Download as pdf or txt
Download as pdf or txt
You are on page 1of 13

Chemical Engineering Journal 390 (2020) 124505

Contents lists available at ScienceDirect

Chemical Engineering Journal


journal homepage: www.elsevier.com/locate/cej

Specific role of aluminum site on the activation of carbonyl groups of methyl T


levulinate over Al(OiPr)3 for γ-valerolactone production
Tingting Zhao, Zhaoyang Ju, Yuxuan Zhang, Lujia Han, Weihua Xiao

Laboratory of Biomass and Bioprocessing Engineering, College of Engineering, China Agricultural University, Beijing 100083, China

HIGHLIGHTS

• Al(OiPr) exhibited high efficiency on conversion of ML to GVL under mild conditions.


• ML
3
was prone to undergo transesterification than direct hydrogenation.
• Al2-Propanol
atom activated ester carbonyl of ML before activating the ketone carbonyl.
• participated in the cyclization process as proton transfer carrier.

ARTICLE INFO ABSTRACT

Keywords: The high-efficiency synthesis of biofuel γ-valerolactone (GVL) from biomass-derived levulinates is a challenging task.
Methyl levulinate The Meerwein-Ponndorf-Verley (MPV) reduction with its extraordinary chemoselectivity is advantageous for the
γ-Valerolactone hydrogenation process, compared to the molecular-hydrogen-based process using noble metal catalysts. Therefore,
Al(OiPr)3 we used a classical Al-based isopropoxide to catalyze transfer hydrogenation (CHT) of methyl levulinate (ML) to GVL.
Kinetic analysis
A high yield of GVL up to 97.6% could be achieved using 2-proponal as the H-donor and solvent under mild
Reaction mechanism
DFT calculations
conditions (150 °C, 30 min). Besides, three reaction stages were observed in the conversion, including transester-
ification, hydrogenation and cyclization. LC/MS analysis and the density functional theory (DFT) caculations re-
vealed that Al atom of Al(OiPr)3 as the electron transfer center activated ester carbonyl of the substrate via four-
membered transition states before activating the ketone carbonyl, resulting in the occurrence of transesterification
prior to the hydrogenation. In addition, 2-propanol as proton transfer carrier assisting the cyclization process was
proved to be the lowest-energy pathway. Our work shed light on the role of Al (OiPr)3 in the MPV reduction of ML,
providing a comprehensive understanding on the metal alkoxide catalysis mechanism for GVL production.

1. Introduction [7], 5-fluorenone [8,9], and long-chain alkanes (e.g., C8-C18 alkanes)
[10,11], which is conducive to alleviate the reliance on fossil fuels in
As one of the top value-added chemical candidates listed in the modern society [12–14].
DOE/NREL report [1], GVL has attracted considerable attention due to Catalytic hydrogenation of biomass-derived levulinic acid (LA) or its
its outstanding physicochemical properties in recent decades. High esters to produce GVL is the widely used method in view of its low
boiling point (approximate 207 °C), low vapor pressure (3.5kpa at energy consumption [15]. A number of heterogeneous catalysts have
80 °C), and good low-temperature fluidity make it easy to be stored and been developed and it was found that noble metal (Ru, Ir, Rh, Pd, Re,
transported in long distance. Due to its low toxicity and fruit flavor, it etc.) based catalysts exhibited excellent performance [16–22]. How-
can be added to food and drink as the additive. In addition, several ever, the limitations of being high cost, prone to deactivation and dif-
reports have demonstrated that GVL as a green solvent is exceedingly ficult to regenerate restricted its practical application [23–25]. Re-
efficient to improve biomass conversion by dissolving biomass and cently, an alternative approach was developed to produce GVL by
humins, facilitating the formation of product and suppressing the pro- catalytic hydrogenation transfer based on MPV reduction Due to its
duct degradation [2,3]. More importantly, it can be used as an ideal highly chemoselectivity, carbonyl compounds can be reduced in mild
precursor for the production of numerous valuable chemicals and bio- conditions while the carboxyl or ester linkage is not affected [26]. In
fuels, such as dimethyl adipate [4,5], 1,4-pentanediol [6], valeric esters addition, using abundant and inexpensive alcohols as the reductant also


Corresponding author.
E-mail address: xiaoweihua@cau.edu.cn (W. Xiao).

https://doi.org/10.1016/j.cej.2020.124505
Received 17 October 2019; Received in revised form 10 January 2020; Accepted 18 February 2020
Available online 20 February 2020
1385-8947/ © 2020 Elsevier B.V. All rights reserved.
T. Zhao, et al. Chemical Engineering Journal 390 (2020) 124505

Table 1
Scheme of the reaction parameters experimental plan.
Entry Temperature (°C) Time (min) Al(OiPr)3 (wt %, based on substrate) H-donor (w:w, based on substrate)

1 130–170 30 50% 20:1


2 optimal value 10–60 50% 20:1
3 optimal value optimal value 12.5% − 100% 20:1
4 optimal value optimal value optimal value 6:1 – 26:1

