SCIENCE CHINA

Chemistry
•ARTICLES• February 2024 Vol.67 No.2: 529–535
https://doi.org/10.1007/s11426-023-1847-x

Molecular solid solution of lanthanide-titanium-oxo clusters with

enhanced photocatalytic hydrogen evolution
1† 1† 1 1 2 1
Rong Chen , Chao-Long Chen , Hui Zhang , Zi-Han Wang , Fu-Li Sun , Ming-Hao Du ,
2 1* 1 1 1*
Gui-Lin Zhuang , Cheng Wang , La-Sheng Long , Lan-Sun Zheng & Xiang-Jian Kong
1
Collaborative Innovation Centre of Chemistry for Energy Materials State Key Laboratory of Physical Chemistry of Solid Surfaces and
Department of Chemistry, College of Chemistry and Chemical Engineering, Xiamen University, Xiamen 361005, China;
2
College of Chemical Engineering, Zhejiang University of Technology, Hangzhou 310032, China

Received September 19, 2023; accepted October 17, 2023; published online December 19, 2023

Molecular solid solutions of metal clusters containing different metal centers with well-defined structures can accurately regulate
the HOMO-LUMO gap, but are rarely available. Herein, a series of colorless lanthanide-titanium-oxo clusters Ln2Ti4(μ2-O)2(μ3-
O)4(Piv)10(THF)2 (Ln2Ti4, Ln = Eu, Gd, Tb, and Ce, HPiv = pivalic acid) were synthesized by the reaction of pivalic acid with
Ln(Ac)3 and titanium isopropoxide. The light yellow crystal of cluster solid solutions Eu2Ti4−xCdx, containing a mixture of Eu2Ti4
2+ 2–
and Eu2Ti3Cd, was obtained by in situ doping Cd and S . Eu2Ti3.92Cd0.08 displays efficient photocatalytic hydrogen evolution
activity without a co-catalyst, which is up to 2.6 times that of Eu2Ti4. Femtosecond time-resolved transient absorption spec-
troscopy and spin-polarized density functional calculations showed that the enhanced photocatalytic performance of Eu2Ti4−xCdx
can be attributed to the narrower HOMO-LUMO gap and lower LUMO position than that of Eu2Ti4. This studyprovides an in situ
doping method to realize the simple preparation of cluster solid solution.
molecular solid solution, lanthanide-titanium-oxo clusters, photocatalytic, hydrogen evolution

Citation: Chen R, Chen CL, Zhang H, Wang ZH, Sun FL, Du MH, Zhuang GL, Wang C, Long LS, Zheng LS, Kong XJ. Molecular solid solution of lanthanide-
titanium-oxo clusters with enhanced photocatalytic hydrogen evolution. Sci China Chem, 2024, 67: 529–535, https://doi.org/10.1007/s11426-023-
1847-x

1 Introduction solution has aroused great interest, mainly because of the

precise molecular structure and the tunability of doping or-
Solid solutions are formed by doping the solute atoms into a ganic ligands [5,6]. At present, metal-organic framework
uniform crystalline matrix and substituting some matrix (MOF)-based molecular solid solutions have made great
atoms. Inorganic solid solutions have received much atten- progress. For example, the substitution of some varied
tion because of their tunable electronic structure and physi- functionalization ligands incorporated into a single phase
cal/chemical properties for their applications in the fields of MOF can achieve effective catalytic or luminescence per-
optics, electricity, catalysis, etc [1–4]. However, due to the formance [7,8]. However, the controllable doping of metal
lack of precise structure model, the theoretical calculation ions in molecular solid solutions still has great challenges.
and accurate performance control of inorganic solid solu- Atomically precise metal cluster is an ideal model for
tions are still very challenging. Recently, the molecular solid studying the structure-activity relationship of inorganic na-
nomaterials [9–14]. For a single cluster, the replacement
position of metal ions is relatively clear and fixed, and the
†These authors contributed equally to this work. clusters doped with specific metal ions also have a precise
*Corresponding authors (email: xjkong@xmu.edu.cn; wangcheng.xmu@xmu.edu.cn)

