Efficient aerobic oxidation of alcohols to esters by acidified carbon nitride photocatalysts

Article abstract of DOI:10.1016/j.jcat.2020.11.021

Photocatalytic aerobic oxidation of alcohols for the direct synthesis of esters has received significant attention in recent years, but the relatively low efficiency and selectivity under visible light irradiation is the main challenge for their practical applications. Here, surface acidic sites were imparted onto metal-free heterogeneous photocatalysts by the protonation of carbon nitride (HMCN) to promote the activity for the esterification reaction through further adsorption and activation of the intermediate aldehyde. The activation of the substrate could be remarkably modulated through tuning the acidic sites on the surface of the photocatalyst, leading to a controllable reactivity of the catalytic reaction. The one-pot process for the direct aerobic oxidative esterification of alcohol exhibits high efficiency and selectivity under mild and additive-free conditions and the apparent quantum yield (AQY) of the photocatalytic esterification reaction is 0.41% at 420 nm. Moreover, a scalable photocatalytic process by the merging of a continuous flow system with the heterogeneous HMCN photocatalyst is demonstrated, combining high catalytic efficiency and stability at ambient temperatures and being promising for larger-scale applications.

Full text of DOI:10.1016/j.jcat.2020.11.021

Journal Pre-proofs
Efficient Aerobic Oxidation of Alcohols to Esters by Acidified Carbon Ni-
tride Photocatalysts
Chong Wang, Qiang Wan, Jiajia Cheng, Sen Lin, Aleksandr Savateev,
Markus Antonietti, Xinchen Wang
PII:
S0021-9517(20)30467-X
https://doi.org/10.1016/j.jcat.2020.11.021
YJCAT 13980
DOI:
Reference:
To appear in:
Journal of Catalysis
Received Date:
Revised Date:
Accepted Date:
27 August 2020
30 October 2020
15 November 2020
Please cite this article as: C. Wang, Q. Wan, J. Cheng, S. Lin, A. Savateev, M. Antonietti, X. Wang, Efficient
Aerobic Oxidation of Alcohols to Esters by Acidified Carbon Nitride Photocatalysts, Journal of Catalysis (2020),
doi: https://doi.org/10.1016/j.jcat.2020.11.021
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 Inc.

Efficient Aerobic Oxidation of Alcohols to Esters by Acidified Carbon Nitride
Photocatalysts
Chong Wanga, Qiang Wana, Jiajia Chenga,*, Sen Lina, Aleksandr Savateevb, Markus Antoniettib,
Xinchen Wanga,*
a
State Key Laboratory of Photocatalysis on Energy and Environment, College of Chemistry,
Fuzhou University, Fuzhou, 350116, P. R. China, http://wanglab.fzu.edu.cn.
b
Department of Colloid Chemistry, Max-Planck Institute of Colloids and Interfaces, Potsdam,
14476, Germany
Email: jjcheng@fzu.edu.cn; xcwang@fzu.edu.cn.
ABSTRACT: Photocatalytic aerobic oxidation of alcohols for the direct synthesis of esters has
received significant attention in recent years, but the relatively low efficiency and selectivity
under visible light irradiation is the main challenge for their practical applications. Here, surface
acidic sites were imparted onto metal-free heterogeneous photocatalysts by the protonation of
carbon nitride (HMCN) to promote the activity for the esterification reaction through further
adsorption and activation of the intermediate aldehyde. The activation of the substrate could be
remarkably modulated through tuning the acidic sites on the surface of the photocatalyst, leading
to a controllable reactivity of the catalytic reaction. The one-pot process for the direct aerobic
oxidative esterification of alcohol exhibits high efficiency and selectivity under mild and
additive-free conditions and the apparent quantum yield (AQY) of the photocatalytic
esterification reaction is 0.41% at 420 nm. Moreover, a scalable photocatalytic process by the
merging of a continuous flow system with the heterogeneous HMCN photocatalyst is
1

demonstrated, combining high catalytic efficiency and stability at ambient temperatures and
being promising for larger-scale applications.
Keywords: photocatalysis, selective oxidation, alcohol, esterification, carbon nitride
2

1. INTRODUCTION:
Esters represent one of the most important chemicals widely used in pharmaceutical, material,
and other fine chemical fields. In practice, esters are generally prepared by the reaction of
carboxylic acid and their derivatives with alcohols (Scheme S1).[1] The oxidative esterification
of aldehydes also proved to be an effective methodology,[2] but this strategy was heavily
restricted by the requirement of environmentally detrimental heavy metal salts as oxidants or
catalysts.[3] As the fermentation product of renewable plant-derived materials, the direct
transformation of alcohols into esters has attracted remarkable attention. The approach could
avoid the use of hazardous acid derivatives and eliminate the generation of undesirable products,
leading to significantly improved process efficiencies.[4]
Among the various alcohol conversion strategies, the direct oxidation of alcohols into esters
using molecular oxygen is undoubtedly attractive.[5] Although most of the reported studies rely
on homogeneous catalytic systems (Scheme S2a), heterogeneous systems based on cobalt, gold,
and palladium salts, with the advantage of high stability and ease of separation and recovery,
have also been investigated for this transformation (Scheme S2b).[6-13] These processes
reported so far are associated with precious or rare metals, and additional acid/base/additive is
usually required to activate the substrate and drive the equilibrium toward product formation.
Consequently, the development of an efficient metal-free catalytic system for the oxidative
esterification of alcohols under mild conditions is very attractive but is still challenging.
As a clean and renewable energy source, sunlight can provide sufficient energy to drive
challenging chemical reactions in an eco-friendly fashion and may play a decisive role in the
future global raw material supply.[14-19] Heterogeneous photocatalyzed selective oxidation of
alcohols is a critical synthetic transformation in the chemical, materials, fragrance, and
3

