Selective hydrogenolysis of biomass-derived sorbitol to propylene glycol and ethylene glycol on in-situ formed PdZn alloy catalysts

Article abstract of DOI:10.1016/j.apcata.2020.117770

Sorbitol hydrogenolysis to industrially important propylene glycol and ethylene glycol receives increasing attention, recently. Here, we developed an efficient and stable PdZn alloy catalyst, in-situ formed from a physical mixture of Pd/ZrO₂ and ZnO, for the sorbitol hydrogenolysis with Mg₃AlO_x as a solid base, and obtained a 54.6% yield of the two target glycols (493 K and 5.0 MPa H₂). The amounts of ZnO and Mg₃AlO_x strongly influenced the activity and selectivity, due to their effects on the formation of the PdZn alloys and the competitive metal-catalyzed dehydrogenation/hydrogenation and base-catalyzed retro-aldol condensation steps. The kinetic isotope effects, combined with the inhibiting effects of H₂ pressure on the activity in a broad range (3.0–8.0 MPa), confirm that the sorbitol dehydrogenation to hexose intermediates is the kinetically-relevant step in the sorbitol hydrogenolysis. This study provides insights into the catalytic functions and reaction parameters for the hydrogenolysis of polyols to the target glycols.

Full text of DOI:10.1016/j.apcata.2020.117770

Journal Pre-proof
Selective hydrogenolysis of biomass-derived sorbitol to propylene glycol
and ethylene glycol on in-situ formed PdZn alloy catalysts
Yuqing Jia (Data curation) (Investigation) (Writing - original draft),
Qianhui Sun (Data curation) (Formal analysis) (Writing - original
draft), Haichao Liu (Project administration) (Resources)
(Supervision)<ce:contributor-role>Writing - reviewand editing)
PII:
S0926-860X(20)30363-X
DOI:
https://doi.org/10.1016/j.apcata.2020.117770
APCATA 117770
Reference:
To appear in:
Applied Catalysis A, General
Received Date:
Revised Date:
Accepted Date:
19 May 2020
28 July 2020
1 August 2020
Please cite this article as: Jia Y, Sun Q, Liu H, Selective hydrogenolysis of biomass-derived
sorbitol to propylene glycol and ethylene glycol on in-situ formed PdZn alloy catalysts, Applied
Catalysis A, General (2020), doi: https://doi.org/10.1016/j.apcata.2020.117770
This is a PDF file of an article that has undergone enhancements after acceptance, such as
the addition of a cover page and metadata, and formatting for readability, but it is not yet the
definitive version of record. This version will undergo additional copyediting, typesetting and
review before it is published in its final form, but we are providing this version to give early
visibility of the article. Please note that, during the production process, errors may be
discovered which could affect the content, and all legal disclaimers that apply to the journal
pertain.
© 2020 Published by Elsevier.

Selective hydrogenolysis of biomass-derived
sorbitol to propylene glycol and ethylene glycol on
in-situ formed PdZn alloy catalysts
Yuqing Jiaa ,1, Qianhui Sun a, b, 1, Haichao Liua,*
a
Beijing National Laboratory for Molecular Sciences, College of Chemistry and
Molecular Engineering, Peking University, Beijing 100871, China
b Research Institute of Petroleum Processing, SINOPEC, Beijing 100083, China
* Corresponding author. E-mail address: hcliu@pku.edu.cn (H. Liu).
1 These authors contribute equally to this work.
Graphical Abstract
Highlights



In-situ formed PdZn alloy catalyst is efficient and stable for sorbitol
hydrogenolysis to propylene glycol and ethylene glycol.
The presence of ZnO facilitates the in-situ formation of PdZn alloys on ZrO2 and
their stability.
The primary kinetic isotope effects with different deuterated sorbitol molecules
validate the kinetically-relevant step in sorbitol hydrogenolysis.

Abstract: Sorbitol hydrogenolysis to industrially important propylene glycol and
ethylene glycol receives increasing attention, recently. Here, we developed an efficient
and stable PdZn alloy catalyst, in-situ formed from a physical mixture of Pd/ZrO2 and
ZnO, for the sorbitol hydrogenolysis with Mg3AlOx as a solid base, and obtained a 54.6%
yield of the two target glycols (493 K and 5.0 MPa H2). The amounts of ZnO and
Mg3AlOx strongly influenced the activity and selectivity, due to their effects on the
formation of the PdZn alloys and the competitive metal-catalyzed
dehydrogenation/hydrogenation and base-catalyzed retro-aldol condensation steps. The
kinetic isotope effects, combined with the inhibiting effects of H2 pressure on the
activity in a broad range (3.0-8.0 MPa), confirm that the sorbitol dehydrogenation to
hexose intermediates is the kinetically-relevant step in the sorbitol hydrogenolysis. This
study provides insights into the catalytic functions and reaction parameters for the
hydrogenolysis of polyols to the target glycols.
Keywords: Sorbitol; Selective hydrogenolysis; PdZn alloy; Propylene glycol;
Ethylene glycol; Reaction mechanism
1. Introduction
Polyols, represented by sorbitol and xylitol, are renewable and largely available
from non-edible cellulose and hemicellulose. Their transformation to propylene glycol
and ethylene glycol via catalytic hydrogenolysis provides an economically viable route
to the two important glycols, widely used as functional fluids and for the synthesis of
polyester resins, fibers, and pharmaceuticals.1-8 Currently, the two glycols are
manufactured mainly by multi-step transformations from petroleum-based propylene
and ethylene.
Because of such importance, the selective hydrogenolysis of sorbitol and xylitol
has been intensively studied on various metal catalysts in the presence of base
promoters.9-21 Zhou et al. used carbon nanofiber-supported Ru catalyst with CaO for
the sorbitol hydrogenolysis, achieving 85.7% sorbitol conversion and 60.8% combined