© Science China Press 2023 chem.scichina.com link.springer.com

530 Chen et al. Sci China Chem February (2024) Vol.67 No.2

structure [15–17]. Therefore, the cluster molecular solid obtained. Anal. calcd. for Eu2Ti4O30C58H110 (FW = 1782.87,
solution can effectively control the doping position of metal yields 43%): C, 39.07; H, 6.22. Found: C, 39.19; H, 6.20.
ions. However, cluster solid solutions are very rare because Anal. calcd. for Gd2Ti4O30C58H110 (FW=1793.44, yields 60
of the weak interaction and hydrogen bonding among clus- %): C, 38.84; H, 6.18. Found: C, 38.86; H, 6.14. Anal. calcd.
ters [18]. The mixing that occurs between clusters requires for Tb2Ti4O30C58H110 (FW=1796.79, yields 42 %): C, 38.77;
an understanding of distortions and spatial orientations, H, 6.17. Found: C, 38.73; H, 5.96. Anal. calcd. for Ce2Ti4
making it very difficult for different chemical components to O30C58H110 (FW=1759.18, yields 40 %): C, 39.60; H, 6.30.
remain in a single crystal phase (Figure 1). Found: C, 39.42; H, 6.14. All of the yields were calculated
Heterometallic lanthanide-titanium-oxo clusters (LTOC) and based on pivalic acid.
have received much attention because of their special
structures and potential applications [19–21]. Herein, we 2.1.2 Synthesis of mesoporous CdS nanosheet
report a series of isomorphous LTOC Ln2Ti4(μ2-O)2(μ3-O)4- The Cd(EDA) was prepared using Cd(NO3)4·4H2O and
(Piv)10(THF)2 (labeled as Ln2Ti4, Ln = Eu, Gd, Tb, and Ce, CH4N2S as precursors and ethylenediamine (EDA) as sol-
HPiv = pivalic acid). Interestingly, a single crystal of Eu2Ti4−xCdx, vent. CdS(EDA) inorganic-organic hybrid nanosheet was
belonging to cluster solid solution [Eu2Ti4]1−x-[Eu2Ti3Cd]x, prepared with the modified method according to the litera-
2+ 2−
was prepared by in situ co-doping of Cd and S (The ture (seen in supporting information). In a typical process,
detailed recrystallization of cluster solid solution using a 1.5 mmol Cd(NO3)4·4H2O and 3.0 mmol CH4N2S were dis-
Soxhlet extractor seen in Supporting Information, Figure solved in 10 mL pure ethylenediamine under ultrasound till
S1). The Eu2Ti4−xCdx shows enhanced water splitting per- all the reagents were dissolved. The solution was then
formance compared to Eu2Ti4 because of the narrower transfer into 25 mL teflon-lined autoclave and maintained at
highest occupied molecular orbital-lowest unoccupied mo- 180 °C for 18 h. The yellow products were rinsed three times
lecular orbital (HOMO-LUMO) gap and lower LUMO po- with distilled water and ethanol respectively, and dried at
sition than that of Eu2Ti4. Ultrafast transient absorption 60 °C overnight in a vacuum oven to evaporate the solvent.
spectroscopy was performed to probe the excited state dy- Mesoporous CdS was prepared with further hydrothermal
namics of Eu2Ti4 and Eu2Ti4−xCdx. methods. Typically, 30 mg CdS(EDA) was dissolved in
10 mL deionized water under ultrasound for a few minutes.
Then the obtained suspension solution was transferred into
2 Experimental 25 mL autoclave and maintained at 140 °C for 12 h. The
yellow products were rinsed three times with distilled water
2.1 Sample preparation and ethanol respectively, and dried at 60 °C overnight in a
2.1.1 Synthesis of Ln2Ti4(μ2-O)2(μ3-O)4(Piv)10(THF)2· vacuum oven to evaporate the solvent.
2H2O (Ln = Eu (1), Gd (2), Tb (3) and Ce (4))
A mixture of Ln(Ac)3 (0.100 mmol), tetrabutyl titanate 2.1.3 Synthesis of H[Eu2Ti4−xCdx(μ2-O)2−ySy(μ3-O)4(Piv)9-
(TBT) (0.12 mmol) and pivalic acid (2.00 mmol) were dis- (THF)]
solved in a mixed solvent with tetrahydrofuran (THF ) (1.50 A mixture of Eu(Ac)3 (0.100 mmol), TBT (0.12 mmol) and
mL) and CH3CN (4.50 mL). The mixture was stirred for 1 h pivalic acid (2.00 mmol) were dissolved in a mixed solvent
and then heated to 85 °C for 6,000 min. After the mixture with THF (1.50 mL) and CH3CN (4.50 mL). Then different
cooled to room temperature, different bulk crystals were weights (0.2, 0.5, 1.2 and 2 mg) of CdS nanosheet were
added to different reaction bottles. The mixture was ultra-
sound for at least 1 h and then heated to 85 °C for 6,000 min.
After the mixture cooled to room temperature, different bulk
crystals were obtained. x and y were calculated by in-
ductively coupled plasma optical emission spectrometer
(ICP-OES) (Table S2, Supporting Information online). Sin-
gle crystal data for clusters solid solution Eu2Ti3.91Cd0.09 (5)
with maximum doping amount were collected and shown in
Table S1.