pharmaceutical industry, as it could provide a direct approach for the preparation of various
chemicals such as aldehydes, ketones, diols, acetals, and esters.[20-24] As an important metal-
free photocatalyst, polymeric carbon nitride (PCN) has been widely used in solar energy
conversion and environmental remediation due to its unique physicochemical properties, high-
stability, non-toxicity, adjustable electronic structure as well as energy band positions.[25, 26]
Notwithstanding enormous efforts have been dedicated to developing PCN photocatalyzed
oxidation of alcohols in the past decades, pristine PCN and its derivatives seem to possess poor
efficiency for ester generation thus far.[27] Aldehyde formation typically prevails in these
reactions, i.e. oxidation stops at the intermediate product.[28] Very few systems for the
conversion of alcohols to the corresponding esters are known, but necessitate the use of metals or
metal oxides, relatively harsh reaction conditions, or the employment of peroxides, rendering
their industrial application rather difficult (Scheme S2b).[29, 30] The catalytic and selective
oxidative esterification of alcohols with molecular oxygen is apparently very difficult, especially
at neutral pH and when employing metal-free catalytic systems. Given these high goals,
promoting the catalytic activity of polymeric carbon nitride represents a promising, direct
approach to meet the requirements of green chemistry (Scheme S2c).
Theoretically, ideal photocatalysts for the aerobic oxidation of alcohols to esters must possess
the desired properties for the common photocatalytic processes, i.e., an appropriate bandgap that
can harvest visible light, suitable locations of conduction, and valance bands that provide
sufficient reducing and oxidizing potentials for the specific target reactions, and an efficient
charge separation/transfer. Most of all, the photocatalysts must exhibit a high affinity for the
educts and intermediates, which is here providing sufficient aldehyde adsorption/activation sites.
We assume that the adsorption and activation of the intermediate aldehyde by photocatalysts is
4

of great significance as it is the key step to promote the oxidative esterification process as well as
to suppress the competitive reactions, particularly acid generation.[31]
On the basis of the fact that catalytic acids have been frequently used as an additive in
esterification processes to activate the carbonyl group and increase the condensation conversion,
we anticipate that a photocatalyst coupled with engineered surface acid sites could boost the
condensation step of the intermediate aldehyde with alcohol, thus improving overall the
photocatalytic performance.[32] In this regard, an integrated photocatalyst design that relies on
the synergy of the visible-light response and surface acid sites should in principle be particularly
effective toward this goal. Up to now, protonation is considered as a facile and effective method
to provide PCN with specific properties such as solubility, ferromagnetism, positive surface
charge, and improved photocatalytic H2 or O2 evolution efficiency.[33-37]
Herein, we demonstrate that a carbon nitride with engineered surface acidic sites is indeed a
promising visible-light photocatalyst for performing metal- and additive-free oxidative
esterification of alcohols under ambient conditions. The protonation of carbon nitride enables the
imparted surface acidic sites to adsorb and activate the carbonyl group of the aldehyde/acetal
intermediate but also up-shifts the conduction band minimum (CBM) of the photo-catalyst to
improve its oxygen reduction potential. An optimized PCN with surface acidic sites thus exhibits
significantly enhanced efficiency in promoting the catalytic oxidation and the subsequent cross-
esterification of alcohols to esters. The observed activity was even higher than the state-of-the-art
transition-metal catalysts (like Ru- and Ir- complexes), further demonstrating the advantage of
light-driven reactions using the surface engineered polymeric carbon nitride materials.
2. EXPERIMENTAL METHODS
5

2.1 Materials: Melamine (C3H6N6, 99%) was purchased from China Sinopharm Chemical
Reagent Co. Ltd. Potassium chloride (KCl, 99%) and lithium chloride (LiCl, 99%) were
purchased from Alfa Aesar Chemicals Co. Ltd. (China). All the materials were used without
further purification.
2.2 Synthesis of bulk carbon nitride (BCN): Melamine (8 g) was heated to 550°C for 4 h at a
rate of 6°C min-1 in a muffle furnace. This sample was denoted as BCN.
2.3 Synthesis of molten-salt carbon nitride (MCN): 600 mg of the prepared BCN sample was
ground with KCl (3.3 g) and LiCl (2.7 g) in a glovebox. Then, the mixture was heated to
different temperatures for 4 h under an N2 atmosphere (400 ml min-1) in a muffle furnace. After
natural cooling, excess salt was removed in boiling distilled water and collected by filtration,
followed by drying at 60 °C under vacuum. This sample was denoted as MCN.
2.4 Synthesis of acid-treated carbon nitride samples (HMCN): 200 mg of the prepared MCN
was placed in different concentrations (0.1, 0.5, 1, 2, 3, 5, and 10 M) of HCl solution (5 mL) and
the solution was stirred at room temperature for 2 h. The obtained sample was then washed with
deionized water and dried at 60 °C under vacuum. These samples were denoted as HMCN-X
(where X refers to the concentrations of HCl).
2.5 Synthesis of the alkali-treated sample (KMCN): 200 mg of the prepared prepared MCN
was placed in 2 M KOH solution (5 mL), and the solution was stirred at room temperature for 2
h. The obtained sample was then washed with deionized water and dried at 60 °C under vacuum.
This sample was denoted as KMCN.
2.6 Synthesis of alkalized protonated carbon nitride sample (KHMCN-2): 200 mg of the
prepared HMCN-2 was placed in 2 M KOH solution (5 mL), and the solution was stirred at room
temperature for 2 h. The obtained sample was then washed with deionized water and dried at 60
6