selectivity to ethylene glycol, propylene glycol and glycerol at 493 K and 8.0 MPa H2.16
Using NaOH as the base promoter, a total yield of 61.4% for ethylene glycol and
propylene glycol was obtained over the amine-functionalized carbon nanotubes
supported Ru catalysts at 483 K and 6 MPa H2.17 Palkovits and coworkers found that
Cu/C in the presence of Ca(OH)2 efficiently catalyzed the full conversion of sorbitol to
ethylene glycol and propylene glycol in a selectivity of as high as 84.5% at 513 K and
5 MPa H2.18 Sun et al. reported the hydrogenolysis of xylitol to ethylene glycol and
propylene glycol in a 67.3% yield at 473 K and 4 MPa on Ni/C in the presence of
Ca(OH)2.19 Liu et al. observed up to 81.0% yield to ethylene glycol and propylene
glycol in xylitol hydrogenolysis on bimetallic NiCu-SiO2 catalysts with Ca(OH)2 at 473
K and 8 MPa H2.20
To avoid the use of soluble base promoters and reduce the formation of carboxylic
acids, typically lactic acid, known competitively via the same intermediates for the
target glycols, bifunctional catalysts with integrated metallic and basic sites have been
explored in hydrogenolysis of sorbitol and xylitol.22-31 Chaudhari and coworkers
investigated the performance of Cu/CaO-Al2O3 in sorbitol hydrogenolysis without
addition of any other bases at 503 K and 6 MPa H2, and observed a high combined
selectivity to ethylene glycol, propylene glycol and glycerol (73.3% at 98.1%
conversion).22 Wang et al. applied MgO-supported Ni, Cu, and Co catalysts in sorbitol
hydrogenolysis, and showed a 66% combined selectivity for ethylene glycol, propylene
glycol and glycerol at 57.2% conversion on Ni/MgO at 473 K and 6.0 MPa H2.23
Shanthi et al. reported 62% selectivity for ethylene glycol and propylene glycol at 64%
sorbitol conversion on Ni/Ca(Sr)-hydroxyapatites at 473 K and 6.0 MPa H2.24 Rivière
et al. carried out xylitol hydrogenolysis on Ru-(Mn-Al)Ox solid base catalysts, and
achieved 58% combined selectivity to the two glycols and glycerol at 80% xylitol
conversion under conditions of 473 K and 6.0 MPa H2.25 Recently, Ni/La2O3/ZrO2
catalyst was found to be efficient for sorbitol hydrogenolysis to propylene glycol and
ethylene glycol with a 47.1% total selectivity at 96.8% sorbitol conversion (at 493 K
and 4.0 MPa).26 Notwithstanding these significant progresses, the solid bases-
containing bifunctional catalysts generally suffer from deactivation, mainly due to the

leaching or transformation of the basic components under the hydrothermal
conditions.22, 25, 27, 29 Therefore, a careful design of stable catalysts is required with
improved hydrogenolysis activity and selectivities to propylene glycol and ethylene
glycol.
As for the reaction mechanism, the dehydrogenation of sorbitol and xylitol is
widely accepted as the first step, for their hydrogenolysis to propylene glycol and
ethylene glycol on different metal catalysts with bases.15, 22, 32-35 As shown in Scheme
1, sorbitol is first dehydrogenated to hexose intermediates, including ketohexose (Path
A) and aldohexose (Path B), on the metal surfaces. In the Path A, ketohexose undergoes
selective cleavage of its C(3)–C(4) bond via retro-aldol condensation with bases to form
tautomeric dihydroxyacetone and glyceraldehyde, which are further hydrogenated to
glycerol on metal surfaces. Competitively, glyceraldehyde can be also dehydrated to
pyruvaldehyde, which then undergoes either hydrogenation to acetol and propylene
glycol or base-involved benzilic acid rearrangement to lactic acid. In the parallel Path
B, the C(2)–C(3) and C(4)–C(5) bonds of aldohexose are successively cleaved via retro-
aldol condensation by bases to form three molecules of glycolaldehyde, which is then
hydrogenated to ethylene glycol on metal surfaces. In this mechanism, the first
dehydrogenation step of sorbitol on metal surface has been generally proposed as the
kinetically-relevant step.15, 22, 32-35 However, few studies have reported on the kinetic
and particularly isotopic evidences to unequivocally validate such hydrogenolysis
mechanism.

Scheme 1. Proposed reaction pathways for sorbitol hydrogenolysis to C2 and C3
products on metal catalysts in the presence of bases.15
Considering the aforementioned progresses and existing problems, this work aims
to develop new catalysts with improved activity and selectivity to propylene glycol and
ethylene glycol in the sorbitol hydrogenolysis, and provide further insights into the
reaction mechanism. Recently, Pd-based catalysts, coupling with ZnO, have shown
promising performance in selective hydrogenolysis of glycerol, which is also
kinetically limited, similar to the sorbitol reaction, by a primary dehydrogenation
step.36-38 For example, we found that physical mixtures of Pd/ZrO2 and ZnO efficiently
catalyze glycerol hydrogenolysis to propylene glycol, as a result of in-situ facile
formation of PdZn alloys on ZrO2. In this work, we present an in-depth study on the
selective hydrogenolysis of sorbitol to propylene glycol and ethylene glycol on the
PdZn alloys in the presence of solid bases. The effects of reaction parameters, including
amount of solid base, temperature and H2 pressure, are examined within the kinetically-
controlled regime. Kinetic isotope effects for different deuterated sorbitol molecules
are measured under actual reaction conditions to probe the kinetic relevance of C-H
bond activation involved in the sorbitol hydrogenolysis.
2. Experimental
Monoclinic ZrO2-supported Pd (Pd/ZrO2) catalysts were prepared by a deposition–

