Article Cite This: Acc. Chem. Res. 2019, 52, 248−257
pubs.acs.org/accounts
Understanding Charge Transport in Carbon Nitride for Enhanced Photocatalytic Solar Fuel Production Mohammad Z. Rahman and C. Buddie Mullins* McKetta Department of Chemical Engineering and Department of Chemistry, Texas Materials Institute and Center for Electrochemistry, The University of Texas at Austin, Austin, Texas 78712-1589, United States
Downloaded via EAST CAROLINA UNIV on January 16, 2019 at 10:29:09 (UTC). See https://pubs.acs.org/sharingguidelines for options on how to legitimately share published articles.
S Supporting Information *
CONSPECTUS: Photocatalytic solar fuel production, for example, production of hydrogen via water-splitting, is an effective means of chemical storage of solar energy and provides a potential option for achieving a zero-emissions energy system. Conveniently, hydrogen can be converted back to electricity either via fuel cells or through combustion in gas turbines, or it can be mixed in low concentrations with natural gas or biogas for combustion in existing power plants. The cornerstone of a practical solar fuel production process is a stable, efficient, and scalable photocatalyst (a semiconductor material that accommodates photon absorption, charge carrier generation and transport, and catalytic reactions). Therefore, the quest for suitable photocatalyst materials is an ongoing process. Recently, carbon nitride (CN) has attracted widespread interest as a metal-free, earth-abundant, and highly stable photocatalyst. However, the catalytic efficiency of CN is not satisfactory because of its poor charge transport attributes. There is a direct relation between the photocatalytic efficiency and charge transport because the basic principle of light-promoted overall photodecomposition of water into H2 and O2 molecules (or, generally speaking, photoredox reactions) relies on separation and subsequent transfer of excited-state electron−hole pairs to relative redox couples. However, the excited states last for a very short time, typically nanoseconds to microseconds in liquids, and unless they are separated within this time frame, the excited-state electron−hole pairs undergo recombination with release of the captured light energy as heat or photon emission. To utilize light in a form other than heat or emitted photons by avoiding the recombination of excited-state electron−hole pairs, charged excitons must be scavenged before the absorption of subsequent photons to sustain a multielectron photoredox reaction. Otherwise, the extraction of charges becomes more difficult. This imposes a potential efficiency-limiting factor. An enhancement in water to hydrogen conversion efficiency in CN therefore requires the use of precious-metal cocatalysts (e.g., Pt) and sacrificial electron donor/acceptors to facilitate multielectron/multiproton transfers to overcome the high kinetic barriers. The use of Pt and sacrificial agents is not consistent with the notion of low-cost and sustainable hydrogen production from water. CN must overcome this dependence to stand out as a truly scalable photocatalyst. To make progress, the foremost requirement is to attain an in-depth understanding of the fundamental charge transport phenomena needed for the rational design of CN-based photocatalysts. In this Account, therefore, our aim is to provide a synopsis of current understanding and progress regarding charge-transportrelated phenomena (e.g., recombination, trapping, transfer of charge carriers, etc.) and to discuss the effects of charge transport in enhancing the apparent quantum yield of hydrogen production in CN. This understanding is necessary to broaden the scope of CN for other catalytic applications, for example, efficient CO2 reduction to methanol or methane, fixation of nitrogen to ammonia, and use as an active material in solar cells. We also identify research gaps and issues to be addressed for a more clear elucidation of charge-transport-related phenomena in CN. Thus, this Account may inspire new research opportunities for tuning the extrinsic/intrinsic photophysicochemical properties of CN by rational design to attain the most favorable properties for improved catalytic efficiency.