°C under vacuum. This sample was denoted as KHMCN-2.
2.7 Physicochemical characterization: Fourier transform infrared (FT-IR) spectra were
acquired on a Nicolet IS-50 instrument. The UV-Vis diffuse reflectance spectra (DRS) of the
samples were conducted on Varian Cary 500 Scan UV-Vis spectrophotometer and PerkinElmer
UV/VIS/NIR Spectrometer Lamdba 950. X-ray diffraction (XRD) measurements were
performed on a powder X-ray diffractometer at 40 kV and 15 mA using Cu Kα radiation
(Miniflex, Rigaku). X-ray photoelectron spectroscopy (XPS) data were obtained on the Thermo
ESCALAB250 instrument with a monochromatized Al Kα line source (200 W). Electron
paramagnetic resonance (EPR) measurements were carried out on a Bruker model A300
spectrometer. Electrochemical measurements were conducted with a BAS Epsilon
Electrochemical System. GC-MS was performed on SHIMADZU GCMS-QP2020. Zeta
potentials and size distribution were measured using a Malven Zetasizer Nano ZS (Malvern,
UK). Thermogravimetric analysis (TG) was performed on Netzsch STA 449F3 simultaneous
thermal analyzer. Temperature programmed desorption (TPD) measurements were carried out on
a Micromeritics AutoChem Ⅱ 2920 Chemisorption Analyzer. The Brunauer-Emmett-Teller
(BET), surface area (SBET), and pore volume (PV) of the samples were determined by N2
adsorption using a Micromeritics ASAP 2046 N2 adsorption apparatus (USA).
2.8 Photo-electrochemical measurement: Electrochemical measurements were conducted in a
standard three-electrode system, using a Pt plate as the counter electrode and a saturated calomel
electrode (SCE) as the reference electrode. The working electrode was prepared on F-doped
SnO2-coated glass (FTO glass) that was cleaned by sonication in ethanol for 30 min and dried at
353 K. The working electrodes were synthesized as follows: 10 mg of the sample was dispersed
in 0.8 mL of dimethylformamide (DMF) by sonication to give a slurry mixture. 10 μL of the
7

slurry was spread onto pretreated FTO glass. The electrolyte was 0.2 M Na2SO4 aqueous solution
without additive (pH = 6.8).
2.9 Photocatalytic test for esterification: Typically, 10 mg photocatalyst and 0.2 mmol
benzylic alcohols were dispersed into 1 mL methanol in a 10 mL round-bottom Pyrex glass flask
with a sealed spigot and a magnetic stirrer. Prior to illumination, the reaction system was
saturated with ultrapure O2 gas (balloon), followed by illumination with a 420 nm LED lamp (50
W) for 24 h. The temperature of the reaction solution was carefully maintained at 25 °C during
the entire experiment with a homemade thermostatic control device. After the irradiation, the
photocatalyst was removed by centrifugation. The chemical structures of products were
confirmed by comparison with standard chemicals and GC-MS. The conversion and selectivity
were determined using biphenyl as an internal standard.
2.10 Computational details: All the spin-polarized calculations were performed using the
Vienna Ab-initio Simulation Packages (VASP), the projected-augmented wave (PAW)
pseudopotentials were utilized to describe the core electrons while a plane-wave kinetic energy
cutoff of 500 eV was used to treat the valence electrons. The exchange-correlation potential was
treated by the Perdew−Burke−Ernzerhof (PBE) version of the generalized gradient
approximation (GGA). The DFT-D2 method of Grimme was used to correlate the van der Waals
dispersion forces between benzaldehyde and catalysts. The K-points mesh setting is 5 × 5 × 1 for
optimization, while 9 × 9 × 1 for electronic structure calculation. All the structures were relaxed
until the forces on each ion were less than 0.01 eV/Å, and the convergence criteria for the energy
was set as 10−5 eV. The adsorption energy (
) was calculated via the following equation:
Whereas the
,
and
is the energy of catalyst with
8

absorbed benzaldehyde, isolated catalyst, and isolated benzaldehyde, respectively.
While the charge density difference ( ) was calculated via the following equation:
In which the
,
and
is the charge density of
catalyst with absorbed benzaldehyde, isolated catalyst and isolated benzaldehyde, respectively.
The model of carbon nitride with two hydrogen and one potassium atoms respectively pre-
adsorbed at three bridge N sites was used to simulate the MCN catalyst, while the model with
four hydrogen atoms pre-adsorbed at three bridge N and one adjacent edge N atoms was adopted
to represent the HMCN catalyst.
2.11 Apparent quantum yield (AQY) measurement: The reaction mixture was prepared
according to the above-mentioned standard procedure. The 1 mL of the reaction mixture was
added into a round-bottom Pyrex glass flask. The flask was capped by a septum and wrapped by
aluminum foil with a 1 cm2 window. The flask was then irradiated by a 420 nm blue LED and a
420 nm band-pass filter for 24 hours. The AQY was calculated using the following equation:
AQY=Ne/Np×100%
where Ne is the number of electrons consumed by the resulting product (methyl benzoate), Np is
the incident photons.
3. RESULTS AND DISCUSSION
3.1 Catalyst Synthesis and Characterization.
The present catalytic system derives from polymeric carbon nitride after molten salt treatment
(MCN), showing good performance in the photocatalytic selective oxidation of benzylic alcohols
to aldehydes.[38] MCN was then treated with hydrochloric acid (HCl), generating an HMCN-
species (Scheme S3). A series of HMCN samples were obtained by treating MCN with different
9

