2.1 Preparation process
As shown in Figure 1a, the amorphous CN is obtained by thermal polymerization of melamine at 550℃ in the air for 3 h. Subsequently, its amorphous structure can be recrystallized by using KCl as structure inducer to furnish K+ implanted CCN.17 Figures 1b-c shows the proposed structure of heptazine-based CN and KCN. The gap between one-dimensional amine-linked heptazine-based melon chains in CN provides a very good diffusion channel for K+ ions. In addition, the diameter of K+ ions (0.276 nm) is smaller than the interlayer spacing of CN (0.32 nm), so its interlayer is also a feasible diffusion channel. It can be predicted that with the increase of diffusion time, the K+ ions concentration gradient will gradually increase. When the heat treatment reaches a certain time, the surface and bulk concentration of K+ ions tend to be uniform, and the gradient will decrease. Therefore, the K+ ions gradient doping in KCN can be constructed by controlling the diffusion time. The structure of KCN consists of one-dimensional amine-linked heptazine-based melon chains, in which K+ ions are inserted in the interstitial region.18,19 The adjacent polymer chains are extended into 2D supramolecular arrays by hydrogen bonding. According to the different maintain time at 600℃, we denote the samples as KCN x (x = 0.5-4), where x is the maintain time. The yield of the prepared samples with the increase of KCl treatment time were shown in Figure S1. The quality of obtained KCN increased first and then decreased. This is because the initial K+ ion embedding leads to the increase of sample mass. Then the thermal decomposition of the KCN causes the mass to begin to decrease. The thermogravimetry (TG, Figure S2) further confirmed our conjecture. All the samples showed similar thermal stability and began to decompose at about 600℃.
2.2 Structure characterization and absorption properties
Elemental analysis (Table S1) shows that the atomic molar ratio of C/N in CN is 0.653, which is close to that of melon (0.67). With the extension of treatment time by KCl, the molar ratio increased from 0.668 (KCN 0.5) to 0.688 (KCN 2) and 0.714 (KCN 4). At the same time, the content of K element in KCN samples boosted from 3.13% (KCN 0.5) to 10.80% (KCN 2) and 15.19% (KCN 4), as estimated by inductively coupled plasma (ICP, Table S2) analysis. These results confirmed a structure composed of heptazine-based melon framework with intercalated K+ ions. The typical X-ray diffraction (XRD) patterns of CN and KCN x (x = 0.5-4) is shown in Figures 2a-b. The CN appears two typical diffraction peaks at 12.9° and 27.6° identified as the (1 0 0) and (0 0 2) facets, respectively. In sharp contrast to CN, the main peak of the (0 0 2) facet experienced a gradual shift from 27.6° to 28.2° with the increase of KCl treatment time, indicating the interlayer spacing narrowed from 0.323 to 0.316 nm owing to the intercalation of K+ ions in heptazine-based melon framework. It should be emphasized that this interlayer spacing narrowing is beneficial to the transfer of photogenerated carriers between adjacent layers. The other two diffraction peaks at 8.0° and 9.9° belong to the (1 1 0) and (0 2 0) facets, respectively.18,19 Besides, those three peaks gradually narrowed and sharpened, indicating the crystallinity has been significantly improved. The full width at half maxima (FWHM) of XRD can reflect the change of crystallinity to a certain extent. As shown in Table S3, the FWHM values narrowed, indicating that the crystallinity of KCN gradually elevated with the thermal treatment. Thus, the structure inducer of solid KCl could recrystallize the amorphous structure during thermal treatment with CN. The color of KCN samples gradually deepens with the increase of annealing time, which vividly indicates the narrowing of bandgap (Figure 2c). The ultraviolet-visible diffuse reflectance spectra (UV-Vis DRS, Figure 2d) demonstrates an obvious red-shift in the absorption edge as the intercalation of K+ ions. The KCN 4 appears the widest absorption region up to 480 nm among all samples, while original CN shows an optical absorption edge at 460 nm. In addition, the bandgap shows about 0.11 eV narrowing from 2.77 eV (CN) to 2.66 eV (KCN 4) observed in Tauc plots (Figure 2e). This narrowing of the bandgap may be due to the expansion of conjugated structure caused by the enhancement of crystallinity.