■
even C3N4 phases in the literature.5,6 Binary CN may also be of two other varieties, namely, carbon subnitrides, C3N4−x, which are nitrogen-deficient, and carbon supernitrides, C3N4+x, which contain excess nitrogen.7,8 These three groups of binary C/N compounds (i.e., C3N4, C3N4−x, and C3N4+x) should be distinguished from ternary C/N/H compounds such as melem
INTRODUCTION
Carbon nitride, C3N4 (denoted as CN), has attracted widespread interest as a photocatalyst and electrocatalyst for the production of hydrogen as a solar fuel from water because of its advantageous physicochemical properties, excellent stability, low cost, and ease of processing.1−4 CN is exclusively a binary compound consisting of C and N. However, various C/N/H polymers that are in many cases comparable to melon (C6N9H3) are erroneously called graphitic carbon nitrides or © 2018 American Chemical Society
Received: October 27, 2018 Published: December 31, 2018 248
DOI: 10.1021/acs.accounts.8b00542 Acc. Chem. Res. 2019, 52, 248−257
Accounts of Chemical Research
■
Article
PHOTOREDOX REACTIONS DEPEND ON CHARGE TRANSPORT Photocatalytic solar fuel production requires a series of interdependent and sequential photophysical, physicochemical, and electrochemical processes.17 Photon absorption by a suspended particulate photocatalyst upon irradiation by a light source or natural sunlight initiates time-dependent nonequilibrium photophysical and photochemical processes, as shown in Figure 1. The absorption of photons commences the
(C6N10H6), melon, poly-s-heptazine imide (C12N17H3), etc., that contain hydrogen as an additional element. The catalytic efficiency of CN is poor.9 Because photocatalytic solar fuel production involves multielectron redox reactions, the vital challenge to increase the photocatalytic efficiency is efficient photogeneration of charge in CN.10 Charge photogeneration comprises sequential steps of generation, dissociation, and diffusion of excitons upon absorption of a photon. Following exciton generation, even though the hole is primarily localized on the highest occupied molecular orbital (HOMO) and the electron on the lowest unoccupied molecular orbital (LUMO), the electron and hole pairs in CN exhibit high Coulombic attraction that is significantly greater than kBT.11 This Coulombic attraction makes dissociation of Coulombically bound electron−hole pairs into free charge carriers difficult.12 An added problem is the recombination and trapping of charge carriers.13,14 Even after efficient exciton quenching, recombination pathways (e.g., geminate recombination and bimolecular recombination) may compete with dissociation of these charges into free charge carriers, which adversely influences the photocatalytic performance.5,15 Because “no charge, no redox reaction” is obvious in photocatalysis, a predictive understanding of the relationship between charge-transport-related phenomena and photocatalytic performance is of the utmost importance. Several important published studies have investigated charge recombination and separation in CN via morphological finetuning from bulk to nanoscale and/or introduction of a heterojunction.9,16 There are also excellent reviews that critically discuss the synthesis, history, physical properties, and applications of CN.2,3,9,10 We have progressed sufficiently that it is now time to dedicate more attention to understanding the charge transport in CN at the molecular or atomic level. Control without deciphering an atomistic view of charge transport can be useful for boosting the quantum yield. However, an atomistic understanding is key for developing design concepts for engineering of the materials to achieve the preferential attributes that would lead to a scalable, sustainable, and highly efficient photocatalyst. In this Account, therefore, our focus is to discuss succinctly the research progress and current understanding of recombination, trapping, and transfer of photogenerated charge carriers in CN. This Account is organized as follows. Following the Introduction, we briefly discuss the inter-relationship between charge transport and photoredox reactions and a model for charge transport kinetics in CN. We then extend our discussion to (i) an atomistic understanding of charge carrier recombination, trapping, and transport in CN, (ii) suppression of recombination and trapping in CN, (iii) the role of excitons in the hydrogen evolution reaction (HER), and (iv) the fate of photogenerated holes at the CN−water or hole−scavenger interface for the oxygen evolution reaction (OER). We conclude with recommendations for future research necessary for a better understanding of charge transport to achieve benchmark photocatalytic efficiency in CN-based materials. We have speculated that molecular tuning of intrinsic properties is necessary to accommodate the best of charge transport phenomena. This has to be accomplished through a combined theoretical and experimental approach. Advanced computational quantum chemistry and state-of-the art nanotechnology should be exploited to optimize the structure, compositions, and electrochemical surface states at the molecular level to attain maximum activities.