concentrations of hydrochloric acid, where the obtained samples were denoted as HMCN-x
(where x refers to the concentrations of HCl). Considering the fact that the surface of carbon
nitride has many basic sites, it is nearby to assume that treatment with acid would lead to the
protonation of base functionalities.[39] More information about the materials synthesis can be
found in the experimental section (Supporting Information).
Fig. 1. a) SEM image of HMCN-2. b) The FTIR spectra of MCN and HMCN-2. c) XRD patterns
of MCN and HMCN-2. d-f) TEM images of HMCN-2.
To show the difference between HMCN-x samples and MCN, we choose in the following
series of characterization experiments HMCN-2 as the representative. The morphology of the
carbon nitride samples is characterized by irregular dense and stacked particles and sheets, as
revealed by the scanning electron microscopy (SEM) observation (Fig. S1 and 1a), and does not
change too much after treatment with acid. The Fourier transform infrared (FTIR) spectra of
10

HMCN-2 and MCN showed approximately identical patterns (Fig. 1b), indicating no significant
change in the bulk structure of the samples after protonation. Additionally, like MCN, HMCN-2
has around 20 wt% water included, based on its thermogravimetric analysis (Fig. S2). X-ray
photoelectron spectroscopy (XPS) analysis (Fig. S3a and S3b) showed that both MCN and
HMCN-2 are mainly composed of C and N. High-resolution XPS analysis supports that MCN
and HMCN-2 both displayed the analogous heptazine structure, which is consistent with FTIR
results.[40] Notably, signals corresponding to the element K were observed with MCN (broad
signal at 293.6 eV, potassium associated with the unreacted cyanamide anion), but not in
HMCN-2, confirming the replacement of surface K ions by protons after treatment with HCl
(Fig. S3c).[41] On the basis of elemental analysis results (Table S1), obviously, the K ions were
removed throughout the material and replaced by protons.
The local structure of HMCN-2 has been characterized by the powder X-ray diffraction (XRD)
patterns, and the sample exhibits identical diffraction peaks as the MCN sample (Fig. 1c).
Compared to MCN, protonated HMCN-2 exhibited sharper XRD diffraction peaks, which is due
to the removal of the irregularly placed K ions in the framework by X-ray invisible protons,
resulting in a clearer view on the framework order of the samples. In addition, the XRD patterns
of the HMCN-2 sample prepared by the solvent evaporation method contains obvious diffraction
peaks of potassium chloride (KCl) (Fig. S4), further proving that the potassium in the samples is
replaced by hydrogen. The transmission electron microscope (TEM) observations (Fig. 1d)
indicate that HMCN-2 consists of several stacked layers of nanosheets. Furthermore, the HR-
TEM images (Fig. 1e and 1f) of HMCN-2 revealed two different lattice fringes of 1.1 and 0.32
nm, corresponding to the in-plane repeat unit of (100) plane and interlayer distance of (002)
plane of carbon nitride. Similarly, the treatment with acid did not obviously change the structure
11

of the layered material. In line with the above results, the energy-dispersive X-ray spectroscopy
(EDS) mapping images of MCN (Fig. S5a) showed a uniform distribution of C, N, and K
throughout the whole selected area, whereas only the distribution of C, N was observed after acid
treatment, again reflecting the replacement of K with the proton of the protonated samples (Fig.
S5b).
Fig. 2. a) Diffuse reflectance spectra of MCN and HMCN-2. Inset: photograph of the samples. b)
Schematic illustration of the band structures.
The diffuse reflectance spectra (DRS, Fig. 2a) indicate that the optical band gap and thereby
the semiconductor properties are maintained but gradually blue-shifted from 460 to 440 nm after
protonation. The color of the sample changed correspondingly from yellow-green to off-white
after acid treatment. The bandgaps from the Tauc plots are 2.85 eV for the HMCN-2 and 2.69 eV
for MCN (Fig. S6). The conduction band potentials of HMCN-2 and MCN are examined to be -
0.74 V and -0.45 V (vs. RHE) by electrochemical Mott-Schottky plots (Fig. S7). Together with
the calculated band gap values, the band position of the obtained HMCN samples could be
schemed in Fig. 2b. It is clear that both conduction band (CB) and valence band (VB) positions
12

