Polymer synthesis and characterization. To demonstrate our design strategy, poly[(9,9-dioctyl-9H-fluorenyl-2,7-diyl)-co-(5-phenylbenzo[b]phosphindole-5-oxide-2,7-diyl)] (PFBPO) with a hydrophobic backbone was chosen as the reference CP for comparison with the DCPs because of its relatively high HER in visible-light-driven photocatalytic reactions27. We designed ethylene glycol (EG)- and ethylene diamine (EA)-based hydrophilic non-conjugated building blocks and incorporated them into the backbone of PFBPO. We synthesized EG- and EA-based DCPs via Pd-catalysed Suzuki−Miyaura coupling polymerization of fluorene-boronic ester and 3,7-dibromo-5-phenylbenzo[b]phosphindole-5-oxide with either EG-based (EG-Br, TEG-Br, and HEG-Br) or EA-based (EA-Br) building blocks. The hydrophilic non-conjugated building blocks were covalently bonded to the backbone of PFBPO at various ratios (5, 10, and 20 mol%), resulting in seven DCP photocatalysts, denoted P-5EG, P-5TEG, P-5HEG, P-10HEG, P-20HEG, P-5EA and P-10EA (Fig. 1b, Supplementary scheme 1). First, the monomers were synthesized and their chemical structures were identified by NMR spectroscopy and mass spectrometry (Supplementary Fig. 1–8). The monomers were then polymerized into the corresponding EG- and EA-based DCPs, the chemical structures of which were investigated in detail using NMR and FT-IR spectroscopy. The 1H NMR spectra (Supplementary Figs. 9–15) showed characteristic signals of the EG- and EA-based building blocks at 3.5–4.5 ppm, indicating that hydrophilic non-conjugated groups were present in the polymer backbones. The intensities of the characteristic signals of HEG-Br and EA-Br in the DCP backbones clearly increased with increasing molar fractions of the HEG-Br and EA-Br monomers during polymerization (Supplementary Figs. 11–15). Characteristic NMR and FT-IR signals of the phosphine oxide groups were observed at 33.80 ppm (Supplementary Fig. 16–22) and 1140–1210 cm-1 (Supplementary Fig. 23), respectively. The molecular weight and polydispersity index of the DCPs were determined using gel permeation chromatography (Supplementary Table 1). The results of thermogravimetric and X-ray diffraction analyzes indicated that all prepared polymers had an amorphous framework that was highly stable in a nitrogen atmosphere (Td > 400 °C; see Supplementary Figs. 24–25).
Optical properties. The optical properties, optical band gap (Eg), and energies of the highest occupied molecular orbital (HOMO) and lowest unoccupied molecular orbital (LUMO) of the DCPs were studied by diffuse reflectance UV–vis spectroscopy (DRS), photoluminescence spectroscopy, and photoelectronic spectroscopy. The DRS of the DCPs was very similar to that of PFBPO (Fig. 2a). The HOMO levels of the DCPs ranged from −5.90 to −5.93 eV, as measured using a photoelectronic spectrometer (Supplementary Fig. 26). The LUMO levels were calculated using ELUMO = EHOMO + Eg, where the Eg values of the DCPs calculated from Tauc plots were close to 2.75 eV (Fig. 2b); the LUMO levels ranged from −3.15 to −3.20 eV, indicating that hydrogen evolution was thermodynamically favourable (Table 1). The differences between the HOMO/LUMO levels and the optical Eg were small, indicating that hydrophilic non-conjugated building blocks could be molecularly engineered into the main-chain of the target CPs using the present approach without significantly affecting the physicochemical and optical properties (Fig. 2c). The photoluminescence spectra show that both DCPs and PFBPO were effectively quenched by adding Pt co-catalysts (Fig. 3a). The photoluminescence quenching of P-5EG, P-10HEG, and P-5EA was higher than that of PFBPO, indicating that the charge transfer from DCPs to the Pt co-catalysts was improved. The time-resolved photoluminescence spectra (Fig. 3b) show that the lifetimes of the excited states of the DCPs were 1.02–1.22 ns. The P-10HEG had the longest lifetime, and all of the DCPs had longer lifetimes than PFBPO (0.84 ns), suggesting that the recombination rates of the electron–hole pairs were effectively reduced.
Table 1. Photophysical properties and HERs of various polymers.
