Metal-free carbon-based materials for electrocatalytic and photo-electrocatalytic CO₂ reduction

Abstract

Nowadays, reducing carbon dioxide emission in the atmosphere is one of the most important environmental issues that must be overcome. At the same time, low-cost and environmentally friendly technologies are necessary to produce renewable fuels able to replace the conventional fossil ones. Electrochemical cells (driven by solar energy) and photo-electrochemical cells (PECs) are among the main efficient technologies to get these challenging goals. Taking into account the PEC working mechanism, two different electrodes, based on photo-electrocatalytic and electrocatalytic materials able to drive reactions both under illumination or in dark conditions, are involved. In this review, recent results on carbon-based materials for electrocatalytic and photo-electrocatalytic carbon dioxide reduction are discussed. The properties and synthesis conditions applied to the preparation of conducting polymer and graphitic carbon nitride (g-C₃N₄) are described and discussed for their application in the photoactive electrodes. As for the electrodes to be applied in the electrocatalytic CO₂ activation and conversion, light heteroelement-doped carbon nanomaterials have been taken into account as highly valuable metal-free candidate to run the process efficiently and selectively. For the latter process, also the influence of the electrolyte and the selectivity towards different reaction products will be discussed. All these data taken together indicate that a lot of work still has to be done to achieve high efficiency with metal-free organic-based electro- and photo-electrocatalysts applied to the carbon dioxide conversion. Anyhow, many seminal outcomes collected in the literature up to now clearly indicate the real possibility to replace highly costly metal-based materials with simply organic ones.

1 Introduction

Energy demand, in the context of increasing global population and energy consumption, combined with steady reduced supply of traditional fossil fuels sources and the effect of carbon dioxide emissions on environmental and climate changes, represents a key issue of our modern society. The research of alternative renewable energy sources to power our planet and the development of new technologies applied to the reduction of carbon dioxide concentration in the atmosphere are among the most urgent priorities to be urgently addressed. In this scenario, CO₂ activation and catalytic conversion represents a highly attractive approach for harvesting a greenhouse gas and a chemical “waste” and converting it into chemicals and fuels of added value. Among the possible methods for CO₂ conversion, (photo)electro-reduction holds great potentialities (Kalamaras et al. 2018; Wang et al. 2018a; Chang et al. 2016; Duan et al. 2017; Liu et al. 2017a). Indeed, with respect to other and more energy consuming technologies [i.e., biological and biochemical transformations (enzyme or algae based), thermochemical conversion and mineralization] (Zheng et al. 2017), CO₂ (photo)electro-reduction lists several advantages such as an effective control on the reaction products by tuning the experimental parameters (i.e., temperature, electrolyte, applied potential, etc.) and mild operative conditions both in terms of reaction temperature and gas pressure. In addition, CO₂ reduction by PEC systems is directly driven by solar energy that is the most abundant (about 10⁵ TW supplied on Earth’s surface every year) renewable energy source available. As a matter of fact, the production of fuels and chemicals from the “photo-induced” reduction of carbon dioxide is a challenging technology and a valuable route to store intermittent renewable energy in an accessible format, while employing current infrastructures (Armaroli and Balzani 2016). Three different configurations for PEC carbon dioxide reduction have been used up to date: photocathode-driven PEC, photoanode-driven PEC and combined photoanode and photocathode (Z-scheme, artificial photosynthesis) cells (Kalamaras et al. 2018; Wang et al. 2018b; Yang et al. 2018; Xie et al. 2016). Thermodynamic and kinetic aspects need to be considered into PEC technology because the reaction rate and products can be controlled through the applied external potential as well as the nature of the catalyst used. Moreover, the kinetic largely depends on other factors such as the reaction temperature and pressure, the reagents concentration and the electrolyte (Xie et al. 2016). At the same time, to produce highly efficient PEC systems, the reactor design must be taken into account as well as the carbon dioxide adsorption and activation mechanism. However, due to the complexity of processes occurring inside a PEC system, it is a critical issue to understand the CO₂ adsorption and conversion route on the electrode materials for the realization of highly efficient PEC technology (Kalamaras et al. 2018; Chang et al. 2016; Yang et al. 2018).

A PEC system for carbon dioxide conversion is characterized by two solar-driven reactions (reduction and oxidation) on the electrodes separated by an electrolyte. The electrodes can work in dark condition (dark electrode) or under illumination (photoelectrode) in the presence of a current collector (Wang et al. 2018a; Yang et al. 2018). Typically, a transparent conductive substrate (fluorine or indium doped tin oxide) is used for the photoelectrode preparation due to its electrical and transparent properties. The space between the electrodes is filled with the electrolyte solution where the reagents are bubbled and, to achieve highly efficient devices, a separation membrane selective for H⁺ migration (such as Nafion) is used for products separation. A stirring system is also required to ensure a good mass diffusion for both the reagents and the products. A reference electrode is finally used for determining potential during PEC tests. Materials with different properties are used for the realization of the three different PEC configurations previously described and reported in Fig. 1.