Table 2 bond activation in carboxylic acids derivatives and exhibited ex-


The effect of catalysts on CHT of EL to GVL. cellence performance in MPV reduction [36]. In this study, classical Al
Entry Catalyst Conversion (%) Yield(%)
(OiPr)3 was chosen as the hydrogenation catalyst, which is much
cheaper and readily available. It has been used for the MPV reduction
1a
ZrO2 95.5 81.5 (ref) of unsaturated aromatic carbonyl compounds several years ago
2 ZrO2 30.8 25.1 [37,38], but used for the reduction of ahenaliphatic ketones was
3 ZrO2 79.2 71.5
4 ZrO2 75.8 58.0
rarely studied. The catalytic hydrogenation mechanisms were pro-
5b Zr(OH)4 93.6 88.5 (ref) posed by some researchers [39–41]. The generally accepted me-
6 Zr(OH)4 67.9 63.7 chanism was the hydride transfer pathway via a six-ring complex in
7 Al(OiPr)3 52.8 48.9 which the carbonyl compound and secondary alcohol were both
8 Al(OiPr)3 97.5 85.2
bound to the Al center. The carbonyl was then activated upon the
9 Al(OiPr)3 98.2 85.6
coordination to the Lewis acidic metal ion, followed by the hydride
Reaction conditions: a: 2 g EL (2-proponal as solvent), 38 g 2-proponal, 1 g transfer from the alcoholate to the carbonyl group [39–40]. One al-
ZrO2, 250 °C, 1 h. 29 Entry 2–4: 1 g EL, 19 g 2-proponal, 0.5 g ZrO2, 180 °C, cohol molecular then replaced the reduction substrate binding to the
reaction times were set as 30, 60, 120 min. b: 2 g EL, 38 g 2-proponal, 1 g Zr Al center, and the reduction product could dissociate in solvents
(OH)4, 200 °C, 60 min, 1 atm N2 [26]. Entry 6: 1 g EL, 19 g 2-proponal, 0.5 g Zr [42,43].
(OH)4, 200 °C, 40 min. Entry 7–9: 1 g EL, 19 g 2-proponal, 0.5 g Al(O-i-Pr)3, In fact, its application was shadowed by some modified catalysts
30 min, reaction temperatures were set as 100, 150, 200 °C. with organic ligand or heterogeneous catalyst with several transition
metals [41,44]. It was partially due to agglomeration of aluminum
shows significant economic advantages [27,28]. Some pioneered works alkoxides, which resulted in the super-stoichiometric amounts dosage
have been done to explore different catalyst system and discover that (1–20 equiv relative to substrate) [40]. In addition, the high recovery
Zr-containing catalytic system could efficiently convert LA or its esters cost restricted the large-scale application [39]. However, we explored
to GVL [26,27,29–31]. For example, Mei Chia et al. reported that 80.4% the catalytic effect of Al(OiPr)3 on GVL production and it exhibited
of GVL could be obtained using ZrO2 as catalyst and secondary alcohol fabulous catalytic activity with less amount than stoichiometric dose
as hydrogen source [29]. Tang et al. found that the conversion of EL under mild conditions and the reaction pathway was different with
with 2-propanol over Zr(OH)4 at 200 °C giving an EL conversion of reported heterogeneous Zr-containing catalysts. This result forced us to
93.6% and a GVL yield of 88.4% [25]. Other Zr-containing catalysts, reassess application of Al(OiPr)3 on hydrogenation process and explore
such as Zr-β [32], Zr-TMPA [33], Al-Zr [34] and UiO-66(Zr) [35] were the detailed catalytic mechanism.
also explored. Most of them were prepared through traditional Therefore, the effect of H-donors, reaction parameters including
methods, e.g. sol–gel method, hydrothermal synthesis method, and co- reaction temperature, residence time, the amount of catalyst, and the
precipitation method, which are relatively safe processes. However, the amount of alcohol on the GVL production were studied, which could
catalyst precursor is more expensive than common metals (Cu, Fe, Ni, provide thermodynamic data for the industrial evaluation in future.
Al) and the preparation processes are time-consuming, which would Furthermore, based on the results of kinetic experiment and LC/MS
increase the cost and efforts. Due to the heterogeneity, a higher tem- data, potential reaction routes of each step were proposed and sup-
perature (≥200 °C) or longer reaction time (≥5h) is still required in plemented with DFT calculations. Finally, a new plausible reaction
the reaction system. pathway for the whole reaction via Al(OiPr)3 was proposed in detail. It
Therefore, we attempted to convert levulinates to GVL with an is expected that our study will remotivate the application of metal
easily available noble-metal-free catalyst under mild conditions. alkoxide on the biofuel synthesis field and simultaneously give a the-
Previous study proved that chiral Al complex was efficient for the C-O oretical support on the reaction mechanism.

Table 3
Conversion of levulinate esters to GVL with various alcohols.
Entry Substrate H-honor Conversion (%) Selectivity (%)

GVL Levulinate esters*

b
1 ML Ethanol 97.5 5.9 ± 0.4 93.9 ± 1.0b
2 ML n-propanol 96.3 15.1 ± 0.1c 81.0 ± 0.2c
3 ML n-Butanol 97.3 ± 0.1 4.7b 90.2 ± 0.7b
4 ML 2-Propanol 100 95.3 ± 0.1a 2.3 ± 1.0a
5 ML 2-Butanol 100 88.7 ± 1.3d –
6 ML tert-Butanol 30.1 ± 0.7 15.1 ± 0.3c 36.2 ± 2.6d
7 EL 2-Propanol 100 82.9 ± 0.5e –
8 BL 2-Propanol 99.7 ± 0.5 87.7 ± 0.3f –

Reaction conditions: 0.5 g substrate, 10 g alcohols, 0.25 g Al(OiPr)3, 150 °C, 30 min.
*Levulinate esters means the transesterification products with corresponding H-honor.
a-f
Different letters above table indicate significant differences between mean values (p < 0.05).
–: Not detected.

2
T. Zhao, et al. Chemical Engineering Journal 390 (2020) 124505

Fig. 1. Effect of reaction conditions on the ML conversion and products selectivity, (a) effect of temperature (time of 30 min, catalyst/ML of 50%, 10 g 2-propanol),
(b) effect of residence time (temperature of 150 °C, catalyst/ML of 50%, 10 g 2-propanol), (c) effect of catalyst dosage (temperature of 150 °C, time of 30 min, 10 g 2-
propanol), (d) effect of solvent dosage (temperature of 150 °C, time of 30 min, catalyst/ML of 50%).

2. Material and methods 2.3. Effect of reaction parameters on the production of GVL

2.1. Chemicals and materials The experiments were carried out in a 35-mL quartz pressure re-
action vessel (PyNN Co. Ltd., USA). The H-donors, including ethanol,
Methyl levulinate (98%), ethyl levulinate (98%) and γ-valerolactone n-propanol, n-butanol, 2-propanol, 2-butanol, tert-butanol were ex-
(98%) were purchased from J & K Technology Co. Ltd. (Beijing, China). plored. The reaction vessel was loaded with 0.5 g ML, 0.25 g Al
Butyl levulinate (99%), methyl levulinate (Standard reagent) and ethyl (OiPr)3, and 10 g alcohols. The vessel was then heated to 150 °C
levulinate (Standard reagent) were obtained from TCI Chemical (within 2 min) and maintained at this temperature for 30 min with a
Industry Co., Ltd. (Shanghai, China). Zirconium oxychloride (99%) and middle magnetic stirring. After the reaction, the vessel was cooled to
aluminum isopropoxide (98%) was supplied by Sinopharm Chemical 26 °C using forced-air cooling equipment. Liquid production was
Reagent Co. Ltd. All other chemicals (AR) were from Beijing Chemical analyzed by gas chromatography (GC) and gas chromatography mass
Works and used without purification. spectrometry (GC–MS), respectively.
Based on the GVL yield, the optimal alcohol was used for the further
study. The reaction parameters of reaction temperature, residence time,
2.2. Comparative study of catalyst on GVL yield the dosages of catalyst and the solvent were then discussed according to
the reaction conditions presented in Table 1.
Comparative experiments of different catalysts were carried out in
Milestone microwave lab station (Italy) equipped with sealed Teflon
reaction vessels (100 mL). ZrO2 and Zr(OH)4 were prepared by pre- 2.4. Kinetic experiments
cursor ZrOCl2·8H2O according to previous studies, respectably [30,27].
The reaction mixture consisted of 1 g ML, 19 g 2-propanol, 0.5 g cat- The 10-mL quartz pressure reaction vessel (PyNN Co. Ltd., USA) was
alyst. The reaction temperature was 180 °C and reaction times were set loaded with 0.1 g ML, 1.6 g 2-propanol and 0.05 g Al(OiPr)3, and the
at 30, 60, and 120 min when ZrO2 was used as catalyst. The reaction mixture was heated to pre-set temperature (110–140 °C). After heating
was carried out at 200 °C in 40 min in the presence of Zr(OH)4. The for a certain time (1–60 min), the vessel was cooled to 26 °C before
reactions were conducted with Al(OiPr)3 for 30 min at 100 °C, 150 °C, being opened. The product was then quantitatively analyzed by GC
and 200 °C, respectively. All of the reaction conditions were set to instrument.
maximize the GVL yield for each catalyst.