Figure 1. Three key dynamics processes (photophysical, photochemical, and electrochemical) involved in overall water-splitting. After excitation across the band gap, the following occur: (1) electron/hole relaxation (tens to hundreds of fs); (2) trapping of electrons/holes into trap states due to defects or surface states (within hundreds of fs to tens of ps); (3) radiative and nonradiative bandedge electron−hole or exciton recombination (within tens of ps); (4) radiative and nonradiative trapped electron−hole or relaxed exciton recombination (within hundreds of ps to a few ns); and (5) nonlinear and nonradiative exciton−exciton annihilation (within hundreds of fs to tens of ps). Here, CB stands for conduction band and VB for valence band.
generation of excitons (i.e., electron−hole pairs). To maintain equilibrium, the excitons can undergo recombination.18 Generation and recombination of excitons happen on the femtosecond time scale. Electron−hole pairs that do not recombine can be separated at the semiconductor−catalyst and semiconductor−solution interfaces. These free electrons and holes are the driving force for useful electrocatalytic redox reactions necessary for solar fuel production, e.g., splitting of water into hydrogen and oxygen molecules. Production of hydrogen from decomposition of water follows a biphotonic photochemical reaction. Photoexcited CN can abstract a hydrogen atom from a water molecule via the sequential transfer of an electron and a proton from water to the triazine/heptazine structure of CN, resulting in a triazinyl/hepazinyl−hydroxyl biradical in the electronic ground state. The excess hydrogen atom of the triazinyl/hepazinyl radical then be photodetached by a second photon, and the two radicals can recombine to form H2.19
■
CHARGE PHOTOGENERATION KINETICS IN CN AFTER LIGHT EXCITATION Photoexcitation of CN generates a singlet exciton that is highly confined to its tri-s-triazine unit because of non-aromatic 249
DOI: 10.1021/acs.accounts.8b00542 Acc. Chem. Res. 2019, 52, 248−257
Article
Accounts of Chemical Research
Figure 2. (a) Simple model for charge carrier kinetics in CN. (b) Cascaded charge transfer kinetics in CN. See ref 23 for a description of the kinetic parameters. (c) Energy diagram depicting the exergonic electron transfer cascade pathway. Reproduced from ref 23. Copyright 2017 American Chemical Society.
Figure 3. (a−f) Light-induced charge transport in CN: (a, b) photoexcitation and generation of singlet excitons (SEs); (c) rapid dissociation ( 1, while the diffusion rate increases.24 Incomplete polycondensation of precursor monomers or doping/adsorption of heteroatoms in C−N networks may result in residual chemical defects (e.g., dangling bonds, interstitials, vacancies, etc.) in the crystal structure.26,27 The surface dangling bonds, due to pendant amine groups, create shallow trap states, while the interstitials and vacancies create deep trap states in the bulk of the CN. These defects are potential sites for charge carrier trapping. Understanding the defect-mediated charge carrier trapping and its impact on the photocatalytic performance is therefore important. Charge trapping leads to spatial electron−hole separation that slows recombination but detrimentally reduces the charge carrier mobility. Interfacial transfer of charge carriers between
Figure 4. Schematic representations of (a) the charge trapping model and (b) the influence of trapping on charge transfer reactions. Reproduced from ref 32. Copyright 2017 American Chemical Society.
The photoexcited electron may transit directly to the conduction band (CB) to create an emissive state (EM) or fall into a deep or shallow trap and end up never reaching the CB. Similarly, an excited electron falling back to the ground state (GS) might also be trapped. In both cases, there could be the illusion of temporary electron accumulation.31 Electron trapping is inversely related to recombination, i.e., if recombination of the electron and hole happens, there will be a decrease in the population of the charge-separated state (CS) and an increase in the energetic barrier to reach the EM. Thermal equilibrium then moves toward the CS. This results in faster decay of the EM compared with the CS.32−34 If the photogenerated charge carriers are trapped in deep trap states within CN, irrespective of the lifetime of charge carriers (μs for long-lived carriers or fs to ns for short-lived ones), this leads to a reduction in driving force for charge transfer reactions.32 As a result, electrons cannot transfer to catalytic active sites for proton reduction. Charge trapping therefore decreases the nominal charge transfer rate. This in turn decreases the reaction rate. Trap states are significantly more likely at energies of 1 eV below the band edges of CN,32 implying that CN should exhibit a gradual decrease in hydrogen quantum yield with increasing wavelength. Overall, trapping reduces the efficiency of surface photocatalytic reactions.