Fig. 1

Schematic illustration of the possible PEC configuration with 2 compartment cells: a photoanode driven; b photocathode driven and c with photoanode and photocathode driven

For the photoanode-driven PEC (Fig. 1a), an active material based on a n-type semiconductor (with an excess of electrons on the conduction band) is required. When the sunlight reaches the semiconductor with a photon of energy equal or higher than the band gap energy (E_g), it creates positive and negative charges (holes and electrons) on its valence (V_B) and conduction bands (C_B), respectively. Due to the n-type semiconductivity, the photogenerated holes can oxidize water; while, the electrons reach the cathode under dark conditions. The water oxidation reaction produces H⁺ ions that selectively move to the compartment where the carbon dioxide is added reducing it into useful products and fuels that can be easily separated. Conversely, a semiconductor material with p-type properties (an excess of holes are present on its valence band) is considered in a photocathode-driven PEC (Fig. 1b). Now, the photogenerated electrons and holes are used to drive water oxidation reaction and carbon dioxide as in the previous case, but the holes are used to produce H⁺ at the dark side; while, the electrons are used to produce solar fuels and products at the photocathode. Finally, to mimic the photosynthetic process, both photoelectrodes can be illuminated and generate simultaneously the electron–hole pairs (Fig. 1c). The majority of the photogenerated charges are recombined at the ohmic contact between the two electrodes and only some charges carriers can be used for oxidation and reduction reactions. However, this is the most promising approach for the development of the next generation PEC due to the possibility to drive the different reactions without applying an external potential. Taking into account Fig. 1 (in particular Fig. 1a, b), the CO₂ conversion occurs by electrocatalytic reactions both under illumination (on photocathode side) and dark conditions (on cathode side). For this reason, the properties of electrode materials for photo-electrocatalytic and electrocatalytic CO₂ conversion (CO₂RR) are key parameters for the development of highly efficient PEC system.

The development of efficient (photo)-electrocatalytic materials for carbon dioxide conversion is a difficult task due to the high thermodynamic stability and chemical inertness of CO₂ (Ponnurangam et al. 2017). Moreover, multiple-electron process is required, involving from 2 up to 12 electrons, thus potentially leading to a large variety of C1 and/or C2 products (Table 1). An accurate control of the process selectivity is, therefore, one of the major challenges in this research field that has not been fully addressed yet (Kalamaras et al. 2018; Chang et al. 2016; Duan et al. 2017; Yang et al. 2018). In addition, when aqueous electrolytes are used, hydrogen evolution reaction (HER) arises as a highly competitive process (Table 1), potentially lowering the overall catalytic efficiency. As a result, the ambitious goal of this research area is to design and develop efficient (photo)-electrocatalysts able to maximize CO₂ conversion towards a target product, while suppressing undesired side processes.

Table 1 Standard potential for water splitting and carbon dioxide reduction reactions

Several metal-based heterogeneous catalysts have been reported to date as valuable promoters for both PEC and CO₂RR. Anyway, they irremediably suffer from common drawbacks such as poor sustainability (from an environmental and economical view-point), scarce selectivity and fast deactivation due to poisoning/leaching or photo-corrosion phenomena. As cheap, environmentally benign and effective alternatives to CO₂ (photo)electro-reduction, metal-free carbon-based materials have been extensively investigated in the last years (Cui et al. 2018), showing often comparable, and to some extent better, catalytic performance with respect to metal-based systems. The importance of this topi was highlighted in a very recenlty special issue on Advanced Materials named “Carbon-based metal-free catalysts (2019)”.

This review aims at highlighting the recent achievements in PEC and CO₂RR processes promoted by metal-free carbon-based materials. Despite that metal-free technologies applied to CO₂ conversion and valorization into commodities and fuels of added value are relatively new, this research field has already scored many interesting results. However, several aspects such as selectivity, productivity and the controversial relationships between structure and performances of carbon-based materials are still pending issues that will be discussed within this review.