The characteristic structure was analysed via Fourier Transform Infrared Spectroscopy (FTIR, Figure 3a) to explore the structure-induced effect of KCl during thermal treatment with CN. The fingerprint signals at 1200-1700 cm−1 attribute to the stretching and bending vibration of conjugated CN heterocycles, while the peak at 810 cm−1 is the out-of-plane bending of the heptazine rings.20,21 The typical C-N vibration signal emerges at 993 and 917 cm−1, and the new peak at 1156 cm−1 can be attributed to the N-H bending vibration. In addition, KCN presents a new peak at 2180 cm−1 ascribed to the stretching modes of cyano group (C≡N), which probably results from the decomposition or incomplete polymerization of heptazine units.18,22 X-ray photoelectron spectroscopy (XPS) was used to characterize the chemical bindings on the sample surfaces. As shown in XPS survey spectra (Figure S3), it confirmed the presence of C, N, O, K and Cl elements in KCN, while only C, N and O elements existed in CN. The residual Cl content in KCN is estimated to be 0.2 wt% (Figure S4). The C 1s spectrum presents three contributions, located at 288.1, 286.4 and 284.8 eV, respectively (Figure 3b). The two peaks around 288.1 and 284.8 eV in KCN are similar to CN, which can be attributed to C atoms in aromatic N-C=N units and graphitic carbon, respectively.3,23 The weak peak in the middle at 286.4 eV originates from C≡N species, consistent with the FTIR analysis.24 As shown in Table S4, the ratio of C≡N in C 1s XPS spectra (Figure 3b) ranges from trace (CN) to 0.07 (KCN 2) and 0.09 (KCN 4). In addition, compared with CN, the proportion of graphite carbon increases remarkably after the intercalation of K+ ions in heptazine-based melon framework, which may be due to the improvement of crystallinity and conjugation. In N 1s XPS spectra (Figure 3c), both CN and KCN present three similar binding energy peaks. The peaks at 398.4 and 399.9 eV could be classified as the N atoms of C-N=C and N-(C)3 in heptazine units.3 Other peaks at 400.8 eV belongs to the N atoms in bridging -NHx or C≡N species. Furthermore, the possible chemical structures were characterized by 13C cross-polarization (CP)-MAS solid-state NMR spectrum (Figure 3d). The two distinct resonance bands at 156.8-157.4 ppm and 163.8-165.3 ppm can be attributed to the C atoms in the C-N3 and CN2-(NHx) groups,25-27 further confirming the existence of heptazine frameworks in CN and KCN. The peak at 168.5 ppm could be assigned to the C atom near the electron-withdrawing group (-C≡N) in KCN.
2.3 Morphology characterization
The field emission scanning electron microscope (FE-SEM) and transmission electron microscopy (TEM) images shows that the as-prepared KCN 2 sample consists of nanoparticles roughly 100-500 nm in diameters (Figures 4a-c), while CN presents the typical amorphous structure of polymer (Figure S5). No lattice fringes can be observed in high-resolution TEM (HR-TEM) image (Inset in Figure S5d) of CN. Contrarily, KCN 2 sample has good crystallinity with obvious lattice fringes and a (1 1 0) interlayer spacing of 1.102 nm (Inset in Figure 4c), further explained that the crystallinity has been greatly improved after the intercalation of K+ ions in heptazine-based melon framework. The nanoparticles were also analysed using Scanning TEM (STEM) coupled with energy dispersive X-ray spectroscopy (EDX) to investigate the elements distribution. The high-angle annular dark-field STEM (HAADF-STEM) image in Figure 4d appears a clear outline of the nanoparticles. The corresponding distribution of C, N, O and K elements is relatively uniform in the whole area (Figures 4e-h). An EDX line from a single nanoparticle from the images in Figures 4i-j is highlighted by a green line, with the elements distribution of the cross section displayed in Figures 4k-l. In a given area of both KCN 2 and KCN 4 samples, the elements content of C and N is relatively homogeneous. Interestingly, the distribution of K element in the two samples is quite different. The distribution of K element in KCN 4 is basically the same in the corresponding region, while its count in KCN 2 gradually decreases from the surface to the centre. This directly indicates that in the process of KCl thermal treatment, a gradient of K+ ions can be formed due to its diffusion from the surface to the bulk. Finally, its concentration on the surface and in the bulk tends to be uniform with the increase of diffusion time. The Brunauer-Emmett-Teller (BET) surface area and pore size distribution are measured by using N2 adsorption-desorption isotherm (Figure S6). With the intercalation of K+ ions in heptazine-based melon framework, the BET surface areas show a slight decrease, which are measured to be 14.5, 13.3, 9.4 and 9.0 m2/g for CN, KCN 0.5, KCN 2 and KCN 4, respectively. The pore size distribution curves present the typical macroporous structure originated from the accumulation of nanoparticles.
