Keywords

1 Introduction

Natural photosynthesis (NPS) is a chemical process conducted by green plants, photosynthetic bacteria, and all other photosynthetic organisms, where CO2 and water are transformed into carbohydrates and molecular oxygen, respectively [1, 2]. This process can be considered as the most efficient and effective method of converting light energy into chemical energy. Over billions of years, NPS has provided an overwhelming amount of energy stock, mostly in the form of fossil fuels, for the world [3]. However, as is widely known, the rate of depletion of this energy stock is much higher than the rate at which nature can replenish it [2, 4]. Consequently, it is only a matter of time before fossil fuels finally run out. With this looming energy crisis, many approaches have been made to imitate NPS by artificially utilizing light energy to drive the thermodynamically uphill reactions of abundant substances to produce energy-containing chemicals, such as hydrogen, hydrocarbons, and ammonia [1, 3]. At present, this method is known as artificial photosynthesis (APS) or the so-called photocatalytic process [5].

As is widely known, the green chlorophyll pigment plays a major role in the NPS reaction. Generally, light energy or photons are absorbed by chlorophyll and used to remove electrons from water molecules to produce oxygen gas as follows:

$$2{\text{H}}_{{2}} {\text{O}}\mathop{\longrightarrow}\limits^{light}{\text{O}}_{2} + 4{\text{H}}^{ + } + 4e^{ - } .$$
(1)

This oxidation reaction occurs in the reaction center of photosystem II. Afterwards, the freed hydrogen and electrons are transferred to photosystem I and trigger the reduction of carbon dioxide (CO2) to produce sugar (C6H12O6) through the Calvin cycle:

$${\text{CO}}_{{2}} + 4{\text{H}}^{ + } + 4e^{ - } \mathop{\longrightarrow}\limits^{light}\frac{1}{6}{\text{C}}_{6} {\text{H}}_{12} {\text{O}}_{6} + {\text{H}}_{{2}} {\text{O}}.$$
(2)

The overall reaction of NPS is written as

$${\text{CO}}_{{2}} + {\text{H}}_{{2}} {\text{O}}\mathop{\longrightarrow}\limits^{light}\frac{1}{6}{\text{C}}_{6} {\text{H}}_{12} {\text{O}}_{6} + {\text{O}}_{2} .$$
(3)

Meanwhile, in APS, a material called a photocatalyst (mostly a semiconductor) is usually employed to harvest light energy [5]. Similar to NPS, this photon energy is used by the photocatalyst to generate charge carriers or excitons, i.e., electrons (e−) and holes (h+). In detail, upon light illumination, e− is excited from the valence band (VB) to the conduction band (CB) of the photocatalyst, leaving behind the h+ in the valence band region [6]. These charge carriers are transferred separately to different reaction centers, where they trigger some endothermic redox reactions; electrons drive the reduction reaction while holes drive the oxidation reaction as follows:

$${\text{A}} + e^{ - } \mathop{\longrightarrow}\limits^{light}{\text{A}}^{ - } ,$$
(4)
$${\text{D + h}}^{+}\stackrel{ \, {\text{light}} \, }{\to }{\text{D}}^{+}.$$
(5)
Fig. 1
figure 1

Comparison between natural and artificial photosynthesis

Full size image

A and D represent an electron acceptor and electron donor, respectively. Figure 1 illustrates a comparison between NPS and APS.

In the past few decades, many studies have been carried out to obtain a full understanding of the reaction mechanisms, kinetics, substances involved, and other important properties of NPS [5]. Even though we do not yet have sufficient understanding to further proceed to the industrial stage, much progress has been made toward applying the key features of NPS to APS [7]. To date, there are three very well known APS reactions that have attracted considerable attention from scientists and researchers worldwide, i.e., water (H2O) splitting, carbon dioxide (CO2) reduction, and nitrogen (N2) fixation; water, CO2, and N2 are very suitable raw materials for APS applications because they are abundant and easily obtained from the natural environment. In this chapter, we provide a summary of the principles, requirements, and latest developments of APS-based reactions.

Before proceeding with the above-mentioned discussion, it is necessary to define a general term that is often used to evaluate photocatalytic efficiency for any APS reaction, i.e., apparent quantum yield (AQY). The concept of AQY has been widely used to make valid comparisons between different experimental and illumination conditions [6]. AQY is evaluated for a certain photon wavelength (monochromatic) and calculated as

$$\text{AQY = }{\text{nr}}/{\text{I}},$$
(6)

where n, r, and I represent the number of e– involved in the photocatalytic reaction, the corresponding production rate, and the rate of incident photons, respectively. Typically, the wavelength-dependent AQY follows the optical absorbance pattern of the semiconductor, which decreases as the irradiation wavelength approaches the absorption edge because of the lower absorption coefficients and larger migration distances for photogenerated carriers [6].

2 Requirements for Artificial Photosynthesis

Since the purpose of APS is to drive some thermodynamically uphill reactions, there are some principal conditions that must be fulfilled so that the reaction may occur effectively and efficiently. These requirements apply to all APS reactions, and therefore, it is important to identify them as the very first stepping stones toward designing any photocatalytic system.