precipitation method. Briefly, ZrO2 powder (Alfa Aesar) was dispersed into an aqueous
solution of PdCl2 (Sinopharm Chemical), containing 3 wt% palladium with respect to
ZrO2. Under vigorous stirring, a Na2CO3 solution (1.0 mol/L) was added to the
palladium solution dropwise until the pH value of the final solution reached 10, which
was then maintained for 1.5 h. The resulting precipitates were centrifugally separated,
thoroughly washed with deionized water until no Cl- ions were detected, and then dried
subsequently in ambient air at 333 K for 12 h and at 393 K for 12 h. Afterwards, the
resulting powders were calcined in flowing air at 773 K for 3 h, and then were reduced
in a 20% H2/N2 (Beijing Haipu) flow at 673 K for 3 h. Following the similar procedure
for Pd/ZrO2, Al2O3 (Alfa Aesar) and TiO2 (Alfa Aesar) supported Pd catalysts were also
prepared. The Pd content in these catalysts was fixed at 3 wt% when prepared. The
actual Pd loadings of Pd/ZrO2, Pd/Al2O3 and Pd/TiO2 catalyst were measured by
inductively coupled plasma atomic emission spectrometry (ICP-AES; Teledyne
Leeman Laboratories), being 2.57%, 3.15% and 2.58%, respectively.
Mg3AlOx (Mg/Al molar ratio of 3/1), was prepared by a co-precipitation method.
Briefly, an aqueous solution (40 mL) of Mg(NO3)2  6 H2O (1.125 mol/L, Beijing Yili
Fine Chemical) and Al(NO3)3  9 H2O (0.375 mol/L, Beijing Yili Fine Chemical) was
dropwisely added simultaneously with another solution containing NaOH (0.34 mol/L,
Beijing Yili Fine Chemical) and Na2CO3 (0.060 mol/L, Beijing Yili Fine Chemical)
into 20 mLdeionized water under vigorous stirring, keeping the pH value of the solution
at 9–10. The resulting precipitates were aged overnight, filtered, and washed thoroughly
with deionized water. The hydroxycarbonate precursor was dried at 383 K overnight
and then calcined in ambient air at 673 K for 4 h to obtain the corresponding Mg3AlOx.
CeO2 was prepared hydrothermally by hydrolysis of an aqueous solution of
(NH4)2Ce(NO3)6 (Sinopharm Chemical) with urea at 413 K for 24 h in a Teflon-lined
stainless-steel autoclave. The resulting yellow precipitates were centrifugally separated
and thoroughly washed with deionized, followed by drying in air at 393 K overnight
and subsequent calcination at 673 K for 3 h.
Transmission electron microscopy (TEM) measurements were carried out on
Philips Tecnai F30 FEG-TEM operating at an acceleration voltage of 300 kV. Before

the TEM measurement, the samples were ultrasonically dispersed in ethanol and the
suspensions were deposited on carbon-coated copper grids. The average sizes of Pd
particles and their size distributions were calculated by measuring at least 300 particles
randomly distributed in the TEM images. XRD patterns were obtained on a Rigaku
D/MAX-2400 diffractometer using Cu K
and 100 mA. The 2θ angle was scanned in the range of 20-60o at a rate of 2o/min. X-
ray photoelectron spectra (XPS) were collected on an Axis Ultra spectrometer (Kratos,
UK) using an Al anode (Al Kα, hv =1486.6 eV) operated at 150 W. The samples were
treated in flowing 20% H2/N2 (50 mL/min) at 493 K for 2 h in a process chamber before
the XPS measurement to eliminate possible oxidation of the metal catalysts caused by
the unavoidable exposure to ambient air. The binding energies were calibrated by
referring to the Zr 3d5/2 signal of ZrO2 at 182.1 eV.38
Sorbitol hydrogenolysis reactions were carried out in a stainless steel autoclave
(100 mL) with vigorous stirring at 750 rpm. Typically, 5 g sorbitol (98%, Alfa Aesar),
0.15 g of catalyst, 0.1 g Mg3AlOx and the specified amount of ZnO (Alfa Aesar) were
introduced into the autoclave containing 45 g H2O. After fully purged with 1.0 MPa H2
(>99.99%, Beijing Haike Yuanchang), the reactor was pressurized with H2 to 5.0 MPa
and then heated to the target reaction temperatures (473-513 K). After the reactor cooled
to room temperature, the catalysts and liquid phase were simply separated by filtration.
The reactants and products in the liquid phase were analyzed by high-performance
liquid chromatography (HPLC, Shimadzu LC-20A) using Bio-Rad Aminex HPX-87H
and HPX-87C columns connected to a RID detector. The activities of Pd catalysts were
calculated by the molar sorbitol conversion rates per mole of Pd loaded per hour (h−1),
while the product selectivities were reported on a carbon basis.
Kinetic isotope effects on the sorbitol hydrogenolysis were examined using
different deuterated sorbitols (including sorbitol-1,1,2,3,4,5,6,6-d8, sorbitol-1,6,6-d3,
sorbitol-1,1-d2, sorbitol-2-d1 and sorbitol-5-d1), in comparison with undeuterated
sorbitol under the same conditions. These deuterated sorbitols were obtained by the
same method described in the literature.15 These hydrogenolysis reactions were carried
out in a Teflon-lined stainless steel autoclave (50 mL) with magnetic stirring at 800 rpm.

In a typical run, 0.05 g of Pd/ZrO2, 0.005 g of ZnO, 0.05 g of Mg3AlOx and 0.22 g of
sorbitol (the same molar amount for deuterated sorbitols) were loaded into the autoclave
containing 15.8 g of H2O. Then, the reactor was charged with 4.0 MPa H2 and heated
to 453 K, which was maintained for 1.5 h.
3. Results and discussion
3.1 Activity and selectivity of sorbitol hydrogenolysis on Pd/ZrO2 and ZnO
Table 1 shows the activities and selectivities in the sorbitol hydrogenolysis on Pd-
based catalysts in the presence of different solid bases at 493 K and 5.0 MPa H2. The
activities and selectivities were compared at similar sorbitol conversions (~30%) within
the kinetically-controlled regime (detailed conversions and reaction time shown in
Table S1, SI). Propylene glycol, ethylene glycol and also glycerol were detected as the
target products, together with by products including lactic acid and C4- C6 polyols (1,2-
butanediol, erythritol, threitol, xylitol, arabitol, mannitol and iditol). As shown in Table
1, physically mixed with a small amount of ZnO (5 mg), Pd/ZrO2 actively catalyzed the
sorbitol hydrogenolysis with an activity of 152.2 h-1, as previously observed for
glycerol hydrogenolysis.38 However, its selectivity for the sum of propylene glycol,
ethylene glycol and glycerol was only 36.8%, while the selectivity for the C6 polyols
(including mannitol and iditol), formed from the hydrogenation of the hexose
intermediates, were as high as 21.4%. Such selectivity trend is consistent with the
absence of stronger basic sites on ZnO, required for catalyzing the retro-aldol
condensation of the hexose intermediates to the C2-C3 products (as shown in Scheme
1). After replacing ZnO with a stronger solid base, Mg3AlOx, the combined selectivity
for the C2-C3 products increased obviously to 64.8%, concurrently with a large decrease
in the selectivity for the C6 polyols (12.1%, Table 1). However, the hydrogenolysis
activity was only 12.2 h-1 on Pd/ZrO2+Mg3AlOx, much lower than that on
Pd/ZrO2+ZnO, demonstrating the promoting effect of ZnO on Pd catalyst in the sorbitol
hydrogenolysis, via the in-situ formation of active PdZn alloys, as previously reported
for glycerol hydrogenolysis,38 which will be further discussed below. Further,