2 Photo-electrocatalytic (PEC) CO₂ reduction

PEC photoelectrodes for carbon dioxide reduction based on organic semiconductors have attracted great attention due to their unique advantages such as low cost, high abundance, environmental friendliness and high charge-carrier mobility (Yao et al. 2018; Hursán et al. 2016). Moreover, organic semiconductors are more stable than the metal-based ones in water and under photo-electrochemical conditions because not influenced by photo-corrosion phenomena. These properties reduce also their payback time as well as the EROI parameter (energy returned of investment). In this review, photoelectrodes for CO₂ conversion based on organic semiconductors thin films are considered as primary light harvesting and charge-transport materials. Conducting polymers (CPs) and graphene-like carbon nitride (g-C₃N₄) were considered, because they can be easily prepared and deposited on the electrode surface as thin films and their optical and electronic properties are suitable for CO₂ conversion within PEC systems. Other organic semiconductor such as graphene, conjugated polymers and materials for organic photovoltaics (OPV) were not taken into account as they are not yet considered due to their low limitation in PEC for CO₂ conversion even if these materials found application in the photocatalytic and photo-electrochemical water splitting applications and photocatalytic carbon dioxide reduction. A deep overview on their properties can be found in different reviews (Yao et al. 2018; Remiro-Buenamañana and García 2019; Joy et al. 2018; Zhang et al. 2018a).

2.1 Conducting polymers (CP)

Conducting polymers (CPS) are a class of organic materials characterized by particular electronic properties such as definite band gap energy and enhanced electronic conduction (Yongfang 2015; Dai 2004). For carbon dioxide reduction in PEC systems, three are three compounds most commonly used: polyaniline, polypyrrole and poly(3,4-ethylenedioxythiophene) (PEDOT), due to their lower overpotential generally required for the electrocatalytic carbon dioxide reduction compared with metal-based catalytic systems. Other CPs such as polythiophene, are considered only for water splitting reaction and are not therefore interesting for this work (Fan et al. 2016; Kao et al. 2018). The monomer chemical structures of the three abovementioned CPs are reported in Fig. 2.

Fig. 2

Chemical structures of monomers for CPs considered for CO₂ PEC technology: a aniline, b pyrrole and c 3,4-ethylenedioxythiophene

Each system is characterized by the presence of heteroatoms (nitrogen, sulphur and oxygen) that can contribute significantly to the electrochemical reduction of carbon dioxide (Li et al. 2017a; Smith and Pickup 2010). Both nitrogen (pyridinic and quaternary) and sulphur induce the formation of positive charge on the polymeric chain that can play an important role in the CO₂ reduction. In this condition, however, the hydrogen evolution reaction can also take place and compete with carbon dioxide reduction. The presence of pyridinic-N, in particular, stabilizes the CO ⁻₂ /*COOH intermediate, that represents an important step for the CO formation (Li et al. 2017b; Li and Bandosz 2018). Besides these chemical properties, CPs electronic properties allow the application of these materials as PEC photoelectrodes. In fact, CPs are widely used for energy production and storage, sensors and separation techniques due to their high conductivity (10⁻³–10³ S/cm), low cost, good stability and high catalytic activity towards a lot of different reactions and redox properties (Yongfang 2015; Sangiorgi and Sanson 2017; Sangiorgi et al. 2019). In addition, CPs have many other interesting characteristics for energy application such as transparency, flexibility, easy availability and low-cost deposition processes (such as electropolymerization). CPs possess the electronic and optical properties of metal inorganic semiconductors but at the same time the flexibility and mechanical properties of polymers (Dai 2004). The distinctive properties of CPs are related to the conjugated molecular structure of the polymeric main chain, where π-electrons are delocalized over the whole polymer. For this reason, these materials exhibit the well-known band structure with π-valence band and π*-conduction band (valence band and the conduction band, respectively). The difference in energy between the valence band and the conduction band (the highest occupied molecular orbital, HOMO, and the lowest non-occupied molecular orbital LUMO, respectively) is the band gap energy that normally stands in the 1.5–3.0 eV range, therefore suitable for absorbing photon energy. The conductivity of CPs can be manipulated by doping during their synthesis. In fact, the charges can flow on them, thanks to the presence of polarons, bipolarons or solitons (degenerated polymers based on two equivalent resonance forms) delocalized on conjugated polymer chain produced after oxidation and reduction doping process (Yongfang 2015; Dai 2004). Via doping process, the conjugated polymers can, therefore, become p- or n- type conductors. In the p-doped state, the main chain is oxidized with a counter-anion that keeps the electron neutrality of the molecule balancing the holes created in the main chains. In the n-doped state, the main chain of the CP is reduced with a counter-cation that helps to maintain the electron neutrality balancing the electrons in the main chains (Yongfang 2015). These types of doping processes can be achieved by two main methods: chemical or electrochemical.