■
SUPPRESSION OF RECOMBINATION AND TRAPPING IN CN Because the origin of trap states in CN is highly dependent on the synthesis method,6,10 engineering of CN to control the concentration of trap states seems feasible. For example, incorporation of cyanimide moieties can result in photo251
DOI: 10.1021/acs.accounts.8b00542 Acc. Chem. Res. 2019, 52, 248−257
Article
Accounts of Chemical Research
Figure 5. (A−C) Probing of trapping, recombination, and charge transport through rise−decay dynamics of transient photocurrent measurement: (A) overall transient photocurrent response; (B) rise time dynamics of the transient photocurrent; (C) decay time dynamics of the transient photocurrent. Reproduced with permission from ref 25. Copyright 2018 Royal Society of Chemistry. (D) Schematic representation of the proposed charge trapping model in g-C3N4 (left) and red P/g-C3N4 (right). (E, F) Normalized fs TA decay kinetics for (E) short and (F) long time scales. Reproduced with permission from ref 37. Copyright 2017 Wiley-VCH.
catalytic activity. It was shown that hydroxylation could substantially promote local spatial charge separation and proton activation of CN.38 Hydroxylation allows the polymeric carbon nitride surface to have a controllable grafting of abundant hydroxyls (−OH) on the −CN sites. This consequently extends the 2D conjugate electron system to a 3D space for local spatial charge separation and polarizes the neighboring N atoms (C−NC) to promote proton adsorption.
synthetic modification of CN that leads to extraction of longlived charge carriers over a period of hours following 30 min of light exposure.35 Recently, a topological CN (TCN) has been reported for localized photon absorption and delocalized charge separation with suppressed recombination and trapping.18,25 Trapping, recombination, and charge transport were probed using rise−decay dynamics in a transient photocurrent measurement (Figure 5a−c).36 In TCN, comparatively high photocurrent was observed, which is evidence of quenched radiative and trap-mediated recombination and transport of photogenerated charge carriers. TCN showed slow trapping but fast detrapping of free charge carriers. Moreover, TCN exhibited a continuing extraction of charge carriers for more than 500 μs after immediate stopping of the illumination.25 In another example, introducing ultrasmall red phosphorus (red P) crystals on CN sheets was shown to be effective in sweeping out the trapped unreactive charges.37 In this case, P atoms reduced the number of defects in the CN structure by forming new chemical bonds and therefore significantly prolonged the lifetime and enhanced the charge separation (Figure 5 d−f). The active charges of red P/g-C3N4 exhibited a 10-fold longer lifetime (1.1 ns) than that of the bare g-C3N4 (0.1 ns), thereby resulting in significantly enhanced photo-
■
IMPACT OF CHARGE PHOTOGENERATION ON WATER-SPLITTING REACTIONS
Role of Exciton Conversion in the Hydrogen Evolution Reaction
There is clearly a lack of substantial studies regarding the elucidation of the photocatalytic mechanism from the perspective of the exciton conversion process in CN. A recent study showed that triplet-to-singlet conversion plays an irreplaceable role in the HER mechanism.39 This report claimed that singlet and triplet excited states extend the carrier recombination lifetime from nanoseconds to microseconds, prompting the transfer of more electrons to active sites for the HER. This triplet-to-singlet conversion arose from a special half-metallic electronic structure in which spontaneous spin 252
DOI: 10.1021/acs.accounts.8b00542 Acc. Chem. Res. 2019, 52, 248−257
Article
Accounts of Chemical Research
Figure 6. (a) Potential energy profile for the adsorption of H+ on different sites. (b) Process of hydrogen production. (c) Free energy as a function of the reaction coordinate of the HER for different active sites. The singlet (red), triplet (black), and optimal process (marked by a black arrow) as functions of the H atom distance are displayed for comparison. The accompanying atom configurations are shown in the bottom panel. (d) Streak image and photoluminescence (PL) spectra experimentally confirm the singlet-to-triplet conversion. Reproduced from ref 39.
