Xylitol Hydrogenolysis over Ruthenium-Based Catalysts: Effect of Alkaline Promoters and Basic Oxide-Modified Catalysts

Article abstract of DOI:10.1002/cctc.201700034

The aqueous-phase hydrogenolysis of xylitol into glycols over Ru/C was performed in the presence and absence of a wide range of concentrations of Ca(OH)₂ to investigate the reaction pathway. Without base, epimerization and cascade decarbonylation were the predominant reactions with high selectivities to C₅ and C₄ alditols and light alkanes at full conversion. Glycol production was obtained by the addition of Ca(OH)₂ to promote the retro-aldol reaction. It competed with reactions without base and became the main reaction for a OH^?/ xylitol molar ratio R_mol(OH/xylitol) of 0.13, and high selectivities to glycols (56 %) and glycerol (16 %) were observed. However, lactate was a byproduct at up to 27 % with a high base amount (R_mol(OH/xylitol)=0.68). Bifunctional Ru/metal oxide/C catalysts (metal: Zn, Sn, Mn, Sr, W) were synthesized and were able to cleave the C?C bond into glycols without a base promoter. The 3.1 wt %Ru/MnO(4.5 %)/C catalyst was the most active (220 h^?1) with reasonable selectivity to glycols (22 %) and glycerol (10 %) and a low production of lactate (<1 %). Nevertheless, metal oxide leaching of the catalyst was observed likely because of the production of traces of lactate.

Full text of DOI:10.1002/cctc.201700034

Accepted Article
Title: Xylitol hydrogenolysis over Ru-based catalysts: effect of alkaline
promoter and basic oxide modified catalysts
Authors: Maxime Rivière, Noemie Perret, Amandine Cabiac, Damien
Delcroix, Catherine Pinel, and Michele Besson
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: ChemCatChem 10.1002/cctc.201700034
Link to VoR: http://dx.doi.org/10.1002/cctc.201700034
A Journal of
www.chemcatchem.org

ChemCatChem
10.1002/cctc.201700034
Xylitol hydrogenolysis over Ru-based catalysts: effect of alkaline promoter
and basic oxide modified catalysts
Maxime Rivière, Noémie Perret, Amandine Cabiac,[b] Damien Delcroix,[b] Catherine
[
a]
[a]
[a]
[a]
Pinel, Michèle Besson*
[a]
Univ Lyon, Univ Claude Bernard, CNRS, IRCELYON, UMR5256, Institut de recherches sur
la catalyse et l’environnement de Lyon, 2 Avenue Albert Einstein, 69626 Villeurbanne, France
[b]
IFP Energies nouvelles, Rond-Point de l’Echangeur de Solaize, BP 3, 69360 Solaize, France
Abstract
Aqueous phase hydrogenolysis of xylitol into glycols over Ru/C was carried out without and in
the presence of a wide range of concentrations of Ca(OH)
pathway. Without base, epimerization and cascade decarbonylation were the predominant
reactions with high selectivities to C and C alditols, and light alkanes at full conversion.
Glycol production was obtained by addition of Ca(OH)
2
in order to investigate the reaction
5
4
2
to promote the retro-aldol reaction. It
competed with reactions without base and became the main reaction for OH / xylitol molar
ratio Rmol( /xylitol) of 0.13; high selectivities to glycols (56%) and glycerol (16%) were
-
OH
observed. However, lactate (LA) was by-produced up to 27% at high base amount
OH
(Rmol( /xylitol)=0.68). Bifunctional Ru/metal oxide/C catalysts (metal: Zn, Sn, Mn, Sr, W) were
synthesized and were able to cleave C-C bond into glycols without base promoter.
-1
3
.1wt.%Ru/MnO(4.5%)/C catalyst was the most active (220 h ) with reasonable selectivity to
glycols (22%) and glycerol (10%) and small production of LA (<1%). Nevertheless, metal
oxide leaching of the catalyst was observed likely due to the production of traces of LA.
Keywords
Biomass; base promoter hydrogenolysis; ruthenium catalysts; xylitol
1
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201700034
1
. Introduction
Due to the growth of worldwide fossil demand for energy and carbon based chemicals along
with the environmental problem awareness, the non-edible biomass appears to be a promising
[
1–3]
alternative resource.
Cellulose and hemicelluloses are the most abundant sources of carbon;
they can be hydrolyzed to sugar monomers (i.e., glucose, xylose) and then further hydrogenated
to the corresponding alditols (i.e., sorbitol, xylitol). In particular, xylitol is listed among the top
building blocks as it can generate numerous high added-value derivatives.[ Hydrogenolysis
of polyols yields primarily important glycol derivatives such as ethylene glycol (EG), propylene
glycol (PG) and glycerol (GLY). The glycols EG and PG are currently produced in high volume
4,5]
[
6]
[7]
(
EG: 25 Mt/y in 2014 and PG: 1 Mt/y in 2013 ) from petroleum-based ethylene and
[7,8]
propylene via their epoxide intermediates.
They are widely used in the manufacture of
polyester resins (i.e., PET or PES), in the cosmetic and pharmaceutical industries, and as heat
[
9,10]
transfer fluids.
Their production from renewable biomass could be a complementary
process for responding to the increasing demand and to the growing interest of major firms and
consumers to introduce biosourced compounds in daily-used products.
The selective catalytic hydrogenolysis of sorbitol[11–31] and xylitol[32–37] into glycols has been
intensively studied in the last two decades over various supported metal catalysts. Typically,
the reaction requires a noble or transition metallic function (Ru, Ni or Cu) for the
dehydrogenation/hydrogenation steps and a base to catalyze the C-C cleavage. The reactions
are usually carried under harsh conditions, i.e. temperature in the range of 160°C to 230°C and
hydrogen pressure in the range of 40 to 120 bar. For instance, a combined EG and PG yield of
5
4-55% was obtained at nearly full conversion during the hydrogenolysis of xylitol (200°C, 60
bar of H ) over 90%Cu-SiO catalysts, prepared by deposition-gel method or by deposition-
precipitation, and in the presence of Ca(OH) A search through the literature on the
2
2
[
34,36]
2
.
[
14]
hydrogenolysis of alditols revealed that Ru-based catalysts are the most active catalysts. High
conversions of sorbitol were obtained over 3%Ru/carbon nanofiber (86%) and 3%Ru/C (71%),
[
14]
in the presence of CaO base promoter.
Moreover the carrier can impact on the catalytic
studied the effect of the support and reported that carbon-
supported Ru catalyst exhibited the highest selectivity to the glycols in the hydrogenolysis of
[
37]
response. Indeed, Liu et al.
xylitol. When working at 200°C, under 40 bar of H
2
and in the presence of Ca(OH) , 56%
2
selectivity towards the glycols was obtained at ca. 25% conversion.[ It is challenging to obtain
high selectivities to glycols, due to the cascade C-C cleavage by decarbonylation followed by
methanation, and the water gas shift reaction generating gaseous by-products and alcohols by
37]
[
16,28]
C-O cleavage.
Most studies report the addition of a base promoter in aqueous phase in order to facilitate the
[
15,35]
[33]
[17]
C-C cleavage via retro-aldol reaction. Ca(OH)
2
, NaOH
and Ba(OH)
2
are usually
employed with a OH /substrate molar ratio in the wide range of 0.07 to 5[33], but mostly
-
[25]
around 0.7.[ It is worth noting that the basic additive Ca(OH)
21]
has been shown to generate
2
better glycols selectivity results than other hydroxide bases.[
21,27]
2
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201700034
The use of large amounts of inorganic homogeneous base requires several separation steps and
final neutralization generates high amounts of invaluable salts.[ Moreover, alkaline additive
favors the formation of lactate as a by-product from glyceraldehyde, at the expense of PG,
35]
[
37]
which limits the selectivity to glycols. To address these issues, recent studies have reported
the use of bifunctional catalysts with basic properties.[
15,18,22–24,26,28,34,35,38]
The conversion of
sorbitol was studied over a co-precipitated Ni-MgO catalyst without alkaline additive in
aqueous phase (200°C, 40 bar of H
2
). Around 60% selectivity to glycols, 20% selectivity to
glycerol (GLY), and a limited amount of LA were obtained at 68% conversion.[ When Ru or
22]
Ni-Ru were supported directly on the alkaline promoter calcium hydroxide, the yield to glycols
(
from sorbitol) was significantly improved, going from 14% over Ru/Al
2
O to 28% and 40%
3
and Ni-Ru/Ca(OH)
, respectively.[ However, the catalysts deactivated
28]
over Ru/Ca(OH)
2
2
quickly during the reaction due to solubilization of the support. Ni-Mg-Al catalyst derived from
[23]
[24]
hydrotalcite-like compounds and co-precipitated Cu/CaO-Al
2
O
3
catalysts were shown to
be stable for the conversion of sorbitol when working at 200°C / 20 bar and 230°C / 76 bar,
respectively; the former one exhibited a selectivity to glycols of 45% and the later one of 61%,
at high conversion. 6%Ni deposited on fly ash support[ mostly composed of SiO
26]
, Al
2
2
3
O and
Fe
2
O
3
, was tested four times for the hydrogenolysis of sorbitol in the absence of base. The
catalyst remained stable and high selectivities to glycols (38%) were reported at 200°C and
[
35]
4
0% conversion. The group of Liu impregnated Ni/C with basic oxides (CaO, BaO, CeO
La , MgO) and tested the catalysts without additional base for xylitol hydrogenolysis at
00°C and under 40 bar H normalized per mole of Ni) was obtained
. The best activity (20 h-1
over Ni-CeO /C with a CeO /Ni molar ratio of 1.5 and the best selectivity to glycols (70% at
5% conversion) over Ni-CaO/C with a CaO/Ni molar ratio of 0.66. Nickel did not leach during
2
,
2
O
3
2
2
,
2
2
1
the reaction, however, the stability of the basic oxide was not addressed.
In this contribution, we considered the modification of a carbon-supported Ru catalyst by
addition of five oxides (ZnO, SrO, SnO
x
, MnO, WO ) that present a range of acid-base surface
x
properties. The objective of this work was to investigate the effect of basic oxides on the
catalytic activity and products distribution in xylitol hydrogenolysis. ZnO is often used for
cellulose or glycerol conversion to EG and PG[
38,39]
; SrO is an alkaline oxide in the same group
as CaO and MgO; although SnO
x
is usually seen as Lewis acid, it has been shown to be an
active phase for the retro-aldol reaction of glucose intermediates from cellulose to acetol.[
40]
We could not find any literature reporting MnO as active species for C-C cleavage, however it
is known to be responsible for the basic character of Ru-MnO
x
/C catalyst in the
[
41]
demethoxylation of guaiacol in aqueous media and for the ring opening reaction of furfuryl
[
42]
alcohol. For comparison, a bifunctional acidic catalyst was prepared using tungsten oxide.
The physicochemical properties of the catalysts (denoted as Ru/MO /C) and their catalytic
performances were compared. We also investigated the mechanism of xylitol hydrogenolysis
over Ru/C catalysis with and without Ca(OH) promoter in order to get further insight on the
x
2
competitive reactions decarbonylation/epimerization and retro-aldol reaction.
3
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201700034
2
. Results and Discussion
2
.1 Catalyst Characterization
Two Ru/C monometallic catalysts were synthetized by wet impregnation with an aqueous
solution of ruthenium(III) nitrosyl nitrate. The bifunctional catalysts, 3wt%Ru/MO (y%)/C,
x
where M stands for W, Sn, Sr, Zn, or Mn and y is the wt% of M deposited, were prepared by
sequential wet impregnation: metal oxide precursor impregnation, calcination, Ru precursor
impregnation, and then reduction. The nominal loading of the second metal M was fixed at
5
wt% except for Ru/SrO/C for which it was 3wt%. Table 1 summarizes the Ru and metal oxide
loadings, the main textural properties, the metal oxide crystallite sizes, and the basic properties
of the catalysts.
Table 1. Metal loadings, total pore volume, BET surface area (SBET), metal oxide crystallite size and total amount
of basic sites of the Ru/MO
x
/C catalysts.
Catalyst
Metal loading[a]
Total pore
volume
[cm .g ]
Crystallite size
Total basic
[
c]
M/Ru
wt%/wt%]
S
BET
2
[
wt%]
[d]
[e]
of MO
x
sites
[µmol g ]
-1
[
[m .g ]
M[b]
-
3
-1
-1
Ru
-
[nm]
n.d
Active carbon
-
0.81
1095
1087
1041
1120
987
-
2
37
2
7
.9%Ru/C
.1%Ru/C
2.85
7.09
3.13
2.63
1.75
1.63
-
-
0.84
-
n.m.
-
-
0.79
-
3
2
2
1
55
61
77
56
3
.1%Ru/MnO(4.5%)/C
.6%Ru/ZnO(4.1%)/C
.8%Ru/SrO(2.1%)/C
.6%Ru/SnO (5.0%)/C
(5.0%)/C
Determined by ICP-OES,
4.48
4.10
2.14
5.00
5.00
1.4
1.6
1.2
3.1
1.8
0.88
14
18
n.d
n.d
n.d
2
0.76
1
0.78
1032
933
1
x
0.74
8
5
2
a]
.7%Ru/WO
x
2.73
0.84
971
[
[b]
M: W, Sn, Sr, Zn, or Mn, [c] relative error: ±5%, [d] Debye-Scherrer equation. n.d: not-
2
-TPD, n.m. not measured.
detectable, [ Determined by CO
e]
The final loadings were close to the nominal loadings for Ru/C and Ru/MO
x
/C catalysts with
M = Mn, Zn and W; the final M/Ru mass ratios were approximately 1.5 for the bimetallic
catalysts, as expected. However Ru was not fully deposited for Ru/SrO/C and Ru/SnO
x
/C
catalysts, with a Ru metal loading around 1.7%.
The N
2
adsorption-desorption isotherms were type IV isotherms (Figure S1)[43] and exhibited
= 0.1, suggesting that a large proportion of the pore channels
very high adsorption below P/P
are micropores. Moreover, an increase of adsorbed volume for P/P
loops in the desorption branch were observed, indicating the presence of mesopores.[
0
[
44]
0
> 0.8 and hysteresis
43–45]
The
deposition of Ru onto the active carbon support did not change the surface area and pore
volume, indicating that the Ru particles did not cause pore blockage (Table 1). The total pore
3
-1
3
-1
volume of the bifunctional catalysts varied between 0.76 cm .g and 0.88 cm .g depending on
4
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201700034
the nature of the second metal. Compared to the Ru/C catalysts, the specific surface area is
2
-1
slightly lower (except for Ru/MnO/C), with values in the range 933-1032 m .g , suggesting
that the textural characteristics of the different catalysts were not strongly affected in spite of
the successive treatments, i.e. impregnation, calcination and reduction steps.
The reduction temperature of 450°C was chosen according to literature. Indeed Ru metal
[
37]
particles with small sizes (< 2 nm) were reported for Ru/C and Ru/ZrO
4
2
after reduction at
[46]
50°C. Accordingly, the powder X-Ray diffraction (XRD) patterns of the two Ru/C catalysts
presented in Figure S2 showed no peaks assigned to metallic Ru (JCPDS 06-0663), suggesting
0
a small crystallite size of Ru even at high loading (7.1wt% Ru).
The particle size for 2.9wt%Ru/C was characterized by TEM (Figure S3). The sample shows
small particles that are well dispersed all over the surface and no agglomerated particles. The
mean particle size was estimated to be 1.65 nm from the size distribution. This corresponds to
a dispersion of 65%, assuming the particles to be cubic with only five accessible faces.
The XRD patterns associated with the Ru catalysts modified with a basic or acidic oxide are
presented in Figure 1. The crystallite mean sizes of the oxides MO
x
are included in Table 1.



