Secondary coordination sphere accelerates hole transfer for enhanced hydrogen photogeneration from [FeFe]-hydrogenase mimic and CdSe QDs in water

Achieving highly efficient hydrogen (H2) evolution via artificial photosynthesis is a great ambition pursued by scientists in recent decades because H2 has high specific enthalpy of combustion and benign combustion product. [FeFe]-Hydrogenase ([FeFe]-H2ase) mimics have been demonstrated to be promising catalysts for H2 photoproduction. However, the efficient photocatalytic H2 generation system, consisting of PAA-g-Fe2S2, CdSe QDs and H2A, suffered from low stability, probably due to the hole accumulation induced photooxidation of CdSe QDs and the subsequent crash of [FeFe]-H2ase mimics. In this work, we take advantage of supramolecular interaction for the first time to construct the secondary coordination sphere of electron donors (HA−) to CdSe QDs. The generated secondary coordination sphere helps realize much faster hole removal with a ~30-fold increase, thus leading to higher stability and activity for H2 evolution. The unique photocatalytic H2 evolution system features a great increase of turnover number to 83600, which is the highest one obtained so far for photocatalytic H2 production by using [FeFe]-H2ase mimics as catalysts.

Nevertheless, these [FeFe]-H 2 ase mimics and QDs based multi-component systems suffer from poor stability with gradual decomposion of both QDs and [FeFe]-H 2 ase mimics during irradiation, which is probably due to the hole transfer rate slower than the electron transfer rate 23,24 . We found, for example, that the electron transfer rate from CdSe QDs to the active site of PAA-g-Fe 2 S 2 is three-order faster than the hole transfer rate from CdSe QDs to H 2 A 23 . In this regard, the cease of H 2 evolution might be a result of accumulating holes in the valence band of CdSe QDs. One may question whether the stability and efficiency of photocatalytic H 2 evolution system could be enhanced by means of facilitating hole removal.
In this contribution, we wish to report that the use of secondary coordination sphere greatly improves the activity and stability of the artificial [FeFe]-H 2 ase-based system, consisting of CdSe QDs, PAA-g-Fe 2 S 2 and H 2 A in water (Fig. 1). Although PAA played a role in stabilizing CdSe QDs to some extent 23 , the protonation of carboxylic groups under acidic condition would reduce the association of PAA with CdSe QDs 25 . Further, the negatively charged carboxyl groups would prevent CdSe QDs from negatively charged HA − in proximity. In this situation, we envision that addition of polyethyleneimine (PEI) would be helpful for the intimate interaction of CdSe QDs with PAA-g-Fe 2 S 2 and H 2 A because PEI has the ability to protect cadmium chalcogenide nanocrystals from aggregation in an extremely wide pH range 26,27 . Unlike the most state-of-the-art approaches [28][29][30][31][32][33] , we take advantage of supramolecular interaction to increase the contact between CdSe QDs and electron donors (HA − ). The positive charge of PEI under acidic condition 34 not only endows CdSe QDs a capacity to associate with PAA-g-Fe 2 S 2 intimately, but also enables HA − to contact closely with CdSe QDs than that in the absence of PEI. Indeed, the as-generated secondary coordination sphere helps with acceleration of hole transfer to a ~30-fold increase, while the electron transfer to PAA-g-Fe 2 S 2 remains unchanged. As a result, the designed system can catalyze proton reduction under visible-light irradiation for 28 h, giving rise to a TON value high up to 83600. This is, to the best of our knowledge, the highest value known to date based on [FeFe]-H 2 ase mimics for H 2 photogeneration.

