Modification of TiO2 Nanoparticles with Organodiboron Molecules Inducing Stable Surface Ti3+ Complex.

As one of the most promising semiconductor oxide materials, titanium dioxide (TiO2) absorbs UV light but not visible light. To address this limitation, the introduction of Ti3+ defects represents a common strategy to render TiO2 visible-light responsive. Unfortunately, current hurdles in Ti3+ generation technologies impeded the widespread application of Ti3+ modified materials. Herein, we demonstrate a simple and mechanistically distinct approach to generating abundant surface-Ti3+ sites without leaving behind oxygen vacancy and sacrificing one-off electron donors. In particular, upon adsorption of organodiboron reagents onto TiO2 nanoparticles, spontaneous electron injection from the diboron-bound O2- site to adjacent Ti4+ site leads to an extremely stable blue surface Ti3+‒O-· complex. Notably, this defect generation protocol is also applicable to other semiconductor oxides including ZnO, SnO2, Nb2O5, and In2O3. Furthermore, the as-prepared photoelectronic device using this strategy affords 103-fold higher visible light response and the fabricated perovskite solar cell shows an enhanced performance.

Introduction

Owing to its abundance, nontoxicity, and stability, semiconductor oxide (TiO2, ZnO, SnO2, etc.) nanoparticles and films have been widely used as wide-band-gap semiconductor photocatalysts for a variety of solar-driven clean energy and environmental technologies, such as photovoltaics and photocatalytic fuel generation (Chen and Mao, 2007, Grätzel, 2001). However, pristine wide band semiconductor oxides is not an appropriate candidate for practical applications since it only adsorbs UV light. The optical response of TiO2 nanocrystal has been tuned to visible region using band gap engineering techniques, including metallic (Dahl et al., 2014, Hoffmann et al., 1995) or nonmetallic (Asahi et al., 2001, Asahi et al., 2014, Chen and Burda, 2008, Khan et al., 2002) impurity doping, solid solution formation (Maeda et al., 2006, Wang et al., 2008), and self-structural modification (Liu and Chen, 2014). Among various self-modification techniques, in situ formation of self-doped Ti3+ in the bulk phase through the introduction of oxygen vacancy at high temperature is an effective strategy for band gap engineering (Figure 1A) (Zuo et al., 2010). Additionally, hydrogenation of TiO2 nanocrystals can also result in visible-light-responsible materials (Chen et al., 2011). The latest studies show that these low-energy absorption materials arouse utilities in visible light water splitting and microwave and terahertz absorption (Green et al., 2018a, Green et al., 2018b, Green et al., 2019a, Green et al., 2019b, Guan and Chen, 2018, Tian et al., 2017). Unfortunately, harsh reaction conditions and long-time treatment have limited the practicality of these methods, especially in the context of on-site fabrication and reprocessing of light harvesting devices. Alternatively, UV irradiation of TiO2 nanoparticles (Schrauben et al., 2012) or visible light irradiation of dye-sensitized systems can also produce visible-light-responsive blue-surface TiO2 (Figure 1B) (Yan et al., 2017). In both cases, organic sacrificial agents are required, and these agents are irreversibly oxidized by photoinduced holes (e.g., alcohol to aldehyde and I− to I2). Moreover, such nascent optically active Ti3+ center is highly unstable toward O2 owing to instantaneous oxidation to Ti4+. To our knowledge, the formation of air-stable surface Ti3+ centers using these techniques has not been previously described. We surmise that this is largely due to the difficulty in stabilizing the key Ti3+‒O2- moiety formed in these processes, as such moiety is highly reactive toward O2 to afford Ti4+‒O‒O−·. In light of these constraints, the ability to access air-stable surface Ti3+ defects within TiO2 would represent a paradigm shift for the field of TiO2-based oxide semiconducting materials. In this report, we show that this goal can be accomplished through the simultaneous modification of surface O2− and Ti4+ sites in TiO2 to afford a persistent, optically active Ti3+‒O−· species.