3
T. Zhao, et al. Chemical Engineering Journal 390 (2020) 124505

Fig. 2. Distribution of products after the reaction with ML. a:110 °C, b:120 °C, c:130 °C, d:140 °C.

Table 4
Reaction rate constant and R-square for the reaction.
No Temperature k1 k2 k3 R12 R1 2 R12
°C L.mol−1.min−1 min−1 min−1

1 110 1.4907 0.0034 0.0632 0.9920 0.9785 0.9522


2 120 3.4100 0.0079 0.1327 0.9806 0.9824 0.9722
3 130 7.6199 0.0268 0.3055 0.9899 0.9914 0.9825
4 140 18.1080 0.1398 1.3837 0.9947 0.9537 0.9665

2.5. Sample analysis Mole of GVL


GVL yield (%) = × 100
Initial mole of substrate
2.5.1. GC/MS and GC analysis
The sample analysis was conducted with an Agilent 5975C GC/MS GVL yield
GVL selectivity (%) = × 100
Instrument with DB-5MS column (30.0 m × 0.32 mm × 0.25 μm) and Substrate conversion
electron impact ionization (EI). Mole of IPL
The compositions of the resultants were quantitatively determined IPLyield (%) = × 100
Initial mole of substrate
by a Shimadzu GC 2014C with a DB-5 column (30.0 m ×
0.32 mm × 0.25 μm) and an FID detector operating at 270 °C. N2 was IPL yield
the carrier gas and the flow rate was 1.0 mL/min. The following pro- IPL selectivity (%) = × 100
Substrate conversion
grammed temperature was used: 50 °C (3 min) – 6 °C/min – 180 °C –
10 °C/min – 280 °C(3 min). The substrate conversion, GVL yield, GVL
selectivity, GVL formation rate, IPL (Isopropyl levulinate) yield, and IPL 2.5.2. LC/MS analysis
selectivity were calculated by the following equations: For the identification of transition states (TSs) products, the sample
was prepared under mild conditions (100 °C, 10 min) and analyzed by
Mole of substrate after reaction LC/MS (Thermos TSQ Quantum Ultra) equipped with a refractive index
Substrate conversion (%) = 1 × 100
Initial mole of substrate detector and a capillary column ultimate XB C18 (3um × 100 mm
length). The mobile phase was the mixture of acetonitrile and water
(80/20, v/v), the flow rate was 1.0 mL/min. MS analysis in the negative

4
T. Zhao, et al. Chemical Engineering Journal 390 (2020) 124505

ZrO2 was stable with presence of air, but it required high temperature
to work efficiently (high energy consumption). Besides, prolonged re-
action time resulted in a decrease in GVL yield (Table 2, entry 3–4), it
might be because more side reactions occurred, such as aldol con-
densation between GVL, EL and acetone (dehydrogenation of iso-
propanol) [30,50]. A progressive deactivation of the catalyst would also
be a reason for the lower catalytic activity [51]. On the other hand, Al
(OiPr)3 could produce 48.9% of GVL at low temperature of 100 °C in
30 min with presence of air. When the temperature increased to 150 °C,
the GVL yield arose to 85.2%, which was comparable to the Zr(OH)4
reported by Tang et al. (Table 2, entry 5, 7–9). The GVL production
with Al(OiPr)3 can be identified as a time-saving and energy-efficient
process. Hence, Al(OiPr)3 with high catalytic activity could be used for
the further study.