were shifted to more negative potentials, and there is especially an increased driving force for
reductive dioxygen activation.
3.2 Photocatalytic Oxidation of Benzyl Alcohol over HMCN-x Catalysts.
Direct esterification of benzyl alcohol (BA) under metal-free photocatalytic conditions as a
green and sustainable process requires an efficient catalyst to activate both oxygen and organic
molecule under mild and base-free conditions. As shown in Fig. 3a, aerobic esterification of BA
and methanol did not proceed in the absence of photocatalyst. Only a trace amount of methyl
benzoate (MB) was observed over MCN and the vast majority of products were benzaldehyde
and benzoic acid (Table S2, entry 2). The alkali-enhanced MCN (noted as KMCN) also shows
no significant increase in the yield of MB (Table S2, entry 3). In comparison, carbon nitride
treated with 0.1% acid (noted as HMCN-0.1) afforded improved yield of MB under identical
conditions. These results reveal that pristine carbon nitride catalysts could not promote the
aerobic esterification, whereas protonated HMCN could considerably prompt the formation of
MB under mild photocatalytic conditions. The samples treated with different concentrations of
HCl are used in the photocatalytic oxidative esterification reactions, and the yield of MB shows a
tendency to increase first and then decrease (Fig. 3a). All these studies evidence that the catalytic
activity of HMCN can be adjusted via tuning the concentration of acidic site in the catalyst
surface. Excellent BA conversion (>99%) and selectivity to MB (94%) are achieved over
HMCN-2 under 420 nm light irradiation. HMCN-5 exhibited a pronounced decrease in catalyst
activity possibly due to structural changes at a higher concentration of acid (Fig. S8). Therefore,
HMCN-2 was utilized as the best-in-class catalyst for the activation of oxygen and organic
molecules in all subsequent experiments.
13

When the reactions were conducted at 5 °C (Table S2, entry 11), the production of MB was
relatively low, and the main product was benzaldehyde. No reaction took place in the absence of
O2 or light (Table S2, entry 12 and 13). Removing the acidic site from the catalyst via KOH
treatment, namely KHMCN-2 (XRD analysis is shown in Fig. S9), significantly lowered its
catalytic activity, forming only a trace amount of MB within 24 h (Fig. 3b). After that, the
effects of acid variation was also investigated (Fig. 4). All reactions proceed smoothly over
samples treated with HCl, HNO3, or H2SO4 to deliver the desired esterification product, whereas
HMCN-2 demonstrated the best performance. Alternatively, samples treated with HCOOH,
CH3COOH, or water (i.e weaker organic acids) display very low selectivity for MB (XRD
analysis is shown in Fig. S10), similar to the untreated MCN sample, and benzoic acid is
obtained as the main product. Compared with the well-known homogeneous and heterogeneous
photocatalysts, HMCN-2 showed the best performance for the production of esters from readily-
available alcohols (Table S2, entries 14-19).
In the process of the catalytic reaction, a significant amount of benzaldehyde and
benzaldehyde dimethyl acetal were observed as intermediates over HMCN photocatalyst (Table
S2, entries 4-11). As shown in Fig. 3c and Table S3, the use of benzaldehyde as substrate leads
to a full conversion to benzoic acid over MCN. In contrast, MB was formed in high yield with
HMCN-2 as the catalyst (Fig. 3c). Moreover, removing oxygen or performing the reaction in the
dark leads from benzaldehyde to the production of benzaldehyde dimethyl acetal in almost
quantitative yield. Similarly, HMCN-2 can catalyze the reaction of benzaldehyde dimethyl acetal
with methanol to form methyl benzoate under oxygen and light (Table S4, entry 4). In this
reaction, MCN could only promote the formation of benzoic acid (Table S4, entry 2). Finally,
the reaction process was monitored with BA as the substrate (Fig. 3d). The experimental results
14

demonstrate that an obvious amount of benzaldehyde or benzaldehyde dimethyl acetal was
formed at the early stage of the reaction. With the proceeding of the reaction, benzaldehyde and
acetal could be further oxidized to MB, resulting in the production of MB in high yield and
selectivity. All the above results indicate that with the assistance of the surface acidic site, the
carbonyl group of benzaldehyde was activated over HMCN, which could finally transform to
MB after subsequent oxidations.
15

Fig. 3. a-b) The photocatalytic reaction of benzyl alcohol with methanol over different catalysts.
c) Photocatalytic reaction of benzaldehyde with methanol under different reaction conditions. d)
Time profile of the photocatalytic oxidation of benzyl alcohol with methanol over HMCN-2.
Reaction condition: substrate (0.2 mmol), catalyst (10 mg), O2 or Ar (balloon), CH3OH (1 mL),
420 nm LED lamp, 25 °C, 24 h.
16

Fig. 4. The photocatalytic reaction of BA with methanol over different catalysts treated with
various acids.
To gain mechanistic insight into the photocatalytic oxidation process, an electron
paramagnetic resonance (EPR) spin trapping experiment was performed, using DMPO (5,5-
dimethyl-1-pyrroline N-oxide) as the spin trap.[42] As shown in Fig. S11, there are no obvious
ESR signals for DMPO-•O2- adduct in the MCN system under visible light irradiation, but a
gradual evolution of clear EPR signals for DMPO-•O2- adducts was observed by using HMCN-2.
Furthermore, when the radical scavenger butylated hydroxytoluene (BHT) was added into the
reaction mixture under standard reaction conditions, the formation of MB over HMCN-2 largely
decreased (Fig. S12). Such a suppressed conversion suggests that the superoxide radical anion
(·O2−) might be involved in the photocatalytic reaction. In light of these results, as depicted in
Scheme 1 and S4, the oxidation of BA first afforded benzaldehyde over protonated carbon
nitride photocatalyst (HMCN-2) under oxygen and illumination conditions. With the adsorption
17