Results
Photocatalytic H 2 generation. The H 2 evolution experiments were carried out in argon saturated aqueous solution of CdSe QDs, PAA-g-Fe 2 S 2 and H 2 A at pH 4.1 under visible-light irradiation. PAA-g-Fe 2 S 2 and CdSe QDs were prepared according to the reported procedures 23 . The characterizations of PAA-g-Fe 2 S 2 and CdSe QDs were described in Supplementary Information (Figs S1 and S2). As shown in Fig. 2, the photocatalytic system, containing CdSe QDs, PAA-g-Fe 2 S 2 and H 2 A at pH 4.1, ceased to evolve H 2 in 4 h, but a significant enhancement of H 2 evolution was achieved in the presence of 0.12 g·L −1 PEI under the same condition ( Supplementary  Fig. S3). Increasing the concentration of PEI to 0.46 g·L −1 improved the rate of H 2 production ( Supplementary  Fig. S4) till higher concentration to 1.84 g·L −1 . The excessive amount of PEI probably retarded the interaction between CdSe QDs and H 2 A and/or PAA-g-Fe 2 S 2 because the average hydrodynamic diameter of CdSe QDs remained unchanged ( Supplementary Fig. S5). Note that the rate of H 2 evolution was positively proportional to the concentration of H 2 A in the range of 0 to 0.1 mol·L −1 ( Supplementary Fig. S6), and negligible H 2 was obtained in the absence of H 2 A (0 mol·L −1 ), PEI and PAA themselves were not possible to serve as electron donors in the designed system. Control experiment, performed under the same concentrations of CdSe QDs, H 2 A, PAA and PEI, evolved much less H 2 , which confirmed the role of Fe 2 S 2 for proton reduction ( Supplementary Fig. S7). To our delight, the rate of H 2 production was linear in 10 h and the lifetime of H 2 evolution could be prolonged to 28 h, yielding an unprecedented TON value to 83600 (Fig. 2). Influence of PEI to the stability of CdSe QDs. The presence of PEI could improve the efficiency and stability of the CdSe QDs and PAA-g-Fe 2 S 2 system for photocatalytic H 2 evolution. To verify the function of PEI, dynamic light scattering (DLS) and high-resolution transmission electron microscope (HRTEM) were employed. As shown in Fig. 3a, CdSe QDs dispersed well without any mutual aggregation under neutral condition. Upon introduction of 0.1 mol·L −1 H 2 A (pH 4.1) to the solution, the average hydrodynamic size of CdSe QDs increased dramatically to ~1400 nm and severe aggregation of CdSe QDs was seen directly from the corresponding TEM image (Fig. 3b), a result of dissociation of surface ligands, mercaptopropionic acid (MPA), from CdSe QDs at acidic condition 35 . When a certain amount of PAA-g-Fe 2 S 2 (0.25 g·L −1 ) was added into the solution, the hydrodynamic size of CdSe QDs decreased from ~1400 nm to ~192 nm. The better dispersion of CdSe QDs in the presence of PAA-g-Fe 2 S 2 was well reflected by TEM image (Fig. 3c), which provided evidence on the stabilization of PAA to CdSe QDs since there are no chemical groups in Fe 2 S 2 cluster can stabilize CdSe QDs. The better dispersity of CdSe QDs caused by the coordination of PAA on the surface could also be confirmed by the enhanced emission intensity and blue-shift of band edge emission ( Supplementary Fig. S8). More strikingly, when PEI (0.46 g·L −1 ) was simultaneously introduced into the solution of CdSe QDs and PAA-g-Fe 2 S 2 , the average size distribution of CdSe QDs decreased to merely ~17 nm, which could be comparable to that in neutral water (Fig. 3d). Also, high-resolution TEM indicated the excellent dispersity of CdSe QDs in the co-presence of PAA and PEI (Fig. 3d). Clearly, the coordination between amino groups on PEI 36,37 with Cd 2+ is so strong even under acidic condition that the presence of PEI allows an excellent dispersion of CdSe QDs at pH 4.1. The solution of the in situ generated PEI and PAA co-stabilized CdSe QDs was found stable for at least 7 days ( Supplementary Fig. S9). The better stability of CdSe QDs in the presence of PEI ensured better interaction with PAA-g-Fe 2 S 2 , which in turn improved the efficiency and stability of this photocatalytic H 2 production system. Consistent with the observation of HRTEM and DLS studies, an apparent blue shift of the CdSe QDs emission in the presence of PEI (Table 1 and Supplementary Fig. S8) suggested that PEI could stabilize CdSe QDs and prevent the formation of large aggregates 27 . Furthermore, with the addition of PEI (0.46 g·L −1 ) into the solution of CdSe QDs and PAA, the emission decay was apparently facilitated from 14.7 ns to 9.6 ns ( Supplementary Fig. S10). With reference to the reports in literature [38][39][40] , the shortened lifetime could be attributed to the delocalization of photo-generated excitons, especially holes, to the surface layer of PEI, which would further benefit hole depletion of CdSe QDs by relaying holes to sacrificial reagents.