2.4 Characterization of K+ ion concentration gradient and BIEF
XPS with different depth etched by argon ion was used to further verify the concentration gradient of K+ ions (Figures 5a-b). Two binding energy around 292.9 and 295.7 eV are attributed to K+ ion,28 which is different from that of metal K (294.7 eV). In addition, the doublet separation of K 2p photoelectron lines is 2.8 eV, which further confirms the presence of K+ ion. With the depth of Ar ion etching from 0 to 100 nm, the strength of K+ ions decreases gradually, and the atomic ratio of K reduces sharply from 13.24% to 4.32%, indicating the existence of K+ ion gradient in KCN 2. Contrarily, in KCN 4 sample, the atomic ratio of K does not change significantly from 14.18% to 11.93% with the increase of etching depth (0-100 nm). This is due to the increase of diffusion time of K+ ions, and its concentration in surface and bulk tends to be relatively uniform. In addition, the etched valence band XPS (VB-XPS) spectra at different depths are used to investigate the changes of VB positions at different depths. As shown in Figures 5c-d, the VB position in KCN 2 and KCN 4 are located at 2.70 eV below Fermi level. With the increase of etching depth, the VB position moves upward gradually. From the surface to 100 nm, the VB position of KCN 2 presents a shift of 0.28 eV, while that of KCN 4 is only 0.12 eV. This difference of VB position between surface and bulk phase, caused by the concentration gradient of K+ ions, is beneficial to generate a BIEF, which promotes the separation of photogenerated carriers and the transport from bulk to the surface.
The intensity of BIEF for KCN samples was estimated using the model established by Kanata,10,11 which indicates that the BIEF strength is determined by the surface potential and surface charge density. Firstly, the surface potential was measured using the Atomic Force Microscopy (AFM) at the surface potential mode (Figures 6a-d). Their two-dimensional field is significantly different among all KCN samples, but relatively uniform for each sample. The surface potential was determined to be 624.7, 1025.2 and 769.8 mV, respectively. Secondly, we measured the Zeta potential of KCN samples in triethanolamine (TEOA) solution. The trend of these results is consistent with the changes of surface potential, which is determined to be -22.7, -32.5 and -29.8 mV (Figure 6e). Subsequently, the surface charge density was calculated using the measured Zeta potential and Gouy-Chapman model.29 Because the surface charge density is determined by the Zeta potential,30 the BIEF of KCN samples can be estimated by surface potential and Zeta potential. Finally, the BIEF strength of KCN 2 and KCN 4 samples is 2.4 and 1.6 times higher than that of KCN 0.5 (Figure 6f). These characterizations demonstrate the existence of BIEF in KCN samples induced by the concentration gradient of K+ ions.
2.5 Testing and interpretation of photocatalytic activity
The enhanced crystallinity and successfully constructed BIEF help to explore their photocatalytic activity under visible light. The 50 mg of samples added with H2PtCl6 (3 wt% Pt) is placed into a 50 mL of 20 vol% TEOA solution. The visible-light irradiation was achieved from a 300 W xenon lamp with a UVCUT-420 nm filter. All tests are controlled at 15℃ by circulating condensate. Figure 7a presents the photocatalytic performance depending on the treatment time of KCl. The H2 production rate of CN was 11.8 μmol/h. Their activity increases sharply with the heat treatment of KCl. Among all samples, KCN 2 presents the highest activity of 261.9 μmol/h, which is ～22 times higher than that of CN. Subsequently, the activity decreases gradually over heat treatment time. In order to illustrate the universality of our proposed strategy to induce the formation of BIEF through thermodynamically feasible diffusion, we synthesized the CN and KCN obtained from different precursors (urea, cyanamide and dicyandiamide) in the same way. As shown in Figures S7-8, similar to the KCN obtained from melamine precursor, the color of other KCN samples is obviously deepened to yellow, and its structure also presents the structure of typical CCN. In addition, those KCN samples showed 4 to 7 times higher activity than the corresponding original CN (Inset of a). These examples fully demonstrate the universality of our strategy.