2.2 The photoelectrochemical analysis

Figure 1 The upper line represents the formation of cluster molecules by
coordination of metal ions and ligands and their assembly into cluster Electrochemical impedance spectra (EIS) data, photocurrent,
single crystals; the following line indicates the in situ substitution of doping
ions during the formation of cluster molecules which form metal-doped and the Mott-Schottky spots were recorded using an Elec-
clusters and assemble them into cluster solid solutions (color online). trochemical workstation (CHI 760E, Shanghai Chenhua).
 Chen et al. Sci China Chem February (2024) Vol.67 No.2 531

The clean and dry indium tin oxide (ITO) glasses served as curve for quantification. The mean results and variance were
the working electrodes. 3 mg catalysts were dispersed in the obtained by testing each sample for three times under the
0.65 mL water, 0.25 mL isopropanol and 100 μL nafion and same conditions, and error bars were plotted.
then sonicated for producing slurry. The conductive tape was The photocatalytic apparent quantum yields were tested
used to adhere to part of ITO glasses to leave an area of 1 with Eu2Ti3.92Cd0.08 at 313 nm. The catalyst was irradiated by
cm×1 cm for depositing 40 μL slurry. And then, the working a 313 nm monochromatic light for 12 h, the intensity of
−2
electrode with 40 μL slurry was dried at room temperature monochromatic light was calibrated to be 70 mW cm , and
2
naturally. The electrochemical measurements were per- the irradiation area was 19.625 cm . The amount of H2 mo-
formed in the cell equipped with three electrodes, namely the lecules generated in 12 h was 98.54 μmol. The apparent
working electrode, counter electrode (Pt plate) and reference quantum efficiency was calculated to be 0.13%±0.01%.
electrode (Ag/AgCl). EIS measurements were carried out in
a three-electrode system and recorded over a frequency range
of 100–200 kHz with ac amplitude of 20 mV at 0 V. A 10% 3 Results and discussion
methanol solution was used as the supporting electrolyte.
The photocurrent was measured in the same three-electrode 3.1 Synthesis and structural determination of Ln2Ti4-
system. A 300 W Xenon light was applied as the light source. (μ2-O)2(μ3-O)4(Piv)10(THF)2·2H2O (Ln = Eu, Gd, Tb, and
A 10% methanol solution was used as the electrolyte. The Ce) and H[Eu2Ti4−xCdx(μ2-O)2−ySy(μ3-O)4(Piv)9(THF)]
Mott-Schottky plots were also measured using the same solid solution
three-electrode system over an alternating current (AC) fre- The crystals of cluster Ln2Ti4 were synthesized by the re-
quency of 1,000, 1,200 and 1,500 Hz. These three electrodes action of pivalic acid, lanthanum acetate, and titanium iso-
were immersed in the 0.2 M Na2SO4 aqueous solution (pH propoxide. The X-ray diffraction analysis shows that the
6.6). Ln2Ti4 are isostructural and crystallize in the triclinic P-1
space group. Herein, we only discuss the molecular structure
2.3 Photocatalytic hydrogen production of Eu2Ti4 in detail. The structure of Eu2Ti4 consists of two
3+ 4+ 2− 2− −
Eu , four Ti , two μ2-O , four μ3-O , ten Piv , and two
3+ 4+ 2−
The photocatalytic H2-production experiments were per- THF ligands (Figure 2a). Two Eu , four Ti , two μ2-O and
2− 10+
formed via a photocatalytic H2-production activity evalua- four μ3-O form an [Ln2Ti4(μ2-O)2(μ3-O)4] metal-core
4+
tion system (CEL-SPH2N, CEAU Light, China) in a 100 mL with a distorted octahedral geometry (Figure 2b). All Ti are
3+
Pyrex flask, and the openings of the flask were sealed with six-coordinated and Eu are seven-coordinated. The Eu–O
silicone rubber septum. A 300 W xenon arc lamp was used as distances are 2.268(7)–2.452(1) Å, while the Ti–O distances
the light source. The focused intensity on the flask was about are 1.847(1)–2.054(8) Å, close to that of the reported LTOC
−2
200 mW cm . The central property of the xenon arc lamp is [19–21]. The yellow crystal of Eu2Ti4−xCdx solid solution
shown in Figure S1a. In a typical photocatalytic H2-pro- was synthesized by the same method with the addition of
duction experiment, 0.025 mmol Ln2Ti4 (Ln = Eu, Gd, Tb, CdS nanosheet (Figure S2). Under heating conditions,
2+ 2–
and Ce) clusters were suspended in 20 mL of H2O with 2 mL Cd and S were controlled to release and generated the
of methanol as the sacrifice agent. Before irradiation, the Eu2Ti4−xCdx cluster solid solution (Figures S3, S4). With
system was vacuumed for 10 min via the vacuum pump to different adding amounts of CdS, the formed solid solution
completely remove the dissolved oxygen and ensure the re- samples were labeled as Eu2Ti3.985Cd0.015, Eu2Ti3.96Cd0.04,
actor was in an anaerobic condition. This is a heterogeneous Eu2Ti3.92Cd0.08, and Eu2Ti3.91Cd0.09, respectively. The certain
reaction since the clusters are stable in water. The solid so- ratios of Eu/(Ti+Cd) were defined by ICP-OES (Table S2).
lutions used in catalytic reactions are obtained by re- The composition of the Eu2Ti4 cluster and Eu2Ti4−xCdx
crystallization (Figure S1b) to remove any possibly formed solid solution was confirmed by electrospray ionization mass
CdS nanoparticles by using a Soxhlet extractor. The photo- spectrometry (ESI-MS) in dichloromethane. Two main
catalytic products were analyzed by gas chromatography groups of peaks in the range from 1,600 to 1,800 were ob-
(GC7920, CEAU Light, China) through an automatic de- served (Figure 2e). The largest mass charge ratios in each
tection system (CEL-SPH2N, CEAU Light, China). The group of peaks are 1664.1763 and 1703.2362, respectively,
peaks with retention time located at 2.0 min were char- in agreement with the calculated values of 1664.1964 for
–
acterized as hydrogen (by pure hydrogen characterization). [Eu2Ti4(O)6(Piv)10(NO3)] and 1703.2691 for [Eu2Ti4(O)6)-
–
In addition, after a specific volume of pure hydrogen was (Piv)11] , suggesting that the Eu2Ti4 maintains its structural
given, a series of standard curves for hydrogen were de- integrity in CH2Cl2. Except for the same groups of peaks of
termined by integration of the resulting peak area. The final Eu2Ti4, two new peaks of 1582.1152 and 1638.1452 ap-
results were obtained by integrating the peak area of the gas peared after CdS doping. These two peaks are corresponding
−
produced by the sample and substituting it into the standard to [Eu2Ti3Cd(O)5(S)(Piv)9] (cal. 1582.0727) and [Eu2Ti3Cd-
532 Chen et al. Sci China Chem February (2024) Vol.67 No.2