Polymer
|
λmax, abs (nm)a
|
Size (nm)b
|
HOMO
(eV)c
|
LUMO
(eV)d
|
Eg
(eV)e
|
Lifetime
(ns)f
|
HER
780 > λ > 380 nm
(µmol h-1)g
|
HER
780 > λ > 380 nm
(mmol h-1 g-1)g
|
PFBPO
|
410
|
471
|
-5.82
|
-3.13
|
2.69
|
0.84
|
23.0 ± 1.85
|
4.60 ± 0.37
|
P-5EG
|
410
|
654
|
-5.92
|
-3.18
|
2.74
|
1.03
|
23.9 ± 2.60
|
4.78 ± 0.52
|
P-5TEG
|
408
|
674
|
-5.91
|
-3.19
|
2.72
|
1.02
|
25.8 ± 1.35
|
5.16 ± 0.27
|
P-5HEG
|
414
|
686
|
-5.91
|
-3.16
|
2.75
|
1.11
|
26.9 ± 1.55
|
5.38 ± 0.31
|
P-10HEG
|
402
|
1129
|
-5.92
|
-3.16
|
2.76
|
1.22
|
34.7 ± 3.45
|
6.94 ± 0.69
|
P-20HEG
|
390
|
1576
|
-5.90
|
-3.20
|
2.70
|
1.11
|
18.7 ± 1.80
|
3.74 ± 0.36
|
P-5EA
|
402
|
957
|
-5.93
|
-3.20
|
2.73
|
1.12
|
30.3 ± 3.05
|
6.06 ± 0.61
|
P-10EA
|
396
|
1434
|
-5.91
|
-3.15
|
2.76
|
1.07
|
19.4 ± 1.95
|
3.88 ± 0.39
|
aDetermined by solid-state UV-vis diffuse reflectance spectroscopy; bDetermined by DLS; cDetermined by photoelectron spectroscopy; dDetermined by means of the equation ELUMO = EHOMO + Eg; eDetermined from Tauc plots; fDetermined by time-resolved fluorescence spectroscopy; gPhotocatalytic conditions: 5 mg of polymers dissolved in a 10 mL mixture solution (H2O/MeOH/TEA) and irradiated by a simulated solar light source (300 W Xe lamp, 1000 W m-2, 780 > λ > 380 nm).
Electrochemical properties. Electron paramagnetic resonance (EPR), electrochemical impedance spectroscopy (EIS), and transient photocurrent measurements were used to characterize the charge generation, transport, and transfer upon irradiation of the DCPs. In the EPR spectra (Fig. 3c, Supplementary Fig. 27), a sharp signal was observed at 3500–3520 G and its intensity increased with light irradiation due to radical generation via photoexcitation. The irradiation-induced signal enhancement for the DCPs was greater than that for PFBPO. The EIS data recorded in the dark showed a semicircular curve (Fig. 3d), where the DCPs had resistance values lower than that of PFBPO due to their enhanced charge transport. Transient photocurrent measurements (Fig. 3e) used to investigate the photoresponse of the DCPs showed that these materials responded quickly to light irradiation and their photocurrents were higher than that of PFBPO. Hence, hydrophilic building blocks may exhibit a suitable interface with water to facilitate charge transport and transfer. Cyclic voltammetry analysis of P-10HEG showed that this material was highly electrochemically stable with a reversible reduction–oxidation cycle in the range from −0.2 V to −1.1 V for at least 15 cycles (Fig. 3f).
Morphology and hydrophilicity. Scanning electron microscopy (SEM; Supplementary Fig. 28–35) images showed that the solid DCPs consisted of small particles with similar morphologies and microscale sizes, independent of the hydrophilic non-conjugated building blocks. The water contact angles of the EG-Br, TEG-Br, HEG-Br, and EA-Br monomers were 95.7°, 82.1°, 47.4°, and 76.7°, respectively (Supplementary Fig. 36), where HEG-Br had the highest hydrophilicity (lowest contact angle). Interestingly, we measured the contact angles of water drop on all of the DCPs films promptly, and observed hydrophobic behaviour with the contact angles of approximately 109° (Supplementary Fig. 37), which were in contrast to most side-chain-engineered CPs that show hydrophilic behavior19-22. This result implies that the type and content of hydrophilic non-conjugated building blocks in the hydrophobic CPs do not significantly affect the hydrophilicity of the DCP films due to packing of the polymers. The particle size distributions of the DCPs in a H2O/methanol (MeOH)/triethylamine (TEA) (1:1:1 by volume) solution were determined by dynamic light scattering (DLS). The DLS results (Supplementary Fig. 38) indicated a wide range of hydrodynamic diameters for PFBPO P-5EG, P-5TEG, P-5HEG, P-10HEG, P-20HEG, P-5EA, and P-10EA of 471, 654, 674, 686, 1129, 1576, 957, and 1434 nm, respectively. All of these values were larger than that of PFBPO, indicating that the DCPs swelled in the aqueous solution28,29. Hence, the hydrophilicity of the DCPs could be tuned by using different types or contents of hydrophilic non-conjugated building blocks to allow water to penetrate the bulk of the polymer in the solution state. To further investigate the morphology of the DCPs, small-angle X-ray scattering (SAXS) measurements were performed by dispersing the DCPs in a 1:1:1 H2O/MeOH/TEA solution and comparing their architectures. The intermediate- and high-q scattering characteristics (q = 0.4−4 nm-1) shown in the SAXS patterns of P-5HEG, P-10HEG, and P-20HEG (Supplementary Fig. 39) were ascribed to a one-dimensional rod geometry with polydispersity in the rod size30. Fitting of the high-q scattering profile indicated that the DCPs formed rod-like bundles in aqueous solution and that their length (2 ± 0.004 nm) was independent of the content of hydrophilic non-conjugated building blocks. However, the low-q scattering profile (q < 0.4 nm-1) of P-10HEG was fitted with the lowest power law exponent of -2.0, revealing that the rod-shaped bundles of P-10HEG can be further organized into an aggregated structure with a random distribution. The lower degree of aggregation implied a loose arrangement of the polymer bundles, resulting in a larger interfacial area.