2.1 Thermodynamic Boundaries and Photon Absorption

To successfully perform any redox reaction, one needs to pay careful attention to the thermodynamic constraints of the reaction. In this case, we mostly use the term reduction potential to quantify the thermodynamic state of a certain reaction. The reduction potential or electrode potential is usually defined as a value for measuring the tendency of some substances to receive electrons, and thus be reduced. It is measured as the energy level, with volt (V) as the unit, with respect to a certain standard potential, such as the standard hydrogen electrode (SHE).Footnote 1 The higher (more negative) the value of this potential, the greater the likelihood of the species acquiring electrons and being reduced. In contrast, the lower (more positive) this value, the greater the tendency for the backward reaction (oxidation) of the species to occur. For a redox photon-induced reaction to occur, the CB energy level of the photocatalyst must be higher than the potential of the reduction reaction, while the VB must be located at a more positive potential than the oxidation reaction. Since different reactions require different thermodynamic boundaries, some photocatalysts will be suitable for a certain reaction but might not be suitable for other reactions. However, it is also possible to shift the position of the energy band by introducing some impurities, known as dopants, to meet different thermodynamic requirements [8]. More details of the thermodynamic constraints of each APS reaction and the corresponding implications will be presented in later sections.

Furthermore, since the ultimate aim of APS is to utilize the energy emitted by the sun, we also need to consider the characteristics of sunlight itself. The sun emits a spectrum of electromagnetic waves with a relatively broad wavelength, which comprises 3–5% UV (<400 nm), 42–43% visible light (400–700 nm), and 52–55% infrared (>700 nm) [9]. Thus, the utilization of visible light, near infrared (NIR), and infrared is essential to achieving substantial solar energy conversion, although this is more challenging. We often use the term band gap (Eg) to evaluate the light-harvesting ability of a semiconductor photocatalytic material. In the electronic structure of solid materials, including semiconductors, the band gap is usually defined as the energy difference between the top of the valence band (valence band maximum, VBM) and the bottom of the conduction band (conduction band minimum, CBM). The value of Eg is usually obtained by applying the Kubelka–Munk transformation and can be used to determine the maximum wavelength of light (λmax) that the photocatalyst can absorb using the following equation:

$$\lambda_{\max } = {{hc} \mathord{\left/ {\vphantom {{hc} {E_{g} ,}}} \right. \kern-\nulldelimiterspace} {E_{g} ,}}$$
(7)

where h is Planck’s constant (6.626 × 10−34 m2 kg/s) and c is the speed of light (3 × 108 m/s). Generally, a photocatalyst can absorb light with a higher energy than its band gap, or in other words, a wavelength less than the value of λmax calculated as above. Hence, to enable the absorption of a larger proportion of the solar spectrum, a semiconductor with a smaller band gap is preferred. Table 1 shows a list of some semiconductors that have been applied in APS and their corresponding band gaps.

Table 1 List of popular and potential semiconductors that have been applied in APS
Full size table

Given that some semiconductors have relatively large band gaps, some approaches to band-gap narrowing have involved inserting some dopant(s) into the molecular arrangement of a semiconductor, thus modifying its electronic structure. The basic concept of this method is to introduce a new energy level below or above the constituent orbital state [10], so that it may change the position of the CBM or VBM, which is also beneficial for meeting different thermodynamic requirements of different reactions as mentioned earlier [8]. For example, one can shift the VBM level of some oxide materials (usually consisting of the O2p state) upwards by adding nitrogen as a dopant, thus introducing the new N2p state, which lies above the O2p state, and generating a higher VBM position [11]. On the other hand, this doping method also has the drawback that the doping sites can act as recombination centers for charge carriers [11]. Therefore, in the design of photocatalytic materials, one must strive to achieve a balance between the modulation of light absorption and the recombination probability [10].

2.2 Charge Carrier Dynamics in Light-Harvesting Material

Following the generation of charge carriers due to photon absorption, the electron–hole pairs need to be separated immediately and transported into different zones; otherwise, they will recombine with each other within the photocatalyst [75, 76]. This electron–hole recombination usually results in a significant loss of energy, thus limiting the quantum efficiency of the photocatalytic system [5]. Consequently, the migration step, which is strongly correlated with the charge carrier dynamics, plays an essential role in an effective photocatalytic system. There are two parameters that are usually used to characterize the charge carrier dynamics of a semiconductor: carrier lifetime and diffusion length [76]. The carrier lifetime can be described as the average time taken for the charge carriers to recombine, whereas the diffusion length is defined as the distance that the carriers can travel before they recombine. From these definitions, it can be easily understood that the higher their values, the higher the probability that the charge carriers reach the surface of the photocatalyst and subsequently participate in the surface reactions. Hence, extensive effort has been made to improve these values, i.e., (1) controlling the shape, structure, and size of the photocatalysts to reduce the distance that the charge carriers must travel before reaching the surface of the photocatalyst; (2) formation of a homojunction and heterojunction to promote the spatial separation of electrons and holes; (3) utilization of a sacrificial agent as a scavenger of particular charge carriers, thus increasing the proportion of oppositely charged carriers in the photocatalyst; and (4) introduction of a cocatalyst on the surface of a photocatalyst that can provide charge accumulation centers and also accelerate interfacial reaction kinetics [5, 75, 77].