Figure 2 (a) The structure of Eu2Ti4(μ2-O)2(μ3-O)4(Piv)10(THF)2. Eu, Ti, O, and C are shown in purple, green, red, and grey, respectively. (b) The metal-oxo
core of Eu2Ti4. (c, d) The crystal photographs of Eu2Ti3.91Cd0.09 solid solution and Eu2Ti4 cluster. (e) The ESI-MS of Eu2Ti4 cluster and Eu2Ti3.91Cd0.09 solid
solution (color online).

− 2+
(O)6(Piv)9(THF))] (cal. 1638.1534), indicating that Cd uniform distribution of Cd and S elements in Eu2Ti4−xCdx
2– 4+ 2−
and S took the place of Ti and O of Eu2Ti4, respectively crystal solid solution (Figure 2b, c).
(Figure 2e). The powder X-ray diffraction (PXRD) patterns
of Eu2Ti4 and Eu2Ti3.91Cd0.09 matched well with their simu- 3.2 Photocatalytic hydrogen production
lated peaks, indicating the good purity of the obtained
clusters and cluster solid solution (Figure S5). The Raman Heterogeneous photocatalytic hydrogen evolution reaction
spectra showed characteristic peaks for Cd–S bond at 303 of Eu2Ti4 cluster and Eu2Ti4−xCdx solid solutions was tested
−1
and 607 cm in Eu2Ti3.91Cd0.09, which is consistent with that without any photosensitizer or co-catalyst under light irra-
of Cd–S bond of CdS (Figures S6, S4b) [22,23]. The Cd–O diation. The photocatalytic H2 amount of Eu2Ti4 cluster with
−1
vibration at 450 cm was observed which corresponds to the methanol as a sacrificial solution under 12 h UV-visible light
transverse optical (TO) + transverse acoustic (TA) combi- illumination was shown in Figure S9. The photocatalytic
−1 −1
nation at the CdO Brillouin zone L point [24,25]. And the activity of Eu2Ti4 is 856 μmol g h , which is the highest
−1
peaks located at 377 and 937 cm are attributed to the 1LO catalytic activity among the reported titanium oxo-clusters
(longitudinal optical) and 2LO phonons stretches of Cd–O (Table S6). As shown in Figure S10, the hydrogen produc-
[24–27]. The appearance of Cd–O bonds in Raman spectra tion of Eu2Ti4 is near a straight line after 48 h illumination
confirms the successful substitution of Cd into the solid so- with turnover number up to 70. To verify the role of the
lution. As shown in Figure S7, the two peaks at 405.3 and lanthanide element in the composites, Ln2Ti4 (Ln = Eu, Gd,
412.0 eV are attributed to Cd 3d3/2 and Cd 3d5/2 of CdS, Tb, and Ce) clusters with different lanthanides and P25 as
respectively [28]. Notably, when compared to CdS, the Cd photocatalysts were studied under the same photocatalytic
3d3/2 and Cd 3d5/2 peaks of Eu2Ti3.91Cd0.09 are shifted in the conditions. The photocatalytic H2 production activities of
negative direction by 0.1 and 0.2 eV, which can be ascribed Eu2Ti4, Gd2Ti4, Tb2Ti4, Ce2Ti4, and P25 were 856, 205, 180,
−1 −1
to the increase of electron density caused by the successful 154, and 90 μmol g h , respectively (Figure 3a and Table
substitution of Cd and S. In addition, no diffraction peaks of S4). Also, Eu2Ti4 exhibited the highest photocatalytic ac-
the CdS nanosheet were observed in PXRD (Figure S5). tivity among other homologous series and showed excellent
These all indicate that Cd and S enter the lattice of the solid stability (Figure 3a, c). The EIS and photocurrent response of
solutions by substitution, rather than by simple surface ad- the isomorphic clusters were measured to study their charge
sorption. This result is also supported by scanning electron separation efficiency. As shown in Figure S11a, b, the Eu2Ti4
microscope (SEM) images of Eu2Ti3.91Cd0.09 (Figure S8a). cluster has the smallest semi-circle and a significantly larger
The energy dispersive X-ray (EDX) mapping elements show photocurrent response among the isomorphisms demon-
that the Cd and S elements are uniformly distributed strating a better charge transport and transfer ability. These
3+
throughout the microcrystals (Figure S8b). Moreover, the results indicated that the Eu in the clusters plays an im-
crystal of Eu2Ti4 is colorless, while the crystal of Eu2Ti4−xCdx portant promoting effect on photocatalytic HER. The high
is light yellow. The change of crystal color also proves the photocatalytic stability of Ln2Ti4 clusters was confirmed by
 Chen et al. Sci China Chem February (2024) Vol.67 No.2 533