valence band (VB) position with regard to the OER potential, CN does not exhibit reliable generation of O2 under visiblelight irradiation.10 Under UV irradiation, O2 evolution is observed, but at much a lower rate than for the HER, indicating that another reaction must also be proceeding to balance the charge electrochemically.42 Therefore, understanding the fate of holes at the interface of CN and water by detailing the altered electronic structure and exciton behavior is essential for identifying the shortcomings of reliable O2 production in CN. Density functional theory calculations revealed that CN can influence the energy level shifts of OH− and H2O at the interface as a result of interfacial electronic alignment for localized molecular orbitals (Figure 7).43 This energy level shift is significantly reduced for molecular orbitals away from the interface, i.e., the influence is stronger when OH− and H2O are closer to the substrate. It was noticed that the electronic levels of OH− surpass the valence band maximum (VBM) of CN, while the orbital energies of H2O molecules remain far below the VBM of CN. This implies that a photogenerated hole generated in CN can relax to OH− instead of neutral H2O and accumulate at the interface, facilitating the OER.44 The higher energy level of OH− warrants the stable trapping of free photogenerated holes on OH− for the OER to proceed. However, a zero-hole transition was observed from the water molecules to CN or within the orbitals of water molecules. When OH− comes into close vicinity of CN, photogenerated electrons in CN tend to transfer to the empty molecular orbitals of OH−, making oxidation of OH− to the hydroxyl
polarization makes the electronic structure around the Fermi level split into spin-up and spin-down states and display metallic features.40 As described in Figure 6, the HER process follows the lowerenergy route. A single H atom appears at S1 spontaneously and adsorbs physically onto the C1 site after overcoming the energy barrier at S2 and a release of ∼1.72 eV at S3. Subsequently, another H atom close to S4 adsorbs onto the C2 site, forming two metastable H* atoms at S5. A photoexcited perturbation leads the two adjacent H* atoms to begin orbital hybridization and to bridge the transition state of H−H at S6. The dissociation of molecular hydrogen from the surface is phonon-assisted. Consequently, the degeneracy of its π2p * orbitals is lifted as a result of the hybridization with the C1 site lone pairs, and the triplet and singlet potential energy surfaces cross (between S6 and S7), while triplet-tosinglet conversion likely takes place at S7 nonradiatively. The red shifts in two time-dependent PL spectra disclose the generation of triplet excited states via the intersystem crossing process (Figure 6d). Finally, the H2 molecule is released from the surface with an energy gain of 1.24 eV at S8. Fate of Photogenerated Holes at the CN−Water or Hole−Scavenger Interface
The OER and the HER are the two half-reactions that complete the water-splitting process. The HER involves the transfer of two electrons, while the OER involves four holes. Therefore, the OER has a higher activation energy barrier than the HER and is more challenging.41 Despite a well-aligned 253
DOI: 10.1021/acs.accounts.8b00542 Acc. Chem. Res. 2019, 52, 248−257
Article
Accounts of Chemical Research
Figure 7. (a) Structures of g-C3N4/water systems. Clockwise from the top left: top and side views of g-C3N4/H2O with no OH−, g-C3N4/OHbulk− with OH− in the bulk of liquid water, and g-C3N4/OHinter− with OH− at the interface. OH− and its H3O+ counterpart are marked with black and green circles, respectively. Carbon, nitrogen, oxygen, and hydrogen atoms are represented by brown, light-blue, red, and pink balls, respectively. (b) Electronic level alignment of valence bands at the g-C3N4−H2O, g-C3N4−OHinter−, and g-C3N4−OHbulk− interfaces at the Γ point. (c) Difference in the charge densities in g-C3N4/H2O, g-C3N4/OHinter−, and g-C3N4/OHbulk−. Reproduced with permission from ref 43. Copyright 2018 Royal Society of Chemistry.
radical difficult. Therefore, oxidation of OH− requires sacrificial reagents to sweep the electrons. In the presence of a sacrificial electron scavenger, the holes in CN migrate to OH− after the photogenerated electrons are depleted. Increasing the concentration of OH− in solution in close vicinity to the CN surface would also enhance the photooxidation reaction rate.43 Therefore, the introduction of sacrificial hole/electron scavengers is now routinely practiced to suppress the recombination of electron−hole pairs to promote the desired photocatalytic reduction reactions. In a competition between the fast electron transfer (typically on the ps to ns time scale) and the subsequent slower electron consumption (typically on the μs to ms time scale), addition of hole scavengers can reduce the loss of photoexcited electrons through the removal of photoexcited holes.45,46 Recently, a reverse hole transfer (RHT) process from the adsorbate to the substrate has been identified as a key factor governing the hole-scavenging ability of different hole scavengers such as methanol and ethylene glycol.47 Methanol has been found to be a superior hole scavenger because of its long-lived RHT rate compared with other molecules (Figure. 8a). In the CH3OH/CN system, upon photoexcitation, energetic holes start relaxing toward the VB edge of CN because of electron−phonon coupling, while interfacial holes are transferred from the CN substrate to the CH3OH adsorbate, which is known as forward hole transfer (FHT). The time scale for this FHT has been found to be