physically mixing Mg3AlOx with Pd/ZrO2 and ZnO led to a significant increase in the
selectivities for propylene glycol and ethylene glycol to 38.9% and 12.7%, respectively,
and a total selectivity of the two polyols and glycerol to 60.1%, while the
hydrogenolysis activity remained to be as high as 136.2 h-1 (Table 1).
Table 1 Effects of solid base and support on activities and selectivities in sorbitol
hydrogenolysis on ZnO-promoted Pd catalysts a
Selectivity (on a carbon basis, %)
Activity
Catalyst
PropyleneEthylene
LacticC4
C5
C6
(h-1)
Glycerol
glycol
152.2 20.2
12.2 31.4
glycol
6.6
acid polyolsb polyolsc polyolsd
Pd/ZrO2+ZnO
10.0
7.6
1.5 3.5
8.7 6.5
3.5 6.9
1.9 6.2
3.9 6.0
4.3 6.1
3.7 5.9
7.3
8.5
1.2
1.3
1.0
1.3
0.8
21.4
12.1
12.7
8.8
Pd/ZrO2+Mg3AlOx
17.1
12.7
13.7
15.4
13.8
12.8
Pd/ZrO2+ZnO+Mg3AlOx 136.2 38.9
Pd/ZrO2+ZnO+La(OH)3 143.7 36.9
8.5
12.1
9.3
Pd/ZrO2+ZnO+CeO2
136.6 35.9
9.2
Pd/Al2O3+ZnO+Mg3AlOx 52.3
Pd/TiO2+ZnO+Mg3AlOx 34.4
40.4
41.6
7.5
9.8
3.6
8.9
a Reaction conditions: 50 g of 10 wt% aqueous sorbitol solution, 0.15 g of Pd/ZrO2 with a 3 wt%
Pd loading, 0.005 g of ZnO or none, 0.1 g of solid base, 493 K, 5.0 MPa H2, ~30% sorbitol
conversion (obtained by varying reaction time). b C4 polyols include 1,2-butanediol, erythritol
and threitol, c C5 polyols include xylitol and arabitol, d C6 polyols include mannitol and iditol.
Resembling Mg3AlOx, other solid bases, such as La(OH)3 and CeO2, can also
efficiently catalyze the retro-aldol condensation involved in the sorbitol hydrogenolysis.
When they were used with Pd/ZrO2+ZnO, the activities were 143.7 and 136.6 h-1, and
the combined selectivities to the two target glycols were 50.6 % and 51.3%, respectively
(Table 1), both essentially identical to those when Mg3AlOx was present. Notably, the
selectivities to lactic acid were about 2-4% in the presence of the three solid bases,
which were much lower than those on other catalysts, for example, Ru/C in the presence
of Ca(OH)2 (~25%)15, implying the superiority of these solid bases to soluble bases in

minimizing the benzilic acid rearrangement of glyceraldehyde-derived pyruvaldehyde
intermediate to lactic acid in the sorbitol hydrogenolysis (Scheme 1). It is also
worthwhile to note that the presence of the solid bases led to very low selectivities (~1%)
for the C5 polyols (Table 1), reflecting another advantage of such catalytic components
in discouraging the metal catalyzed C-C cleavage via decarbonylation of the carbonyl
intermediates to C5 and C1 byproducts in the sorbitol hydrogenolysis, as reported on
Ru-based catalyst15.
For the Pd catalysts, the support effects were also examined with a similar particle
size of Pd sites (~3 nm, see Figs. S1 and S2, SI) at 493 K and 5.0 MPa H2. As shown in
Table 1, the product distributions on the three supported Pd catalysts (i.e. Pd/ZrO2,
Pd/Al2O3 and Pd/TiO2) were similar at similar sorbitol conversions (~ 30%), showing
about 60% combined selectivities to propylene glycol, ethylene glycol and glycerol and
~4% selectivities to lactic acid. But Pd/ZrO2 exhibited a much higher activity (136.2 h-
1) than Pd/Al2O3 (52.3 h-1) and Pd/TiO2 (34.4 h-1), mainly attributed to their different
metal-support interaction and the consequent effect on their sorbitol dehydrogenation
activity39, 40
.
The promoting effect of ZnO on the sorbitol hydrogenolysis was further examined
with its varying amounts, physically mixing with Pd/ZrO2 and Mg3AlOx (Fig.1, detailed
conversions and reaction time shown in Table S2, SI). Without addition of ZnO, the
activity was as low as 12.2 h-1 , and the selectivities to propylene glycol, ethylene glycol,
glycerol and lactic acid were 31.4%, 17.1%, 7.6% and 8.7%, respectively, as discussed
above (Table 1). As the amount of ZnO increased to 5 mg, the activity increased sharply
to 136.2 h-1, and then decreased slightly to 96.3 h-1 with increasing the ZnO amount to
20 mg. Within this range of the ZnO amounts, the selectivity to propylene glycol first
increased to 40.8% at 2.5 mg ZnO, and then gradually decreased to a constant value of
37.5% with increasing the amount to above 7.5 mg. The selectivity to ethylene glycol
declined from 19.1% to 12.2% when the amount of ZnO increased from 1.0 mg to 7.5
mg, which then remained essentially constant. Meanwhile, the selectivity to glycerol
varied in a narrow range of 7.0%~9.1%. Differently, the selectivity to lactic acid
decreased to the minimum value of 3.5% at 5 mg ZnO, and then slightly increased to