Figure 4 In situ EPR characterization and ultrafast transient absorption.

Figure 3 The H2 production of (a) Eu2Ti4, Gd2Ti4, Tb2Ti4, Ce2Ti4, P25 (a) The in situ EPR intensity of Eu2Ti4 and Eu2Ti3.91Cd0.09 in dark and UV-
and (b) Eu2Ti4, Eu2Ti3.985Cd0.015, Eu2Ti3.96Cd0.04, Eu2Ti3.92Cd0.08, and visible light irradiation. Femtosecond transient absorption spectra of (b)
Eu2Ti3.91Cd0.09 in 20 mL H2O with 2 mL of methanol as the sacrificial agent Eu2Ti4 with 310 nm excitation; (c) Eu2Ti3.91Cd0.09 with 310 nm excitation
under UV-visible light illumination. (c) The H2 production of Eu2Ti4 with and (d) Eu2Ti3.91Cd0.09 with 450 nm excitation (color online).
three cycles and (d) the H2 production of Eu2Ti3.92Cd0.08 solid solution with
11 cycles (12 h per cycle). Methanol sacrificial agent was supplemented in
every two cycles (color online). 3+
magnetic Ti generated by ligand-to-metal charge transfer
(LMCT) (Figure 4a) [29–33]. In addition, there is a signal at
2+
PXRD patterns and infrared radiation (IR) spectra (Figures 2.01 with a characteristic fine structure assignable to Eu ,
3+
S12, S13). Among the Eu2Ti4−xCdx solid solution with dif- which suggests an efficient charge transfer to Eu in the
ferent doping amounts, the Eu2Ti3.92Cd0.08 exhibited the Eu2Ti4 cluster and Eu2Ti3.91Cd0.09 solid solution [34,35].
−1 −1
highest photocatalytic activity with 2,236 μmol g h , Consistently, the photoluminescence spectroscopy (PL) also
3+ 2+
which was about 2.6 times that of Eu2Ti4. However, with the demonstrated that Ti and Eu were generated in Eu2Ti4 and
continued addition of CdS, the photocatalytic activity re- Eu2Ti3.91Cd0.09 under light illumination (Figure S16).
mained almost at this level without further improvement and Furthermore, the femtosecond time-resolved transient ab-
the loading amount has barely increased. The photocatalytic sorption (fs-TA) spectroscopy was studied to probe the ul-
H2 production cycles (12 h per cycle) of Eu2Ti3.92Cd0.08 were trafast excited state dynamics of Eu2Ti4 and Eu2Ti3.91Cd0.09
3+
tested under the same light illumination. In the reactions, a (Figure 4b–d and Figure S17). Sequential formation of Ti
2+
certain amount of methanol sacrificial agent was supple- and Eu can be revealed from the spectra. A positive signal
mented every two cycles. The stable performance after 11 of Eu2Ti4 was observed between 450 and 650 nm im-
cycles (Figure 3d) and the PXRD and IR spectra before and mediately after excitation at 310 nm, which can be attributed
3+ 3+
after experiments (Figure S14) demonstrate the good stabi- to the absorption of Ti . The Ti formation is due to charge
2– 4+ − 3+
lity of crystal solid solution. ICP-OES studies showed that transfer from the μ3-O to the Ti to obtain O -Ti species
3+
less than 0.3% of clusters leached into the solution after by LMCT. This Ti absorption in Eu2Ti4 and Eu2Ti3.91Cd0.09
132 h reaction. The apparent photocatalytic quantum yield of then decayed within 500 ps with concomitant growth of a
Eu2Ti3.92Cd0.08 at 313 nm was 0.13%±0.01%. Eu2Ti3.92Cd0.08 negative signal centered at 450 nm, which is assigned to the
2+
photocatalyst was also tested under visible light irradiation stimulated emission of Eu [36]. These observations suggest
3+ 3+
(λ≥420 nm); however, only trace H2 production was detected an electron transfer from Ti to the Eu center via the
3+ 3+ 2+
(Figure S15). Ti -O-Eu route. The generated Eu may be located away
−
from the O and thus avoid fast recombination. On the other
3.3 In situ EPR characterizations and ultrafast tran- hand, the negative signal of Eu2Ti3.91Cd0.09 centered at 490
sient absorption nm can be attributed to the bleaching signal of Cd–S [37,38].
A positive signal centered at about 450 nm is assigned to the
− 2– 2+
To investigate the HER mechanism, the in situ electron absorption of S after an S -to-Cd charge transfer. It
paramagnetic resonance (in situ EPR) measurements on should be noted that Eu2Ti3.91Cd0.09 and bulk CdS semi-
pristine Eu2Ti4 and Eu2Ti3.91Cd0.09 in methanol solution were conductors have similar ultrafast spectra at 450–490 nm. The
conducted at 100 K under light irradiation. Upon 40 min of positive signal between 550 and 650 nm of Eu2Ti3.91Cd0.09
UV-visible light illumination, an intense EPR signal centered (excited at 310 nm) is of the same origin.
at 1.93 was observed, which can be attributed to para- The dynamics can be globally fitted to a three-component
534 Chen et al. Sci China Chem February (2024) Vol.67 No.2