Figure 1

Conventional Methods and Present Work for the Preparation of Self-Doped Ti3+-TiO2 Materials

(A) Ti3+ generation as a result of oxygen vacancy formation. High-temperature combustion with reductant leads to oxygen vacancy in TiO2 bulk phase (Zuo et al., 2010).

(B) Ti3+ generation by photo irradiation. By removal of photoinduced holes using organic sacrificial agents, electrons are cumulated at the conduction band of TiO2, resulting in the reduction of Ti4+ to Ti3+ (Schrauben et al., 2012, Yan et al., 2017).

(C) Present work. Diboron compound (B2Pin2) adsorption on the surface of TiO2 nanoparticles induces spontaneous electron injection for the formation of optically active Ti3+−O−· species. The light blue, blue, red, orange, earth yellow, green, and pink balls stand for the Ti4+, Ti3+, O2−, O−·, C, B, and H atom, respectively.

(D) Four commercially available diboron compounds used in this study: B2Pin2 (bis(pinacolato)diboron), B2(OH)4 (tetrahydroxydiborane), B2Cat2 (bis(catecholato)diboron), and B2(NMe2)4 (tetrakis(dimethylamino)diboron).

(E) Photograph comparing TiO2 (P25), B2Pin2 (B1), B1-TiO2-N2 sample under an inert atmosphere (mixture of solid TiO2 and B2Pin2) and B1-TiO2 under air after heating at 80°C under vacuum for 3 h.

(F) EPR spectra of B1-TiO2 at various temperatures.

(G) HRTEM of B3-TiO2. A short-dashed curve is applied to outline a portion of the interface between the crystalline core and the outer layer (marked by white arrows).

(H) The corresponding HADDF-STEM image of B3-TiO2 with distribution of the C element mapping.

Organodiboron compounds constitute a class of stable and highly versatile reagents commonly used in organic synthesis (Neeve et al., 2016). Despite their unique reactivity, to date, the use of these diboron species in the modification of inorganic materials remains surprisingly scarce. Herein, we report a new method for the facile generation of Ti3+ defects on TiO2 surface under mild conditions (<80°C) enabled by the use of diboron reagents (Figure 1C). In this process, facilitated by the intimate interaction of the diboron reagent's B center with the surface bridging O2c of TiO2, the adsorption of organodiboron reagents onto TiO2 nanoparticles leads to spontaneous electron transfer, resulting in a stable, blue surface Ti3+-TiO2. Notably, this process does not require complicated synthetic manipulations such as anaerobic photo irradiation or high temperature calcination, which are commonly employed by previously developed techniques.

Previously developed strategies for TiO2 surface modification have mainly focused on the modification of surface Ti sites with organic nucleophiles (e.g., phenol, alcohol, and carboxylic acids). In contrast, surface bridging oxygen (O2c) sites have seldom been functionalized with electrophilic organic reagents. We posited that electrophilic organic adsorbates with an appropriate reduction potential might favorably interact with the surface O2c sites. Importantly, this binding event might overcome the barrier of electron transfer from O2− sites to adjacent Ti4+ sites and further stabilize the primary charge separation state. We were particularly interested in the use of organic diboron reagents because of their unique Lewis acidity and reducing ability (vide supra). Previous work in the area of synthetic organic chemistry showed that, upon binding to a Lewis basic oxygen atom, these organic diboron species can function as single electron reducing agents, thus allowing for various important transformations (Liu et al., 2019, Mo et al., 2010, Mo et al., 2018, Pietsch et al., 2015, Wang et al., 2016, Zhang and Jiao, 2017). Based on these reasons, we envisioned that, upon the coordination of such diboron compounds with the surface oxygen atom in metal oxide materials, the formation of surface diboron-oxygen Lewis pair may induce single electron transfer from the ipso-O2c site to the adjacent Ti site. Furthermore, the interaction of diboron species with the oxygen atom may stabilize the resulting low-valent Ti‒high-valent O pair.