2.3 Existence of Reactive Sites

Last but not least, a photocatalyst material should have active catalytic sites on its surface for the APS reaction to occur. However, most photocatalyst surfaces are not typically designed to catalyze redox reactions [76]. To address this problem, a cocatalyst that can provide reactive sites for a certain APS reaction is frequently loaded and dispersed on the surface of the photocatalyst [78]. For this purpose, cocatalyst components are often adapted from the components of electrocatalysts applied for the same reactions [10]. In other cases, some surface defects are introduced on the surface of the photocatalyst, which can unexpectedly provide reactive sites for APS reactions, especially CO2 reduction [59, 79, 80] and N2 fixation reactions [69, 71, 73].

To establish effective surface reactions, the reactive sites must be designed according to the Sabatier principle,Footnote 2 which states that the catalytic surface must have a trade-off between being reactive and not being poisoned by the reaction intermediates [81]. Hence, the interactions between the sites and the substances (either the reactant or the product) should be modest: neither too weak so that the sites can adequately bind the reactant and the surface reaction (through the electron transfer process) may take place, nor too strong so that the product can easily desorb from the catalytic surface [82,83,84]. Secondly, if a cocatalyst is employed, it should form a good coordination with the semiconductor to ensure smooth charge migration into and from the reactive sites [10, 50]. This is because, for any catalytic reaction, a sustained charge carrier supply is the key to a continuous and effective surface reaction [85]. Accordingly, a suitable loading method and a precursor of the cocatalyst are crucial. Finally, increasing the number of reactive sites by either altering the morphology of the material to a low-dimensional structure [48, 86, 87], exposing more active edges of the cocatalyst [88, 89], or reducing the size of the cocatalyst into clusters [90,91,92] or single atoms [64, 93,94,95] is also an effective way of enhancing the photocatalytic activity.

Recently, an advanced strategy to satisfy such requirements has been demonstrated by downsizing transition-metal-based cocatalysts to the molecular scale, thus maximizing the catalytic activity [50]. By anchoring well-dispersed single-site Co/Ni species on a sulfide semiconductor, excellent photocatalytic H2 production was achieved, with AQY as high as 56.2 and 67.5% at 420 nm for Co and Ni species, respectively. These outcomes indicate the importance of the good dispersion of cocatalyst species as well as good coordination between the cocatalyst and the semiconductor, which will result in the full utilization of reactive sites, efficient charge transfer, and favorable kinetics for better catalytic performance [50].

3 Water Splitting

Following the pioneering work of photo-electrochemical water splitting by Fujishima and Honda in the early 1970s [12], much effort has been made toward developing an effective and efficient water-splitting photocatalytic system. Water splitting itself is an uphill redox reaction with an increase in Gibbs free energy (ΔG°) of about 237.13 kJ/mol [78], which can be written as

$$2{\text{H}}_{{2}} {\text{O}}\mathop{\longrightarrow}\limits^{light}2{\text{H}}_{2} + {\text{O}}_{2}$$
(8)

The water splitting reaction can be divided into two half reactions, i.e., a hydrogen evolution reaction (HER) and an oxygen evolution reaction (OER):

$$2{\text{H}}^{ + } + 2e^{ - } \to {\text{H}}_{{2}} ,$$
(9)
$$2{\text{H}}_{{2}} {\text{O}} + 4h^{ + } \to {\text{O}}_{2} + 4{\text{H}}^{ + } .$$
(10)

As already explained for the thermodynamic constraints, the CBM of the semiconductor must be located at a more negative level than the H+/H2 reduction potential (0 V vs. SHE at pH 0), while the VBM should be more positive than the O2/H2O energy level (1.23 V vs. SHE at pH 0). Consequently, the theoretical minimum band gap energy required to drive water splitting is 1.23 eV. Nevertheless, considering the possible thermodynamic losses (0.4 eV) and overpotentials (0.3–0.4 eV), the practical minimum band gap energy of the semiconductor for this reaction should be around 1.9 eV [5, 61]. Generally, if one of these thermodynamic requirements is not fulfilled, no water splitting will take place. Therefore, using a semiconductor with suitable band positions is indispensable in this case.

To assess the water splitting performance of any photocatalytic system, the concept of solar to hydrogen efficiency (STH) is also often used together with the AQY evaluation. STH is defined as the product of the hydrogen production rate and the increase in Gibbs free energy (ΔG°) for the water splitting reaction divided by the total energy of incident solar irradiation:

$${\text{STH = }}\frac{{{\text{output}}\;{\text{energy}}}}{{{\text{incident}}\;{\text{solar}}\;{\text{light}}\;{\text{energy}}}} = \frac{{r_{{H_{2} }} \times \Delta G^{ \circ } }}{{P_{{{\text{sun}}}} \times S}},$$
(11)

where Psun and S represent the flux energy of solar irradiation and the surface area of the system available for incident light, respectively. STH is considered as an absolute and practical standard to evaluate the water splitting performance of any photocatalyst under solar light irradiation [6]. An AM 1.5 G solar simulator (100 mW/cm2) is commonly used as the source of solar light to evaluate STH.