sequential model for Eu2Ti4 and to four-component se-

quential models for Eu2Ti3.91Cd0.09 (Figure S18 and Table S5)
[39]. From the first component to the second component, the
3+
absorption intensity of Ti of Eu2Ti4 and Eu2Ti3.91Cd0.09
decays. However, the first component of Eu2Ti3.91Cd0.09 de-
cays faster (τ1 = 1.32±0.01 ps) than that of the Eu2Ti4 (τ1 =
3.54±0.03 ps). On the other hand, the increased intensity of
the Cd–S bleaching signal of Eu2Ti3.91Cd0.09 suggested that
3+ 2+
the electron of Ti may transfer to Cd . The lifetime of the
last component for Eu2Ti3.91Cd0.09 is the lifetime of the se-
parated charge before recombination. Interestingly, the life-
time (τ4 = 6.38 ± 0.11 ns) of the last component of
Eu2Ti3.91Cd0.09 after 310 nm excitation is much longer than
that (τ4 = 2.73 ± 0.04 ns) of Eu2Ti3.91Cd0.09 after 450 nm
excitation. This can be rationalized as different lifetimes of a
+ −
state with an electron on Cd and a hole on O after 310 nm
+ −
excitation vs. a state with an electron on Cd and a hole on S
after 450 nm excitation. The longer distance separation be-
Figure 5 (a) The HOMO and LUMO positions (vs. NHE) of Eu2Ti4 and
tween the electron and the hole in the Eu2Ti3.91Cd0.09 leads to Eu2Ti4−xCdx. (b) Free energy profile of HER. Proposed charge separation
a prolonged lifetime of the charge-separated state that can be and transfer in photocatalytic HER mechanism of (c) Eu2Ti4 and (d) Eu2Ti4−xCdx.
3+ 2+ Eu, Ti, O, Cd, and S are shown in purple, green, red, blue, and yellow,
responsible for the high quantum efficiency. The Ti to Cd
respectively (color online).
electron transfer in Eu2Ti3.91Cd0.09 thus resulted in improved
charge separation that led to enhanced HER performance.
Moreover, thanks to the contribution of synergistic effect
3+ 4+
between Eu and Ti , the Eu2Ti4 cluster shows higher ef- lations, the photocatalytic HER mechanism can be proposed
ficiencies than Gd2Ti4, Tb2Ti4, and Ce2Ti4 clusters. as below. Under light irradiation (Figure 5c, d), a photo-
2– 4+
generated electron was transferred from the O to the Ti ,
− 3+
3.4 DFT calculations forming the O -Ti species by an LMCT process. Next, the
3+ 3+
electron can transfer from Ti to the Eu center, which
To further understand the different photocatalytic properties, avoids the fast electrons and holes recombination. The gen-
3+ 2+ +
spin-polarized density functional calculations were carried erated Ti and Eu further reacts with the H to produce H2.
out at the theoretical level of PBE+U by utilizing the VASP The photoinduced holes then oxidized the electron donor
program [40,41]. The calculation revealed that Eu2Ti3Cd CH3OH to complete the catalytic cycle, forming for-
bears the smaller HOMO-LUMO gap (calculated: 2.33 eV), maldehyde. As is expected, the rate of formaldehyde for-
in contrast to that of Eu2Ti4 (calculated: 3.44 eV) (Figure 5a). mation is equal to the rate of hydrogen production (Figures
3+
HOMO-LUMO position results indicate that compared to S20, S21). When doping the Cd–S, the electron of Ti can
2+ +
pristine Eu2Ti4 (calculated: EHOMO = 2.73 eV and ELUMO = transfer to Cd to generate Cd for Eu2Ti4−xCdx, resulting in
−0.71 eV), the HOMO position (EHOMO = 1.57 eV) of improved charge separation and lowered LUMO position,
Eu2Ti3Cd moves upward about 1.16 eV, due to the con- leading to a better HER performance than that of Eu2Ti4.
tribution from the S 3p and Cd 4d orbitals, while that (ELUMO
= −0.76 eV) of LUMO position slightly moves upward about
0.05 eV, dominated by the Ti 3d orbitals (Figure S19). In- 4 Conclusions
spection of PDOS reveals that LUMO concentrates on the Ti
3d states, O 2p states and C 2p states. The μ2-O atom between In summary, we have reported heterometallic lanthanide ti-
two Ti(IV) ions, analogous to lattice oxygen on the surface of tanium-oxo clusters Ln2Ti4 and prepared a cluster solid so-
TiO2 [42], serves as a catalytically active site for HER. The lution Eu2Ti4−xCdx (a mixture of Eu2Ti4 and Eu2Ti3Cd) by
+ − 2+ 2–
values of ΔG in the Volmer step (H + e → *H) are −0.35 in situ doping of Cd and S . The Eu2Ti4 cluster exhibited
eV for Eu2Ti3Cd and −0.58 eV for Eu2Ti4, the difference of excellent photocatalytic water splitting activity and stability
which matches the trend of photocatalytic results (Figure without any photosensitizer or co-catalyst, and the
5b). The lower LUMO position makes an important con- Eu2Ti4−xCdx solid solution performed better than Eu2Ti4.
tribution to the enhanced photocatalytic HER performance of Femtosecond time-resolved transient absorption spectro-
Eu2Ti4−xCdx compared to Eu2Ti4 [43]. scopy and spin-polarized density functional calculations of
Based on the above experimental results and DFT calcu- Eu2Ti4−xCdx show that the enhanced photocatalytic perfor-
 Chen et al. Sci China Chem February (2024) Vol.67 No.2 535