~5.0% at the larger amounts. Notably, the catalytic performance of Pd/ZrO2 can be
largely improved by optimizing the amount of ZnO in the sorbitol hydrogenolysis.
Such trends of the activities and also selectivities to C2-C3 products are very similar
to those previously observed in glycerol hydrogenolysis on Pd/ZrO2+ZnO, where PdZn
alloys were formed on ZrO2, via the in-situ reduction of dissolved Zn2+ ions on Pd
surfaces under the similar hydrogenolysis conditions, and found to be very effective in
the kinetically-relevant step of glycerol dehydrogenation38. The Pd/ZrO2+ZnO mixture
was characterized by XPS before and after the sorbitol hydrogenolysis. Relative to the
binding energies of the Pd 3d3/2 and 3d5/2 signals for Pd/ZrO2 (340.6 and 335.3 eV, Fig.
2a), the values for the used catalyst after the sorbitol hydrogenolysis shifted by 0.7 eV
to 341.3 and 336.0 eV, respectively (Fig. 2b), reflecting electronic transfer from Pd to
Zn and their strong interaction. These values are identical to those for the as-prepared
PdZn alloys on ZrO2 (Fig. 2c).38, 41 Taken together, these results suggest that the PdZn
alloys were also formed in-situ on ZrO2 from the mixture of Pd/ZrO2+ZnO during the
sorbitol hydrogenolysis in the presence of Mg3AlOx, which acted as the active
dehydrogenation sites and contributed to the observed higher activities.
Figure 1. Effects of ZnO addition on selectivities and activities in sorbitol
hydrogenolysis on Pd/ZrO2 with Mg3AlOx (50 g of 10 wt% aqueous sorbitol solution,

0.15 g of Pd/ZrO2 with a 3 wt% Pd loading, 0.1 g of Mg3AlOx, 493 K, 5.0 MPa H2,
~30% sorbitol conversion obtained by varying reaction time).
Figure 2 XPS spectra of Pd 3d and Zr 3p signals for (a) Pd/ZrO2, (b) a physical mixture
of 0.15 g 3 wt% Pd/ZrO2 and 0.1 g ZnO after sorbitol hydrogenolysis reaction (493 K,
5.0 MPa H2, 2h, 50 g of 10 wt% aqueous sorbitol solution) and (c) fresh PdZn/ZrO2
prepared by the method reported in the Reference [38].
It is noted that the high combined selectivity for the two glycols on Pd/ZrO2+ZnO
with Mg3AlOx can be obtained at nearly 100% sorbitol conversions. As shown in Fig.
3, with increasing sorbitol conversion from 20.3% to 96.0% at 493 K and 5.0 MPa H2,
the selectivity for propylene glycol and ethylene glycol increased from 37.6% and 12.6%
to 43.3% and 13.6%, respectively, indicative of their high stability under the reaction
conditions, leading to a 54.6% yield of the two target glycols on Pd/ZrO2+ZnO with
Mg3AlOx. The glycerol selectivity increased slightly from 7.6% to 8.5%, and then
declined to 6.8%, while the lactic acid selectivity declined from 3.9% to 2.4%, a very
low value at nearly full conversion of sorbitol. In addition, the total selectivity for the
C6 polyols (mannitol and iditol) decreased monotonically from 15.6% to 2.3% with
increasing sorbitol conversion in the range of 20.3–96.0%, indicating that the two
sorbitol epimers are also reactive to undergo hydrogenolysis to propylene glycol and

ethylene glycol under the identical reaction conditions.
Figure 3. Dependence of product selectivities on sorbitol conversion on Pd/ZrO2+ZnO
with Mg3AlOx (50 g of 10 wt% aqueous sorbitol solution, 0.15 g of Pd/ZrO2 with a 3
wt% Pd loading, 0.005 g of ZnO, 0.1 g of Mg3AlOx, 493 K, 5.0 MPa H2, 1-17.5 h).
The stability and recyclability of Pd/ZrO2+ZnO+Mg3AlOx were also examined. As
shown in Fig. 4, the sorbitol conversions remained essentially unchanged for five
successive cycles at 493 K and 5.0 MPa H2. The selectivities for propylene glycol and
ethylene glycol were also stable, being 42.7%-43.6% and 13.3%-13.8%, respectively,
while the selectivities for glycerol and lactic acid declined slightly from 6.5% and 2.2%
to 5.2% and 1.5%, respectively. Consistently, no leaching of Pd or Zr was detected by
ICP-AES in the filtrate after the recycling experiments. These results reveal that the
PdZn alloys derived from the physical mixture of Pd/ZrO2 and ZnO with Mg3AlOx is
efficient and stable for catalyzing the sorbitol hydrogenolysis under the conditions in
this work.

Figure 4. Conversions and selectivities (to propylene glycol, ethylene glycol, glycerol
and lactic acid) for the five cycles of sorbitol hydrogenolysis on Pd/ZrO2+ZnO with
Mg3AlOx (50 g of 10 wt% aqueous sorbitol solution, 0.50 g of Pd/ZrO2 with a 3 wt%
Pd loading, 0.02 g of ZnO and 0.2 g of Mg3AlOx in the first cycle, 0.005 g of ZnO and
0.05 g of Mg3AlOx in cycles 2-5, 493 K, 5.0 MPa H2, 10 h).
3.2 Effects of reaction parameters on sorbitol hydrogenolysis
Fig. 5 shows the effects of the Mg3AlOx amount on the sorbitol hydrogenolysis
activities and selectivities on Pd/ZrO2+ZnO at 493 K and 5.0 MPa H2. Mg3AlOx, as a
mild solid base, plays a key role in providing basicity and catalyzing the retro-aldol
condensation of sugar intermediates that leads to the selective cleavage of their C–C
bonds. With increasing the amount of Mg3AlOx from 0.06 g to 0.3 g, the activity
increased from 120.0 to 160.0 h-1 (detailed conversions and reaction time shown in
Table S3, SI), implying the favored hydrogenolysis at the higher basicity on
Pd/ZrO2+ZnO. This increase in the amount of Mg3AlOx led to the slight decrease in the
selectivities for propylene glycol and glycerol from 39.4% and 9.1% to 37.5% and 7.8%,
respectively, with the concurrent increase in the lactic acid selectivity from 3.1% to
7.5%. Such selectivity change is consistent with the proposed mechanism that the three
C3 products are formed via the same glyceraldehyde intermediate, which undergoes
favorable benzilic acid rearrangement to lactic acid over its hydrogenation reaction to