Influence of PEI to the electron transfer rate of CdSe QDs. The reductive potential of PAA-g-Fe 2 S 2 was determined to be − 0.43 V (vs NHE), demonstrating an exothermic electron transfer from excited CdSe QDs to PAA-g-Fe 2 S 2 because the conduction band potential of CdSe QDs is more negative than − 0.43 V (vs NHE) 23 . Spectroelectrochemical and time-resolved absorption spectra were then used to study the electron transfer and hole transfer process in the designed system [41][42][43][44] . Either in the absence or presence of PEI, electrochemical reduction of PAA-g-Fe 2 S 2 led to the formation of a new absorption peak around 400 nm, which could be assigned to Fe I Fe 0 species (Supplementary Fig. S11) [45][46][47] . Under the same condition, electrochemical reduction of PAA and PEI couldn't result in the formation of similar signals ( Supplementary Fig. S12). The results suggested that the introduction of PEI would not influence the active site of PAA-g-Fe 2 S 2 very much. Time-resolved transient absorption spectra were measured to further evaluate the electron transfer from CdSe QDs to PAA-g-Fe 2 S 2 . As PAA-g-Fe 2 S 2 was added into the aqueous solution of CdSe QDs, the bleaching recovery of CdSe QDs was apparently facilitated from 36.9 ns to 11.5 ns, a result of electron transfer from excited CdSe QDs to the active site of PAA-g-Fe 2 S 2 (Fig. 4, Supplementary Table S1) 23,41 . The rate of electron transfer (k e ) was determined 41   QDs to varied extent. Combining with the emission decay with and without PEI ( Supplementary Fig. S10), the rate of hole transfer (k h ) from CdSe QDs to HA − was determined as 1.5 × 10 10 M −1 ·s −1 and 4.6 × 10 8 M −1 ·s −1 (Table 1, Fig. 5, Supplementary Figs S13 and S14, see details in Supplementary Information). Surprisingly, a ~30-fold enhancement of hole transfer was obtained with the addition of PEI. The presence of PEI facilitated the hole-extraction pathway of CdSe QDs by HA − . Considering the coordination of amino groups of PEI on the surface of CdSe QDs, the superficial electric charge of CdSe QDs would be altered. Indeed, the zeta potentials (ζ ) of CdSe QDs in H 2 A (0.1 mol·L −1 ) solution, and in H 2 A (0.1 mol·L −1 ) and PAA (0.25 g·L −1 ) solution were measured to be − 4.7 mV and − 8.7 mV, respectively (Table 1, Supplementary Table S2). After adding PEI (0.46 g·L −1 ) into the solution, the ζ value changed to + 25.7 mV, indicating the extremely strong positive charge of PEI modified CdSe QDs. On the one hand, the strongly positive charge ensured the excellent stability of CdSe QDs in water due to the strong electrostatic repulsion, vide ante ( Supplementary Fig. S9). On the other hand, the inversed charge of CdSe QDs surface provided a driving force to interact with negatively charged HA − by electrostatic interaction.  The as-generated secondary coordination sphere strengthened the association of HA − with CdSe QDs, which in turn accelerated the hole transfer from CdSe QDs to HA − 48 . The accelerated hole transfer process is well-manifested by the model of electrical double layer (EDL) to estimate the concentration of adsorbed counterion (see details in Supplementary Information) 49 . For PAA and PEI/ PAA stabilized CdSe QDs, the surface adsorbed HA − ions were determined as 0.07 mol·L −1 and 0.27 mol·L −1 , respectively. The secondary coordination sphere of HA − to CdSe QDs, simply by addition of PEI, contributed a ~30-fold increase of hole transfer from CdSe QDs to HA − and a significant decrease of k e /k h ratio from ~740 to ~30, which benefited the balance of excitons consumption and protected CdSe QDs from photo-oxidation during visible-light irradiation. As a result, the CdSe QDs and PAA-g-Fe 2 S 2 system that would decompose in 4 h irradiation showed a remarkably enhanced activity and stability for photocatalytic H 2 evolution.

Discussion
On the basis of above results, we speculated that the secondary coordination sphere is crucial for the enhancement of H 2 photogeneration from [FeFe]-H 2 ase mimic and CdSe QDs in water. Upon photoexcitation, CdSe QDs induces electrons promoting to the conduction band and leaving holes in the valence band to form separated electron/hole pairs. Photoexcited electron in the conduction band of CdSe QDs transfers to the Fe 2 S 2 core of PAA-g-Fe 2 S 2 and generates a one-electron reduced Fe I Fe 0 species. This active species further reacts with a proton in catalytic cycle to produce H 2 . At the same time, the leaving hole in the valence band of CdSe QDs is captured by electron donor HA − to regenerate CdSe QDs. Because two electrons are required to produce each molecular H 2 , the regeneration of both CdSe QDs and Fe 2 S 2 active site of [FeFe]-H 2 ase mimic is imperative. So the better balance of electron transfer and hole transfer of CdSe QDs to PAA-g-Fe 2 S 2 and HA − is, the higher efficiency of H 2 evolution would be. In this typical PAA-g-Fe 2 S 2 -based system, the in situ PEI-modification can greatly improve the stability of CdSe QDs to a well-dispersed colloidal solution under acidic condition (Fig. 1), which ensures smooth electron transfer from CdSe QDs