(a)





(
b)
c)
d)
e)
f)
80
(
(
(
(
10
20
30
40
50
60
70
2θ / °
x
Figure 1. XRD patterns of bifunctional Ru/MO /C catalyst with M: a) Mn; b) Zn; c) Sr; d) Sn; e) W, f) active
carbon. () MnO phase (JCPDS 07-230) and () ZnO phase (JCPDS 36-1451).
After impregnation of the support with the metal oxide precursor, a calcination step was
conducted at 200°C under air in order to generate the oxide; the temperature of this heat
treatment was limited by the nature of the carrier, i.e. active carbon.
5
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201700034
x
The XRD patterns for reduced Ru/SnO /C (Figure 1.d) and Ru/SrO/C (Figure 1.c) did not
exhibit diffraction peaks associated with tin or strontium oxide, indicating either a crystallite
size below the XRD detection threshold (ca. 3 nm) or amorphous phase properties. In addition,
a Ru/SnO
synthesized and calcined at two temperatures (200°C under air or 500°C under Ar) however the
formation of Ru, RuSn alloy or SnO crystallites were still not observed (Figure S4). TEM with
EDX analysis of Ru/SnO /C showed that Ru and Sn elements were well distributed over the
solid. However the presence of chlorine suggests that the precursor was not totally decomposed
to SnO (Figure S5(a)). Some larger particles, probably SnCl , were detected; they were moving
under the electron beam and were transformed into SnO (JCPDF 88-0287) (Figure S5(b)). The
absence of diffraction peak is surprising since Deng and Liu reported the formation of PtSn
x
/C catalyst with a higher Ru (6.6 wt%) and tin oxide (13.5 wt% Sn) loading was
x
x
x
2
2
alloy (by XRD) in addition to SnO
x
species (revealed by XPS), using similar conditions of
support and a Sn/Pt atomic ratio exceeding 1.5.[ Besides,
47]
catalyst synthesis with an Al O
2
3
bimetallic catalysts prepared by incipient wetness impregnation of a 5wt%Ru/C catalyst with
an aqueous solution of tin chloride (final catalyst containing 1.6 wt% and 5.5 wt% Sn), followed
by heating in air at 80°C for 2 h and reduction in H at 450°C demonstrated multiple phases:
2
[
48]
Ru, Ru
2
Sn
3
, Ru
3
Sn , β-Sn.
7
For Ru/ZnO/C (Figure 1.b), diffraction peaks were observed and attributed to ZnO (JCPDS 36-
451) at 2θ = 31.5° (100), 34.2° (002), 36.0° (101), 47.2° (102), and 56.2° (110). The mean
crystallite size of ZnO was calculated and was of 18 nm.
1
The XRD pattern of Ru/WO
x
/C (Figure 1.e) did not show any diffraction peak of crystalline
2
WO
x
species. Below a monolayer coverage (ca. 5 W/nm ) tungsten oxide is known to be
molecularly dispersed on oxide support with high surface area and poorly dispersed on active
[
49]
carbon. In the latter case, the diffraction peaks of WO
wt%.
3
are usually not observed below 6.0
[
49]
Raman spectroscopy would be suitable for the discrimination of the molecular
structures of supported tungsten oxide species such as isolated surface WO species,
4
crystalline nanoparticles.[
50]
However a more detailed
polytungstate species, WO
3
characterization of that catalyst was not judged necessary for this study.
The XRD pattern of Ru/MnO /C catalyst (Figure 1.a) exhibited peaks at 34.8°, 40.7°, 58.7°
x
attributed to the (111), (200) and (220) cubic planes of MnO phase (JCPDS 07-230); the mean
crystallite size was 14 nm. The TEM characterization of the MnO-promoted catalyst shown in
Figure S6 confirms the pattern observed by XRD. Small particles of Ru were deposited on both
x
the carbon support and large particles of MnO . The oxide phase displayed a particle size around
2
5 nm.
Ru/MnO/C catalysts with different MnO weight loadings in the range 1.3-19.3 wt% were
prepared and characterized. A Ru/MnO catalyst was obtained by impregnation of Ru on a
synthesized MnO support. The physicochemical characteristics of these catalysts are included
x
x
in Table 2 and the XRD diffraction patterns are presented in Figure 2.
6
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201700034
(a)