One of the most interesting applications of CPs in PEC technology is the use of polyaniline photocathode to reduce CO₂ to alcohols (Hursán et al. 2016). Polyaniline can in fact photochemically reduce carbon dioxide without adding specific catalysts also as a consequence of a surface with good selectivity for chemical interaction with CO₂. This is due to the N-rich polymeric chains that are generally considered as the best solid-state adsorbents for CO₂. PANI films were prepared by electrochemical polymerization on gold electrode. Under illumination (300 W Xe–Hg arc lamp) and with CO₂-saturated electrolyte, PANI electrode shows a cathodic photocurrent equal to 170 µA/cm² instead of the 80 µA/cm² obtained with nitrogen-saturated electrolyte. When the PANI was prepared with MWCNTs as substrate, the photocurrent reaches 0.5 mA/cm² due to the improved charge separation and transport processes (Fig. 3). Electrolysis tests under potentiostatic conditions with CO₂ were done with an applied potential of − 0.4 V vs Ag/AgCl (NaCl 3 M) and the acquired stationary photocurrent after 7000 s was equal to 70 µA. In these conditions, the obtained products were methanol and ethanol (with H₂ present in gas phase) with a Faradic efficiency equal to 43% and 20%, respectively. The comparison between PANI photoelectrodes and the ones based on polypyrrole (PPy) and PEDOT (prepared by electrochemical polymerization) confirms that the nitrogen atoms present inside the polymers were found to control the photo-electrochemical CO₂ reduction affecting its final conversion.

Fig. 3

Adapted from Hursán et al. (2016) with permission from The Royal Society of Chemistry

Photocurrent acquired for PANI electrodes under a low (100 W) and b high (300 W) illumination conditions at pH 7.0 and c related to PANI/MWCNTs under 300 W at pH 4.0.

For PEDOT and PPy, where the N-rich polymeric chain is less available (PPy) or not present (PEDOT) ,the photocatalytic activity was found to be definitively lower. Hursán et al. (2016) proposed a dedicated mechanism for the interaction between PANI and CO₂ (Fig. 4).

Fig. 4

Reproduced from Hursán et al. (2016) with permission from The Royal Society of Chemistry

Proposed mechanism for carbon dioxide adsorption and photo-electrochemical reduction on PANI electrode.

Kormányos et al. (2017) reported the application of PANI electrodes for solar fuels production and described the properties of the nanohybrid photoelectrodes based on PANI/SiC. The polymer was prepared by electrochemical method (potentiodymanic cycling) producing a PANI film with band gap energy of 2.9 eV suitable for harvesting visible portion of the light spectrum. Moreover, PANI was able to reduce CO₂ photo-chemically (under 300 W Hg–Xe arc lamp) producing a correspondent photocurrent of 15 µA/cm² (in CO₂-saturated NaHCO₃ 0.1 M solution). The performances of PANI considerably enhanced when used in combination with SiC. The photocurrent obtained with the hybrid (PANI/SiC) electrode was in fact doubled with respect to that obtained with the plain PANI. The influence of PANI film thickness was studied changing the amount of current applied on the electrode during the electro-polymerization process and the highest photocurrent (150 µA/cm²) was obtained with the highest amount of polymer deposited (40 mC/cm²). The electrolysis tests on CO₂ photoreduction show the presence of H₂ and CO in the case of plain PANI electrode; while for PANI/SiC, the same amount of H₂ together with a higher amount of CO and a small quantity of methanol and ethanol