to PAA-g-Fe 2 S 2 . More importantly, the in situ PEI modification induces the secondary coordination of negatively charged sacrificial reagents (HA − ) to the positively charged surface of the modified CdSe QDs, which contributes a 30-fold enhancement of hole transfer rate. Benefitting from these advantages, the photocorrosion of CdSe QDs is avoided to a great extent by facilitating hole extraction, and thus greatly enhancing the efficiency and stability of CdSe QDs and PAA-g-Fe 2 S 2 system for photocatalytic H 2 evolution. In summary, we have demonstrated for the first time that secondary coordination sphere enables to facilitate hole transfer for efficient H 2 photogeneration. In terms of supramolecular interaction, the performance of photocatalytic system, i.e. CdSe QDs, PAA-g-Fe 2 S 2 , and H 2 A at pH 4.1, has been improved by PEI to 83600 turnovers, which is the highest value known to date for photocatalytic H 2 evolution by using [FeFe]-H 2 ase mimics as catalysts. Studies on steady-state and time-resolved spectroscopy reveal that the presence of PEI greatly enhances the rate of hole transfer up to ~30-fold as compared with the same system without PEI. As a result, the ratio (k e /k h ) of electron transfer and hole transfer from CdSe QDs to PAA-g-Fe 2 S 2 and HA − decreases from 740 to 30, which is advantageous to the balance of excitons generated by photoexcited CdSe QDs to synergistically improve the efficiency of H 2 evolution. Our results imply that environment surrounding each component, for example, CdSe QDs photosensitizer, Fe 2 S 2 catalyst and H 2 A sacrifical electron donor and proton source in this case, might cause significant activity difference of photocatalytic system. The crucial role of PEI suggests that to create efficient H 2 evolution systems based on [FeFe]-H 2 ase mimic, one would need to mimic not only the active site structure of [FeFe]-H 2 ase but also the kinetic balance for electron transfer and hole transfer in a real H 2 evolution system. It is anticipated that the use of secondary coordination sphere would be a promising alternative to enhance the stability and efficiency of artifical [FeFe]-H 2 ase-based system for H 2 evolution, which is reminiscent of natural [FeFe]-H 2 ase buried in protein matrix.

Methods
H 2 evolution experiments. A typical procedure for H 2 production was described as follows. Certain amounts of PAA-g-Fe 2 S 2 , MPA-CdSe QDs, H 2 A and PEI were dissolved in ultrapure water respectively, to make a solution at certain concentration. Then, certain volumes of the solutions for PAA-g-Fe 2 S 2 , MPA-CdSe QDs, H 2 A and PEI were taken to mix in a Schlenk tube. The pH value of the solutions were determined by a pH meter and adjusted by aqueous NaOH or HCl solution. The total volume of the mixed solution was diluted with ultrapure water to 5 mL. Thereafter, the sample was saturated with argon gas and 1000 μ L CH 4 was injected as the internal standard for quantitative GC-TCD analysis. The light source was a blue LED lamp (3 W, λ = 450 nm) equipped with an agitator and a cooling apparatus. After a certain period of irradiation time, 500 μ L mixed gas was taken from the sample tube and injected into the GC for analysis. The response factor for H 2 was 8.47 and the response factor for CH 4 was 2.09 under the experimental condition, which were established by calibration with known amounts of H 2 and CH 4 , and determined before and after a series of measurements.  50 . Briefly, selenium powder (40 mg) was added into Na 2 SO 3 aqueous solution (100 mL, 1.5 mmol). The mixture was refluxed at 130 degree until the solid powder of selenium disappeared and then colourless transparent Na 2 SeSO 3 solution was obtained. The Na 2 SeSO 3 solution (10 mL) was extracted and injected into the argon saturated CdCl 2 solution (46 mg CdCl 2 ·2.5H 2 O in 190 mL ultrapure water) at pH 11 in the presence of 3-mercaptopropionic acid (26 uL) as stabilizing agent. The resulting mixture was refluxed at 120 degree to control the growth of CdSe QDs. The size of QDs was monitored by UV-vis absorption spectrum during refluxing.
Synthesis of PAA-g-Fe 2 S 2 . The water soluble catalyst PAA-g-Fe 2 S 2 was synthesized according to our previous procedure 23 . The color of pure PAA is white and the color of pure Fe 2 S 2 active site is red. After modification, the color of PAA changed from white to light red, indicating the successful anchor of Fe 2 S 2 on the chain of PAA. 1 H-NMR, UV/Vis spectrum and FTIR spectroscopy further confirmed the modification of the Fe 2 S 2 active site onto the chain of PAA. The grafting amount of Fe 2 S 2 was determined as 2.8 × 10 −6 mol·g −1 for PAA-g-Fe 2 S 2 by inductively coupled plasma-atomic emission spectrometry (ICP-AES) based on the relative content of Fe in samples. Spectroelectrochemical and time-resolved absorption spectra experiments employed PAA-g-Fe 2 S 2 with larger grafting amount of Fe 2 S 2 site (7.1 × 10 −4 mol·g −1 ) as electron capture to obtain more obvious signals.
Details of the instruments, chemicals and the computing methods used in this work are given in the Supplementary Information.