Results and Discussion

Spectroscopic Characterizations

We use P25 TiO2 to study the surface modification behavior, for P25 TiO2 is typical TiO2 material applied in photochemical applications and shows enhanced performance based on fine nanoparticles and heterointerface between anatase and rutile phases (Xia et al., 2013, Xia et al., 2014). Mechanical mixing of commercial Degussa TiO2 P25 and bis(pinacolato)diboron (B2Pin2, B1) in a nitrogen-filled glovebox at room temperature resulted in a rapid color change from white to blue within 1 min (Figure 1E and Video S1). This blue titania, labeled as B1-TiO2-N2, was highly sensitive to O2 and faded immediately upon exposure to air. Unexpectedly, heating B1-TiO2-N2 at 80°C under vacuum for 3 h resulted in a blue sample that is stable under air, which we labeled as B1-TiO2. This sample could be stored outside the glovebox for weeks, and the blue color persisted. Similar transformations were observed with three other diboron compounds as shown in Figures 1 and S1. To fully disperse the diboron compound onto the TiO2 nanoparticles, this heterogeneous reaction was performed in a diboron-soluble solvent, such as methanol or diethyl ether. Same blue samples were obtained after removal of organic solvents under vacuum. On the basis of previous studies (Iorio et al., 2012), the blue color was indicative of the formation of Ti3+ defects. To probe whether B1 was oxidized in this process, the newly prepared blue B1-TiO2 sample was extracted with CDCl3, and B1 was found to be the only boron-containing species present in the extract (Figure S2). Additionally, the use of this recovered B1 from the extract for TiO2 modification also gave rise to the same blue titania. Thus, these experiments suggested that diboron compound B1 is not a one-off sacrifice agent. Based on these findings, a new mechanism must be responsible for the formation of this blue TiO2.

Video S1. Mechanical Mixing of Two Fine White Powders of Commercial Degussa TiO2 P25 and bis(pinacolato)diboron (B2Pin2, B1) in a Glove Box and in Air Were Recorded by a Smart Phone Camera, Related to Figure 1E

Samples change color from white to blue within one minute. Please watch the video for detail.

To probe the existence of Ti3+ species, electron paramagnetic resonance (EPR) spectra (Zuo et al., 2010) of the B1-TiO2 sample were acquired. Low-temperature (2 and 100 K) EPR data feature transitions with g = 1.97–1.99, whereas the room temperature EPR is silent (Figure 1F). This behavior is characteristic of Ti3+ species as documented by previous studies (Li et al., 2008). The observed g values show the presence of a strong anisotropic paramagnetic Ti3+ center, indicating that the local symmetry of Ti3+ is vastly broken. Furthermore, this Ti3+ EPR signal would be split if B replaces the bridging O or is directly bound to Ti (Gopal et al., 2008). However, in all the cw-EPR measurements, we did not observe the hyperfine coupling originating from the 10B or 11B nucleus. This suggests either no or very weak coupling between the B atom and the Ti3+ center. Moreover, we found that B3-TiO2 shows a g signal at 2.003 at room temperature (Figure S3), which could be characteristic of a Ti3+-O−· radical. We also applied the ACTEM method to study the existence and status of diboron molecule on the surface of TiO2. In the high-resolution transmission electron microscopy (HRTEM) image of B3-TiO2 (Figure 1G), we found that the surface of TiO2 nanocrystal is wrapped by an outer layer with ~1-nm thickness. In addition, the EDS mapping of C element under the HADDF model (Figure 1H) shows a carbon-rich surface shell representing adsorbed organic diboron molecule on the TiO2 surface.