and activation of the carbonyl group by the surface acidic site of the photocatalyst, benzaldehyde
could further react with methanol to form acetal or hemiacetal. As trace amounts of water are
present in the reaction mixtures (as impurities, but also as a reaction product), an equilibrium
among benzaldehyde, hemiacetal, and acetal is established, and the second oxidation probably
proceeds via the hemiacetal. The active oxygen species, the superoxide radical anion, is in-situ
generated on the surface of HMCN, presumably as a surface exciton radical pair. This
localization ensures high efficiency and selectivity for the production of MB via a tandem
catalytic process, the cross-dehydrogenative-coupling of the benzaldehyde with methanol.[43] In
contrast with the pristine MCN or alkaline-treated KMCN, the desorption of benzaldehyde takes
place, due to the lack of surface acidic sites, and oxidation of benzaldehyde in the solvent by free
superoxide radicals leads to the formation of benzoic acid.
Scheme 1. Possible reaction process for the oxidative esterification.
3.3 Surface Acidic Sites to Activate the Carbonyl Group and Promote the Catalytic
Performance
Further analysis of the structure of the HMCN catalyst was conducted to collect additional
evidence on the surface acidic sites to activate the carbonyl group for the selective oxidation of
BA. The similar texture of HMCN-2 and MCN excludes the obvious effects of the particle size
and nanostructure of the materials on the final activity (Fig. 1a and 1b). A higher specific surface
18

area to effect (Table S5) the final catalytic activity can also be excluded here as the specific
surface area of the protonated samples does not change much. The Fourier transform infrared
spectroscopy (FTIR, Fig. S13) analysis exhibited obvious peaks near 2150 cm-1 and 1000 cm-1 in
the MCN sample which could be attributed to the cyano group and symmetric and asymmetric
vibrations of metal-NC2 bonds.[44, 45] The intensity of these peaks decreases with HMCN
samples, suggesting the hydrolysis of the cyano group and the exchange of K+ with protons
under acidic conditions.
19

Fig. 5. a) CO2-TPD profiles of MCN and HMCN-2. b) NH3-TPD profiles and the selectivity of
MB of MCN, HMCN-1, and HMCN-2. c) O2-TPD profiles of MCN and HMCN-2.
The CO2-TPD measurements demonstrate that MCN exhibits a strong CO2 desorption peak at
around 75 °C (Fig. 5a), whereas no significant CO2 desorption peak was observed with HMCN-
2. The above results confirmed the presence of abundant basic sites on the surface of the MCN
catalyst, while these surface basic sites are generally neutralized with acid treatment. Besides,
20

HMCN-2 has the highest NH3 desorption peak (Fig. 5b), centered at a temperature of
approximately 150 oC, quantifying the high acid strength and density on this protonated material.
Notably, the strength and density of acidic sites show a positive correlation relation with the
selectivity of the overall transformation (Fig. 5b), well illustrating the determinative role of
surface acidic sites for the superb performance of the oxidative esterification process. Moreover,
according to the O2-TPD spectrum as shown in Fig. 5c, the O2 desorption peak of HMCN-2
shifts to a higher temperature in comparison with MCN, suggesting enhanced chemisorption and
activation of oxygen after protonation. Along with protonation of the catalyst and C-NHx
bonding, the surface charge, as measured by zeta potential of HMCN-2 dispersion in water (Fig.
5d) was shifted after contact with HCl from -55 to -40 mV, again proving a positively charged,
acidic surface. Furthermore, the propagation of acidic sites on the catalyst surface could
effectively inhibit the hydrogen evolution reaction, further reducing the side reaction and
improving the overall reaction selectivity. All the above-mentioned results suggest that the
introduction of engineered surface acidic sites could both realize the activation of the substrate
compounds and reduction of the undesirable side reactions, leading to remarkably enhanced
photocatalytic efficiency for the oxidative esterification reactions.
21

Fig. 6. Optimized configurations of benzaldehyde adsorption over a) MCN and b) HMCN with
the calculated charge density difference diagrams c, d), respectively. The blue section represents
electron depletion while the yellow section represents electron accumulation. e) The total density
of states (TDOSs) of isolated benzaldehyde molecule (Ⅰ), benzaldehyde absorbed over MCN
(Ⅱ), and benzaldehyde absorbed at HMCN (ⅡⅠ). The Fermi level is set to 0 eV. Color mode:
22

N, blue; C, grey; O, red; H, white; K, purple. The isosurface level of charge density is 0.0015
e/bohr3.
In addition, DFT calculations were carried out to further investigate the key role of surface
acidic sites on the activation of the carbonyl group of intermediate benzaldehyde. It can be seen
from Fig. 6a and 6b that the benzaldehyde molecule can adsorb both on the MCN and HMCN
via forming hydrogen bonds. On MCN, the adsorbed benzaldehyde is almost vertical to the
catalyst plane and the calculated distance of the hydrogen bond is about 1.87 Å, generating
binding energy of -0.87 eV. While on HMCN, the adsorbed benzaldehyde is parallel to the
catalyst surface and two hydrogen bonds with distances of 2.01 and 1.72 Å are formed. This
configuration is more stable than that on MCN due to a calculated higher binding energy of -1.08
eV for benzaldehyde.
To obtain a deeper understanding of the interaction between benzaldehyde and the catalyst,
charge density difference analysis is performed, and the diagrams are plotted in Fig. 6c and 6d. It
is clear to see that the electron transfer mainly takes place between the carbonyl and the edge
hydrogen atoms on the surface N atoms of both catalysts. For HMCN, the electron transfer
between the benzaldehyde molecule and the catalyst is more significant, compared to that for
MCN, due to more hydrogen bonds. The electrons of oxygen are mainly localized at the side that
interacts with the N-H species and enhancement of the electron-withdrawing ability of carbonyl
carbon relative to the case on MCN. Based on this, the hydrogen bound to the carbonyl carbon
loses more electron in comparison to the hydrogen in the MCN case (Table S6). Moreover, we
carried out the density of state analysis to verify the electrophilicity of carbonyl carbon. As
shown in Fig. 6e, only a single occupied state exists near the Fermi level (EF) for the C1 in the
isolated benzaldehyde. After adsorption onto the MCN surface, the p states of C1 move towards
23