A large specific surface area is conducive to better contact between the photocatalyst and the reaction solution and the generation of more active sites. In order to exclude the effect of specific surface area change on the activity and better explain the relationship between KCl treatment time and activity, the H2 production rate normalized by specific surface area was calculated. As shown in Figure 7b, their normalized activities also exhibit a trend of first increasing and then decreasing with the treatment time of KCl. KCN 2 shows the highest rate of 557.2 μmol/h/m2, which is ～34 times higher than CN (16.3 μmol/h/m2). In addition, the change of photocatalytic performance is consistent with the BIEF strength shown in Figure 6f. These results fully demonstrate that the BIEF induced by the concentration gradient of K+ ions promote the photocatalytic activity. Photocatalytic cycling experiments are used to illustrate the stability of the materials. As shown in Figure 7c, after five photocatalytic cycles, the activity only declines by 8.4%, which fully demonstrates the photostability of KCN 2. Figure 7d displays the apparent quantum efficiency (AQE) test. The AQE of KCN 2 can reaches 13.5% at 420 nm, which is ～22 times higher than CN (～0.6%). At 470 nm, its AQE is still as high as 2.4%, while trace H2 can be detected for CN. This red-shift of the photoresponse wavelength is matched with the change of UV-Vis DRS spectra observed in Figure 2d. We compared the reported CCN to objectively evaluate the activity of our materials. In order to exclude the great difference in activity caused by different test conditions, such as test temperature, light power, catalyst quality, concentration and volume of sacrificial agent, Pt loading, etc., the improvement multiple of H2 production activity of CCN compared with that of corresponding CN was taken as the standard. As shown in Figure 7e, our H2 production activity is 34 times higher than original CN, and is at a relatively good level in reported CCN.17,24,27,31-39
Surface photovoltage spectroscopy (SPS) and steady-state and transient-state photoluminescence (PL) characterizations are used to investigate the BIEF induced transport and separation of photogenerated carriers. As shown in Figure 8a, a positive surface photovoltage (SPV) response band in the range of 350-500 nm is observed, which can be attributed to the band to band transition. After the intercalation of K+ ions in heptazine-based melon framework, the SPV signal of KCN samples is sharply enhanced due to the improved crystallinity. Above all, the stronger BIEF induced by the concentration gradient of K+ ions leads to the higher SPV response of KCN 2 than that of KCN 4. Figure S9 shows the TEM images of the photo-deposited Pt nanoparticles on KCN 2 and KCN 4 surface. In contrast, the Pt nanoparticles are relatively dispersed on the surface of KCN 2, but agglomerated on KCN 4, further confirming the BIEF induced charge separation. Compared to a band-to-band emission peak at ～454 nm, the PL peak gradually redshifts to 472 nm in KCN samples (Figure 8b), which is consistent with the narrowed bandgap. In addition, the PL and BIEF intensity presents a similar trend with first decrease and then increase, as the extension of KCl processing time. This may be due to the BIEF greatly promotes the dissociation of excitons, resulting in the decrease of PL intensity.32,40 The strongest PL emission peak of KCN 4 sample can be attributed to the weak BIEF. Similarly, the fluorescence lifetime also presents a similar change with fluorescence intensity. As shown in Figure 8c, KCN 2 exhibits faster fluorescence decay with an average lifetime of 2.33 ns than CN (2.79 ns) and KCN 4 (2.75 ns). The concentration gradient of K+ ions induced BIEF can be described as shown in Figure 8d. By controlling the diffusion of K+ ions from the surface to bulk in CN, its concentration gradient in KCN samples was successfully constructed, resulting in the continuous change of energy band position from surface to bulk. Most of all, this continuous change of energy band position can generate a BIEF, which promotes the separation of photogenerated carriers and the transport from bulk to the surface.