Photocatalytic hydrogen evolution. The photocatalytic hydrogen evolution of the DCPs was measured at 25 °C under visible-light irradiation (λ = 380–780 nm) and compared to that of PFBPO. Fig. 4a shows that the HER values of the DCPs with 5–10 mol% hydrophilic non-conjugated building blocks were clearly higher than that of PFBPO. In particular, a stable HER of P-10HEG was observed with increasing reaction time, and an excellent HER of 34.7 μmol h−1 was obtained, which was approximately 50% higher than that of PFBPO (23.0 μmol h−1). The hydrophilic non-conjugated building blocks facilitated the swelling of the polymer aggregates in the presence of water and thus enhanced the HER of the DCPs (Table 1). However, the HER of P-20HEG (with 80 mol% BPO active building blocks and 20 mol% HEG hydrophilic building blocks) was lower, with a value of 18.7 μmol h−1. Therefore, the optimal content of hydrophilic HEG building blocks on the backbone of DCPs was approximately 10 mol%. Considering the SAXS results, the fact that P-10HEG exhibited the highest HER was closely related to its loose aggregates, which provide a high interfacial area for hydrogen evolution.
The P-5EA sample also showed a high HER of 30.3 μmol h−1, which was higher than those of its counterparts with 5 mol% EG-based hydrophilic building blocks (e.g., P-5EG, P-5TEG, and P-5HEG). The EA-based hydrophilic building blocks with N-containing functional groups experienced stronger hydrogen bonding with water than the EG-based hydrophilic building blocks with oxygen-containing functional groups. In addition, when 10 mol% EA was present, the corresponding P-10EA exhibited a relatively low HER of 19.4 μmol h−1. In the EA-based DCPs, the optimal content of EA hydrophilic building blocks on the backbones of the DCPs was approximately 5 mol%. As seen from the DLS data, P-10HEG and P-5EA were significantly swollen with water in the H2O/MeOH/TEA solution, forming aggregates with a diameter of approximately 1000 nm, which were suitable for water-based photocatalytic hydrogen evolution. It was demonstrated that the HER of polymer photocatalysts can indeed be increased by diffusing more water into the active building blocks via main-chain engineering of hydrophilic building blocks within the backbones of hydrophobic CPs.
Similar results were found when the concentrations of CPs and DCPs in the photocatalytic solution were increased. The HER of PFBPO was 23.0 μmol h-1 in a 5 mg/10 mL solution, and it increased only 1.3–1.6 times in 10–15 mg/10 mL solutions (Fig. 4b, Supplementary Table 1). However, the HER of P-10HEG increased 1.5, 2.9, and 3.5 times in 5 mg/10 mL (HER = 34.7 μmol h-1), 10 mg/10 mL (HER = 66.1 μmol h-1), and 15 mg/10 mL (HER = 81.6 μmol h-1) solutions, respectively, compared to that of PFBPO in the 5 mg/10 mL photocatalytic solution. Notably, P-20HEG exhibited a higher HER than PFBPO in 10 mg/10 mL and 15 mg/10 mL solutions, demonstrating that the optimized concentration saturation of the DCPs was much higher than that of the respective CPs. This finding suggests that increasing the concentration of CPs in a photocatalytic solution enhances their aggregation, so that the inner polymer chains of the CPs cannot interact effectively with the water. Likewise, a linear increase in the HER was observed with increasing quantity of photocatalytic solutions in the absence of a Pt co-catalyst (Fig. 4c, Supplementary Fig. 40), demonstrating that, at a fixed concentration, the total amount of hydrogen production could be increased by increasing the reactor size. Apparent quantum yield (AQY) values were obtained under standard photocatalytic conditions using a light source with a bandpass filter (λ = 420, 460, 500, 550, or 600 nm). Without a Pt co-catalyst, P-10HEG exhibited high AQYs of 18.19% and 17.82% at 420 and 460 nm, respectively (Fig. 4d, Supplementary Table 2). The trend in AQY values as a function of wavelength was very similar to that of the UV-vis absorption spectra of P-10HEG, indicating that photocatalytic hydrogen evolution occurred via light harvesting. Taking advantage of the strong interaction between the hydrophilic non-conjugated building blocks and the Pd co-catalyst, which can enhance the charge separation and transfer at the interfacial area26,31, the HER of P-10HEG with 5 wt.% Pt co-catalyst was increased to 54.1 μmol h−1 (10.82 mmol h−1 g–1) (Supplementary Fig. 41).