3.2. The effect of H-donor

Various alcohols as H-donors were evaluated on GVL production in


this study (Table 3). The ML conversion with different alcohols except
tert-butanol were all above 95%, and the secondary alcohols including
Fig. 3. Arrhenius plot for each step in the conversion of ML to GVL from 110 °C 2-propanol and 2-butanol exhibited excellent performances with the
to 140 °C. GVL selectivity of 95.3% and 88.3%, respectively (Table 3, entry 4–5).
In addition, the outstanding performance of 2-propanol was still ob-
ions mode was performed on a mass spectrometer equipped with an ESI served when the substrate was replaced by EL and BL (Table 3, entry
ion source. The ESI spray volt was set at 3.5 kV and the mass scan range 7–8). When the primary alcohols were employed, most of ML was
was 50–600 m/z. The heated capillary was set to 370 °C. converted into corresponding levulinate esters through transesterifica-
tion (Table 3, entry 1–3). This result indicated that Al(OiPr)3 could
efficiently catalyze both of the transesterification and hydrogenation.
2.6. Computational method
The variation on the GVL production might be attributed to the dif-
ferences between the reduction potentials of alcohols used. The re-
The computational calculation was performed to provide mechanistic
duction potential was represented as ability of providing H-atoms and
understanding of the chemistry of the desired reaction catalyzed by Al
defined as the difference of the standard molar enthalpy of formation
(OiPr)3. All of the calculations were performed with Gaussian 16
between the alcohol and the corresponding carbonyl compound. Dif-
package. The geometry optimizations in this study were carried out using
ferent alcohols based on the reduction potentials descended in the
the M062X-D3 method [45] with the basis set of def2-TZVP [46]. In the
following order: n-propanol (87.3 KJ/mol) > ethanol (85.4 KJ/
M062X-D3 method, the D3 term represented a dispersion correlation
mol) > n-Butanol (79.7 KJ/mol) > 2-Propanol (70.0 KJ/mol) ≈ 2-
[47]. All the structures were fully relaxed, and vibrational frequencies
butanol (69.3 KJ/mol) [52], which was in line with our results except n-
were calculated. The initial geometries of transition structures were
propanol. In this case, alkyl chain length might be the dominate factor.
manually guessed and further verified by the presence of a single ima-
The sum of the selectivity of GVL and levulinate esters was less than
ginary frequency and IRC (intrinsic reaction coordinate) calculations
100%, a small amount of by-products probably generated via aldol
[48]. To account for the effects of solution environment around the
condensation, dehydration or derivative of intermediate and GVL (Figs.
catalytic active site, the calculations were performed in a 2-propanol
S1 and S2) [27,50,53]. The reaction with tertiary alcohol i.e. tert-Bu-
dielectric using the SMD solvation model at the M062X level of theory
tanol gave a poor result of a ML conversion of 30.1% and a GVL se-
with the same basis sets as used for the geometry evaluation [49].
lectivity of 15.1% (Table 3, entry 6). It was probably due to the steric
effect suppressing the formation of transition state for its high stereo
3. Results and discussion structure and electron density, and more side reactions occurred during
the process.
3.1. Comparison of catalyst based on MPV reduction
3.3. The effect of the reaction parameters
To evaluate the catalytic activity of Al(OiPr)3, the comparative ex-
periments with ZrO2 and Zr(OH)4 based on MPV reduction were carried The effect of reaction parameters including reaction temperature,
out. As shown in Table 2, 63.7% of GVL yield was obtained when Zr residence time, the dosages of catalyst and the solvent was thoroughly
(OH)4 was used as catalyst in the absence of N2, which was much lower investigated with 2-propanol being the H-donor.
than the result obtained in N2 atmosphere reported by Tang et al. As shown in Fig. 1a, the conversion of ML was complete under
(Table 2, entry 5–6) [27]. This observation revealed that the reaction 130 °C. The GVL selectivity linearly increased from 67.4% to 95.3% as
system with Zr(OH)4 might be sensitive to oxygen. When ZrO2 was used the reaction temperature increased from 130 °C to 150 °C. Further in-
as catalyst, the maximum EL conversion (79.2%) and GVL yield crease the temperature resulted in a slight drop in GVL yield, which
(71.5%) were both higher than that with Zr(OH)4, it might be due to might be due to the acceleration of side reaction [27]. Residence time
more catalytic sites (i.e.,Zr) existed with the same dosage of different was proved to be another dominate factor (Fig. 1b). Along with the
catalysts (Table 2, entry 3,6). Howerve, these results was still lower increase of time from 10 min to 60 min, a maximum selectivity of GVL
than that reported by Tang (Table 2, entry 1,3) [30]. It indicated that 95.3% was achieved at 30 min. It was noteworthy that IPL was

5
T. Zhao, et al. Chemical Engineering Journal 390 (2020) 124505

Table 5
The interpretation of main fragment ions in mass spectra.
Compound Chemical formula Chemical structure m/z

1 C6H10O3 129.95

2 C8H14O3 157.96

3 C5H8O2 100.97

4 C9H21AlO3·H+ 205.05

5 C9H21AlO3·H3O+ 222.95

6 C12H28AlO4·H3O+ 283.00

7 C9H18O4·K+ 229.15

10 C17H35AlO6·Na+·H3O+ 404.86

(continued on next page)

6
T. Zhao, et al. Chemical Engineering Journal 390 (2020) 124505

Table 5 (continued)

Compound Chemical formula Chemical structure m/z

+
11 C14H29AlO5·H3O 322.79

12 C8H16O3·H3O+ 179.05

13 C11H23O4 218.95

generally formed at lower temperature or shorter time, and its se- dCML I ±1
=-k1CML CIPA
1 I ±1
k3CML CIPA
3
lectivity declined as the GVL selectivity increased until it reached a dt
constant value below 2%. Besides, the sum of the two products se-
dCIPL
lectivity is generally above 90%. This phenomenon indicated that IPL I ±1
=k1CML CIPA
1 I ±2
k2CIPL CIPA
2
dt
might be an intermediate which was then converted to GVL.
The catalyst dosage is an important parameter to measure the cat- dCGVL I ±2
=k2CIPL 2 + k C I ± 1C 3
CIPA
alytic activity and the feasibility of industrial production. Hence, the dt
3 ML IPA

experiments of various catalyst dosage were carried out in our study


and the results were shown in Fig. 1c. A dramatic increase in GVL se- where Ci terms were the molar concentration for species i, and ki was
lectivity from 35.8% to 95.3% was observed as the catalyst dosage in- the apparent rate constant of step i. αi was the reaction order of ML or
creased from 12.5% to 50%. At the same time, the selectivity of IPL IPL in each step, and βi was for 2-propanol (IPA).
decreased from 64.1% to 2.3%. These results revealed that transester- In excess of 2-propanol, its difference of concentration could be
ification could occur with lower dosage while the formation of GVL negligible, and then the model equations could be simplified as follows:
required high dosage. Taking the economic principle of industrial dCML I ±1 I ±1
=-k1CML k3CML
production into account, the solvent dosage also needs to be in- dt
vestigated (Fig. 1d). ML was mostly converted and GVL selectivity of
97.6% was higher than other dosages at the solvent loading of 16:1. Too dCIPL I ±1 I ±2
=k1CML k2CIPL
low or too high loading would generate a certain amount of IPL, leading dt
to the decrease in the selectivity of GVL. Hence, 16:1 was found to be dCGVL I ±2 I ±1
the optimized dosage in this study. =k2CIPL + k3CML
dt
The Matlab was used to optimise the rate constants and the results
3.4. Kinetic analysis was summarized in Table 4. Transesterification was fitted well to quasi-
second-order model, and the two hydrogenation-cyclization steps were
The kinetics experiments were carried out over a range of tem- fitted to a first-order equation. In particular, the reaction rate constant
perature (Fig. 2, 110-140°C, 10 °C for an interval). Previous studies hold of the transesterification was 8–20 times of the direct hydrogenation of
that ML underwent hydrogenation and cyclization via an unstable in- ML to GVL at the same temperature. Compared to the ML, the hydro-
termediate, i.e. methyl 4-hydroxypentanoate, and IPL was formed as the genation of IPL to GVL was much slower. These results indicated that
by-product [54,55]. However, our kinetic experiments showed that a these two reaction pathways were both existed but the two-step process
large amount of IPL was generated in early stage, and its rates of for- was the dominate one. As shown in Fig. 3, the apparent activation
mation and removal were both proportional to temperature, which energies (Ea) of hydrogenation-cyclization was higher than that of
indicated that ML was more rapidly to undergo transesterification than transesterification (38.8 vs 23.5 kcal/mol), indicating that the fomer
hydrogenation. In addition, IPL could be efficiently converted to GVL as was the rate-limiting step.
ML dose (Fig. S3). Based on these results, two plausible reaction
pathways were proposed in Fig. S4. 3.5. Reaction mechanism
Due to the rapid nature of the methyl 4-hydroxyvalerate and iso-
propyl 4-hydroxyvalerate (4-HIPL) cyclization to GVL, hydrogenation 3.5.1. Potential reaction routes
and cyclization were combined as one step here. The kinetic equations To elucidate the whole reaction pathway, LC-MS was employed to
for the reaction network were shown as follows: detect the possible intermediate compounds of the reaction mixture.