In a recent review, it was estimated that for any photocatalytic water splitting system to compete with the conventional methane steam reforming process, it must attain STH of 10% and a lifetime exceeding 10 years [6, 10, 96]. Figure 2 shows the relationship between STH and available photon wavelength in an AM 1.5 G simulator at different AQYs [6]. For example, to achieve an STH of 10% when using solar irradiation with a wavelength of less than 600 nm, an AQY of 60% is needed. For UV utilization (<400 nm), the maximum possible STH is 1.7% even at an AQY of 100%. These calculations once again show the need to develop a photocatalytic system that can operate using a wider range of wavelengths of the solar spectrum.

Fig. 2
figure 2

Relationship between STH conversion efficiency and photon wavelength at different AQYs for photocatalytic water splitting. Reprinted with permission from Springer Science + Business Media New York: Springer US [6], Copyright (2014)

Full size image

3.1 Overall Water Splitting

To perform successful overall water splitting (OWS), there is a critical issue related to the rapid backward reaction (reformation of H2O), which is thermodynamically predominant over the forward reaction. Consequently, deactivating the reverse reaction must be the top priority in designing any photocatalytic system for OWS [97]. One of the most popular methods of dealing with this problem is to apply a surface nanolayer coating [10]. This layer should be able to function as a molecular sieve, selectively allowing H2 and water molecules to permeate through the layer while preventing O2 molecules from diffusing and reaching the surface of the cocatalyst [98]. Since the reactive sites for backward and forward reactions are typically the same, this method effectively hinders the backward reaction. For example, a Cr2O3 nanolayer has been applied in many studies and found to be effective for inhibiting the reverse reaction, thus improving the photocatalytic OWS performance [22, 99]. The Cr2O3 nanolayer can be photodeposited on the surface of a cocatalyst by reducing the CrO42− ions where it encapsulates the cocatalyst, thus producing a core–shell structure [99]. Figure 3 illustrates the mechanism by which the Cr2O3 nanolayer coating suppresses the backward reaction in photocatalytic OWS [100]. The Ni/NiO core/shell cocatalyst and hydrophilic nanolayers of amorphous oxyhydroxides of group IV and V transition metals (Ti, Nb, Zr, and Ta) are other successful examples of this technique [101, 102].

Fig. 3
figure 3

Schematic illustration of the cocatalyst-Cr2O3 core–shell for suppressing the backward reaction in photocatalytic OWS. Reprinted with permission from Maeda and Domen, J. Phys. Chem. Lett., 1, 2,655–2,661 (2010) [100]. Copyright (2010) American Chemical Society

Full size image

As mentioned earlier, a photocatalyst with suitable band positions is necessary to fulfill the thermodynamic requirement of any APS reaction. However, it is quite difficult to find a suitable semiconductor with CBM and VBM levels straddling the redox potentials of H+/H2 and O2/H2O. An appealing solution to this problem was demonstrated by Domen and his group, who used two semiconductors to form a two-step photoexcitation system, so-called Z-scheme water splitting (see Fig. 4), achieving STH exceeding 1% [22]. This method is inspired by natural photosynthesis, which also involves two photosystems where two different reactions occur separately as previously explained. Similarly, in the Z-scheme system, one semiconductor acts as a hydrogen evolution photocatalyst (HEP), while the other acts as an oxygen evolution photocatalyst (OEP). The remaining holes and electrons in the HEP and OEP respectively recombine through a solid-state electron mediator (e.g., Au, Rh, Ni, Ag, Ir, and RGO) [22, 24, 103,104,105] or an aqueous redox mediator (e.g., Fe3+/Fe2+ and IO3−/I− solutions) [23, 106]. With this technique, semiconductors with unsuitable band positions (CBM lower than the H+/H2 redox potential or VBM higher than the O2/H2O potential) and a relatively small band gap can be utilized, thus expanding the choice of semiconductors [5, 9]. Furthermore, this system also provides a larger driving force for water splitting than the conventional photocatalytic system (one semiconductor) [10]. However, the major drawback of Z-scheme water splitting is the reduced amount of H2 and O2 produced, since half of the charge carriers recombine within the mediator [77].

Fig. 4
figure 4

Schematic of Z-scheme OWS on Ru-modified SrTiO3:La,Rh/Au/BiVO4:Mo particulate photocatalyst sheets. Reprinted with permission from Springer Nature [22], Copyright (2016)

Full size image

3.2 Hydrogen Production from Sacrificial Water Splitting

Achieving OWS using a simple photocatalytic system is difficult in practice owing to many strict requirements that must be simultaneously satisfied [75]. Therefore, some researchers have attempted to apply a sacrificial agent as an irreversible electron donor or acceptor to perform an HER or OER separately, thus greatly reducing the complexity of the reaction. This sacrificial water splitting (SWS) method is relatively effective and efficient for assessing the suitability of a certain photocatalytic system for performing water splitting in terms of band positions, photogenerated charge dynamics, compatibility of the cocatalyst, and other properties [75]. Furthermore, since the reverse reaction (water formation) can be annulled in SWS, it is common to obtain a higher apparent production rate and quantum yield than in the case of OWS.