mance is attributed to the improved charge separation and Chem Soc, 2023, 145: 16098–16108
17 Yin HQ, Zhang ZM, Lu TB. Acc Chem Res, 2023, 56: 2676–2687
transportation, the narrower HOMO-LUMO gap and the 18 Chorazy S, Stanek JJ, Nogaś W, Majcher AM, Rams M, Kozieł M,
lower LUMO position than that of Eu2Ti4. This study pro- Juszyńska-Gałązka E, Nakabayashi K, Ohkoshi S, Sieklucka B,
posed and designed a molecular cluster solid solution and Podgajny R. J Am Chem Soc, 2016, 138: 1635–1646
provided a simple method to prepare it by in situ doping. 19 Lu DF, Kong XJ, Lu TB, Long LS, Zheng LS. Inorg Chem, 2017, 56:
1057–1060
20 Zheng H, Du M, Lin S, Tang Z, Kong X, Long L, Zheng L. Angew
Acknowledgements This work was supported by the National Natural Chem Int Ed, 2018, 57: 10976–10979
Science Foundation of China (21871224, 92161104, 92161203, 21721001). 21 Du M, Xu S, Li G, Xu H, Lin Y, Liu W, Long L, Zheng L, Kong X.
Angew Chem Int Ed, 2022, 61: e202116296
Conflict of interest The authors declare no conflict of interest. 22 Xu Y, Ye Y, Liu T, Wang X, Zhang B, Wang M, Han H, Li C. J Am
Chem Soc, 2016, 138: 10726–10729
Supporting information The supporting information is available online at 23 Chen R, Yan Z, Kong X, Long L, Zheng L. Angew Chem Int Ed, 2018,
chem.scichina.com and link.springer.com/journal/11426. The supporting 57: 16796–16800
materials are published as submitted, without typesetting or editing. The 24 Popović ZV, Stanišić G, Stojanović D, Kostić R. Phys Stat Sol (b),
responsibility for scientific accuracy and content remains entirely with the 1991, 165: K109–K112
25 Cuba M, Qamhieh N, Saleh N, Kumar RTA, Mousa HA, Mahmoud
authors.
ST. J Lumin, 2018, 198: 289–295
26 Cuscó R, Ibáñez J, Domenech-Amador N, Artús L, Zúñiga-Pérez J,
1 Millet MM, Algara-Siller G, Wrabetz S, Mazheika A, Girgsdies F, Muñoz-Sanjosé V. J Appl Phys, 2010, 107: 063519
Teschner D, Seitz F, Tarasov A, Levchenko SV, Schlögl R, Frei E. J 27 Kumar S, Ojha AK, Singh RK. J Raman Spectr, 2014, 45: 717–722
Am Chem Soc, 2019, 141: 2451–2461 28 Wei RB, Huang ZL, Gu GH, Wang Z, Zeng L, Chen Y, Liu ZQ. Appl
2 Dalafu HA, Rosa N, James D, Asuigui DRC, McNamara M, Kawa- Catal B-Environ, 2018, 231: 101–107
shima A, Omagari S, Nakanishi T, Hasegawa Y, Stoll SL. Chem 29 Horiuchi Y, Toyao T, Saito M, Mochizuki K, Iwata M, Higashimura
Mater, 2018, 30: 2954–2964 H, Anpo M, Matsuoka M. J Phys Chem C, 2012, 116: 20848–20853
3 Li Q, Meng H, Zhou P, Zheng Y, Wang J, Yu J, Gong JR. ACS Catal, 30 Snoeberger III RC, Young KJ, Tang J, Allen LJ, Crabtree RH, Brudvig