We next investigated the structures and properties of these diboron-TiO2 nanoparticles with powder X-ray diffraction (PXRD), solid-state UV-vis spectroscopy (UV-vis), in situ attenuated total reflectance-Fourier transform infrared spectroscopy (in situ ATR-FTIR) and solid-state 11B magic angle spinning nuclear magnetic resonance (11B MAS NMR). PXRD analysis (Figure S4) shows that the commercial P25 sample is a mixture of anatase-phase and rutile-phase TiO2. The PXRD pattern of TiO2 remained the same upon adsorption of the diboron reagent, indicating that no phase change occurred in this process. Thus, this result suggests that the blue species is likely related to surface engineering rather than bulk phase modification. The solid-state UV-vis spectra in Figure 2A showed that neither the pristine white TiO2 nor the diboron compound absorbs visible light (>400 nm). First, absorption bands beginning at 400 nm and steadily growing into the near-infrared (NIR) region were observed for all four B-TiO2 samples. These adsorption bands were different from that arising from metal to ligand charge transfer of the surface complex between Ti4+ sites and organic ligands commonly observed in previous work (Lang et al., 2012). The UV-vis absorption spectra associated with Ti3+ materials can differ between different preparation methods (Howe and Gratzel, 1985). In our case, the diboron-modified TiO2 samples were found to absorb visible light with higher intensity in the NIR region. This suggested that the Ti3+ species in our case is different from the previously reported Ti3+-O2- structures. In the in situ ATR-FTIR spectra (Figure 2B), the emergence of new signals corresponding to the diboron compound was observed over the time, clearly demonstrating the adsorption process. In the FTIR spectra (Figure 2C), signals at 1280 and 1170 cm−1 are assigned to two different types of B−O vibration. Based on density functional theory (DFT) calculations, the B−B stretching vibration appears at ca. 1000 cm−1 (Figure S5). As shown in Figure 2C, this signal is absent in B1 owing to the C2 symmetry of this diboron molecule. In contrast, a new peak at 1010 cm−1 corresponding to B−B stretching emerged in the B1-TiO2 sample, thereby demonstrating the formation of the oxygen-diboron Lewis pair.

Figure 2

Spectroscopic Characterizations

(A) UV-vis spectra of the white TiO2 P25 and diboron-adsorbed TiO2 P25. The inset shows the UV-vis spectra of four free diboron compounds.

(B) In situ ATR-FTIR spectrum of B1 onto TiO2 over 15 min.

(C) FTIR spectra of free B1 and B1-TiO2.

(D) Solid-state 11B MAS NMR of diboron compounds B and B-TiO2.

(E) XRD pattern of the diboron compounds modified TiO2.

(F) Ti 2p, O 1s, and valence band structure X-ray photoelectron spectroscopy spectra of TiO2 and B1-TiO2.

(G) Normalized Ti-L edge XANES spectra of TiO2 and B1-TiO2.

(H) Normalized O-K edge XANES spectra of TiO2 and B1-TiO2.

Solid-state 11B MAS NMR spectroscopy was next used to elucidate the binding details of the diboron reagent with the TiO2 surface. Upon adsorption, the B(sp2)-B(sp3) adducts should exhibit two distinct boron environments in the solid-state NMR spectra. As shown in Figure 2D, in all four cases, new peaks appear and shift upfield, indicating changes in the chemical environment of the two B atoms. Previous studies revealed that the 11B signal of sp3 hybridized tetracoordinate boron undergoes an upfield shift relative to the sp2 hybridized tricoordinate one (Nöth and Wrackmeyer, 1978). Thus, we attribute the peaks close to 0 ppm to B atom bound to the surface O atom of TiO2 nanoparticles. X-ray diffraction (XRD) was applied to examine the structure of the TiO2 by diboron compound modification (Figure 2E). According to the XRD results, no change of the lattice was characterized. To further our understanding of the surface interaction between the diboron molecule and TiO2 nanoparticles, we carried out XPS and SXANES studies. The Ti 2p spectra shows that the peaks were shifted for 0.2 eV toward the lower-binding-energy region; this also happened in O 1s spectra (Figure 2F). The valence band spectra show that the band level shifts to lower energy, a shift from 2.5 to 2.16 eV. The Ti 2p, O 1s spectra change means that the surface modification shows significant influence on the surface electron structure of the TiO2. As shown in Figure 2G, the t2g (L3) peak of Ti-L edge shows that the B1-TiO2 is slightly lower than that of pristine TiO2 and the peak of eg (L2) is shifted to higher energy, which can be attributed to the existence of Ti3+ (Kronawitter et al., 2011, Stewart et al., 2006). The O-K edge originates from the hybridization of O2p ligand-hole states with the coordinating atoms s, p, d states (Karvonen et al., 2010). In our case, the O-K edge shows that the peaks of B1-TiO2 that index to the Ti‒O bond around 545 eV and the O 2p around 539 eV are much higher than that of pristine TiO2 (Figure 2H), demonstrating more electron location in O 2p orbital and Ti‒O bond instead of the O eg orbitals (de Groot et al., 1989). Meanwhile, the Ti-K edge line shows no significant change between B1-TiO2 and pristine TiO2 in the pre-edge region and the EXANE region (Figure S6). The reason is that the diboron compound is bonded with surface ipso-O2- atoms. Moreover, the B 1s XPS also provides consistent results indexing of B species with different binding energies (Figure S7). On the basis of these spectroscopic evidences, we propose that the coordination of diboron compound with surface O sites on TiO2 facilitates the electron transfer from the ipso-O2- to the adjacent Ti4+ site and this coordination further stabilizes the newly formed Ti3+-O−· species. We note that the activation of inert lattice O2− and the stabilization of surface Ti3+ enabled by the formation of ≡B‒O−·‒Ti3+ moiety is a novel process that has not been reported in TiO2 modification.