the lower energy region relative to the EF, causing the EF to locate in the energy gap between the
highest occupied state and the lowest unoccupied state. More interestingly, when the
benzaldehyde forms two hydrogen bonds with HMCN, the p states further shift to a deeper
energy region, causing the EF quite close to the lowest unoccupied state, suggesting that the
lowest unoccupied p states of C1 are facile to interact with electron-rich species. This also
suggests that benzaldehyde adsorbed on HMCN is more likely to be attacked by nucleophiles
relative to the case on MCN. This trend well agrees with our experimental observation that
benzaldehyde dimethyl acetal was readily formed over HMCN, which is probably initiated via a
nucleophilic attack of methanol on the activated carbonyl group of benzaldehyde
As water is generated during the photocatalytic aerobic oxidation process, the hydrophilicity
of the catalyst surface may have a certain effect on the catalytic process. The untreated MCN
exhibits good hydrophilicity, and the hydrophilicity of alkali-treated KMCN-2 is further
improved. However, the hydrophobicity of HMCN-2 (Fig. S14) is most improved in comparison
with MCN and KMCN-2, which then can take up the water generated during the oxidative
esterification from the rection site, thereby improving the catalytic efficiency.
3.4 Photocatalytic Cross-Esterification of Different Benzylic Alcohols with Methanol,
Scaling-up Reaction, and Reusability of Catalyst.
The HMCN-2 photocatalyst can be used to drive a series of photocatalytic oxidative
esterification of alcohols under metal-free conditions (Table 1). In general, the reaction exhibits
great photocatalytic performance with electron-rich benzylic alcohols, forming the corresponding
esters in excellent yield. With electron-donating methyl, methoxy, and ethyl group on the para-
positions, the benzyl alcohol reacted smoothly to give the corresponding ester derivatives with
high selectivity (Table 1, entries 1-4). The reaction scope could also be extended to benzylic
24

alcohols substituted with electron-withdrawing Cl and Br groups, which gives the desired esters
in high conversion and selectivity (Table 1, entries 5-6). Additionally, heterocyclic substituted
alcohol such as thiophene-2-methanol can also be used as a suitable substrate in the
photocatalytic oxidative esterification reaction, and the heterocyclic carboxylic acid ester could
be afforded incomplete conversion and 88% selectivity (Table 1, entry 7). The advantages of
HMCN over noble-metal-based catalysts with respect to being more resistant against poisoning
by N or S atoms are demonstrated. It is worth noting that when 1,4-benzenedimethanol is used as
the substrate, dimethyl terephthalate is obtained in lower yield under the standard experimental
conditions (Table 1, entry 8). The conversion and selectivity of dimethyl terephthalate can
however be significantly improved with the extension of reaction time to 72 h (Table 1, entry 9).
All these results indicate the high chemoselectivity and functional group tolerance of HMCN for
metal-free aerobic esterification of alcohols using molecular oxygen.
Table 1. Photocatalytic synthesis of methyl esters from benzylic alcohols.
O
CH3
OH
HMCN-2, O2, 25 oC
LED (420 nm), 24 h
O
R
+
CH3OH
R
Entry[a]
Conv. [%][b]
Sel. [%][c]
Benzylic alcohol
Methyl ester
O
CH3
O
1
>99
>99
94
OH
O
CH3
O
2
98
OH
H3C
H3C
25

O
O
CH3
O
O
3
>99
94
OH
H3C
H3C
CH3
4
5
6
7
>99
>99
>99
>99
>99
96
OH
H3C
H3C
O
O
O
O
CH3
O
OH
Cl
Br
Cl
Br
CH3
O
93
OH
88
OH
O
CH3
O
S
S
O
CH3
O
8
45
76
58
83
OH
O
HO
HO
H3C
O
O
O
CH3
O
9[d]
OH
O
H3C
[a] Reaction conditions: benzylic alcohols (0.2 mmol), HMCN-2 (10 mg), O2 (balloon), CH3OH (1 mL), 420
nm LED lamp, 25 °C, 24 h. [b] Determined by GC-MS analysis. [c] All yields are calculated using GC-MS.
[d] Reaction time changes to 72 h.
The photocatalytic oxidative esterification process employing the HMCN-2 photocatalyst
shows favorable potential for scaling-up in terms of the starting concentration of the reactants,
26