(b)






(c)
(d)






(e)
*
*
*
30
40
50
60
70
80
2θ / °
Figure 2. XRD patterns of: a) 3.0%Ru/MnO(1.3%)/C; b) 3.1%Ru/MnO(4.5%)/C; c) 2.9%Ru/MnO(13.4%)/C; d)
3
2
.1%Ru/MnO(19.3%)/C and e) 2.8%Ru/MnO
4-0734).
x
3 4
. () MnO phase (JCPDS 07-230) and (*) Mn O phase (JCPDS
Table 2. Ru and Mn loadings, BET surface area (SBET), and mean metal crystallite sizes of Ru/MnO/C and
Ru/MnO catalysts.
x
Catalyst
Ru
loading
Mn
loading
Mn/Ru
S
BET
-1
Crystallite size of
[a]
[a]
2
[b]
[wt%/wt%] [m .g ] Mn oxide [nm]
[
wt%]
[wt%]
3
3
.0%Ru/MnO(1.3%)/C
.1%Ru/MnO(4.5%)/C
2.99
3.13
2.90
3.10
2.75
1.31
0.4
1.4
4.6
6.2
-
992
1120
834
720
24
n.d
14
14
14
n.d
4.48
13.37
19.32
-
2
3
.9%Ru/MnO(13.4%)/C
.1%Ru/MnO(19.3%)/C
2
.8%Ru/MnO
x
[a]
Determined by ICP-OES, [b] Calculated by Debye-Scherrer, n.d: not-detected
The specific surface area decreased with increasing MnO loading from 1.3% to 19.3%. The
manganese species remained as MnO (Figure 2) and the crystallite size of MnO was constant
2
-1
at 14 nm (Table 2). The Ru/MnO
two phases of MnO (Figure 2.e): peaks with the higher intensities were assigned to MnO
JCPDS 07-230), while the smaller peaks at 32.5°, 36.0° and 59.9° were assigned to tetragonal
Mn (JCPDS 24-0734). As for Ru/C, no peaks attributed to Ru species were observed,
x
catalyst prepared on MnO
x
carrier (24 m .g ) demonstrated
x
(
3
O
4
x
suggesting a high dispersion of Ru on MnO-promoted or MnO -supported Ru catalysts, even
with low surface area of the latter one.
7
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201700034
The basic properties were measured by CO
2
-TPD experiments. The CO
2
-TPD are
comparatively presented in Figure S7, the amounts of CO
2
desorbed are included in Table 1. It
is important to note that the detection used was a thermal conductivity detection but without
MS. All solids, except the pristine carbon, showed a large desorption peak centered at
approximately 425°C. The peak shift to higher temperature observed for Ru/MnO/C and
Ru/SnO /C indicates that these solids should exhibit relatively stronger basic sites. In addition,
x
a broad signal was shown at 175°C corresponding to basic sites of medium strength for ZnO,
SrO and MnO additives. A similar trend for TPD signals was observed in the comparison with
[41]
literature for Ru/C and Ru-MnOx/C. Among the studied materials, the total number of basic
sites varied as follows: Ru/MnO/C > Ru/SrO/C > Ru/ZnO/C > Ru/C > Ru/SnO /C >
Ru/WO /C.
x
x
To summarize, characterization of the catalysts confirmed the small Ru crystallite size of
monometallic Ru/C catalysts even at 7.1 wt% Ru. Based on XRD, the nature of Sr, Sn and W
species over active carbon support could not be defined, whereas metal oxide peaks of the
second metal were identified for Ru/ZnO/C and Ru/MnO/C. An increase in MnO loadings while
keeping Ru loadings constant was associated with a slight decrease in specific surface area of
the catalyst, whereas the crystallite size of MnO remained constant. For Ru/MnO
mixture of two manganese oxide phases, MnO and Mn , were observed, on which Ru was
well dispersed. The solids presented different basicity properties.
x
catalyst, a
3
O
4
2
.2 Hydrogenolysis of xylitol over Ru/C without and in the presence of Ca(OH)
2
base
promoter
The hydrogenolysis of polyol to glycols is usually conducted in the presence of a metal catalyst
and a basic promoter added to the reaction medium. The simplified and most widely accepted
mechanism for the hydrogenolysis of xylitol in the presence of a base is presented in Scheme
1
. First, xylitol is dehydrogenated on the 1- position (1) to xylose. The aldose can then undergo
either a terminal C-C cleavage via decarbonylation reaction (2) into threitol, or an enolization
and hydrogenation (3) into arabitol, an epimer of xylitol, or also a C-C cleavage via retro-aldol
reaction (4) into glyceraldehyde and glycolaldehyde. These aldehydes can then be hydrogenated
respectively into glycerol (GLY) and EG. A dehydration reaction of glyceraldehyde may also
occur (5) and yields pyruvaldehyde; the latter is either hydrogenated into PG or yields lactate
(
LA) by Cannizzaro-type disproportionation in basic media (6). The retro-aldol reaction cleaves
the C-C bond according to the dehydrogenation position of the hydroxyl group. Typically,
dehydrogenation of xylitol on the 1-, as (4), and 2- positions (not shown) yields C and C
products, while dehydrogenation of the 3-position yields C and C products (not shown).
2
3
1
4
[51]
Tajvidi et al. proposed also an isomerization of the aldose to ketoses via the so-called Lobry
de Bruyn-Alberda van Ekenstein reaction to explain the product distribution they observed over
O
catalyst in the absence of a base.[
51]
Cu/ZnO/Al
2
3
8
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201700034
Xylitol
OH OH
3
HO
OH
2
2
1
1
OH
+
H2
- H2
(1)
OH
OH OH
OH OH
-
CO
+ H2
OH
O
OH
HO
OH
HO
-
H2
OH
Threitol
OH
OH
Arabitol
(2)
(3)
-
OH
(4)
+
-
H2
-
H2
OH
OH
HO
O
+
O
OH
HO
OH
H2
OH
+ H2
OH
GLY
EG
(
1)
(1)
-
H O
2
(5)
O
O
+
2 H2
-
+
OH
(1)
(6)
LA
OH
HO
O
O
-
OH
PG
Scheme 1. Reaction scheme of xylitol hydrogenolysis in the presence of a base, proposed by Sun and Liu.[37]
EG: ethylene glycol, GLY: glycerol, PG: propylene glycol, LA: lactate. Reaction (1) equilibrium
dehydrogenation/hydrogenation, (2) decarbonylation, (3) enolization and hydrogenation, (4) retro-aldol
reaction, (5) dehydration, and (6) Cannizzaro-type disproportionation.
Through the use of C
3
-C alditol stereoisomers, Deutsch et al. showed that dehydrogenation of
6
the terminal primary alcohols was favored compared to internal hydroxyl groups, and was
[
16]
followed by decarbonylation. This was confirmed by H/D exchange of sorbitol in D O and
2
the order of dehydrogenation was the following: 1- > 2- >> 3-position.[ By kinetic isotopic
52]
study on deuterated sorbitol, Jia and Liu showed that the preferential activation step was on the
[
29]
5
-position of sorbitol over Ru/C catalyst in the presence of Ca(OH)
2
.
Hence, sorbitol is
epimerized more rapidly to mannitol and iditol than to galactitol and allitol. Therefore, from
the point of view of kinetics, xylitol would preferentially give arabitol than adonitol (Scheme
[
53]
2
), as reported recently.
9
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201700034
OH OH
OH OH
3
3-position
OH OH
2
OH
-positions
HO
HO
OH
HO
OH
2
2
1
1
OH
OH
OH
Arabitol
Adonitol
Xylitol
2
Scheme 2. Possible formation of epimers from xylitol under H pressure, depending on the position of
the dehydrogenation step.
We investigated the role of the base promoter by conducting the reaction without Ca(OH)
2
additive and in the presence of a wide range of concentrations of the base.
First, no conversion was observed in the blank xylitol hydrogenolysis reactions at 200°C under
H2 = 60 bar, with or without 0.25 M Ca(OH) base, in the absence of a catalyst.
P
2
Figure 3 and Figure 4 present the results obtained for xylitol hydrogenolysis as a function of
time (a) and as a function of conversion (b), over 2.9%Ru/C in the absence of base (Figure 3)
and in the presence of the highest concentration of base investigated, i.e. 0.25 M Ca(OH)
2
(Figure 4), respectively.
-1
When no basic additive was introduced, the initial activity was 727 h (normalized to mole of
Ru, as described in the experimental section) and xylitol conversion attained 99% after 24 h
(
Figure 3(a)). Epimers of xylitol (arabitol and adonitol), as well as C polyols (threitol and
4
erythritol) were the main products analyzed in liquid phase within the first 3 h. Afterwards, the
concentration of arabitol dropped and those of other products decreased slightly. Compounds
with a smaller chain of carbons such as glycerol (GLY), propylene glycol (PG) and also to a
lesser degree ethylene glycol (EG), and butanediols (1,2-BDO and 2,3-BDO) were formed.
They became the major products in liquid phase at complete xylitol conversion. Traces of
methanol were also detected in the aqueous phase. However the low carbon balance in liquid
phase reaching only 45% after 24 h indicated that most products were in gas phase. The analysis
of the gas phase revealed that light alkanes including CH
4
(83% of carbon in gas phase
products), C (9.9%), C (5.2%), C 10 (1.2%) and C 12 (0.2%), as well as 0.1% of CO
2
H
6
3
H
8
4
H
5
H
2
were detected, whereas CO was not present. The amount of carbon measured in the gas phase
roughly corresponds to the carbon mass balance defect measured in the aqueous phase. The
results are consistent with previous studies which reported that methanation of CO and cleavage
[
13,16,28]
[54]
of C-O bonds may occur from xylitol
and xylose over Ru based catalysts. In addition,
these results show that the formation of the gaseous products occurred mainly after 80% xylitol
conversion (Figure 3(b)), suggesting that Ru is active for the C-O hydrogenolysis of the
products formed from xylitol, i.e. C
4
polyols, GLY, PG and EG, into C
4
-C light alkanes.
1
1
0
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201700034
(a)
100
0.40
80
0.30
60