glycerol and propylene glycol at higher basicity, as discussed above (Scheme 1).
Meanwhile, the selectivity for the C6 polyols decreased monotonically from 14.5% to
9.2%, apparently as a consequence of the favored base-catalyzed retro-aldol
condensation of the hexose intermediates over their epimerization. Such effects show
the importance in finely tuning the basicity to maximize the selectivity for the two target
glycols.
Figure 5. Effect of Mg3AlOx addition on selectivities and activities in sorbitol
hydrogenolysis on Pd/ZrO2+ZnO ((50 g of 10 wt% aqueous sorbitol solution, 0.15 g of
Pd/ZrO2 with a 3 wt% Pd loading, 0.005 g of ZnO, 493 K, 5.0 MPa H2, ~30% sorbitol
conversion obtained by varying reaction time).
Fig. 6 shows the effects of the reaction temperature on the activities and
selectivities in sorbitol hydrogenolysis on Pd/ZrO2+ZnO with Mg3AlOx. In the
kinetically-controlled regime (at ~30% sorbitol conversions, detailed conversions and
reaction time shown in Table S4, SI), as expected, the activities increased dramatically
from 48.2 to 392.8 h-1 with increasing temperature from 473 to 513 K. Accordingly,
estimated by the Arrhenius equation, the apparent activation energy for the sorbitol
hydrogenolysis was 105±3 kJ/mol at 473-513 K on PdZn alloy (Fig. S3, SI), close to
the value (123±5 kJ/mol, at 463-513 K) for glycerol hydrogenolysis38, reflecting their

similar kinetically-relevant step of the polyol dehydrogenation. In this temperature
range, the propylene glycol selectivity increased from 32.5% to the maximum value of
38.9% at around 493 K, and then declined slightly to 35.6%. In contrast, the glycerol
selectivity decreased monotonically from 13.3% to 6.2%. The lactic acid selectivity
increased slightly from 2.3% to 4.7%, consistent with the fact that the basicity of the
Mg3AlOx increased at higher temperature and thus facilitated the benzilic acid
rearrangement. Meanwhile, the selectivity for C6 polyols decreased from 14.5% to 11.
2% with the concurrent increase in the selectivity for the C2-C3 products, indicating that
the retro-aldol condensation reaction of the hexose intermediates is faster than their
hydrogenation at higher temperatures. It is also noted that the acetol intermediate was
detected above 493 K, and its selectivity increased up to 5.0% at 513 K. Such lower
activity for acetol hydrogenation to propylene glycol indicates the structural change of
the active PdZn alloys derived from Pd/ZrO2 and ZnO at the higher temperatures. Such
temperature-sensitive activity provides a useful direction for optimizing the parameters
in the polyol hydrogenolysis reactions on the PdZn alloy-based catalysts.
Figure 6. Effect of temperature on selectivities and activities in sorbitol hydrogenolysis
on Pd/ZrO2+ZnO with Mg3AlOx (50 g of 10 wt% aqueous sorbitol solution, 0.15 g of
Pd/ZrO2 with a 3 wt% Pd loading, 0.005 g of ZnO, 0.1 g of Mg3AlOx, 5.0 MPa H2, ~30%
sorbitol conversion obtained by varying reaction time).

The effects of H2 pressure on sorbitol hydrogenolysis activities and selectivities
were examined at 493 K on Pd/ZrO2+ZnO with Mg3AlOx in kinetically-controlled
regime (at ~30% sorbitol conversions, detailed conversions and reaction time shown in
Table S5, SI). As shown in Fig. 7, under a very low H2 pressure (1.0 MPa), acetol was
the main by-product in the reaction with a 20.3% selectivity, very close to the selectivity
for propylene glycol (19.1%), because of the inefficient hydrogenation of acetol to
propylene glycol. With increasing the H2 pressure to 8.0 MPa, acetol disappeared, and
the selectivities for propylene glycol and ethylene glycol increased significantly to a
plateau around 35.9% and 12.7%, respectively. Concurrently, the glycerol selectivity
increased monotonically from 2.6% to 12.2% while the lactic acid selectivity decreased
slightly from 5.0% to 3.1%. Such a change in the selectivities for the C3 products
reflects their strong dependence on the relative rates of the competitive reactions of the
glyceraldehyde intermediate on the PdZn alloy and basic sites (Scheme 1). These results
also reflect the accelerated hydrogenation of the unsaturated intermediates at higher H2
pressures in the sorbitol hydrogenolysis, together with the monotonic increase in the
selectivities for the C6 polyols (from 6.9% to 18.4%).
As for the sorbitol hydrogenolysis activity, it increased from 106.3 to 126.9 h-1 with
increasing the H2 pressure from 1.0 to 3.0 MPa, because of the inhibition to the
dehydrogenation/hydrogenation activity caused by the strong adsorption of the
unsaturated intermediates on the active metal sites at the low H2 pressures42. However,
further increasing the H2 pressure to 8.0 MPa led to a monotonic decrease of the activity
to 106.7 h-1. Such inhibiting effect on the sorbitol hydrogenolysis in this broad H2
pressure range (3.0-8.0 MPa), as also observed on La-promoted Ni/ZrO2 by Cai et al.26,
reflects the proposition that the sorbitol hydrogenolysis is kinetically limited by its
dehydrogenation15, a thermodynamically unfavorable step at higher H2 pressures. This
effect is consistent with the kinetic isotope effects for the sorbitol hydrogenolysis on
PdZn alloys, as discussed below.

Figure 7. Effect of H2 pressure on selectivities and activities in sorbitol hydrogenolysis
on Pd/ZrO2+ZnO with Mg3AlOx (5 g of sorbitol, pretreated catalysts, 0.1 g of Mg3AlOx,
493 K, 8 MPa-balance with N2, ~30% sorbitol conversion obtained by varying reaction
time; Hydrothermal pretreatment conditions of catalysts: 0.15 g of Pd/ZrO2 with a 3 wt%
Pd loading, 0.005 g of ZnO, 45.0 g of water, 493 K, 5.0 MPa H2, 1h).
3.3 Kinetic isotope effects for sorbitol hydrogenolysis
Recently, we studied the kinetic isotope effects (KIE) with undeuterated and
deuterated sorbitol molecules on Ru/C to experimentally confirm the primary
dehydrogenation of sorbitol as the kinetically-relevant step in the sorbitol
hydrogenolysis15. However, because of the interference of the facile H-D exchange
between the deuterated sorbitol and H2O/H2 molecules on Ru/C under the
hydrogenolysis conditions, the measurements of the KIE values were actually
conducted under N2, instead of H2. Differently, PdZn alloys are inert to the H-D
exchange between the C–H bonds of polyols and H2O/H2 molecules, as observed in
glycerol hydrogenolysis38, which renders it viable to rigorously measure the intrinsic
kinetic isotope effects and probe the kinetically-relevant step in the sorbitol
hydrogenolysis.
Undeuterated and deuterated sorbitol molecules (including sorbitol-1,1,2,3,4,5,6,6-
d8, sorbitol-1,6,6-d3, sorbitol-1,1-d2, sorbitol-2-d1 and sorbitol-5-d1) were used to