DFT Calculation Results

We next investigated the surface Ti3+‒O−· structure in the blue TiO2 nanoparticles by simulating the adsorption of B1 on the TiO2 surface and the accompanying charge transfer between them using DFT calculations. According to the literature, the {101} facet in anatase TiO2 is the dominant facet, and we set up the adsorption model of anatase TiO2 {101} facet (Lazzeri et al., 2001). Our calculations showed that the Fermi levels in no adsorption is located in the TiO2 band gap (Figures 3A and 3C). Importantly, the diboron-adsorbed TiO2 model shows a clear charge transfer from B1 to the TiO2 {101} surface (Figure 3B). Furthermore, the Bader charge calculation showed substantial electron transfer from the B1 atom to the bridging O atom. Furthermore, those electrons (light blue region) are localized between the bridging O atom and the neighbor surface Ti atom according to the calculated charge difference density mapping. The calculated projected density of state (PDOS) plots showed that the Fermi level rises to the bottom of the TiO2 conduction band, which is mainly derived from Ti 3d states (Figure 3D). Some Ti 3d and O 2p states below the Fermi level appear, which are filled with electrons. The filled Ti 3d states indicate the partial reduction of the surface Ti sites, which explains the existence of the Ti3+ species observed in our EPR experiments. Moreover, the adsorption model of B2, B3, and B4 on TiO2 {101} surface was also calculated (Figures S8A–S8C and S9). These data showed that a new electron-filled band-gap state consisting of Ti 3d appears in all the three models (Figures S8D–S8F and S9). This strongly suggested that the adsorption of organic diboron compounds can lead to the formation of Ti3+ species on the TiO2 surface. The adsorption model based on rutile TiO2 shows similar results (Figures S10 and S11). The adsorption model of the interface between anatase TiO2 and rutile TiO2 was also set up to evaluate the heterojunction in P25 nanoparticles (Figures S12 and S13). To interpret the stability of the surface Ti3+ in B-TiO2 in the air atmosphere, the oxygen adsorption models were set. The calculation result shows that O2 molecule is hesitant to adsorb to the adjacent Ti5c site of the molecule (energy rising process), which represents the protection role of the molecule (Figure S14 and Table S1).

Figure 3

DFT Calculation Results

(A) Optimized geometric structures and charge difference density mappings for no adsorption.

(B) Optimized geometric structures and charge difference density mappings for adsorption of B1 on the TiO2 {101} surface. The numbers in the figure stand for the Bader charges on the various atoms. The isosurfaces of the electron density difference plots are all 0.001 e Å−3. The yellow and light blue surfaces represent electron depletion and accumulation. The blue, red, earth yellow, green, and pink spheres stand for the Ti, O, C, B, and H atoms, respectively.

(C) Projected density of state (PDOS) plots for no adsorption.

(D) Projected density of state (PDOS) plots for adsorption of B1 on the TiO2 {101} surface (the insets on the right are the magnification of the inter band at near the Fermi energy level). The dashed line shows the Fermi level.