0.20


4
0
20
0








0
.10







































0.00 
0
5
10
15
20
25
time / h
(b)
60
4.0




5
0







3.0
40



30
2.0


20

1
.0














10


































0.0
0
30
40
50
60
70
80
90
100
xylitol conversion / %
Figure 3. Hydrogenolysis of xylitol over Ru/C without base promoter: (a) xylitol conversion, concentrations
profiles, and TOC as a function of time; (b) products distribution as a function of xylitol conversion. () TOC
measured; () Xylitol conversion; () Arabitol; () Adonitol; () Threitol; () Erythritol; () BDO
(
butanediols); () GLY (glycerol); () PG (propylene glycol); () EG (ethylene glycol); () gas products; ()
/C products molar ratio.
Reaction conditions: xylitol 0.7 mol.L (10 wt%); 135 mL H
C
3
2
-1
2 2
O; 0.5 g 2.9% Ru/C catalyst; 60 bar H ; 200°C.
1
1
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201700034
The evolution of the product distribution (on a carbon basis, %) towards the different products
versus xylitol conversion without base addition is shown in Figure 3(b). The high selectivity to
arabitol, up to 55% at 50% xylitol conversion, decreased to 7% at nearly 100% conversion of
xylitol. The selectivities to adonitol and threitol remained approximately constant at 12% and
8
%, respectively, before they declined for conversion above 90%. The selectivity to erythritol
became significant beyond 50% conversion and attained 12%, as for threitol, before decreasing
after 90% conversion. Meanwhile, the selectivity to products in the gas phase evolved in inverse
proportion to arabitol selectivity, i.e. it increased slowly until 80% conversion and then sharply
up to 55% at complete conversion. In parallel, the selectivities to glycols and glycerol were
1
7% at 99% conversion with a gradual increase of the C
3
/C products molar ratio. It should be
2
noted that Ru leaching after reactions was below the detection limit of the ICP-OES analyzer
-1
(
< 0.2 mg.L , < 0.15%). No modification of the XRD pattern of Ru/C catalyst after use was
observed suggesting no sintering of the Ru nanoparticles.
The addition of 0.25 M of Ca(OH)
2
base significantly modified the concentration profiles and
the products distribution over Ru/C (Figure 4). The reaction proceeded efficiently, however,
-1
-1
the activity decreased from 727 h without base to 223 h with 0.25 M Ca(OH)
2
. As shown in
Figure 4(a), xylitol conversion was nearly complete after 27 h. A similar trend in activity was
[
37]
observed by Sun and Liu for hydrogenolysis of xylitol over 5wt%Ru/C with the addition of
a base. It could be explained by the equilibrium reaction of retro-aldol reaction in the
mechanism, which slowed down xylitol conversion. As concerns the pH, the initial value of
1
2.0 decreased sharply to 7.5 after 10h.
(a)
100







0.40


80

0.30











60
40
20
0





0
.20




.10 
0













 





  






 


0
.00 

  
  
  


0
5
10
15
20
25
30
35 45 50
time / h
1
2
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201700034
(b)
60
2.0
50

1.5

4
0













1.0
30














20





0.5

1
0












  





0.0

  
0



 

26 28 60
70
80
90
100
xylitol conversion / %
2
Figure 4. Hydrogenolysis of xylitol over 2.9%Ru/C with 0.25 M of Ca(OH) base promoter: (a) concentrations
profiles and xylitol conversion and TOC was a function of time; (b) products distribution as a function of xylitol
conversion. () TOC measured; () Xylitol conversion; () Threitol; () Erythritol; () BDO; () GLY; ()
PG; () EG; () FA (formate); () LA (lactate); () gas products; ()C
3
/C
2
products molar ratio. Reaction
-1
conditions: xylitol 0.7 mol.L (10 wt%); 135 mL H
2
O; 0.5 g 2.9%Ru/C catalyst; 2.5 g Ca(OH)
2
; 60 bar H ; 200°C.
2
No epimer of xylitol was generated when adding Ca(OH)
2
; on the other hand, traces of C
4
products were observed. The main products were EG, LA and PG. EG and LA reached a plateau
at 0.43 M and 0.27 M, respectively, after 8 h. When the reaction time was extended, EG was
not stable under the reaction conditions and was gradually consumed after 27 h (i.e. total
conversion) whereas LA concentration remained constant. Meanwhile, the PG concentration
increased continually up to 52 h, overtaking LA concentration, while GLY concentration
remained around 0.05 M up to 24 h and declined afterwards. A small amount of formate (FA)
was generated at the beginning of reaction, it reached a maximum of 0.03 M after 3 h and was
completely consumed thereafter.
The evolution of selectivities as a function of xylitol conversion in the presence of Ca(OH)
shown in Figure 4(b). Product distribution was essentially unaltered until 80% conversion of
xylitol. The C /C products molar ratio was approximately constant at 1.25. As the reaction
proceeded, the selectivity towards EG, LA, and GLY decreased from 33%, 32% and 4% to
3%, 24% and 1% at full conversion, respectively, whereas the selectivity to PG and products
in the gas phase increased sharply from 22% and 10% to 26 and 25%, respectively.
Consequently, the C /C ratio gradually reached 1.6. A lower production of gaseous products
should be noted by comparison with the reaction without base additive (Figure 3(b) and Figure
(b)); indeed, in the presence of Ca(OH) the selectivity towards gaseous products was 20% at
total xylitol conversion compared to 55% in the absence of base. Furthermore, GC analysis of
2
is
3
2
2
3
2
4
2
the gas phase indicated that the degradation products consisted mainly of CH (94.6% of carbon
4
1
3
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201700034
in the gas phase products), C
was not detected. After reaction, the solid was recovered by filtration, washed, dried and
analyzed by XRD. The phase CaCO was detected in addition to the catalyst. During the
reaction in basic medium, Ca(OH) must react with CO and formed CaCO
2
H
6
(2.8%) and more CO
2
(1.6%). Presence of C
4
and C alkanes
5
3
2
2
3
.
1
4
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201700034
(a)
100
0.40


80


0.30




60











0.20
40





0
.10

2
0















 







25








0.00      


0
0
5
10
15
20
30
time / h
(b)
60

6.0
5
.0
50


2.0

40



1.5
1.0




3
0



 











 
20







0.5
10








 






 0.0


0


  
30
40 50 80
90
100
xylitol conversion / %
Figure 5. Hydrogenolysis of xylitol over 7.1%Ru/C with addition of 0.25 M Ca(OH)
2
base: (a) xylitol conversion,
concentration profiles, and TOC as a function of reaction time; (b) products distribution as a function of xylitol
conversion. () TOC measured; () Xylitol conversion; () BDO; () GLY; () PG; () EG; () FA; ()
LA; () gas products; () C3/C
Reaction conditions: xylitol 0.7 mol.L (10 wt%); 135 mL H
; 200°C.
2
products molar ratio.
-1
2
2
O; 0.5 g 7.1%Ru/C catalyst; 2.5 g Ca(OH) ; 60 bar
H
2
In order to examine the stability of the formed products in the presence of Ru-based catalyst in
aqueous solution, the hydrogenolysis of xylitol was conducted over the catalyst containing a
1
5
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201700034
higher Ru loading (7.1 wt%) under the same reaction conditions in the presence of 0.25 M
-1
Ca(OH) . The results are shown in Figure 5. The activity (202 h , normalized to molar Ru) was
2
-1
quite similar to that of 2,9wt% Ru/C (223 h ) and the reaction was complete after 27 h.
However, up to 80% conversion, the selectivities to EG (27%), LA (29%) and PG (20%) over
7
,1%Ru/C were slightly lower than the ones observed previously over 2.9%Ru/C (33%, 32%,
and 22%, respectively). The C /C ratio was 1.45, whereas the selectivity to gaseous products
3
2
was higher (20% instead of 7% at 80% conversion of xylitol). Beyond 80% conversion, the
selectivity to EG dropped, and in parallel the selectivity to gaseous products increased,
confirming that EG was converted to gas products over Ru/C.
To further investigate the degradation of EG and identify the products formed, additional
reactions were performed using EG as reactant. The reactions were conducted over 2.9%Ru/C
with and without Ca(OH)
2
and the results are shown in Figure 6. The catalytic activity observed
-1
for the conversion of EG was three times lower in the presence of the base (67 h ) than without
-1
base (194 h ); a similar ratio was obtained for xylitol reaction. In neutral medium, EG generated
methanol (MeOH) as the sole product in liquid phase. MeOH was also observed by Ooms et al.
during the conversion of aqueous solutions of sugars over Ni-W C/AC at 245°C under 60
2
[
55]
bar. In this study, MeOH was then degraded to products which were transferred to the gas
phase. Accordingly, TOC analysis of the reaction medium did not detect any carbon left after
7
h, showing that all carbon initially introduced was transferred to the gas phase. In alkaline
medium, EG yielded MeOH, small amounts of FA and traces of glycolic acid. Once again,
when analyzing the solid residues at the end of the reaction, we could identify CaCO3.
100