the reaction volume, and the stability, as representatively demonstrated by the photocatalytic
synthesis of MB. As shown in Fig. S15a and S15b, a gram-grade reaction was conducted at the
identical conditions in a batch and continuous-flow reactor system. The experiments
demonstrated that while a 56% yield of MB could be obtained after 48 hours under batch
photochemical reaction conditions, the reaction outcome could be greatly improved to 87% yield
within the continuous-flow protocol.
Fig. 7. Recycling of the HMCN-2 photocatalyst.
We also examined the stability and reusability of the HMCN-2 catalyst by recovering the used
catalyst for further characterizations and catalytic reactions. The HMCN-2 catalyst could be
reused at least five times (Fig. S16) with only a slight attenuation of activity, which could mainly
be attributed to the reduction of acid sites on the catalyst surface (Fig. S17). The crystal structure
of the sample does not change much after 5 cycles, as demonstrated by the XRD characterization
results (Fig. S18). Moreover, a protonation process, in which the used HMCN-2 catalyst was
27

washed with 2 M HCl solution, was carried out to ensure its activity, again suggesting the
protonation process is the key step in the whole aerobic esterification reaction of BA over the
HMCN heterogeneous catalyst (Fig. 7). All the above results unambiguously confirmed the
necessity of acidic sites on the catalyst surface for the performance of the photocatalytic
oxidative esterification process.
4. SUMMARY
A metal-free, visible light-responsive HMCN photocatalyst with surface acidic sites is
presented to drive a photocatalytic aerobic oxidative esterification reaction under additive-free,
mild conditions. The surface acidification promotes the adsorption and activation of the carbonyl
group of the intermediate aldehyde, thus giving rise to high efficiency and selectivity for the
direct photoredox catalytic oxidation of alcohols to esters. It is anticipated that the favorable
performance and high stability also shown in a scaled-up flow reactor makes the simple, metal-
free, cost-effective, and sustainable HMCN a promising candidate in streamlining and
diversifying industrial applications of photochemistry.
Conflicts of interest
The authors declare no conflicts of interest.
Acknowledgment: This work was financially supported by the National Key R&D Program of
China (2018YFA0209301), the National Natural Science Foundation of China (21425309,
21861130353, 21761132002, 21673040 and 21973013), and the 111 Project.
References
28

[1] J. Otera, Chem. Rev., 93 (1993) 1449-1470.
[2] B. Xu, X. Liu, J. Haubrich, C.M. Friend , Nat. Chem., 2 (2010) 61-65.
[3] K. Suzuki, T. Yamaguchi, K. Matsushita, C. Iitsuka, J. Miura, T. Akaogi, H. Ishida, ACS
Catal., 3 (2013) 1845-1849.
[4] R.L. Oliveira, P.K. Kiyohara, L.M. Rossi, Green Chem., 11 (2009) 1366-1370.
[5] Z. Shi, C. Zhang, C. Tang, N. Jiao, Chem. Soc. Rev., 41 (2012) 3381-3430.
[6] S. Gowrisankar, H. Neumann, M. Beller, Angew. Chem. Int. Ed., 50 (2011) 5139-5143.
[7] C. Liu, J. Wang, L. Meng, Y. Deng, Y. Li, A. Lei, Angew. Chem. Int. Ed., 50 (2011) 5144-
5148.
[8] C. Liu, S. Tang, A. Lei, Chem. Commun., 49 (2013) 1324-1326.
[9] W. Zhong, H. Liu, C. Bai, S. Liao, Y. Li, ACS Catal., 5 (2015) 1850-1856.
[10] H. Su, K.-X. Zhang, B. Zhang, H.-H. Wang, Q.-Y. Yu, X.-H. Li, M. Antonietti, J.-S. Chen,
J. Am. Chem. Soc., 139 (2017) 811-818.
[11] M. Liu, Z. Zhang, H. Liu, Z. Xie, Q. Mei, B. Han, Sci. Adv., 4 (2018) eaas9319.
[12] X.-F. Bai, F. Ye, L.-S. Zheng, G.-Q. Lai, C.-G. Xia, L.-W. Xu, Chem. Commun., 48 (2012)
8592-8594.
[13] R.V. Jagadeesh, H. Junge, M.-M. Pohl, J. Radnik, A. Brückner, M. Beller, J. Am. Chem.
Soc., 135 (2013) 10776-10782.
[14] S. Chen, T. Takata, K. Domen, Nat. Rev. Mater., 2 (2017) 17050.
[15] X.-B. Li, C.-H. Tung, L.-Z. Wu, Nat. Rev. Chem., 2 (2018) 160-173.
[16] I. Ghosh, T. Ghosh, J.I. Bardagi, B. König, Science, 346 (2014) 725-728.
[17] X. Lang, X. Chen, J. Zhao, Chem. Soc. Rev., 43 (2014) 473-486.
[18] A. Savateev, M. Antonietti, ACS Catal., 8 (2018) 9790-9808.
[19] S. Gisbertz, B. Pieber, ChemPhotoChem, 4 (2020) 456-475.
[20] L.-M. Zhao, Q.-Y. Meng, X.-B. Fan, C. Ye, X.-B. Li, B. Chen, V. Ramamurthy, C.-H.
Tung, L.-Z. Wu, Angew. Chem. Int. Ed., 56 (2017) 3020-3024.
[21] H. Zhang, Z. Zhu, Y. Wu, T. Zhao, L. Li, Green Chem., 16 (2014) 4076-4080.
[22] J. Wang, Y.-X. Feng, M. Zhang, C. Zhang, M. Li, S.-J. Li, W. Zhang, T.-B. Lu, CCS
Chem., 2 (2020) 81-88.
[23] W. Huang, B.C. Ma, H. Lu, R. Li, L. Wang, K. Landfester, K.A.I. Zhang, ACS Catal., 7
(2017) 5438-5442.
29