7
T. Zhao, et al. Chemical Engineering Journal 390 (2020) 124505

Scheme 1. Proposed mechanisms for the transesterification of ML to IPL over Al(OiPr)3.

The chromatogram spectrum and the mass spectra of different retention property. The oxygen of 2-propanol was then connected to the ester
time were shown in Fig S3 and Fig S4, respectively. The interpretation carbonyl carbon and the hydrogen of 2-propanol attached to the ester
of main fragment ions in mass spectra was shown in Table 5. As we can oxygen forming an unstable TS1(Scheme 1a) [56]. Alternatively, the Al
see, the ML molecular was bound to 2-proponal forming unstable (OiPr)3 combined with 2-propanol and then the Al atom directly par-
complex, which was with K+ to give a specie at m/z 229.34. But the ticipated in the formation of TSs after attached to the ester carbonyl
experiments have proven that no IPL was generated during the reaction oxygen. In this case, the ester carbonyl carbon conected to the alkoxy
without catalyst. Moreover, despite the fragments with m/z 205.05 and oxygen of aluminum complex forming a four-centered TS2. After the
222.95 were attributed to the combination of Al(OiPr)3 with H+ and electron transfer, a new linkage of C-O-Al was formed. The methoxy
H3O+, the catalyst could also be attached to the 2-proponal forming a oxygen was then attached to the Al atom and a new four-centered TS3
four-coordinate aluminum complex, which was with H3O+ to give a was generated [57,58]. Through the electron transfer, IPL was finally
specie at m/z 282.36. Based on these results, plausible transesterifica- formed (Scheme 1b). In addition, the Al atom of the complex could be
tion routes were proposed in Scheme 1. Due to the electrophilicity, the also connected to the ester oxygen. As shown in Scheme 1c, a four-
Al atom of Al(OiPr)3 was readily attract to the ester carbonyl oxygen centered TS4 was probably formed via the connection of ester carbonyl
which led the carbonyl carbon exhibiting relative strong electropositive carbon to alkoxy oxygen of aluminum complex [42].

8
T. Zhao, et al. Chemical Engineering Journal 390 (2020) 124505

Scheme 2. Proposed mechanism for the hydrogenation of IPL to 4-hydroxypentanoate isopropyl over Al(OiPr)3 using 2-proponal as hydrogen donor.

IPL combined with one molecular of Al(OiPr)3, Na+ and H3O+ 3.5.2. Computational results of possible reaction routes
giving an unstable peak at m/z 404.46. It was inferred that hydride To find out the optimal route of the dominate pathway and in-depth
probably transferred from the alkoxy carbon of isopropoxide to the understanding of the reaction mechanism, computational simulation
ketone carbonyl carbon of IPL through the TS1 (Scheme 2). The specie was employed in our study. The free energy of each route of transes-
further released acetone and Na+ forming a specie with m/z 322.79. terification and cyclization process was shown in Fig. 4a and Fig. 4b,
Subsequently, a solvent molecule attaching to the Al atom provided a respectively.
proton via a four-centered TS2 that lead to the formation of 4-HIPL Fig. 4a showed that the activation energy with route a was
(isopropyl 4-hydroxyvalerate, m/z 179) and Al(OiPr)3. The mechanism 48.0 kcal/mol, which was the highest energy barrier of these three
of hydrogenation step was also in agreement with previous studies for routes. Between the two reaction routes assisted by aluminum complex
the hydrogenation of ketones through MPV reduction [42,43]. (Route b and c), two-step process was completed with a 28.2 kcal/mol
Although the kinetic model of 4-hydroxy pentanoic acid or its esters energy barrier, which was lower than the one-step process of 38.2 kcal/
converted to GVL has been explored by previous studies, to the best of mol.
our knowledge, the clear mechanism of ring-closure has not been re- The calculated free energies of four routes for the cyclization were
ported yet [21,52]. Based on the experimentally inferred intermediate shown in Fig. 4b. The intramolecular cyclization of 4-HIPL should
species and earlier studies, four possible routes were explored and overcome 48.5 kcal/mol energy barrier, which was higher than the 2-
proposed in scheme 3. The first route was the intramolecular cycliza- propanol route (37.8 kcal/mol, route b). Comparing to the Al(OiPr)3
tion of 4-HIPL to GVL, which was in line with the easily reported studies and bis-molecule cyclization, 2-propanol route was also proved to be
for the catalytic conversion of LA to GVL in aqueous medium the optimal pathway. These observations revealed that the cyclization
(Scheme 3a) [20,21]. In addition, the cyclization with the assistance of was properly carried out with the synergistic effect of 2-propanol, and
2-propanol was confirmed by LC/MS data (Table 5, entry 13). As an other situations were almost impossible to occur.
electron transfer carrier, there was greater probability of 2-proponal Based on the above conclusions, the reaction pathway could be
involved in the formation of stable six-centered TS2 and promoted the completely drawn. The free energy and enthalpy profile were all shown
ring-closure process (Scheme 3b). On the basis of this finding, Al(OiPr)3 in Fig. 5. As we can see from the Fig. 5, transesterification was an
was expected to be an alternative electron transfer carrier participating exothermic reaction including two steps (P0 → P2). Al(OiPr)3 was
in the reaction (Scheme 3c). The last possible route was the inter-mo- firstly coordinated with one solvent molecule, followed by the iso-
lecular cyclization between two molecules of 4-HIPL (Scheme 3d). The propoxy transfer from the aluminum complex to the ester carbonyl
electron transfers probably occurred via a symmetric eight-center TS4 carbon of ML (P0-P1) through the TS1. Subsequently, the methoxy
that produced two molecules of GVL. oxygen was connected to the Al atom and then the methoxy was
Unlike some other studies, the intermediates and TSs of direct hy- transferred to the Al(OiPr)3 (P1-P2). The hydrogenation was an ex-
drogenation of ML were not detected in this reaction system. ergonic and exothermic process (P2 → P5). Hydrogen transfer (R3)
Nevertheless, the possibility of direct hydrogenation still existed, and a from the 2° carbon of isopropoxide to the ketone carbonyl carbon of IPL
detailed discussion would be done in the subsequent computational occurred forming the intermediate P3. Subsequently, the solvent mo-
simulation section. lecule replaced acetone to form intermediate R4, the –OH of solvent