8
0


60

40


20



















0

0
1
2
3
4
5
6
7
8
9
time / h
1
6
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201700034
Figure 6. Hydrogenolysis of EG over 2.9%Ru/C in the absence (dotted line) or presence of 0.25 M of Ca(OH)
base (full line): () EG; () methanol (MeOH) () FA; () products in gas phase. Reaction conditions: EG 0.45
2
-1
2 2 2
mol.L (2.7 wt%); 135 mL H O; 0.5 g 2.9%Ru/C catalyst; 0 or 2.5 g Ca(OH) ; 60 bar H ; 200°C.
Based on these results, we propose a reaction network under neutral conditions depicted in
Scheme 3. First, the results clearly confirm that epimerization and decarbonylation reactions of
alditols are predominant reactions in neutral medium over Ru/C.[
16,37,52]
Both reactions start by
an initial dehydrogenation step of xylitol on metallic sites.
Xylitol
OH OH
HO
3
OH
2
2
1
1
OH
2-positions
3-position
2'-
position
3-position
CO
-
H2
OH OH
OH OH
4
'-
position
HO
3'
5' OH
HO
3
OH
+
H2
2'
4'
2
2
1
'
2-
positions
1
1
OH
OH
Adonitol
CH4
Arabitol
-
CO
- CO
-
CO
-
H2
- H2
- H2
OH
OH
OH
OH
HO
HO
OH
Erythritol
OH
Threitol
BDO + GLY + PG + EG+ CO
H2
C H + C H + C H + CH
4
4
10
3
8
2
6
Scheme 3. Reaction network in the absence of a base: possible epimerization and decarbonylation reactions of
xylitol. NB: the deoxygenation reaction yielding C 12 is not shown due to the very low selectivity observed.
5
H
Let us consider the differences in selectivity between the different positions for epimerization.
For sorbitol, Hausoul et al. showed that the formation of an aldose intermediate by
dehydrogenation of the hydroxyl groups located at terminal carbons was easier than the
formation of the 2-ketose, and even more than the formation of the 3-ketose, i.e.
dehydrogenation of an internal hydroxyl group.[ Similar reactions for xylitol dehydrogenation
would lead to xylose, xylulose and 3-pentulose, which would then be readily hydrogenated
towards xylitol, arabitol and adonitol, respectively. In the present study, the higher selectivity
to arabitol compared with the selectivity to adonitol confirms the proposed epimerization
mechanism.
52]
1
7
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201700034
Further, the different decarbonylation reactions occurring during xylitol reaction require that
the aldose intermediates adsorbed on Ru undergo terminal C-C bond cleavage to yield carbon
[
16]
monoxide and a shorter-chain alditol. Accordingly, the formation of threitol was observed
from the beginning of xylitol reaction, while the formation of erythritol occurred when the
selectivity to arabitol started declining, at around 50% conversion (Figure 3(a)). Moreover, we
should note that threitol was the major product relative to erythritol, which can be explained as
following. Threitol is formed either by xylitol or arabitol decarbonylation; on the other hand,
erythritol can be formed by arabitol or adonitol decarbonylation (Scheme 3). Moreover, since
the data evidenced that C alditols were produced with different yields, the lower production of
4
adonitol consequently leads to a lower extent of decarbonylation to erythritol. Subsequently,
the formation of compounds with shorter carbon chains, such as GLY or EG, results from
terminal chain decarbonylation, as proposed by Van der Klis et al.[ Such reactions result in
high production of gaseous products, as observed at the end of reaction. In addition, the
54]
formation of BDO is explained by the subsequent dehydration reaction of C alditols (threitol
4
and erythritol) via butanetriols; the latter were observed in very small amounts, making
quantification difficult. Similarly, PG was produced by dehydration/hydrogenation of glycerol
or dehydrogenation/ decarbonylation of 1,2,3-butanetriol.
In the presence of the base, the retro-aldol reaction is considered as the main reaction in the
hydrogenolysis of polyols (Scheme 4).[
27,37,56]
A dehydrogenation step is also required before
the C-C bond cleavage, which is catalyzed by the hydroxide ions. Starting from xylitol, the
dehydrogenation in 1- (step 1.c) and 2- positions (step 1.b) produces C and C final products,
while dehydrogenation in 3-position (step 1.a) yields C and C final products. According to
2
3
[51]
1
4
[
16]
Deutsch et al., the retro-aldol reaction occurring on 3-pentulose is negligible; combined with
the slow dehydrogenation rate in 3- position of xylitol (step 2.a), this explains the low formation
of C
discriminate the predominance of the retro-aldol reaction in 1-position or 2-position of xylitol
steps 2.b and 2.c) since both of them yield C and C products.
4
and C products in the present study (Figures 4 and 5). However, it was not possible to
1
(
2
3
1
8
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201700034
OH OH
HO
OH
OH
+
H2
- H2
Xylitol
-
H2
+ H2
+
H2
- H2
(1.a)
(1.c)
OH OH
OH OH
(
1.b)
O
OH
HO
OH
O
OH
HO
OH
OH
O
Xylose
3-pentulose
OH
Xylulose
-
-
OH
(2.a)
OH
(2.c)
-
OH
(2.b)
OH
O
O
O
+
OH
OH
H
O
HO
+ HO
+
O
OH
OH
(1.d)
OH
GOA
OH
H
O
DHA
GOA
(1.f)
GEA
(1.h)
(
4.a)
(
1.g)
(
1.e)
(1.f)
(3)
-
H2
+ H2
+
H2
-
- H2 + H2
- H2 + H2
- H2
+
H2
- H2 + H2
- H2
+
OH
- H O
2
H
-
O
HO
OH
HO
O
OH
OH
MeOH
+ MeOH HO
HO
OH
O
HO
OH
O
FA
OH
OH
OH
EG
EG
GLY
GLY
OH
(
1.i)
(4.b)
+ OH
-
Threitol -
Erythritol
- 2 H2
+
2 H2
HO
OH
O
O
CO2
CO2
CaCO3
+ PG
OH
-
+
PG
CH + C H
2
6
LA
4
Scheme 4. Possible reaction pathways occurring during hydrogenolysis of xylitol over Ru/C catalyst with 0.25 M
Ca(OH) base. (1) dehydrogenation/hydrogenation equilibrium, (2) retro-aldol reaction, (3) dehydration, (4)
Cannizzaro-type disproportionation. GOA: glycolaldehyde, DHA: dihydroxyacetone, GEA: glyceraldehyde.
2
Upon prolonged reaction times, the C
3
/C
2
molar ratio was higher than 1 owing to the
degradation of EG. We propose that EG was dehydrogenated on Ru metal sites and cleaved into
[
57]
two molecules of formaldehyde, as described by Shabaker et al. Ooms et al. also observed
the conversion of EG into gaseous products (carbon selectivity 86%) and methanol (MeOH)
(
selectivity 14%) over a Ni-W
2
C/AC catalyst in neutral aqueous solution under similar
[
55]
temperature and pressure conditions.
Formate (FA) was formed by Cannizzaro-type
disproportionation (step 4.a) in basic media while MeOH was formed by hydrogenation of
formaldehyde (step 1.d) or as a by-product from the step 4.a. MeOH and FA were then
converted to gaseous products, such as CO
methanation. The concentration of formate decreased after 5 h and was no more present when
the pH became neutral; formaldehyde was then converted to CH and CO , as confirmed by the
increase of selectivity to gaseous products at 72% xylitol conversion (Figure 5). Furthermore,
the C /C molar ratio remained constant around 1.25 and then increased at the end of the
reaction beyond 72% xylitol conversion. This results in EG degradation, while C products were
stable. The use of a catalyst with a higher Ru loading in the presence of 0.25 M Ca(OH) clearly
showed EG degradation and the increase of the C /C molar ratio after full conversion of xylitol
Figure 5).
2
(partly converted to CaCO
3
) and CH by
4
4
2
3
2
3
2
3
2
(
1
9
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201700034
2
.3 Influence of the amount of Ca(OH) additive over monometallic Ru/C catalyst
2
We investigated the catalytic performances in term of activities (Table 3) and selectivities
Figure 7) of 2.9%Ru/C for the hydrogenolysis of xylitol in the presence of different
(
concentrations of Ca(OH)
2
-
. The base concentrations were in the range of 3.1 mM (31 mg) to
OH
5
0 mM (500 mg), i.e. OH / xylitol molar ratio Rmol( /xylitol) varied from 0.009 to 0.13. The
results were compared to those obtained previously in neutral medium and in the presence of
.25 M Ca(OH) . The selectivities were given at 80% xylitol conversion, before the formation
of significant amounts of gaseous products due mainly to the degradation of EG.
0
2
Upon introduction of even a very low amount of Ca(OH)
2
(3.1 mM), the catalytic activity
-1
-1
significantly decreased from 727 h (in the absence of base) to 67 h (Table 3). The activity
remained approximately at that value upon addition of increasing amounts of the base up to 25
-1
-1
mM. It then increased to 93 h with addition of 50 mM of Ca(OH) and to 223 h with
2
introduction of 250 mM of Ca(OH)
reported that the addition of Ca(OH)
range 0.07-1 during sorbitol hydrogenolysis over Ni/C.[
2
. This is in agreement with the literature where it has been
OH
2
enhanced the conversion for Rmol( /sorbitol) ratio in the
15,27]
The addition of small amounts of
base inhibited the reaction, mostly the epimerization reaction to arabitol and adonitol. However,
higher amounts increased xylitol conversion by catalyzing the retro-aldol reaction.
Table 3. Activity, C
3
/C
2
products molar ratio and carbon balance in the presence of different amounts of Ca(OH)
2
base[
a]
Carbon balance[b] /%
Ca(OH)
mM]
2
Activity
[h ]
C
3
/C
2
products
Rmol(OH
/
)
xylitol
-
1
[b]
[