measure the KIE (kH/kD) values on PdZn alloys (Pd/ZrO2+ZnO) under H2 (4.0 MPa H2,
453 K). As shown in Table 2, irrespective of their deuteration degree, these sorbitol
molecules showed similar selectivities to the C2-C3 products (including propylene
glycol, ethylene glycol, glycerol and lactic acid), suggesting that they undergo the
hydrogenolysis reactions via the same mechanism on PdZn alloys. At 453 K and 4.0
MPa H2 in H2O, the activity of undeuterated sorbitol was 14.7 h-1, while the activity of
fully deuterated sorbitol-1,1,2,3,4,5,6,6-d8 was 6.9 h-1 (Table 2). Consequently, the
kH/kD value between sorbitol and sorbitol-d8 was 2.13, clearly demonstrating that
cleavage of C–H bonds is involved in the kinetically-relevant steps of the sorbitol
hydrogenolysis. This normal KIE is consistent with the observed inhibiting effects of
H2 pressure (Fig. 7), and provides an intrinsic evidence for the dehydrogenation
mechanism of the sorbitol hydrogenolysis. On the other hand, for the partially
deuterated sorbitols (Table 2), the corresponding kH/kD values for sorbitol-1,6,6-d3,
sorbitol-1,1-d2 and sorbitol-2-d1 were all close to unity (1.03, 1.09 and 1.09,
respectively), while sorbitol-5-d1 showed a higher KIE value of 1.26. Such a
comparison implies that the activation of the C(5)–H bond is more preferred when
compared to the C(1)–H, C(2)–H and C(6)–H bonds in sorbitol molecules as also found
15
with Ru/C in the presence of Ca(OH)2 . This proposed mechanism is consistent with
the fact that the selectivities for C3 products via the retro-aldol condensation of
ketohexose intermediate (Path A, Scheme 1) are much higher than that for C2 product
(i.e. ethylene glycol) via aldohexose intermediate from dehydrogenation at the C(1) or
C(6) position (Path B, Scheme 1). This mechanism also indicates the significant effects
of the configurations of hexitols on their hydrogenolysis activity and the limiting factors
for the formation of the two target glycols.
Table 2. Activities and selectivities for hydrogenolysis of undeuterated and different
a
multi-deuterated sorbitols on Pd/ZrO2+ZnO with Mg3AlOx under H2
Selectivity (on a carbon basis, %)
Conversion Activity
KIE
Reactant
(h-1)
(kH/kD)
Propylene
glycol
Ethylene
glycol
Lactic
acid
(%)
Glycerol

Sorbitol
22.6
10.6
14.7
6.9
-
18.6
18.3
6.6
7.2
11.3
14.2
3.6
3.1
Sorbitol-
1,1,2,3,
4,5,6,6-d8
Sorbitol-
1,6,6-d3
Sorbitol-
1,1-d2
2.13
21.9
20.7
20.8
17.9
14.3
13.5
13.6
11.7
1.03
1.09
1.09
1.26
18.1
20.1
18.3
16.7
6.7
7.2
8.1
8.3
12.7
10.6
11.3
9.6
3.1
3.8
4.2
4.2
Sorbitol-
2-d1
Sorbitol-
5-d1
a Reaction conditions: ~1.18 mmol reactant, 15.8 g H2O, 0.05 g of Pd/ZrO2 with a 3 wt% Pd
loading, 0.005g of ZnO, 0.05 g of Mg3AlOx, 453 K, 4.0 MPa H2, 90 min.
4. Conclusions
Pd/ZrO2 physically mixed with ZnO is highly active and stable for the sorbitol
hydrogenolysis to propylene glycol and ethylene glycol in the presence of Mg3AlOx as
a solid base, via the efficient dehydrogenation of sorbitol to the hexose intermediates
and their retro-aldol condensation, catalyzed respectively by in-situ formed PdZn alloys
on ZrO2 and basic sites. The selectivity to the two target glycols reaches 56.9%, with a
significantly low selectivity to lactic acid (2.4%) at a nearly full conversion of sorbitol.
For this catalyst system of Pd/ZrO2+ZnO+Mg3AlOx, the activities and selectivities
depend strongly on the amount of ZnO and solid base, due to their effects on the
formation of the active PdZn alloys and the competitive metal- and base-catalyzed steps
involved in the sorbitol hydrogenolysis. Together with the observed strong inhibition
of H2 pressure to the hydrogenolysis activity in a broad range (3.0-8.0 MPa), the kinetic
isotopic studies confirm that the primary dehydrogenation of sorbitol, via preferential
cleavage of its C(5)–H bond to the hexose intermediates, is the kinetically relevant step
in the sorbitol hydrogenolysis. This study provides fundamental understanding useful
for the design of more effective catalysts and optimization of reaction parameters
toward the synthesis of the target propylene glycol and ethylene glycol from different
polyols.

CRediT authorship contribution statement
Yuqing Jia: Data curation, Investigation, Writing - original draft.
Qianhui Sun: Data curation, Formal analysis, Writing - original draft.
Haichao Liu: Project administration, Resources, Supervision, Writing - review& editing.
Declaration of Competing Interest
The authors declare no competing financial interest.
Acknowledgments
This work was supported by the National Key Research and Development Program of
China (2016YFB0701100), the Natural Science Foundation of China (21690081,
21832001, 21821004, and 21805309), and Beijing National Laboratory for Molecular
Sciences (BNLMS-CXXM-201905).
Appendix A. Supplementary data
TEM micrographs and histograms of Pd particle size distribution for Pd/Al2O3 and
Pd/TiO2; XRD patterns of monoclinic ZrO2 and Pd/ZrO2; Apparent activation energy
for the sorbitol hydrogenolysis on PdZn alloy. Detailed conversion and reaction time
for each experiment in Table 1, Figure 1, and Figure 5-7.
REFERENCES
[1] Corma, S. Iborra, A. Velty, Chem. Rev. 107 (2007) 2411-2502.
[2] J. ten Dam, U. Hanefeld, ChemSusChem 4 (2011) 1017-1034.
[3] C. H. Zhou, X. Xia, C. X. Lin, D. S. Tong, J. Beltramini, Chem. Soc. Rev. 40 (2011) 5588-
5617.
[4] D. M. Alonso, S. G. Wettstein, J. A. Dumesic, Chem. Soc. Rev. 41 (2012) 8075-8098.
[5] A. M. Ruppert, K. Weinberg, R. Palkovits, Angew. Chem. Int. Edit. 51 (2012) 2564-2601.
[6] J. Zhang, J. B. Li, S. B. Wu, Y. Liu, Ind. Eng. Chem. Res. 52 (2013) 11799-11815.
[7] B. Zada, M. Y. Chen, C. B. Chen, L. Yan, Q. Xu, W. Z. Li, Q. X. Guo, Y. Fu, Sci. China-

