R. Zhu et al.
Applied Catalysis A, General 608 (2020) 117888
process of decarbonylation, so we collected the gas after the reaction for
GC analysis. It could be seen from Fig. S3ꢀ 1 that the peak at 0.57 min
was attributed to the reaction gas hydrogen, the peak at 2.24 min was
determined as carbon monoxide, and no carbon dioxide was detected.
These results confirmed our assumption. In addition, H. A. Bate [59] has
reported that oxonium ion intermediates can be produced by decar-
bonylation of THFCA and its derivatives by using concentrated sulfuric
acid (Brønsted acid). Based on this report, we supposed that the triflate
in the present co-catalytic system, as a Lewis acid, might also played a
similar role in decarbonylation. Therefore, we used reaction in-
termediates as substrates for further reaction studies to investigate the
specific effect of triflate on the reaction process.
From the perspective of theoretical calculation, the steric resistance
of the lanthanide metal is too large, resulting in a low computational
efficiency. And the transformation from furancarboxylic acid to buta-
nediol is a multi-stage tandem reaction, which is not able to be easily
explained by simple theoretical calculations. Therefore, we designed the
following experiments (Scheme 3) to explain the reaction mechanism.
We firstly used intermediate THFCA as the substrate to carry out the
reaction under the optimal reaction conditions, obtaining a BDA yield of
76.4 %. In order to further explore the specific role of Pd/C and triflates
in the decarbonylation process, we tried the reaction under H2 atmo-
sphere with Pd/C as catalyst alone, and the results showed that THFCA
was almost not converted. While, when using La(OTf)3 as the catalyst
alone, all of THFCA was converted and amounts of CO was detected but
no BDA was detected. Therefore, it was speculated that the presence of
Lewis acid La(OTf)3 resulted in the decarbonylation of THFCA. How-
ever, due to the lack of hydrogenation catalyst in the system, the oxo-
nium ions produced by THFCA decarbonylation could not be rapidly
hydrogenated to THF and some polymers would be generated, thus the
last step esterification ring opening process could not occur. After that,
we used Pd/C + La(OTf)3 co-catalytic system to carry out the reaction
under N2 atmosphere. The conversion of THFCA was complete, but there
was no formation of BDA. A large amount of reactants was polymerized,
which further validated our conjecture. The role of La(OTf)3 (Lewis
acid) in decarbonylation was similar to that of concentrated sulfuric acid
(Brønsted acid). Therefore, we believed that THFCA first produced
oxonium ions by decarbonylation under the effect of Lewis acid La
(OTf)3, and then rapidly hydrogenated to THF by Pd/C catalyst. More-
over, 90.4 % yield of BDA was obtained by using tetrahydrofuran as
substrate at 180 ◦C and 2 MPa N2, indicating that La(OTf)3 has an
excellent catalytic activity for esterification ring opening of tetrahy-
drofuran ring (Scheme 3).
Fig. 4. Effect of reaction time. Reaction conditions: 1 mmol FCA, 1 mol% Pd/C,
2 mol% metal triflate, 2 MPa H2, 5 mL acetic acid, 180 ◦C. The conversion was
detected by HPLC and the yield was detected by GC and the biphenyl was used
as the internal standard. The reaction time is recorded when the temperature of
the device reaches 180 ◦C.
excessive hydrogenolysis to form by-products BA and butane. In order to
further study the reaction process, the distribution of intermediates and
products during the reaction was also analyzed by 1H NMR
spectroscopy.
The characteristic peaks of possible intermediates were initially
assigned in the 1H-NMR spectra. The 1H-NMR spectra of FCA, THFCA,
THF and BDA are shown in Figure S4ꢀ 1. THF and butane was not
analyzed because of the low boiling point, and they were lost completely
in the post-treatment process of the reaction solution. The characteristic
peaks of FCA, THFCA and BDA were at δ = 6.6, 4.5 and 2.05 ppm,
respectively. Solutions were then analyzed by 1H-NMR spectroscopy at
different reaction times (Figure S4ꢀ 2). Details of the treatment pro-
cedure of the reaction solution at different times were given in the
Supporting Information. From the above 1H-NMR spectrum, we can see
that FCA was completely converted after initial heating for 30 min and
the characteristic peak of THFCA appeared (4.5 ppm) in the meanwhile.
This result indicated that the saturated hydrogenation of the furan ring
in the FCA was a very fast process. Although the presence of THF could
not be seen from 1H-NMR spectrum due to post-treatment factors, we
used GC to quantitatively analyze THF in the reaction solution. It was
detected by GC that only a small amount of THF was contained in the
reaction solution all the time. The characteristic peak of BDA (2.05 ppm)
began to appear after 30 min of reaction time, indicating that the in-
termediate THFCA began to decarbonylate rapidly to form THF at this
temperature, after which conversion to BDA was continued. As pro-
longed constant heating times, the peaks corresponding to THFCA
gradually decreased (δ = 4.5 ppm), whereas those for BDA gradually
increased (δ = 2.05 ppm). Only the peak of BDA could be observed after
120 min of reaction, which was consistent with the results of GC
analysis.
In summary, we speculated that the continuous reaction of FCA to
BDA underwent three stages (Scheme 4). Initially, FCA was hydroge-
nated to the intermediate THFCA over Pd/C. After that, THFCA was
decarbonylated to form oxonium ions under the effect of La(OTf)3, and
then rapidly hydrogenated to form THF by Pd/C. However, due to the
fact that the oxonium ion was not stable enough under high temperature
conditions, partial polymerization might occur, which might be one of
the reasons why the yield of the final product BDA cannot be further
improved. There was no decarboxylation reaction in the process of FCA
to THF. Finally, THF was hydrogenated into BDA by the effect of La
(OTf)3. The excessive hydrogenolysis of BDA resulted in the formation of
by-product BA and butane.
Recycling experiments were also carried out under optimal condi-
tions. The results are listed in Fig. 5. After the reaction, all the reaction
liquid was taken out from the reactor, and tetradecane was added as an
internal standard for quantitative analysis. Thereafter, the solvent acetic
acid and butanediol diacetate were removed by distillation under
reduced pressure. The internal standard tetradecane, Pd/C and La(OTf)3
remained in the system. The residue in the round bottom flask was then
dissolved using acetic acid and transferred to autoclave, and another
batch of the substrate was added for the next cycle.
Originally, we assumed that furan might be generated by the
decarboxylation of FCA and then furan was hydrogenated to THF, but
furan has not been detected in the intermediate capture process. Another
possible reaction route is that FCA was first hydrogenated to form
THFCA and then decarboxylated. However, in terms of organic reaction
mechanism, aliphatic carboxylic acids are more difficult to decarbox-
ylate than aromatic carboxylic acids [58]. So the sp3 hybridized carbon
on THFCA should be more difficult to decarboxylate than the sp2 hy-
bridized carbon on FCA. Therefore, we speculated that the conversion of
THFCA to the intermediate THF does not undergo decarboxylation but a
The results indicated that the yield of BDA remained good after
cycling for twice, which showed certain stability. The yield of BDA
5