For HER, different semiconductors require different sacrificial agents (as an electron donor or hole scavenger) to obtain good performance. This phenomenon is due to the wide range of ability of semiconductors to oxidize a sacrificial agent [107]. For example, the semiconductor graphitic carbon nitride (g-C3N4) can exhibit superior performance when triethanolamine (TEOA) is used as a hole scavenger but has a rather poor performance when a different sacrificial agent is applied. Considering the cost and efficiency, the realization of this sacrificial photocatalytic hydrogen production strategy by utilizing biomass and abundant compounds in nature or industrial waste as the sacrificial agent will be of great significance [75].

CdS is an important and popular semiconductor which is often used in sacrificial HER. Although CdS has a very suitable band position for water splitting with a visible-light-responsive band gap (2.4 eV), it is ineffective for OWS owing to the self-photocorrosion phenomenon, which is often encountered with the use of sulfide materials [97]. The sulfur (S2–) in CdS, instead of H2O, is very susceptible to oxidation by photogenerated holes accompanied by the elution of the Cd2+ cation, as shown by the following equation [75]:

$${\text{CdS}} + 2h^{ + } \to {\text{Cd}}^{2 + } + {\text{S}}.$$
(12)

Owing to this problem, a mixture of Na2S and Na2SO3 is often utilized as a sacrificial agent, which not only protects the sulfide semiconductor from photocorrosion, thus producing a stable photocatalyst, but also greatly boosts its photocatalytic HER performance [48].

An interesting study about nature-inspired environmental phosphorylation to boost the sacrificial hydrogen production of g-C3N4 and several other semiconductors has been reported [62]. In NPS, phosphates play vital roles in establishing the Calvin cycle, acting as the transporter and pump for charge carrier transfer during light-dependent reactions and as mediators in carbon fixation during the dark reactions of photosynthesis. In this study, it was found that the addition of a phosphate component (KH2PO4) markedly boosted the H2 generation of Pt/g-C3N4 by establishing a similar proton reduction mechanism to the one found in NPS (see Fig. 5), achieving a very high AQY of 26.1% at a wavelength of 420 nm for a g-C3N4-based photocatalyst. This finding may provide a promising and facile approach to highly efficient photocatalysis for both SWS and OWS in the future.

Fig. 5
figure 5

Copyright Wiley–VCH Verlag GmbH & Co. KGaA. Reproduced with permission

Schematic illustration of photocatalytic H2 evolution via the proton reduction mechanism involving HPO42−. From Liu et al., Angew. Chem. Int. Ed., 54, 13,561–13,565 (2015) [62].

Full size image

4 CO2 Reduction

In modern society, owing to the rapidly increasing demand for energy, CO2 emissions have escalated, resulting in a very serious state with NPS no longer compensating for CO2 emission [108]. CO2 is generally regarded as the main greenhouse gas, with excess CO2 causing environmental problems, such as global warming, melting of the polar ice caps, sea level rises, and worsening droughts [108, 109]. These concerns have led to attempts to control the amount of CO2 by converting CO2 into harmless and useful chemical feedstocks [110]. Regarding this approach, solar-driven CO2 conversion through APS is considered as one of the most reasonable and ideal approaches not only to reduce the imbalance between NPS and CO2 emission, but also to provide a new type of energy feedstock, thus increasing the capacity of the global energy storage system [83, 111].

Since the carbon atom in CO2 molecules possesses the highest valence, the photoreduction of CO2 may result in a wide variety of carbon products with different oxidation states, ranging from gaseous CO and CH4 to higher hydrocarbon species (e.g., C2H4, C3H8), as well as liquid-phase oxygenated hydrocarbons, such as CH3OH and HCOOH (but not complex products such as carbohydrates) [112]. Table 2 shows the standard electrode potentials (E°) for the transformation of CO2 into various products in aqueous media at pH 7 with respect to SHE. Unfortunately, CO2 photoreduction is difficult to realize owing to the extreme stability of the CO2 molecule, ascribed to its two symmetric C = O bonds, which possess a high dissociation energy of about 750 kJ/mol [83]. In this case, the activation of CO2 molecules very likely becomes the rate-determining step for the CO2 reduction reaction [112]. Moreover, in the presence of water, CO2 reduction must compete with proton reduction owing to the similar reduction potential and also the preferential adsorption of water compared with CO2, thus resulting in the low selectivity and efficiency of the desired reaction [83].

Table 2 Standard reduction potentials (pH 7) for the reactions of CO2 reduction to different products in aqueous media
Full size table

From the kinetic perspective, the CO2 reduction reaction is also challenging. The first step in CO2 reduction is generally considered to be the formation of an intermediate CO2− (through a single electron transfer (ET) to a CO2 molecule), whose reduction potential is very high (about −1.9 V) [112], as shown in Table 2. Consequently, a high overpotential is necessaryFootnote 3 for the CO2 reduction reaction to further proceed. Given this, an alternative reaction route involving a series of multiple proton-coupled electron transfer (PCET)Footnote 4 processes can be employed to bypass the formation of CO2−, thus avoiding the large activation barrier and the formation of unstable and high-energy intermediates [4, 83] (see Table 2). However, the PCET process is kinetically dependent on both the concentration of protons in the solution and the surface electron density at the reactive sites [112]. Moreover, since most of the CO2 reduction products require multielectron transfer, the corresponding CO2 reduction reactions are less kinetically favorable than the reduction of water, which only involves two electrons [112].