Chem. 60 (2017) 853-869.
[8] P. Sudarsanam, E. Peeters, E. V. Makshina, V. I. Parvulescu, B. F. Sels, Chem. Soc. Rev.
48 (2019) 2366-2421.
[9] J. Y. Sun, H. C. Liu, Green Chem., 2011, 13 () 135-142.
[10] M. Banu, P. Venuvanalingam, R. Shanmugam, B. Viswanathan, S. Sivasanker, Top. Catal.
55 (2012) 897-907.
[11] L. M. Ye, X. P. Duan, H. Q. Lin, Y. Z. Yuan, Catal. Today 183 (2012) 65-71.
[12] Z. Huang, J. Chen, Y. Jia, H. Liu, C. Xia, H. Liu, Appl. Catal., B 147 (2014) 377-386.
[13] H. Liu, Z. Huang, C. Xia, Y. Jia, J. Chen, H. Liu, ChemCatChem 6 (2014) 2918-2928.
[14] Y. Jia, H. Liu, Chin. J. Catal. 36 (2015) 1552-1559.
[15] Y. Jia, H. Liu, Catal. Sci. Technol. 6 (2016) 7042-7052.
[16] J. H. Zhou, G. C. Liu, Z. J. Sui, X. G. Zhou, W. K. Yuan, Chin. J. Catal. 35 (2014) 692-
702.
[17] X. Guo, H. Dong, B. Li, L. Dong, X. Mu, X. Chen, J. Mol. Catal. A: Chem. 426 (2017)
79-87.
[18] X. Wang, A. K. Beine, P. J. C. Hausoul, R. Palkovits, ChemCatChem 11 (2019) 4123-4129.
[19] J. Sun, H. Liu, Catal. Today 234 (2014) 75-82.
[20] H. Liu, Z. Huang, H. Kang, X. Li, C. Xia, J. Chen, H. Liu, Appl. Catal., B 220 (2018) 251-
263.
[21] A. K. Beine, A. J. D. Kruger, J. Artz, C. Weidenthaler, C. Glotzbach, P. J. C. Hausoul, R.
Palkovits, Green Chem. 20 (2018) 1316-1322.
[22] X. Jin, J. Shen, W. Yan, M. Zhao, P. S. Thapa, B. Subramaniam, R. V. Chaudhari, ACS
Catal. 5 (2015) 6545-6558.
[23] X. Wang, X. Liu, Y. Xu, G. Peng, Q. Cao, X. Mu, Chin. J. Catal. 36 (2015) 1614-1622.
[24] R. Vijaya Shanthi, R. Mahalakshmy, K. Thirunavukkarasu, S. Sivasanker, Mol. Catal. 451
(2018) 170-177.
[25] M. Rivière, N. Perret, D. Delcroix, A. Cabiac, C. Pinel, M. Besson, Catalysts 8 (2018) 331.
[26] C. Cai, H. Wang, H. Xin, C. Zhu, Q. Zhang, X. Zhang, C. Wang, Q. Liu, L. Ma, RSC Adv.
10 (2020) 3993-4001.
[27] I. Murillo Leo, M. López Granados, J. L. G. Fierro, R. Mariscal, Appl. Catal., B 185 (2016)
141-149.
[28] X. G. Chen, X. C. Wang, S. X. Yao, X. D. Mu, Catal. Commun. 39 (2013) 86-89.
[29] W. Du, L. Zheng, X. Li, J. Fu, X. Lu, Z. Hou, Appl. Clay Sci. 123 (2016) 166-172.
[30] M. Rivière, N. Perret, D. Delcroix, A. Cabiac, C. Pinel, M. Besson, ACS Sustainable Chem.
Eng. 6 (2018) 4076-4085.
[31] S. Li, Y. Zan, Y. Sun, Z. Tan, G. Miao, L. Z. Kong, Y. Sun, J. Energy Chem. 28 (2019)
101-106.
[32] K. L. Deutsch, D. G. Lahr, B. H. Shanks, Green Chem. 14 (2012) 1635-1642.
[33] P. J. C. Hausoul, A. K. Beine, L. Negahdar, R. Palkovits, Catal. Sci. Technol. 7 (2017) 56-
63.
[34] B. Gumina, F. Mauriello, R. Pietropaolo, S. Galvagno, C. Espro, Mol. Catal. 446 (2018)
152-160.
[35] B. Kuhne, H. Vogel, R. Meusinger, S. Kunz, M. Kunz, Catal. Sci. Technol. 8 (2018) 755-
767.

[36] X. Li, C. Zhang, H. Cheng, W. Lin, P. Chang, B. Zhang, Q. Wu, Y. Yu, F. Zhao,
ChemCatChem 7 (2015) 1322-1328.
[37] X. Li, B. Zhang, Q. Wu, C. Zhang, Y. Yu, Y. Li, W. Lin, H. Cheng, F. Zhao, J. Catal. 337
(2016) 284-292.
[38] Q. Sun, S. Wang, H. Liu, ACS Catal. 7 (2017) 4265-4275.
[39] W. J. Shen, M. Okumura, Y. Matsumura, M. Haruta, Appl. Catal., A 213 (2001) 225-232.
[40] Y. Shao, Z. Y. Xu, H. Q. Wan, H. A. Chen, F. L. Liu, L. Y. Li, S. R. Zheng, J. Hazard.
Mater. 179 (2010) 135-140.
[41] N. Iwasa, N. Takezawa, Top Catal. 22 (2003) 215-224.
[42] S. Wang, K. H. Yin, Y. C. Zhang, H. C. Liu, Acs Catal. 3 (2013) 2112-2121.