9
T. Zhao, et al. Chemical Engineering Journal 390 (2020) 124505

Scheme 3. Proposed mechanisms for the cyclization of 4-HIPL.

provide a proton to complete a new hydrogen transfer. After the pro- As mentioned above, the direct hydrogenation of ML was explored
cess, Al(OiPr)3 regenerated and the intermediate of 4-HIPL was formed here. The reaction route was confirmed based on the hydrogenation of
(P4). Finally, GVL was generated through the cyclization, which was IPL. As shown in Fig. 6., the energy barrier (16.6 kcal/mol) was a little
accomplished with one solvent molecule as a carrier of hydrogen higher than that of IPL (14.3 kcal/mol), but lower than the transes-
transfer (P4-P5). The energy barrier of 37.4 kcal/mol of the cyclization terification process as shown above. This phenomenon confirmed the
made the hydrogenation-cyclization a rate-limiting step during the probability of direct hydrogenation of ML on the theoretical perspec-
whole reaction, which was also in agreement with the experiment data. tive. Taking the experimental data and LC/MS analysis the into

10
T. Zhao, et al. Chemical Engineering Journal 390 (2020) 124505

Fig. 6. Free energy profile for the gas phase hydrogenation of ML or IPL to
4HIPL.

account, direct hydrogenation of ML might account for an small part.


This finding could be helpful for some other theoretical studies on
catalytic hydrogenation of levulinate esters by alkoxides.

4. Conclusion

Al(OiPr)3 exhibited a high performance on the catalytic hydro-


genation transfer of ML to GVL. The process optimization showed that
97.6% of GVL was obtainable when 2-propanol was the H-donor and
solvent without inert gas protection at 150 °C in 30 min. The transes-
terification occurred prior to the hydrogenation-cyclization, and its
reaction activation energy were 23.5 kcal/mol. The hydrogenation-cy-
clization was regarded as the rate-limiting step with an activation en-
ergy of 38.8 kcal/mol. LC/MS data and computational results revealed
that the transesterification was preceded through a two-step electron
Fig. 4. Computed free energy of three routes for the transesterification of ML to transfer with the activation of ester bond with Al atom as the transfer
IPL (a) and four routes for the cyclization of 4-HIPL to GVL (b). carrier, and then a two-step proton transfer occurred through carbonyl

Fig. 5. Computed (M062X) free energy profile and enthalpy profiles for the catalytic synthesis of ML to GVL by Al(OiPr)3. Note that all energies are reported in Kcal/
mol and a solvation model for 2-proponal is used in all the step.