were found. The enhancement of photocurrent was mainly due to the higher specific surface area of PANI/SiC electrode and the positive catalytic effect of PANI on electrons transfer from SiC to the substrate. Other than PANI, polypyrrole (PPy) films were used as efficient photoelectrode for solar fuels production (Beladed et al. 2013; Won et al. 2015). Beladed et al. (2013) reported PPy prepared by chemical synthesis showing a direct band gap energy of 1.02 eV and p-type conductivity properties with energy levels suitable for photocathode application in PEC technology. The obtained E_g shows the ability of PPy to absorb the visible part of the solar spectrum. Stability tests were done (under 200 W lamp) in different pH conditions and the results show the high stability of PPy electrodes to corrosion phenomena. In the paper by Won et al. (2015), PPy was prepared by electrochemical polymerization (cyclic voltammetry) controlling the film thickness with the amount of charges that flow on the metal electrode during the synthesis. The PPy films were able to reduce CO₂ to formic acid (under 450 W Xe lamp) at − 0.3 V vs RHE of applied potential with 21.1% Faradic efficiency. This value was improved by using 40-nm PPy film electropolymerized on p-type ZnTe substrate. The composite electrode reaches at − 0.2 V vs RHE the Faradic efficiencies of 37.2% and 13.8% for formic acid and carbon monoxide, respectively. The production rates were 0.11 and 0.04 µmol h⁻¹ cm⁻² for formic acid and carbon monoxide, twice the values obtained for bare ZnTe photocathode. In these conditions, also hydrogen was detected as co-product but in a minor extent compared to bare ZnTe. The positive effect of PPy on ZnTe properties was due to the decrease of electron–hole recombination on the interface PPy/ZnTe (observed by electrochemical impedance experiments) and the corresponding improvement of the electron transfer. The presence of PPy enhances the selectivity on formic acid production and the polymer reduces the recombination of surface adsorbed protons. Finally, only few articles report the application of poly(3,4-ethylenedioxythiophene) (PEDOT) as photoelectrode for PEC application (Hursán et al. 2016; Ghosh et al. 2018). Ghosh et al. (2018) studied the properties of PEDOT nanofibers prepared by chemical polymerization and the influence of gold nanoparticles (Au NPs) added to PEDOT for the realization of new nanohybrid photoelectrochemical active material. During the linear sweep voltammetry (LSV) analyses, the bare PEDOT photoelectrode was able to produce a photocurrent of 3.9 ± 0.05 mA/cm² at potential of 0.8 V vs Ag/AgCl under visible light irradiation (100 mW/cm²). On the other hand, when the nanohybrid PEDOT/Au NPs electrode was tested in the same conditions, the photocurrent was enhanced to 11.2 ± 0.09 mA/cm² reducing the onset potential if compared with bare PEDOT. The stability tests performed on the nanohybrid electrodes show high stability with degradation of just 8% after 240 min. The highest performance obtained for PEDOT/Au NPs was due to the properties of the interface created between these two materials that enhanced the light harvesting, thanks to the co-sensitization of plasmonic Au NPs by multiple states of PEDOT coming from the oligomeric and polymer chain unit. In general, CPs possess a great potential for their application as photoelectrode in PEC systems for CO₂ conversion, but their electronic and catalytic properties must be further enhanced to make this materials competitive with the metal-based ones.