In the following sections, we will briefly discuss the strategies for addressing some challenges in CO2 reduction reactions: the activation of CO2 molecules and controlling the selectivity of CO2 reduction.

4.1 Activation of CO2 Molecules

The activation of a CO2 molecule is closely related to its adsorption by the surface atoms of a photocatalyst. The chemisorption of a CO2 molecule induces the formation of a partially charged CO2δ•− adsorbate, which no longer possesses a linear symmetrical structure [114]. The bending of CO2 molecules is beneficial for lowering the barrier for electron transfer into CO2 owing to the decrease in the lowest unoccupied molecular orbital (LUMO) level [114]. Hence, the basic strategy for driving the activation of CO2 molecules is to promote the chemisorption of CO2. Figure 6 shows some possible binding modes for the adsorption of a CO2 molecule on the surface of a metal oxide (such as TiO2), a popular photocatalyst for the photoreduction of CO2 [115].

Fig. 6
figure 6

Reproduced from Ref. [115] with permission from The Royal Society of Chemistry

Modes of adsorption for CO2 on the surfaces of regular and modified metal oxides.

Full size image

As approaches to activating CO2 molecules, several practical strategies have been experimentally demonstrated, i.e., increasing the available surface area for adsorption, establishing surface defects, introducing surface basic sites, and employing a cocatalyst on the photocatalyst surface. In general, increasing the surface area of the photocatalyst, either by modifying the structure and morphology of the photocatalyst or by using a porous material, will provide more reactive sites for CO2 adsorption [116,117,118,119]. Surface defects, such as oxygen vacancies [25, 79, 80] and sulfur vacancies [57,58,59], have attracted considerable attention in the last decade owing to their potential for generating an active trapping surface for CO2 molecules. The challenge in this strategy is to prevent the formation of bulk defects, which usually occurs alongside the formation of surface defects, because bulk defects can act as recombination centers, which adversely affect the photocatalytic performance [75]. The formation of surface basic sites is also beneficial for CO2 chemisorption owing to the Lewis acidity of the C atom, which enables CO2 molecules to interact with the alkaline surface, thus leading to the formation of carbonate or bicarbonate intermediates [120,121,122]. Finally, a cocatalyst with a large work function, such as a noble metal, can also be employed, which may induce electron backdonation from metal d-orbitals into the (C–O) π* orbital of CO2, causing the formation of a CO2δ●− adsorbate [109, 112].

It has also been reported that CO2 molecules can be polarized and subsequently activated by utilizing highly energetic electrons excited by the localized surface plasmon resonance (LSPR) of some plasmonic nanometals (e.g., Au, Pt, Pd, etc.), assisted by thermal energy [123,124,125,126,127]. The LSPR effect is defined as the resonant photon-induced collective oscillation of valence electrons, established when the frequency of the photons matches the natural frequency of the surface electrons oscillating against the restoring force of positive nuclei [128]. This phenomenon generates hot electrons with very high energy, which are very effective for disrupting the stability of CO2 molecules [129]. Figure 7 illustrates a schematic and the reaction mechanism of the LSPR effect over a Pt-Au/SiO2 catalyst for the photoreduction of CO2 [127]. Following these studies, the photothermal effect through the self-heating process was observed on an illuminated boron catalystFootnote 5 [130]. It is well known that CO2 molecules are more reactive at high temperatures. In this case, the photothermal effect of boron can induce a high local temperature, which facilitates CO2 activation. It also triggers the self-hydrolysis of boron particles, which in situ produces H2 as the active proton source and electron donor for CO2 reduction, as well as boron oxides that function as promoters of CO2 adsorption onto the surface of boron catalysts. All these effects synergistically promote the CO2 reduction reaction.

Fig. 7
figure 7

a Schematic of LSPR effect generating light-excited hot electrons for the activation of stable compounds, such as CO2 and CH4. b Proposed reaction mechanism for photocatalytic CO2 reduction over Pt-Au/SiO2 by utilizing the dry reforming of methane (DRM) reaction. Reprinted with permission from Song et al., ACS Appl. Mater. Interfaces, 10, 408–416 (2018) [127]. Copyright (2018) American Chemical Society

Full size image

4.2 Controlling the Selectivity of CO2 Reduction

As mentioned earlier, CO2 reduction has several possible reaction pathways, depending on the number of electrons involved in the reactions, leading to various carbon products. Therefore, in addition to the competition with proton reduction, it is also important to ensure that the CO2 reduction selectively proceeds through the desired reaction. In short, the binding mode of CO2 during the adsorption state (see Fig. 6) determines the route that the reaction takes [112, 115]. Accordingly, tuning the surface features of the photocatalyst to favor the formation of a particular adsorbate intermediate, which in turn will be converted into the desired carbon product, is the key to controlling the selectivity of CO2 reduction.