We further verified the performance of the DCPs for photocatalytic hydrogen evolution in the absence of MeOH (i.e., in a H2O/TEA solution, Fig. 4e). PFBPO exhibited a very low HER (6.79 μmol h−1), while P-10HEG and P-5EA exhibited values of up to 22.8 μmol h−1 and 18.6 μmol h−1, respectively, in the H2O/TEA solution, and these values were close to that of PFBPO in the presence of MeOH (23.0 μmol h−1). Therefore, the introduction of 5–10 mol% HEG-Br or EA-Br hydrophilic building blocks on the main-chain of DCPs could replace the use of the typically used 33 vol.% MeOH. In addition, compared to side-chain engineering of CPs (containing 50 mol% hydrophilic building blocks), our approach required a smaller amount of hydrophilic building blocks (5–10 mol%) and could effectively increase the interaction between water and the inner active sites of the CPs to increase the HER. To further demonstrate the advantages and applicability of our design approach, we synthesized a side-chain-engineered CP (denoted as PFTEGBPO) to compare its photocatalytic HER with that of the P-10HEG DCPs (Fig. 4f, Supplementary Fig. 42). The results proved that our main-chain engineering strategy utilized the inner active sites of CPs more effectively than the side-chain engineering strategy. Furthermore, to demonstrate the universality of our approach, we extended the hydrophilic HEG-Br building block to polymerize with other donor–acceptor CPs. A PF8BT-10HEG sample was synthesized by the main-chain engineering of 10 mol% HEG-Br into the backbone of a PF8BT, which exhibited an enhanced HER compared to that of PF8BT (Supplementary Fig. 43), proving that this approach can be easily applied to different CPs.
Furthermore, we coated the PFBPO and P-10HEG to obtain uniform films on silicon wafers by only single dropcast cycle (Fig. 5a,b). As shown in Fig. 5c and Supplementary Table 3, the P-10HEG showed a record high HER of 16.6 mmol m-2 h-1 in a film state. Interestingly, the HER of P-10HEG showed 2-times enhancement compared to that of PFBPO, which is higher than the enhancement in solution state (1.5-times). We also measured the water contact angle of film photocatalysts over a period of time to determine the wetting effect of these films (Fig. 5d,e). In the case of the P-10HEG film with the initial contact angle of 107.9° and then decreased to 65.5° over the period of 20 minutes. While for the PFBPO film, the initial water contact angle of 112.0° decreased to 79.3° after 20 minutes. Since the molecular packing of polymer in film state is severe than that in solution state, resulting the inside of the polymer film may not react with water efficiently, this result demonstrated the advantage of our main-chain engineered DCP as film photocatalysts.
Molecular dynamics simulation provides a microscale understanding of the interaction between water and DCP. Improvement in hydrogen results from a higher possibility of hydrogen bond formation. To further examine the mechanisms at the microscale, we considered three different models: a non-conjugate block located at the middle, one-third, and the end of the polymer chain. Here, we consider that the DCPs are composed mainly of these three configurations with different weight percentages. Fig. 6a shows the average number of hydrogen bond formations for different DCPs. The statistic of hydrogen bounds suggest that the system composed of 80% of the DCP with conjugate monomer located at the center, 15% at 1/3 of the overall chain, and 5% at the end, in agreement with the trend of experimental findings. Fig. 6b plots the radial distribution function and the morphology of each DCP chain. One can observe that DCP with 10 mol% HEG-modified exhibits the highest possibility of hydrogen bond formation and the probability of radial distribution function. In contrast, a 10 mol% EA-modified DCP shows a relatively low possibility of hydrogen bond formation. Simulation results also reveal that 10 mol% HEG-modified and 5 mol% EA-modified DCPs were entangled more than other DCP systems where other DCPs remained in single-chain formation.