molar ratio
TOC
HPLC
None
0.00
727
3.2
83
80
3
6
.1
.7
0.009
0.018
0.033
0.066
0.13
67
65
1.7
1.5
1.4
1.4
1.4
1.3
79
83
85
86
93
91
74
77
77
81
81
90
1
2
5
2.2
4.4
0.0
67
75
93
2
50.0
0.68
223
[
a]
Reaction conditions: xylitol 10 wt%, 135 mL H
bar H , 200°C.
at 80% xylitol conversion.
2 2
O, 0.5 g 2.9%Ru/C, 0-2.5 g (0-250 mM Ca(OH) ), 60
2
[b]
The product distributions at different Ca(OH) loadings are compared in Figure 7. The
2
detailed data are reported in Table S1. The addition of the base using a hydroxyl/xylitol molar
ratio of 0.009 was sufficient to initiate C-C cleavage; the yield of glycols (EG + PG) was six
times larger than without base. Moreover, the combined selectivity to C
adonitol) and C alditols (threitol + erythritol) was divided by two (32% instead of 65%).
Thereafter, with increasing addition of Ca(OH) , this combined selectivity continued to
decrease. The concentrations of C alditols and C products were nearly zero at 50 mM Ca(OH)
when glycols reached a maximum of selectivity of 56%. The introduction of higher quantities
5
alditols (arabitol +
4
2
5
4
2
2
0
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201700034
of base in the medium (Rmol(OH/xylitol) > 0.13) induced the production of LA via Cannizzaro-
type disproportionation at the detriment of PG and GLY. It may also be noted that increasing
the amount of base significantly improved the carbon balance from 79% to 91%.
Without base addition, the quantity of products followed the order C
indicates that epimerization and decarbonylation reactions were favored. Moreover, the C
molar ratio (Table 3) was 3.2, indicating that no retro-aldol reaction occurred. Even with a very
low amount of base promoter (3.1 mM), the selectivity to C and C products declined and the
molar ratio C /C decreased to 1.7 as a result of the competition between epimerization and
decarbonylation reactions on one hand, and retro-aldol reaction on the other hand. With
increasing amounts of Ca(OH) this ratio decreased slowly until 1.3 for 250 mM of Ca(OH)
No by-products from decarbonylation and epimerization were then observed. Liu et al. studied
the hydrogenolysis of xylitol over Cu-SiO and obtained a similar C /C molar ratio of 1.2 (at
0% conversion) when working under similar reaction conditions: Ca(OH) base
5
> C
4
> C
3
> C
2
which
3
/C
2
5
4
3
2
2
2
.
2
3
2
9
2
OH
[34]
(Rmol( /xylitol) = 0.42), 200°C under 60 bar H .
2
EG
PG
GLY
LA
C4 alditols
BDO
C5 alditols
100
80
60
40
20
0
0.0
3.1
6.7
12.2
24.4
50.0
250.0
Ca(OH) / mM
2
Figure 7. Selectivities in liquid phase at 80% xylitol conversion using different amounts of Ca(OH)
alditols: arabitol and adonitol; C alditols: threitol and erythritol. Reaction conditions: xylitol 10 wt%, 135 mL H
.5 g Ru/C, 0-2.5 g (0-250 mM) Ca(OH) base, 60 bar H , 200°C.
2
promoter. C
5
4
2
O,
0
2
2
Therefore, the use of a high Rmol(OH/xylitol) promotes the production of C
and C products in
2
3
the hydrogenolysis of xylitol. However, it is also associated with a switch in selectivity towards
LA to the detriment of PG. To avoid or to minimize the production of LA, the hydrogenolysis
reaction of xylitol was performed on bifunctional catalysts, by modifying the Ru/C catalyst with
a basic oxide.
2
1
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201700034
2
.4 Promotion of xylitol hydrogenolysis by Ru/metal oxide/C catalyst without Ca(OH)
2
addition
Lactate was produced as a major by-product during xylitol hydrogenolysis under alkaline
conditions, obtained when the base promoter Ca(OH) was added to the Ru/C system.
According to previous reports, co-deposition of Ni and a solid base (CaO, CeO ) on active
carbon was an efficient way to promote the C-C cleavage of xylitol into glycols with negligible
2
2
[
35]
LA production. In this study, we synthesized catalysts by sequential deposition of a metal
oxide and ruthenium and we investigated their catalytic response for the hydrogenolysis of
xylitol with and without the addition of the base Ca(OH)
.9%Ru/C and bifunctional Ru/metal oxide/C catalysts and the results, in term of activity and
selectivity (on a carbon basis) to the different products, are given in Table 4. Among the
2
. The reactions were conducted over
2
[
38,39]
[41,42]
promoters investigated, ZnO,
SrO and MnO
are considered as basic oxides, WO as
x
an acidic oxide and SnO as a Lewis acid oxide which can promote the retro-aldol reaction of
x
[
40]
aqueous glucose solution.
Without addition of Ca(OH)
2
promoter, with regard to the effect of the metal oxide, the
modified catalysts were less active in xylitol hydrogenolysis compared with the Ru/C catalyst.
All the oxide-promoted catalysts exhibited an activity at least three times lower than Ru/C: 220
-1
-1
-1
-1
h for Ru/MnO/C, 115 h for Ru/WO
x
/C, 84 h for Ru/SrO/C and 13 h for Ru/ZnO/C,
compared with 727 h for Ru/C. No clear direct correlation between initial reaction rate (Table
) and total basicity (Table 4) could be established for the oxide-modified Ru catalysts. The
-1
1
MnO-modified catalyst, which revealed the highest number of total basic sites was the most
active; however, the ZnO modified catalyst with significant basicity demonstrated a very low
activity. Further characterization of the materials should be performed to identify the active
sites. We should note that Ru/SnO
conversion. The reason could be due to the presence of SnCl
decomposed to tin oxides, as shown previously by TEM-EDX analysis. Another reason could
be the pre-treatment conditions (reduction of catalyst by H gas at 450°C) that could reduce tin
oxide to some metallic tin. Indeed, it has been shown that TPR experiments of Sn catalysts
x
/C catalyst displayed no catalytic activity towards xylitol
2
which was not completely
2
[
48,58]
[59]
supported on active carbon
, alumina or zirconia showed H uptakes in the temperature
2
4+
2+
2+
0
range of 270°C to 500°C, corresponding to the reduction of Sn to Sn and Sn to Sn . This
could explain the absence of reaction in the presence of the SnO-promoted catalyst.
As for the products distribution, the selectivity profiles were significantly different and
dependent on the nature of the metal oxide. Ru/ZnO/C catalyst significantly favored the C-C
cleavage as it exhibited the best selectivity to C
and the lowest selectivity to arabitol (17%). This behavior is in agreement with the relatively
high amount of basic sites measured by CO -TPD (Table 1). However, the results should be
2
and C
3
products (64% combined selectivity)
2
taken with caution. The selectivities to the different products depend strongly on xylitol
conversion (Table 4). Or, due to the very low activity of this catalyst, the data are given at 40%
xylitol conversion, instead of 80%. Ru/MnO/C and Ru/SrO/C, with significant amount of basic
sites (Table 1), were also able to cleave C-C bond to produce glycols with high combined
2
2
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201700034
selectivity to the glycols (23% and 18%, respectively) and to glycerol (8% and 7%,
respectively), whereas the selectivity to arabitol was 17% and 20%, respectively. In contrast,
x
Ru/WO /C favored the isomerization reaction to arabitol, the decarbonylation to threitol and
erythritol, and the cyclization of xylitol to 1,4-anhydroxylitol by a dehydration reaction, known
[60]
to occur on acidic WO sites.
x
When the hydrogenolysis reaction was performed in the presence of 0.25 M Ca(OH)
2
, the
addition of a metal oxide modifier on the solid also diminished the activity of the Ru catalyst
-1
-1
-1
-1
but to a lesser extend: from 223 h (Ru/C) to 178 h (Ru/WO
(
x
/C), 90 h (Ru/MnO/C), 88 h
Ru/SrO/C), and 13 h (Ru/ZnO/C). Nevertheless, the influence on the C-C bonds cleavage to
and C products was moderate in the presence of Ca(OH) base, with a global selectivity to
-1
C
2
3
2
the expected glycols remaining approximately 53% for all catalysts. However, Ru/ZnO/C
catalyst behaved differently and produced slightly more LA and less glycols than the other
catalysts.
In summary, in the absence of base, most of the metal oxide-promoted catalysts investigated in
this work were active and more selective than Ru/C for the hydrogenolysis of xylitol to glycols.
However, their selectivity to glycols remained lower than in the presence of Ca(OH) base. The
2
interesting point is the formation of only traces of LA under neutral conditions. Unfortunately,
an important leaching of the metallic oxide was observed in non-alkaline solutions at the end
of the reaction, i.e. 15% of W, 60% of Zn, 71% of Mn, and up to 100% of Sr, making
problematic the catalyst recycling.
Table 4. Activity and product distribution in the hydrogenolysis reaction of xylitol over Ru/metal oxide/C catalysts
in the absence or presence of Ca(OH) base
2
Carbon selectivity
[%]
Carbon balance[a]
[%]
Activity
Catalyst
-
1
[
h ]
Glycols
EG+PG)
GLY
LA
Arabitol Others [b]
TOC
HPLC
(
2
1
2
1
2
3
2
2
1
2
3
.9%Ru/C
727
0
4
5
-
0
-
30
-
32