Photodetectors and Perovskite Solar Cells Performance

To further explore the optoelectronic properties of this diboron-adsorbed TiO2 material under visible-light excitation, thin films were then fabricated into photodetectors (denoted as FTO/TiO2/diboron/Spiro-OMeTAD/Au, Figure 4A). The time-dependent current curves (I-t) for these diboron-type photodetectors under visible-light illumination (>400 nm) exhibited excellent on-off switching repetitions through five cycles (interval = 10 s) (Figure 4B). The photocurrent for these four diboron-type detectors increased linearly with the light intensity (Figures 4C and S15 and Table S2). In contrast, the observed photocurrent for TiO2 without diboron modification did not change regardless of the on-off state, which was ascribed to the intrinsic band gap of anatase TiO2 (3.2 eV). Furthermore, the charge transport process in diboron-modified devices was investigated using electrochemical impedance spectroscopy measurements in the dark at 0 V (Figure 4D). The larger semicircle radius of Nyquist plot in the low-frequency region for the B3 device compared with the B1, B2, and B4 devices indicated less electron recombination (Kim et al., 2012). This observation agreed well with the highest photocurrent for the B3 device (Figure 4B). The IPCE (incident photon-to-electron conversion efficiency) was also evaluated, and the B3-TiO2 device shows the best performance (Figure S16). The B-TiO2 photodetectors show good response toward long-time on-off tests compared with the TiO2 photodetector (Figures S17 and S18), and the response of the photodetectors does not attenuate after being stored in a dry air atmosphere for at least 7 months (Figure S19). We also fabricated solar cells based on the diboron molecule interface modulation, and the J-V performance of the device was characterized showing an improved solar cell performance (Figures 4E–4G). The data show that the B3-treated device based on the B3-SnO2 interface has a better performance compared with the reference device of SnO2-based perovskite solar cell.

Figure 4

Photodetectors and Perovskite Solar Cells Performance

(A) Schematic of the FTO/TiO2/diboron/Spiro-OMeTAD/Au photodetector.

(B) Time-dependent current curves for the detectors under light illumination (λ > 400 nm) with 10 s on-off switching intervals.

(C) Time-dependent current curves for these photodetectors under illumination intensities (λ > 400 nm) of 19.5, 26.5, 35.0, and 44.5 mW·cm−2.

(D) Nyquist plots from the photodetectors in the dark.

(E) Schematic of the ITO/SnO2/diboron/Perovskite/Spiro-OMeTAD/Au solar cell.

(F) Current density-voltage curves (J-V) of the diboron-modified perovskite solar cell.

(G) Photovoltaic parameters of the diboron-modified perovskite solar cell. FF, fill factor; PCE, power conversion efficiency.

We note that previously developed solar cells such as dye-sensitized solar cells and perovskite solar cells inevitably involve the use of expensive dyes and/or toxic and unstable reagents. Thus, our approach to generating visible-light active center via the adsorption of stable and inexpensive diboron compounds holds great advantages over these traditional techniques. Furthermore, a diverse range of diboron compounds can be conveniently synthesized from B4, potentially allowing for the fine-tuning of semiconductor-based materials. More importantly, the present strategy is not limited to TiO2. In our studies, we have determined that a broad range of other common semiconducting oxides, including ZnO, SnO2, Nb2O5, and In2O3, were also successfully modified by diboron reagents, thus clearly demonstrating the generality of this method for semiconducting metal oxide modification (Figure S20).

Limitations of the Study

In the current study, perovskite solar cells assembled from B-TiO2 display relatively poor performance (Figure S21). Many factors such as device fabrication could influence the performance of such device. Additionally, in our present work, mesoporous TiO2 was used in B-TiO2-based devices, whereas planar SnO2 was used in B-SnO2-based devices. We are uncertain if this deviation led to the contrasting device performance based on B-TiO2 and B-SnO2 materials. Nonetheless, the three orders of magnitude improvement observed in the B-TiO2-based photodetector clearly demonstrates the utility of this diboron-based modification strategy. Currently, we are actively pursuing other applications based on diboron-modified semiconductor oxide materials featuring unique surface defects.

Methods

All methods can be found in the accompanying Transparent Methods supplemental file.