2.2 Graphitic carbon nitride

Graphitic carbon nitride (g-C₃N₄) represents an emerging polymeric semiconductor suitable for solar fuels production. In particular, a lot of work has been devoted to photocatalytic, electrocatalytic and photoelectrochemical hydrogen production; while, an enormous interest for carbon dioxide reduction has recently emerged (Volokh et al. 2019; Wang et al. 2018b; Shen et al. 2018; Lu et al. 2018; Sun et al. 2018; Fang et al. 2018a; Cao et al. 2017; Wen et al. 2017; Ma et al. 2018; Dong et al. 2016; Fang and Wang 2018; Fang et al. 2018b). g-C₃N₄ is a 2D material constituted by a single-layer of sp² hybridized graphitic nanosheets based on carbon and nitrogen atoms bound by van der Waals forces. Moreover, g-C₃N₄ is composed of two cheap, abundant, biocompatible and nontoxic elements with a stable structure based on tri-s-triazine amino rings as the basic units and a minor amount of residual amino groups on the edge of the structure. The chemical structure of g-C₃N₄ is reported in Fig. 5.

Fig. 5

Structures of a tris-s-triazine as the building block of b g-C₃N₄ structure

Figure 5 shows the presence of a nitrogen-rich structure that is suitable for adsorption and activation of a high amount of carbon dioxide. In particular, the triazine nitrogen and amino groups act as Lewis and Brönsted basic sites that are expected to favor carbon dioxide capture and activation. Moreover, the 2D structure can facilitate the electron transfer during the photo-electrochemical reaction and the tunable structure can be modified by changing the precursors and polymerization conditions. Wang et al. (2009) reported C_B and V_B levels at − 1.3 V and + 1.6 V vs NHE at pH 7 indicating the great potentiality of this material for carbon dioxide photocatalytic reduction. The high negative energy position of C_B suggests that the photogenerated electrons possess a high thermodynamic driving force to reduce water, oxygen and carbon dioxide. About CO₂ reduction mechanism, the activation of carbon dioxide was one of the most important step that influences the overall final efficiency. Goettmann et al. (2007) showed the ability of g-C₃N₄ material to reduce CO₂ to CO and recently, Azofora et al. (2016) and Zhu et al. (2017) performed DFT experiments to deeply understand the photocatalytic mechanism. Low Gibbs free energy for the first reaction (H⁺/e⁻) was found for the corrugate g-C₃N₄ with respect to the planar structure, thanks to a reduced electronic repulsion. Moreover, the corrugate structure was found to be selective for the production of CO/CH₃OH. DFT analyses demonstrated that CO₂ molecules adsorb preferably at the two-coordinated N atom instead of the other nitrogen groups. Finally, the band gap value was found to be around 2.7 eV (corresponding to a wavelength of 460 nm), therefore suitable to capture a portion of visible sunlight. The light absorption process occurs by π–π* transition in the conjugated aromatic structure. The g-C₃N₄ can be easily synthesized through polymerization processes of different nitrogen-rich precursors such as melamine, dicyanamide, cyanamide urea and thiourea. On the other hand, the g-C₃N₄ nanosheets should be prepared by an exfoliation procedure from the bulk material (Volokh et al. 2019; Wang et al. 2018a). The precursors and the synthesis approach deeply influence the final properties of the g-C₃N₄ in terms of energy band position, defects, surface area and electronic properties. Thanks to these properties, as described before, a lot of papers and reviews were dedicated to the use of g-C₃N₄ for water splitting reaction and carbon dioxide reduction. In this work only the use of g-C₃N₄ as thin film for PEC application will be discussed. Nowadays, the preparation of high quality g-C₃N₄ film represents one of the main issue for the development of highly efficient PEC systems for carbon dioxide reduction based on this material (Volokh et al. 2019). In fact, normally the g-C₃N₄ is obtained as a powder or in the bulk form that do not allow to obtain films with suitable photoelectrode properties (good substrate adhesion, good electrical connection and electron transport, homogeneity, continuity, stability). To overcome this issue, very promising results were obtained considering a bottom–up approach to prepare the g-C₃N₄ film directly on the substrate. Wang et al. (2018b) reported an accurate review on the different deposition processes. Focusing on carbon dioxide reduction driven by PEC processes, Saraga et al. (2016) studied the properties of g-C₃N₄ thin film and the influence of doping with boron. Considering the n-type semiconductivity, g-C₃N₄ was doped with different boron content (0.5–1.5–3.0 at%) and the as-obtained materials were deposited on FTO (fluorine doped tin oxide) substrate by electrophoretic deposition. For all the samples prepared, the absorption edge was found around 550 nm and the corresponding band gap energy for the sample with boron at 1.5 at% was 2.59 eV. Moreover, thanks to the doping effects, the semiconductivity was changed and the B-doped g-C₃N₄ samples show a p-type conductivity. Mott–Schottky analyses in NaHCO₃ 0.5 M solution (pH 8.6) show a flat band potential for B-doped g-C₃N₄ equal to 2.19 V vs NHE (at pH 0) that corresponds to the valence band. Considering the band gap value, the conduction band was then located at − 0.44 V vs NHE. These energy levels are different from the ones obtained for the pristine g-C₃N₄ film [1.64 V and − 1.03 V vs NHE (at pH 0) for valence and conduction band, respectively] as a consequence of the influence of boron on the electronic properties of this material. The band gap value and energy levels position of B-doped g-C₃N₄ (1.5 at%) show the great potentiality of this thin film as photoelectrode for CO₂ reduction in PEC technology. LSV were done on these photoelectrodes under AM (air mass) 1.5 solar-simulated light at 100 mW/cm² with CO₂-saturated NaHCO₃ 0.5 M solution and the results show the production of photocurrents equal to 4 and 6 µA/cm² at − 0.4 V vs Ag/AgCl for pristine and doped g-C₃N₄, respectively. Long-time tests were performed and no photocorrosion or reduction phenomena were found for both materials (confirmed by XRD analyses) with ethanol as the main reduction product (78% Faradic efficiency) with a small amount of CO and without the presence of hydrogen. Finally, different co-catalysts were added to the films enhancing the photocurrents and the products generation due to the reduced number of recombination and/or the enhancement of ethanol selectivity produced by multi-electrons mechanisms. Another example of doped g-C₃N₄ is related to the work of Tang et al. (2019) where the influence of Eu(III) as doping agent on the properties of g-C₃N₄ was studied. Eu can work as electrons target to suppress the recombination of photogenerated electrons and holes enhancing the photocatalytic activity. As described before, the high rate of charges recombination is one of the most important issues related to g-C₃N_4. Tang et al. (2019) created photoelectrodes on FTO substrate with pristine or Eu-doped (2, 4, 6, 8, 10 wt%) g-C₃N₄ chemically synthesized. All the samples prepared show a strong characteristic absorption in the range between 400 and 800 nm starting from bare g-C₃N₄ with a band gap energy of 2.65–2.60 eV for the Eu-doped sample (8%). Moreover, through cyclic voltammetry analyses, the active surface area was determined for each sample and the results show the highest value of this parameter for Eu-doped g-C₃N₄. Mott–Schottky analyses on all of the samples prepared showed a n-type semiconductive behavior with europium doping enhancing the charge-carrier concentration. Transient photocurrent was acquired for each sample under 300 W Xe lamp with 0.5 M Na₂SO₄ solution under reduction potential. The higher photocurrent was obtained for the sample doped with 8% of europium (4 µA/cm²) with respect to the one obtained for the pristine g-C₃N₄ (2.5 µA/cm²). The main products detected were CH₄ and H₂. The positive effect of Eu-doping was due to the enhanced electrons transfer that finally increases the photocurrent produced. Moreover, the presence of europium improves the charge separation and finally increases the transfer rate of the photogenerated charges. To enhance the photo-electrochemical activity of g-C₃N₄, a nanocomposite with CeO₂ was proposed by Li et al. (2016b). The authors deposited the thin film of pristine g-C₃N₄ and the related nanocomposite with different amount of CeO₂ by electrophoretic process and studied the properties of these photoelectrodes. The band gap energy changed from 2.72 eV for bare g-C₃N₄ to 2.76 eV for the highest amount of CeO₂ passing thorough 2.65 eV with 3% of oxide. All of these materials are, therefore, suitable for solar-driven application. The transient photocurrents were acquired at 0.3 V vs Ag/AgCl under illumination (300 W Xe lamp) with 0.2 M Na₂SO₄ electrolyte. The obtained photocurrent of nanocomposite was 0.75 µA/cm² and this value was higher than the one obtained for the pristine g-C₃N₄ photoelectrode (0.18 µA/cm²). These differences were due to the highest charge transport for the CeO₂/g-C₃N₄ and to the highly efficient separation and transport of the photogenerated charges enhanced by the redox ability of cerium (Ce^3+/4+). The main products detected for all samples were CO and CH₄ with the highest amount obtained for the nanocomposite material (only CO was found when bare g-C₃N₄ was used). Liu et al. (2018a) proposed phosphorous-doped g-C₃N₄ nanotubes prepared by thermal reaction starting from g-C₃N₄ nanosheets as efficient material for carbon dioxide photo-reduction. As expected, the obtained band gap energy for pristine g-C₃N₄ was 2.70 eV, while the value for the phosphorous-doped sample dropped to 2.58 eV. The doping process induces a red-shift adsorption, hence increasing the portion of light spectrum that can be harvested. Photoelectrodes were prepared on metal substrate by simply drop-casting method and the electrochemical and photo-electrochemical properties were determined with 300 W Xe lamp with 0.5 M Na₂SO₄ electrolyte. The Mott–Schottky analyses showed for phosphorous-doped g-C₃N₄ nanotubes and pristine g-C₃N₄ n-type conductivity; while the valence band and conduction band were found 1.80 and 1.75 and − 0.90 and − 1.07 V vs NHE, respectively. These results indicate that both materials tested possess suitable thermodynamic energy for carbon dioxide and hydrogen reduction reactions. Phosphorous-doped g-C₃N₄ nanotubes show higher photocurrent (30 µA/cm²) than the pristine g-C₃N₄ (15 µA/cm²). The electrochemical impedance analyses (EIS) show that the phosphorous doping increases not only the optical absorption but also the charge separation and surface charge transfer. The main product obtained from carbon dioxide reduction was found to be CH₄ derived from CO. At the same time also H₂ was detected.