-
83
100
87
81
90
84
91
96
95
97
90
80
100
86
72
80
71
90
88
95
90
90
.6%Ru/SnO
x
(5.0%)/C
-
.6%Ru/ZnO(4.1%)/C[a]
.8%Ru/SrO(2.1%)/C
13
56
18
12
22
53
42
53
52
53
8
trace
trace
trace
trace
27
17
20
23
17
0
5
84
7
27
40[c]
22
4
.7%Ru/WO
.1%Ru/MnO(4.5%)/C
.9% Ru/C + Ca(OH)
.6%Ru/ZnO(4.1%)/C + Ca(OH)
.8%Ru/SrO(2.1%)/C + Ca(OH)
.7%Ru/WO (5.0%)/C + Ca(OH)
.1%Ru/MnO(4.5%)/C + Ca(OH)
x
(5.0%)/C
115
220
223
32
5
10
4
2
2
0
39
0
0
2
88
0
32
0
0
x
2
178
90
4
30
0
0
2
0
34
0
0
Reaction conditions: 10 wt% xylitol; 200°C; 60 bar H
at 41% xylitol conversion after reaction time 32 h.
butanediols, threitol, and erythritol
2
; 0.5 g catalyst, Ca(OH)
2
0 and 0.25 M, at 80% xylitol conversion.
[a]
[b]
[c]
other products include butanediols, threitol, erythritol, adonitol, and 1,4-anhydroxylitol.
2
3
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201700034
2
.5 Further insight in Mn promotion
From the screening of bifunctional catalysts in neutral medium, it appeared that
.1%Ru/MnO(4.5%)/C exhibited the highest activity and an interesting selectivity to glycols
3
and glycerol, when working in neutral conditions. In order to get greater understanding of the
nature of the active Mn species and the effect of MnO on the parameters responsible for C-C
bond cleavage and to get a better control and prevent the leaching, the influence of Mn loading
of 3%Ru/MnO(y%)/C (y in the range 0-19.3 wt%) was investigated without Ca(OH)
2
additive.
The results are displayed in Table 5 and Figure 8.
Table 5. Activity, C
3
/C
2
products molar ratio, carbon balance, and Mn leaching of 3%Ru/(y%)MnO/C (y=0-19.3
base.
wt%) as a function of MnO loading, in the hydrogenolysis of xylitol without addition of Ca(OH)
2
Carbon balance[a]
[%]
Molar ratio
/C
products
Mn
Leaching
[%]
Activity
Catalyst
C
3
2
-
1
[
h ]
[a]
TOC
HPLC
2
.9%Ru/C
727
123
220
344
384
n.m.
3.2
83
80
-
3
3
.0%Ru/MnO(1.3%)/C
.1%Ru/MnO(4.5%)/C
1.9
1.5
1.4
1.5
86
84
85
71
63
71
47
68
15
2
3
.9%Ru/MnO(13.4%)/C
.1%Ru/MnO(19.3%)/C
MnO(4.7%)/C
83
77
83
83
-
100
100
[
a]
data taken at 80% xylitol conversion. n.m. not measurable.
2
Reaction conditions: xylitol 10 wt%, 135 mL H O, 0.5 g catalyst, 60 bar H , 200°C.
2
MnO(4.7%)/C catalyst displayed a very low conversion (5% conversion after 30 h). Compared
-1
to Ru/C, the activity dropped down to 123 h with 1.3% MnO loading on the carbon support.
-1
Afterwards, it gradually improved with increasing MnO loading and reached 384 h over
%Ru-MnO(19.3%)/C. As far as we know, this catalyst is one of the most active catalysts
reported in the literature in absence of base for Ru based-catalysts in hydrogenolysis of
3
[21]
alditols. As it was noted previously in free-base conditions for the different additives, the pH
value decreased from ca. 7 to ca. 4.3 for Ru/MnO(y%)/C, while it decreased only to 6 in the
presence of MnO(4.7%)/C when no reaction occurred.
The selectivities to glycols and glycerol (Figure 8, Table S2) increased also with increasing
MnO content, going from 19% to 35% for the glycols (EG+PG) and from 8% to 18% for
glycerol. Meanwhile, the selectivity to C
5
alditols dropped from 40% to 10% and the selectivity
to C alditols decreased to 10%. These results corroborate the ability of MnO-promoted catalyst
4
to selectively cleave the C-C bonds. Furthermore, MnO/C catalyst without Ru demonstrated no
activity, suggesting, as expected, that MnO has no role on the initial dehydrogenation reaction.
Notably, only traces of LA were measured, whatever the Mn loading. The percentage of Ru
leaching at the end of reaction was less than 0.15% for all reactions, whereas the leaching of
2
4
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201700034
Mn was in the range 47 to 75% independently of the amount of MnO introduced, except for
MnO(4.7%)/C where only 15% of Mn leaching was observed.
EG
PG
GLY
LA
C4 alditols
BDO
C5 alditols
100
80
60
40
20
0
2.9%Ru/C
1.3
4.5
13.4
19.3
Ru/MnO(y%)/C
Figure 8. Selectivities to products over 3%Ru/ MnO (y%)/C (y=0-19.3 wt %) without addition of Ca(OH)
alditols (arabitiol+adonitol) and C alditols (threitol+erythritol). Reaction conditions: xylitol 10 wt%, 135 mL
O, 0.5 g 3%Ru- MnO(y%)/C catalyst, 60 bar H , 200°C, 80% xylitol conversion.
2
base.
C
5
4
H
2
2
The high level of Mn leaching limits the reusability of the MnO-promoted catalyst. After
washing the catalyst with water and drying it at 60°C under N , a second run was conducted.
2
-1
-1
A drop in activity (Figure S8), from 220h to 80h , was observed. The product distribution
was similar but 88% of the Mn remaining at the end of the first run leached during the second
run.
Therefore we decided to investigate in more details the nature and the effect of Mn species on
the hydrogenolysis of xylitol and more specifically on Mn leaching. The catalytic performances
of physical mixtures of Ru/C catalyst with MnO, MnO
2
, and a Mn(NO
3
)
2
aqueous solution, are
summarized in Table 6. All reactions were performed introducing the same amount of Mn
x
species, i.e. 22.5 mg, in the reaction medium, except for Ru/MnO . The results are shown in
Table 6.
Table 6. Influence of Mn species (22.5 mg Mn) on the performance of the Ru catalyst in the hydrogenolysis
reaction of xylitol.
Carbon selectivity[a]
[%]
Carbon balance[a]
[%]
Mn
leaching
[%]
Activity
Catalyst
-
1
[
h ]
Glycols
EG+PG)
C
5
C
4
GLY
LA
BDO
TOC
HPLC
(
alditols
alditols
2
5
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201700034
3
.1%Ru/MnO(4.5%)/C
220
150
170
160
300
22
26
24
30
27
10
13
12
15
10
traces
traces
traces
traces
3
19
15
15
10
7
17
19
19
14
13
3
-
84
85
85
81
72
71
78
75
78
66
71
64
66
-
2
.9%Ru/C + MnO
2
.9%Ru/C + MnO
2
-
2
.9%Ru/C +Mn2+ [b]
-
c]
2
.8%Ru/MnO [
x
-
63
Reaction conditions: 10 wt% xylitol; 200°C; 60 bar H
at 80% conversion. introduction of 1.71 mL Mn(NO ) aqueous solution (14,5 g L ); introduction of 0.3 g catalyst (227.5 mg Mn).
2
; 0.5 g catalyst.
3 2
[a]
[b]
-1 [c]
-1
The catalytic activities remained in the range of 150-220 h when the different Mn species were
introduced with the Ru/C catalyst. The products distributions were also similar and only traces
of LA were detected. The epimerization selectivity was higher when Mn was deposited on the
support as Mn oxide (19%) than when Mn was added as a physical mixture of Ru/C and MnO
or MnO (15%), or as a salt in the aqueous reaction medium (10%). Mn leaching was important
in the range of 66% to 71%, regardless of the amount of Mn introduced.
-1
Over the Ru/MnO catalyst, the activity was higher (300 h ), suggesting a better proximity and
x
cooperativity between Ru and Mn. Regarding the products distribution, the selectivities towards
glycols and glycerol were similar to the other catalysts, while the selectivity to LA was
detectable and attained 3%. No Ru leaching was observed, although the Mn leaching was of
6
3%.
2
+
These results confirm that the association of Mn ions with Ru/C may be an efficient catalytic
system for the C-C bond cleavage of xylitol into C -C products.
2
3
The Mn species responsible for the synergy of activity for C-C cleavage may be either MnO on
the solid or Mn ions leached into the reaction medium. Over the MnO(4.7%)/C catalyst,
leaching of Mn was of only 15% and no conversion of xylitol could be observed, suggesting
that the operating conditions were not responsible for the elution. In contrast, over the
Ru/MnO/C catalysts, when xylitol was converted, leaching was high suggesting that Ru is
required for selective reaction and that Mn leaching was due to the formation of some reaction
products. Leaching of MnO has been reported previously under similar conditions.[
28,41]
Indeed,
in the hydrodeoxygenation of guaiacol in aqueous phase at 160°C under 15 bar H
2
over a Ru-
MnO
presence of the acidic substrate and intermediates (phenolic compounds).[ Recently, Murillo
Leo et al. similarly explained the high leaching of Ca(OH) used as the support for Ni and Ru
x
/C catalyst, Ishikawa et al. attributed the rapid Mn elution during the reaction to the
41]
2
catalysts during sorbitol hydrogenolysis to the formation of acidic by-products (as demonstrated
by the change of pH at the end of the reaction).[
28]
The comparison of the concentration of LA produced and the amount of Mn leached as a
function of time during the reaction over a 3%Ru/MnO(5.6%)/C catalyst (Figure 9) shows some
correlation between the two values. The amounts of Mn leached and of LA produced attained
a plateau after 5 h and remained constant up to 30 h of reaction. The molar ratio LA/Mn varied
in the range 1-1.8. In addition, it was observed that the introduction of the Ru/MnO/C catalyst
2
6
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201700034
in a LA aqueous solution (3.81 mM) at room temperature immediately resulted in the leaching
of Mn. These results demonstrate the link between the formation of acidic products and Mn
leaching, even at very low concentration of lactate (selectivity < 1%).
0.6