The cocatalyst is also important for not only kinetically driving the multielectron reaction of CO2 reduction (since it may serve as an electron-trapping center), but also tuning the selectivity of the reaction [109]. The final product of the reaction may vary with the adsorption properties of the intermediate, electron transfer (for proton reduction and CO2 activation) overpotentials, loading method, elemental composition, dispersion, particle size, structure, and exposed sites of the cocatalyst [109, 112, 131]. Therefore, the careful selection and design of the cocatalyst are indispensable for regulating the reaction mechanism [109]. Figure 8 summarizes the general selectivity trends of CO2 reduction over various metal-based cathodes [132]. In addition, it is also necessary to tune the experimental conditions. Among the many experimental parameters, the pH of the solution can considerably affect the CO2 reduction reaction mechanism [131]. pH not only determines the concentration of H+ ions in the solution (which is related to the rate of H2 evolution), but also affects the solubility of CO2 and the relative concentrations of CO32− and HCO3−, which are strongly correlated with the modes of CO2 adsorption [112, 131].

Fig. 8
figure 8

Section of the periodic table depicting the primary products in CO2 reduction over some metal-based electrodes. Reprinted with permission from White et al., Chem. Rev., 115, 12,888–12,935 (2015) [132]. Copyright (2015) American Chemical Society

Full size image

5 N2 Fixation

The N2 fixation reaction can be regarded as one of the most crucial chemical processes in nature, considering the importance of fixed nitrogen compoundsFootnote 6 for living organisms on this planet, owing to its integral role in producing basic biological building blocks [85, 133, 134]. In particular, ammonia (NH3), the main product of N2 fixation, is a very important raw material for the production of fertilizers and also as a promising carbon-free energy carrier [135,136,137]. The first work on photocatalytic nitrogen fixation in sterile soils containing natural abundant oxides was performed by Dhar and Pant in the early 1940s [138]. Subsequently, around 30–40 years later, Schrauzer and Guth provided more thorough reports of photon-driven nitrogen fixation over natural materials and as-synthesized TiO2 [19, 139]. These early studies triggered numerous extensive works to investigate nitrogen fixation over some well-designed synthetic photocatalysts [133]. Unfortunately, in the following decades, the studies on semiconductor-based photocatalytic nitrogen fixation were considerably hindered (as indicated by the substantially fewer reported studies) by the difficulties in N2 adsorption and activation over the catalytic surface, which were a stumbling block to further development and achieving competitive performance with other methods [133, 136, 137]. Nevertheless, owing to the recent advancement of nanotechnologies and nanomaterials, which can provide solutions to such kinetic limitations, photocatalytic N2 fixation has returned to the main stage in the field of APS in the past few years [85].

Being the ‘youngest’ of the APS applications, photon-driven N2 fixation still has a long way to go before an ideal catalytic N2 photofixation system can be devised, inasmuch as the comprehensive knowledge and understanding of the mechanistic principles of the reaction and also suitable materials are still lacking [85]. In this section, we briefly discuss the basic mechanism of N2 photofixation (especially to NH3), the requirements and challenges in achieving effective reaction processes, and the possible approaches to deal with them. The discussion herein is expected to contribute to filling the knowledge gap in the study of APS-based N2 fixation.

5.1 Fundamentals of Photon-Driven N2 Fixation

The N2 fixation and CO2 reduction reactions actually have many things in common. Similar to the CO2 molecule, the N2 molecule is also very stable owing to its extremely strong N≡N bond, with a dissociation energy of 945 kJ/mol for the direct cleavage of the triple bond [85]. As a result, a single ET (−4.16 V vs. SHE) and PCET (−3.20 V vs. SHE) to the N2 molecule (for its activation) are almost thermodynamically imposible with currently available semiconductors owing to their very high reduction potentials. Therefore, in a similar way to the CO2 reduction reaction mechanism, multiple PCET processes can also be applied in the N2 reduction reaction to avoid the principal energy barrier while bypassing the formation of high-energy intermediates [140, 141]. Multiple PCET, depending on the number of protons and electrons involved, may lead to various partially reduced and stable intermediate species (e.g., N2H2 and N2H4) other than NH3, thus resulting in competition between them and also with the water reduction reaction [140]. Table 3 provides a list of possible N2 reduction reactions in aqueous media and their corresponding standard reduction potentials.

Table 3 Standard reduction potentials (pH 0) for the reactions of N2 reduction to NH3 in aqueous media
Full size table

The extremely high stability of the N2 molecule is also a major disadvantage regarding the kinetic feasibility of the N2 fixation reaction. Because of the high stability, the adsorption and activation of the N2 molecule can be considered as the rate-limiting steps of the overall reaction [142]. For the same reason, it still remains a major challenge to enhance the kinetic feasibility of the N2 photofixation reaction. Therefore, even though there are other concerns that must be looked into and dealt with carefully, in the next section, we will only discuss the adsorption and activation of N2, the most crucial yet challenging steps in the N2 fixation reaction.

