu, G. Ji, Z. Liu, Z. Liu Green Chem., 2017,
1
9, 3525–3529
38] C. Ortiz-Cervantes, M. Flores-Alamo, J. J. García Tetrahedron Lett. 2016,
7, 766–771
5
39] X.-L. Du, L. He, S. Zhao, Y.-M. Liu, Y. Cao, H.-Y. He, K.-N. Fan Angew.
Chem. 2011, 123, 7961–7965
40] Under solventless conditions (without water, experiments not shown in
Table 1), a further slight increase of the pyrrolidone amount up to 35%
was observed at complete conversion, thereby confirming the
detrimental role of water for the reductive amination of LA.
41] J. D. Vidal, M. J. Climent, P. Concepción, A. Corma, S. Iborra, M. J.
Sabater ACS Catal., 2015, 10, 5812–5821
[
[
42] J. D. Vidal, M. J. Climent, P. Concepción, A. Corma, S. Iborra
ChemSusChem 2017, 10, 119 – 128.
[43] R. W. Layer Chem. Rev. 1963, 63, 489–510
[44] A. E. Aksoylu, M. Madalena, A. Freitas, M. F. R. Pereira, Jose L.
Figueiredo Carbon 2001, 39, 175–185.
[
[
[
45] M. N. Nagornaya, G. I. Razdyakonova, S. Ya. Khodakova Procedia Eng.
2
016, 152, 563–569.
46] J. L. Figueiredo, M. F. R. Pereira, M. M. A. Freitas, J. J. M. Orfao, Carbon
999, 37, 1379–1389
47] J. J. Venter, M. A. Vannice Carbon 1988, 26, 889-902
1
This article is protected by copyright. All rights reserved.

ChemSusChem
10.1002/cssc.201900925
FULL PAPER
Entry for the Table of Contents (Please choose one layout)
Layout 1:
FULL PAPER
Multiphase reactions of LA
Author(s), Corresponding Author(s)*
occurred with selective synthesis
of GVL or pyrrolidones in aqueous
solution and complete recovery of
Ru/C in hydrocarbon phase
Page No. – Page No.
Title
Layout 2:
FULL PAPER
Author(s), Corresponding Author(s)*
((Insert TOC Graphic here; max. width: 11.5 cm; max. height: 2.5 cm))
Page No. – Page No.
Title
Text for Table of Contents
This article is protected by copyright. All rights reserved.