11
T. Zhao, et al. Chemical Engineering Journal 390 (2020) 124505

bond activation with Al atom to obtain 4-HIPL, followed by cyclization zeolite in water, Appl. Catal. A: General 470 (2014) 215–220.
with the assistance of one molecular of 2-propanol to accomplish the [20] S. Li, Y. Wang, Y. Yang, B. Chen, J. Tai, H. Liu, B. Han, Conversion of levulinic acid
to gamma-valerolactone over ultra-thin TiO2 nanosheets decorated with ultrasmall
reaction. Ru nanoparticle catalysts under mild conditions, Green Chem. 21 (2019) 770–774.
[21] A.S. Piskun, H.H. van de Bovenkamp, C.B. Rasrendra, J.G.M. Winkelman,
Declaration of Competing Interest H.J. Heeres, Kinetic modeling of levulinic acid hydrogenation to γ-valerolactone in
water using a carbon supported Ru catalyst, Appl. Catal. A: General 525 (2016)
158–167.
The authors declare that they have no known competing financial [22] D. Zhao, Y. Wang, F. Delbecq, C. Len, Continuous flow conversion of alkyl levuli-
interests or personal relationships that could have appeared to influ- nates into gamma-valerolactone in the presence of Ru/C as catalyst, Mol. Catal. 475
(2019) 110456.
ence the work reported in this paper. [23] H. Zhou, J. Song, X. Kang, J. Hu, Y. Yang, H. Fan, Q. Meng, B. Han, One-pot con-
version of carbohydrates into gamma-valerolactone catalyzed by highly cross-
Acknowledgments linked ionic liquid polymer and Co/TiO2, RSC Adv. 5 (2015) 15267–15273.
[24] H. Li, Z. Fang, S. Yang, Direct conversion of sugars and ethyl levulinate into γ-
valerolactone with superparamagnetic acid-base bifunctional ZrFeOx nanocatalysts,
The authors would like to acknowledge the financial support from ACS Sustainable Chem. Eng. 4 (2015) 236–246.
the National Key R&D Program of China (2016YFE0112800), the [25] H.Y. Luo, D.F. Consoli, W.R. Gunther, Y. Román-Leshkov, Investigation of the re-
National Natural Science Foundation of China (Grant No. 31671572), action kinetics of isolated Lewis acid sites in Beta zeolites for the
Meerwein–Ponndorf–Verley reduction of methyl levulinate to γ-valerolactone, J.
and the National Higher-education Institution General Research and Catal. 320 (2014) 198–207.
Development Funding (No. 2018TC027). [26] J. Song, L. Wu, B. Zhou, H. Zhou, H. Fan, Y. Yang, Q. Meng, B. Han, A new porous
Zr-containing catalyst with a phenate group: an efficient catalyst for the catalytic
transfer hydrogenation of ethyl levulinate to gamma-valerolactone, Green Chem. 17
Appendix A. Supplementary data (2015) 1626–1632.
[27] X. Tang, H. Chen, L. Hu, W. Hao, Y. Sun, X. Zeng, L. Lin, S. Liu, Conversion of
Supplementary data to this article can be found online at https:// biomass to γ-valerolactone by catalytic transfer hydrogenation of ethyl levulinate
over metal hydroxides, Appl. Catal. B: Environ. 147 (2014) 827–834.
doi.org/10.1016/j.cej.2020.124505. [28] A. Corma, M.E. Domine, S. Valencia, Water-resistant solid Lewis acid catalysts:
Meerwein–Ponndorf–Verley and Oppenauer reactions catalyzed by tin-beta zeolite,
References J. Catal. 215 (2003) 294–304.
[29] M. Chia, J.A. Dumesic, Liquid-phase catalytic transfer hydrogenation and cycliza-
tion of levulinic acid and its esters to γ-valerolactone over metal oxide catalysts,
[1] T. Werpy, G. Petersen, A. Aden, J. Bozell, J. Holladay, J. White, A. Manheim, D. Chem. Commun. 47 (2011) 12233–12235.
Eliot, L. Lasure, S Jones, DTIC Document: 2004. [30] X. Tang, L. Hu, Y. Sun, G. Zhao, W. Hao, L. Lin, Conversion of biomass-derived ethyl
[2] K. Krommyda, C. Panopoulou, C. Moustani, E. Anagnostopoulou, K. Makripidi, levulinate into γ-valerolactone via hydrogen transfer from supercritical ethanol
G. Papadogianakis, A remarkable effect of aluminum on the novel and efficient over a ZrO2 catalyst, RSC Adv. 3 (2013) 10277–10284.
aqueous-phase hydrogenation of levulinic acid into γ-valerolactone using water- [31] J. Song, B. Zhou, H. Zhou, L. Wu, Q. Meng, Z. Liu, B. Han, Porous zirconium-phytic
soluble platinum catalysts modified with nitrogen-containing ligands, Catal. Lett. acid hybrid: a highly efficient catalyst for Meerwein-Ponndorf-Verley reductions,
149 (2019) 1250–1265. Angew. Chem. Int. Ed. 54 (2015) 9399–9403.
[3] D. Liu, L. Zhang, W. Han, M. Tang, L. Zhou, Y. Zhang, X. Li, Z. Qin, H. Yang, One- [32] L. Bui, H. Luo, W.R. Gunther, Y. Román-Leshkov, Domino reaction catalyzed by
step fabrication of Ni-embedded hierarchically-porous carbon microspheres for le- zeolites with brønsted and lewis acid sites for the production of γ-valerolactone
vulinic acid hydrogenation, Chem. Eng. J. 369 (2019) 386–393. from furfural, Angew. Chem. Int. Ed. 52 (2013) 8022–8025.
[4] J. Lange, J.Z. Vestering, R.J. Haan, Towards‘bio-based’Nylon: conversion of γ-va- [33] Y. Xie, F. Li, J. Wang, R. Wang, H. Wang, X. Liu, Y. Xia, Catalytic transfer hydro-
lerolactone to methyl pentenoate under catalytic distillation conditions, Chem. genation of ethyl levulinate to γ-valerolactone over a novel porous Zirconium tri-
Commun. 2007 (2007) 3488–3490. metaphosphate, Mol. Catal. 442 (2017) 107–114.
[5] Y. Yang, X.R. Wei, F.X. Zeng, L. Deng, Efficient and sustainable transformation of [34] J. He, H. Li, Y. Lu, Y. Liu, Z. Wu, D. Hu, S. Yang, Cascade catalytic transfer hy-
gamma-valerolactone into nylon monomers, Green Chem. 18 (2016) 691–694. drogenation–cyclization of ethyl levulinate to γ-valerolactone with Al–Zr mixed
[6] D. Sun, T. Saito, Y. Yamada, X. Chen, S. Sato, Hydrogenation of γ-valerolactone to oxides, Appl. Catal. A: General 510 (2016) 11–19.
1,4-pentanediol in a continuous flow reactor, Appl. Catal. A: General 542 (2017) [35] W. Ouyang, D. Zhao, Y. Wang, A.M. Balu, C. Len, R. Luque, Continuous flow con-
289–295. version of biomass-derived methyl levulinate into γ-valerolactone using functional
[7] C.E. Chan-Thaw, M. Marelli, R. Psaro, N. Ravasio, F. Zaccheria, New generation metal organic frameworks, ACS Sustain. Chem. Eng. 6 (2018) 6746–6752.
biofuels: γ-valerolactone into valeric esters in one pot, RSC Adv. 3 (2013) [36] P. Nandi, A. Solovyov, A. Okrut, A. Katz, AIII–Calix[4]arene catalysts for asym-
1302–1306. metric Meerwein–Ponndorf–Verley reduction, ACS Catal. 4 (2014) 2492–2495.
[8] J.C. Serrano-Ruiz, D. Wang, J.A. Dumesic, Catalytic upgrading of levulinic acid to 5- [37] K. Flack, K. Kitagawa, P. Pollet, C.A. Eckert, K. Richman, J. Stringer, W. Dubay,
nonanone, Green Chem. 12 (2010) 574–577. C.L. Liotta, Al(OtBu)3 as an effective catalyst for the enhancement of
[9] F.D. Pileidis, M. Titirici, Levulinic acid biorefineries: new challenges for efficient Meerwein–Ponndorf–Verley (MPV) reductions, Organic Process Res. Development
utilization of biomass, ChemSusChem 9 (2016) 562–582. 16 (2012) 1301–1306.
[10] K. Yan, Y. Yang, J. Chai, Y. Lu, Catalytic reactions of gamma-valerolactone: a [38] C.F. De Graauw, J.A. Peters, H. van Bekkum, J. Huskens, Meerwein-Ponndorf-
platform to fuels and value-added chemicals, Appl. Catal. B: Environ. 179 (2015) Verley reductions and oppenauer oxidations: an integrated approach, Synthesis-
292–304. Stuttgart 1994 (1994) 1007–1017.
[11] H. Li, Z. Fang, S. Yang, Direct conversion of sugars and ethyl levulinate into γ- [39] J. Geboers, X. Wang, A.B. de Carvalho, R. Rinaldi, Densification of biorefinery
valerolactone with superparamagnetic acid–base bifunctional ZrFeOx nanocata- schemes by H-transfer with Raney Ni and 2-propanol: a case study of a potential
lysts, ACS Sustain. Chem. Eng. 4 (2016) 236–246. avenue for valorization of alkyl levulinates to alkyl γ-hydroxypentanoates and γ-
[12] Q. Long, I.T. Horváth, Catalytic conversion of fructose to γ-valerolactone in γ-va- valerolactone, J. Mol. Catal. A: Chem. 388–389 (2014) 106–115.
lerolactone, ACS Catal. 2 (2012) 2247–2249. [40] C.R. Graves, E. Joseph Campbell, S.T. Nguyen, Aluminum-based catalysts for the
[13] P.A. Son, S. Nishimura, K. Ebitani, Production of γ-valerolactone from biomass- asymmetric Meerwein–Schmidt–Ponndorf–Verley–Oppenauer (MSPVO) reaction
derived compounds using formic acid as a hydrogen source over supported metal manifold, Tetrahedron: Asymmetry 16 (2005) 3460–3468.
catalysts in water solvent, RSC Adv. 4 (2014) 10525–10530. [41] J.S. Cha, Recent Developments In Meerwein-Ponndorf-Verley and related reactions
[14] M.A. Mellmer, D.M. Alonso, J.S. Luterbacher, J.M.R. Gallo, J.A. Dumesic, Effects of for the reduction of organic functional groups using aluminum, boron, and other
γ-valerolactone in hydrolysis of lignocellulosic biomass to monosaccharides, Green metal reagents: a review, Org. Process Res. Dev. 10 (2006) 1032–1053.
Chem. 16 (2014) 4659–4662. [42] R. Cohen, C.R. Graves, S.T. Nguyen, J.M.L. Martin, M.A. Ratner, The mechanism of
[15] S. Song, S. Yao, J. Cao, L. Di, G. Wu, N. Guan, L. Li, Heterostructured Ni/NiO aluminum-catalyzed Meerwein-Schmidt-Ponndorf-Verley reduction of carbonyls to
composite as a robust catalyst for the hydrogenation of levulinic acid to γ-valer- alcohols, J. Am. Chem. Soc. 126 (2004) 14796–14803.
olactone, Appl. Catal. B: Environ. 217 (2017) 115–124. [43] R.S. Assary, L.A. Curtiss, J.A. Dumesic, Exploring Meerwein–Ponndorf–Verley re-
[16] X. Li, J. Li, X. Liu, Q. Tian, C. Hu, The promoting effect of ce on the performance of duction chemistry for biomass catalysis using a first-principles approach, ACS Catal.
Au/CexZr1−xO2 for γ-valerolactone production from biomass-based levulinic acid 3 (2013) 2694–2704.
and formic acid, Catalysts 8 (2018) 241. [44] G.I. Nikonov, New tricks for an old dog: aluminum compounds as catalysts in re-
[17] A.M. Hengne, N.S. Biradar, C.V. Rode, Surface species of supported ruthenium duction chemistry, ACS Catal. 7 (2017) 7257–7266.
catalysts in selective hydrogenation of levulinic esters for bio-refinery application, [45] Y. Zhao, D.G. Truhlar, Density functionals with broad applicability in chemistry,
Catal. Lett. 142 (2012) 779–787. Accounts Chem. Res. 41 (2008) 157–167.
[18] S. Wang, H. Huang, V. Dorcet, T. Roisnel, C. Bruneau, C. Fischmeister, Efficient [46] A. Schäfer, C. Huber, R. Ahlrichs, Fully optimized contracted Gaussian basis sets of
iridium catalysts for base-free hydrogenation of levulinic acid, Organometallics 36 triple zeta valence quality for atoms Li to Kr, J. Chem. Phys. 100 (1994) 5829–5835.
(2017) 3152–3162. [47] S. Grimme, S. Ehrlich, L. Goerigk, Effect of the damping function in dispersion
[19] J.M. Nadgeri, N. Hiyoshi, A. Yamaguchi, O. Sato, M. Shirai, Liquid phase hydro- corrected density functional theory, J. Comput. Chem. 32 (2011) 1456–1465.
genation of methyl levulinate over the mixture of supported ruthenium catalyst and [48] C. Gonzalez, H.B. Schlegel, Improved algorithms for reaction path following:

12
T. Zhao, et al. Chemical Engineering Journal 390 (2020) 124505

higher-order implicit algorithms, J. Chem. Phys. 95 (1991) 5853–5860. adjustable Lewisand Brønsted acid sites, Appl. Catal. B: Environ. 214 (2017) 67–77.
[49] A.V. Marenich, C.J. Cramer, D.G. Truhlar, Universal solvation model based on so- [54] H.Y. Luo, D.F. Consoli, W.R. Gunther, Y. Román-Leshkov, Investigation of the re-
lute electron density and on a continuum model of the solvent defined by the bulk action kinetics of isolated Lewis acid sites in Beta zeolites for the Meerwein-
dielectric constant and atomic surface tensions, J. Phys. Chem. B 113 (2009) Ponndorf-Verley reduction of methyl levulinate to γ-valerolactone, J. Catal. 320
6378–6396. (2014) 198–207.
[50] K. Dong, J. Zhang, W. Luo, L. Su, Z. Huang, Catalytic transfer hydrogenation of [55] L. Negahdar, M.G. Al-Shaal, F.J. Holzhäuser, R. Palkovits, Kinetic analysis of the
ethyl levulinate into γ-valerolactone over zirconium oxide derived from zirconyl catalytic hydrogenation of alkyl levulinates to γ-valerolactone, Chem. Eng. Sci. 158
nitrate, J. Biobased Mater. Bio. 11 (2017) 543–552. (2017) 545–551.
[51] P.B. Vásquez, T. Tabanelli, E. Monti, S. Albonetti, D. Bonincontro, N. Dimitratos, [56] G.W. Parshall, S.D. Ittel, Homogeneous Catalysis, Wiley interscience, New York,
F. Cavani, Gas-phase catalytic transfer hydrogenation of methyl levulinate with 1992.
ethanol over ZrO2, ACS Sustain. Chem. Eng. 7 (2019) 8317–8330. [57] D.C. Bradley, R.C. Mehrotra, D.P. Gaur, Metal alkoxides, Academic Press, London,
[52] J.C. Van der Waal, P.J. Kunkeler, K. Tan, H. Van Bekkum, Zeolite titanium beta: a 1978.
selective catalyst for the gas-phase Meerwein–Ponndorf–Verley, and Oppenauer [58] I. Shigemoto, T. Kawakami, H. Taiko, M. Okumura, A quantum chemical study on
reactions, J. Catal. 173 (1998) 74–83. the polycondensation reaction of polyesters: the mechanism of catalysis in the
[53] F. Li, L.J. France, Z. Cai, Y. Li, S. Liu, H. Lou, J. Long, X. Li, Catalytic transfer polycondensation reaction, Polymer 52 (2011) 3443–3450.
hydrogenation of butyl levulinate to γ-alerolactone over zirconium phosphates with

13

You might also like