2013, 3: 882–889 GW, Coppens P, Batista VS, Benedict JB. J Am Chem Soc, 2012, 134:
4 Maeda K, Takata T, Hara M, Saito N, Inoue Y, Kobayashi H, Domen 8911–8917
K. J Am Chem Soc, 2005, 127: 8286–8287 31 Negre CFA, Young KJ, Oviedo MB, Allen LJ, Sánchez CG, Jar-
5 Lusi M. CrystEngComm, 2018, 20: 7042–7052 zembska KN, Benedict JB, Crabtree RH, Coppens P, Brudvig GW,
6 Tsunashima R. CrystEngComm, 2022, 24: 1309–1318 Batista VS. J Am Chem Soc, 2014, 136: 16420–16429
7 Mohammad-Pour GS, Hatfield KO, Fairchild DC, Hernandez-Burgos 32 Schneider J, Matsuoka M, Takeuchi M, Zhang J, Horiuchi Y, Anpo M,
K, Rodríguez-López J, Uribe-Romo FJ. J Am Chem Soc, 2019, 141: Bahnemann DW. Chem Rev, 2014, 114: 9919–9986
19978–19982 33 Li Z, Xiao JD, Jiang HL. ACS Catal, 2016, 6: 5359–5365
8 Newsome WJ, Ayad S, Cordova J, Reinheimer EW, Campiglia AD, 34 Kim TJ, Cho YH, Choi IK, Kang JG, Jee KY. J Lumin, 2007, 127:
Harper JK, Hanson K, Uribe-Romo FJ. J Am Chem Soc, 2019, 141: 731–734
11298–11303 35 Park YJ, Kim TJ, Cho YH, Jung Y, Im HJ, Song K, Jee KY. Bull
9 Ullman AM, Liu Y, Huynh M, Bediako DK, Wang H, Anderson BL, Korean Chem Soc, 2008, 29: 127–129
Powers DC, Breen JJ, Abruña HD, Nocera DG. J Am Chem Soc, 2014, 36 Fu L, Wu D, Xiao Y, Zou W, Kuang J, Liang Y, Du F, Peng J, Ye X. J
136: 17681–17688 Alloys Compd, 2021, 878: 160354
10 Evangelisti F, Moré R, Hodel F, Luber S, Patzke GR. J Am Chem Soc, 37 Rawalekar S, Kaniyankandy S, Verma S, Ghosh HN. J Phys Chem C,
2015, 137: 11076–11084 2010, 114: 1460–1466
11 Fang WH, Zhang L, Zhang J. J Am Chem Soc, 2016, 138: 7480–7483 38 Berr MJ, Vaneski A, Mauser C, Fischbach S, Susha AS, Rogach AL,
12 Fan X, Wang J, Wu K, Zhang L, Zhang J. Angew Chem Int Ed, 2019, Jäckel F, Feldmann J. Small, 2012, 8: 291–297
58: 1320–1323 39 van Wilderen LJGW, Lincoln CN, van Thor JJ. PLoS ONE, 2011, 6:
13 Tamang SR, Singh A, Bedi D, Bazkiaei AR, Warner AA, Glogau K, e17373
McDonald C, Unruh DK, Findlater M. Nat Catal, 2020, 3: 154–162 40 Kresse G, Furthmüller J. Comput Mater Sci, 1996, 6: 15–50
14 Pan ZH, Weng ZZ, Kong XJ, Long LS, Zheng LS. Coord Chem Rev, 41 Kresse G, Hafner J. Phys Rev B, 1997, 55: 7539–7548
2022, 457: 214419 42 Han F, Zhou Z, Zhang X, Huang Z, Li M, Guo L. J Phys Chem C,
15 Guo S, Kong L, Wang P, Yao S, Lu T, Zhang Z. Angew Chem Int Ed, 2018, 122: 2546–2553
2022, 61: e202206193 43 Hu W, Lin L, Zhang R, Yang C, Yang J. J Am Chem Soc, 2017, 139:
16 Li N, Lin JM, Li RH, Shi JW, Dong LZ, Liu J, He J, Lan YQ. J Am 15429–15436