5.2 Adsorption and Activation of N2 Molecules

Similar to the CO2 molecule, the activation of the N2 molecule is also strongly associated with the adsorption of the species on the catalytic surface, since electron transfer can only occur when there is an intimate interaction between the chemical species and the catalyst [142]. Generally, a photocatalyst with a high surface area is favorable for heterogeneous catalytic reactions, including N2 fixation, owing to its capability of adsorbing more reactants [144]. Some other potential approaches for promoting the adsorption of N2 molecules by employing surface defects, such as nitrogen vacancies [69, 145] and oxygen vacancies [74, 146], have been succesfully demonstrated for various nitride and oxide semiconductors, respectively. The vacancy species on the surface of a semiconductor can be occupied by N atoms, thus enabling swift electron donation to the adsorbed N2 owing to the direct contact between the photocatalyst and the N2 molecules. Such a phenomenon is essential for the N2 activation step and will be discussed later. To suppress the faster H2 reduction reaction, employing a metal-based catalyst with stronger adsorption of N atoms than that of H atoms might be a promising strategy worth investigating and exploring more thoroughly [147, 148].

Regarding N2 activation, it is beneficial for the adsorbing sites to be able to synergically accept electrons from and backdonate to N2, which simultaneously weakens the N2 bondFootnote 7 and strengthens the bond between the sites and N2, as illustrated in Fig. 9 [135]. As a result, the adsorbed N2 is more susceptible to electron donation owing to the reduced resistance and activation energy barrier of the attenuated N2 bond. Subsequently promoting charge exchange between the catalyst and N2 is essential to provide an accessible kinetic pathway towards the activation step and later stages of the reaction [85]. Interestingly, the above-mentioned strategies can be simultaneously employed by generating vacancy species on the surface of a semiconductor, which is a powerful method of promoting the photocatalytic N2 fixation reaction, as mentioned earlier, owing to the N2-capturing ability and nature of the vacancy as an electron localization center [74]. When the surface vacancy successfully binds a N2 molecule, the backtransfer of charges from the vacancy to the adsorbed N2 occurs, rendering electron depletion on the vacancy and electron accumulation on the N2 molecule [71].

Fig. 9
figure 9

Simplified schematic of bonding in well-known end-on bound transition metal N2 complexes (left) and monovalent boron species (right). From Légaré et al., Science, 359, 896–900 (2018) [135]. Reprinted with permission from AAAS

Full size image

Even though a vacancy possesses great potential for N2 adsorption and activation, a surface vacancy can be easily consumed through oxidation or a similar phenomenon, leading to the deactivation of the photocatalyst [149]. Vacancy stabilization through the self-regeneration of the vacancy under light irradiation is therefore an ideal solution to this problem [74, 149]. Very recently, an excellent example of light-switchable oxygen vacancy circulation for sustainable N2 fixation on an ultrafine Bi5O7Br nanotube has been reported, the scheme of which is shown in Fig. 10 [74]. In addition, similar to the CO2 molecule, the N2 molecule can also be activated by utilizing hot electrons produced through the LSPR effect [26, 27, 150].

Fig. 10
figure 10

Copyright Wiley–VCH Verlag GmbH & Co. KGaA. Reproduced with permission

Schematic illustration of photoinduced oxygen vacancy (OV) circulation for N2 fixation on Bi5O7Br. From Wang et al., Adv. Mater. 29, 1,701,774 (2017) [74].

Full size image

6 Summary and Outlook

Owing to the rapid advancement of human civilization, both global energy consumption and environmental pollution have increased significantly and almost uncontrollably. To make matters worse, the regeneration of fossil fuels, our current primary energy resource, is extremely slow. Thus, the realization of artifical photosynthesis (APS) technology to exploit and convert the limitless solar energy into renewable, clean, and sustainable energy feedstocks has become an urgent task requiring intensive and extensive effort and study. For this reason, many APS-based scientists and researchers have attempted to acquire in-depth and comprehensive knowledge on the APS reaction mechanism as well as practical and efficient approaches to devise an ideal photocatalytic system.

In this chapter, we have summarized the basic concepts in designing an effective photocatalytic system by examining the general steps in the APS process, i.e., photon absorption, charge carrier migration, and the surface reaction. Furthermore, the fundamentals of three important APS reactions (water splitting, CO2 reduction, and N2 fixation) as well as the specific challenges of each reaction and some possible strategies to address them were briefly introduced. We also discussed several effective strategies for improving the performance of some photocatalytic materials in diverse applications, such as a surface nanolayer coating, the Z-scheme system, the sacrificial photocatalytic reaction, the localized surface plasmon resonance (LSPR) phenomenon, and surface defects (self-regeneration).

While considerable advances in APS research have recently been accomplished, efficiency, stability, and cost are still the main issues that must be simultaneously overcome to enable APS technology to enter industrial use. The development of novel photocatalytic materials, either by modifying currently available materials or by discovering completely new materials, may lead to a significant breakthrough in these problems. Correspondingly, surface and interface engineering are indispensable for the fabrication of outstanding materials with distinctive and useful features, and thus should be considered and further extended. Last but not least, it is also important to carry out an in-depth study of the mechanism through the combination of in situ observation and a theoretical approach (commonly by computational methods) to obtain a full, accurate, and detailed understanding of the mechanism behind all APS processes.