0.4


0.2
0
.0 
0
5
10
15
20
25
30
35
time /h
Figure 9. Comparison between leached Mn () and produced LA () as a function of time for the hydrogenolysis
of xylitol over Ru/MnO/C catalyst.
Reaction conditions: xylitol 10 wt%; 135 mL H O; 0.5 g 3%Ru/MnO(5.6%)/C catalyst; 60 bar H ; 200°C.
2 2
3
. Conclusions
In summary, the addition of a homogeneous base has a strong effect on the xylitol
hydrogenolysis mechanism over Ru/C. Without base, fast epimerization and decarbonylation
reactions occurred until 80% xylitol conversion. Afterwards, the formed epimers underwent
cascade decarbonylation which then produced gaseous light alkanes at full conversion. The
addition of Ca(OH)
retro-aldol reaction. The C
2
base, even in small quantities, limited these reactions and promoted the
/C molar ratio of 1.25 is superior to the theoretical value of 1 as EG
3
2
reacts over Ru catalyst and forms light alkanes such as CH . Increasing the amount of base, up
4
OH
to Rmol( /xylitol) = 0.68 improved the selectivity to glycols (EG+PG) but the selectivity to by-
product LA (27%) exceeded those to PG and GLY (25% and 24%, respectively). To avoid the
production of LA, Ru/MnO/C catalyst, which exhibits the highest amount of basic sites (CO
2
-
-
TPD), was the most efficient bifunctional catalyst in absence of base, with high activity (220 h
1
)
and selectivity to glycols (22%) and glycerol (10%), and only traces of LA. The catalytic
response, in term of activity and selectivity towards glycols, improved with the amount of MnO.
However, Mn leaching up to 70% was observed, associated with the traces of LA formed even
in the absence of added base. The leaching of metal promoter remains a challenge for further
works, it is essential to reduce the formation of acidic derivatives.
2
7
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201700034
4
. Experimental Section
4
.1 Preparation of Catalysts
The monometallic Ru/C catalysts were prepared by wet impregnation using RuNO(NO
>31.3%, Alpha Aesar). An aqueous solution containing 0.15 g or 0.35 g of Ru in 15 mL of
distilled water (for 3wt%Ru/C or 7wt%Ru/C, respectively) was added to a suspension of active
3
)
3
.xH
2
O
(
2
-1
3
-1
carbon (L3S CECA, Table 1, SBET = 1095 m .g and Vpores = 0.81 cm .g ), 5 g in 20 mL water,
in a 250 mL round-bottom flask. After stirring for 6h at room temperature, water was removed
in a rotary evaporator under 100 mbar at 70°C. The resulting powder was dried at 60°C
overnight in an oven under N
2
-1
atmosphere. The solid was reduced under H
2
flow (100 mL.min-
gas flow for 30 min at
1
)
at 450°C for 3 h (2°C.min ) and finally passivated under 1% O
2
/N
2
room temperature.
The bifunctional Ru/metal oxide/C (denoted as Ru/MOx(y)/C, where the y in parenthesis
represents the measured M wt%) catalysts were prepared by successive wet impregnation. First,
the active carbon in suspension in water was impregnated with an aqueous solution of the
metallic precursor salt ([SnCl
Zn(NO .6H O], 99% AppliChem; [Sr(NO
After evaporation of water and further drying overnight at 110°C, the solid was calcined at
4
.5H
2
O], 98% Aldrich; [(NH
4
)
6
2
H W
12
O40.xH
2
O], 99% Fluka;
[
3
)
2
2
3
)
2
], 99% Fluka; [Mn(NO
3
)
2
.4H
2
O], 98% Merck).
-1
-1
2
00°C for 4 h (2°C.min ) under air flow (100 mL.min ). Subsequently, the wet impregnation
of Ru and the reduction were carried out as described above. The amount of Ru was calculated
to obtain a M/Ru mass of approximately 1.5. Ru/MnO/C catalysts containing higher Mn
contents were synthetized by increasing the concentration of Mn(NO
3
)
2
.4H O in the aqueous
2
solution.
A Ru/MnO
procedure described elsewhere. The MnO
using 45 mL of 0.5 M Mn(NO .4H O aqueous solution and 50 mL of 0.5 M Na
x
catalyst was prepared by incipient wetness impregnation according to the
[
42]
x
support was first prepared by co-precipitation
CO (99.95%,
3
)
2
2
2
3
Sigma Aldrich) aqueous solution at 80°C (pH 9) for 2 h. After filtration and washing with hot
demineralized water, in order to remove any trace of sodium, the solid was dried at 110°C
-1
-1
overnight and calcined under air (100 mL.min ) at 500°C for 5 h (4°C.min ). Then, an aqueous
solution of RuNO(NO .xH O (19 mg of Ru in 1.80 mL) was impregnated on the MnO
support for 6 h at room temperature. The solid was dried at 110°C overnight and reduced under
flow and passivated as previously.
3
)
3
2
x
H
2
4
.2 Catalyst characterization
2
8
This article is protected by copyright. All rights reserved.

ChemCatChem
10.1002/cctc.201700034
The metal loadings of the catalysts were determined after complete dissolution in aqua regia at
50°C for 12h and analyzed by inductively coupled plasma optical emission spectroscopy (ICP-
OES) analysis (Activa Jobin-Yvon). X-ray diffraction (XRD) patterns were recorded using a
Bruker D8A25 diffractometer with CuK radiation and multi-channel fast detector LynxEye
0.04°.s over the range 5 ≤ 2θ ≤ 80°). The average crystallite sizes were determined according
to the Debye Scherrer equation from the full width at half maximum of the peak with the highest
intensity. N physisorption isotherms were obtained at -196°C using a Micrometrics ASAP
020 sorption analyzer and used for the determination of specific area (BET method) and total
1
α
-1
(
2
2
pore volume (at P/P
0
=0.99). Prior to the measurement, the sample was degassed under vacuum
-4
-1
(<10 mbar) at 350°C for 7 h (2°C.min ). TEM images were taken on a JEOL 2010 instrument
operated at 200 keV. Samples were prepared by dispersing the solid in ethanol and then placed
onto carbon-coated grids. The average size of Ru particles and their size distribution were
determined by measuring ca. 300 particles randomly distributed in the images. Energy
dispersive X-ray (EDX) analysis was carried out with an Oxford Link Isis spectrometer. The
surface basicity of the catalysts was determined through temperature-programmed desorption
of CO
2
(CO
2
-TPD). In a typical run, the reduced sample (about 30 mg) was first pre-treated in
-1
He gas (50 mL.min ) at 450°C for 1 h, and then cooled to 100°C. Subsequently, the sample
was saturated with 5% v/v CO
-1
2
/He (25 mL.min ) at that temperature for 30 min. The TPD
profiles were recorded with a thermal conductivity detector (TCD) when heating the sample at
-1
-1
8
°C.min from 100°C to 450°C under a continuous flow of He (50 mL.min ).
4
.3 Hydrogenolysis reaction
Xylitol hydrogenolysis was performed in a 300 mL Hastelloy Parr autoclave. In a typical
experiment, 135 mL of a 10 wt% xylitol (99%, Sigma Aldrich) aqueous solution containing the
Ca(OH) base (96%, Merck) and 0.5 g of reduced catalyst were introduced into the reactor. The
2
sealed autoclave was purged three times with 60 bar of Ar to remove the residual air, pressurized
with hydrogen, and then heated to 200°C under stirring at 700 rpm. Once the reaction
temperature was reached, the pressure was adjusted to 60 bar H
rpm) was applied in order to eliminate the diffusion effects. This corresponds to reaction time t
0. During the reaction, liquid samples were periodically collected through a sampling valve;
2
and vigorous stirring (1000
=
they were filtered over PVDF membranes (0.45 µm) before analysis. After reaction, the
autoclave was cooled down and the catalyst was separated by vacuum filtration over PVDF
membranes (0.45 µm).
4
.4 Products analysis
Liquid reaction samples were analyzed using two separation methods on two Shimadzu HPLC
2
+
apparatus. The first one had a Rezex RCM-Monosaccharide Ca 8% (700 x 7.80 mm) column
2
9
This article is protected by copyright. All rights reserved.