3 Electrocatalytic CO₂ reduction (CO₂RR)

Electroreduction represents a valuable and highly investigated approach for CO₂ conversion into products of added value. The CO₂ molecule is extremely stable, which makes its electrochemical conversion into target chemicals rather difficult. As a result, relatively high overpotentials are generally required to accomplish the first CO₂RR step, this latter being commonly recognized as the rate-determining step that involves the re-arrangement of a linear CO₂ molecule to a bent radical anion CO ^·−₂ (Duan et al. 2017). Beyond large research efforts devoted to metal-based catalysts, recent years have boosted metal-free heterodoped carbon-based nanomaterials as alternative, sustainable and highly efficient catalytic systems to promote CO₂RR (Genovese et al. 2018; Duan et al. 2017; Liu et al. 2017b; Zheng et al. 2017). While plain C-based materials display only negligible CO₂RR, the inclusion of heteroatoms in their honeycomb structure with the subsequent alteration of their pristine electroneutrality has unveiled a rich and to some extent unexpected electrocatalytic activity towards CO₂ reduction. Both electronic and morphological properties of C-nanomaterials change with the doping using heteroatoms forming high-energy surface sites suitable for CO₂ activation (Chai and Guo 2016).

Due to the multiple-electrons nature of CO₂RR, the process potentially leads to different C1 and/or C2 products (see also Table 1). The control on the nature of CO₂ reduction products necessarily passes through the comprehension of the structure/reactivity relationships of carbon-based materials in the process. Unfortunately, due to the structural and chemico/physical complexity of hetero-doped materials, it appears rather difficult to get unambiguous evidences on the nature of the active sites directly engaged in the transformation. For instance, the structural complexity of N-doped carbon-based catalysts, where different N-configurations coexist at the same time, hamper a clear-cut comprehension of the nature of active sites responsible of the observed electrocatalytic CO₂RR performance. The join combination of experimental outcomes and DFT calculations has been exerted by several research teams with the aim to shed light on this matter often coming to controversial conclusions. Several authors have claimed pyridinic nitrogen sites as the N-configuration of choice for the process to occur due to the basic properties of pyridinic sites capable to foster CO₂ coordination. On the other hand, other reports indicate either quaternary nitrogens or N-neighboring carbon atoms as the positively charged sites for the stabilization of the CO ^·−₂ radical anion coming from the first kinetically sluggish electron-transfer process.

It is evident that further research efforts are still necessary to enlarge the knowledge on this matter before getting a mature catalyst technology opened to the rational design of carbon-based metal-free electrocatalysts and featured by high performance and controlled selectivity.

3.1 Reaction environment: the choice of the electrolyte

The first contribution on the use of carbon-based metal-free electrocatalysts for CO₂RR was reported by Kumar et al. (2013). In their seminal work, authors demonstrated the ability of N-doped carbon nanofibers to promote the selective CO₂ reduction to CO with a Faradaic efficiency (FE) up to 98%. The reactions were carried out in an ionic liquid (IL) (EMIM-BF₄) whose ability to actively participate into the process through the formation of an EMIM-CO₂ intermediate was considered as the driving-force fostering CO₂ reduction to CO. Even if the reaction mechanism could not be elucidated due to the complexity of the N-doped material, high-resolution XPS analyses on the material before and after catalysis led authors to identify C atoms adjacent to pyridinic sites as the active species engaged into the process. More recently, Sun et al. (2016) prepared a N-doped carbon graphene-like material capable to produce methane (CH₄) using ILs as electrolyte. Authors obtained CH₄ with extremely high FE (93.5%) pointing out pyridinic and pyrrole/pyridonic groups as the active sites for the process, while stressing the importance in the use of ILs for driving the process towards methane as reaction target. This work represents the first and to date the unique example of a metal-free material for the CO₂ selective electroreduction to methane. Besides ILs as electrolytes, the only other non-aqueous reaction medium explored in the process was based on methanol (MeOH). Nakata and Einaga’s group developed boron-doped nanodiamonds (B-NDs) capable of promoting CO₂RR in MeOH giving formaldehyde as the major product with a Faradic efficiency up to 74% (Nakata et al. 2014). The choice of MeOH as electrolyte was due to the high solubility of CO₂ (up to five times higher than in water) that obviously fosters the reduction process.

Since these contributions, other examples of doped carbon-based materials have been developed and applied to CO₂RR by giving increased attention to the choice of the reaction environment. In particular, the use of more benign aqueous electrolytes has been addressed to make the process more sustainable, cheap and feasible for future and large-scale applications. Despite of several advantages coming from of the use of an aqueous medium, other critical issues still remain to be properly addressed and overcome as to render this chemical environment really convenient from an electrochemical view-point. Indeed, CO₂ shows only a moderate solubility in water and CO₂RR in aqueous media occurs at higher overpotentials, thus making the side hydrogen evolution reaction (HER) a highly competitive process. As a result, Faradaic efficiency (FE) for CO₂RR and catalyst productivity can be significantly affected, thus resulting in moderate electrocatalytic performance. It is, therefore, mandatory to design catalysts able to reduce CO₂ efficiently already at relatively low potential values, to minimize as much as possible the occurrence of HER. The first example of CO₂RR conducted in aqueous medium (KHCO₃ solution) has been reported by Zhang et al. (2014). They prepared N-doped carbon nanotubes coated with a thin polyethyleneimine (PEI) film as highly selective electrocatalysts for formate production. The combination of N-CNTs with the polymer was found to reduce markedly the overpotential value at which CO₂RR starts (200 mV less respect to plain N-CNTs), thus enhancing FE to formate up to 87% with a current density value of 9.5 mA/cm². The PEI overlayer was supposed to play a double role: (1) concentrate CO₂ at the electrode surface and (2) stabilize the CO ^·−₂ radical anion, the latter being again the presumed intermediate formed after the first rate-determining electron transfer. Authors’ conclusion was that the basic character of PEI-coated N-CNTs was mandatory to ensure a relatively high CO₂ concentration at the catalyst surface, a key pre-requisite for the CO₂RR process to occur efficiently.

Among aqueous electrolytes, bicarbonate solutions are the most commonly used for running CO₂RR. However, its use has raised several remarks regarding its effective role in the process. Indeed, bicarbonate can be considered itself as a CO₂ source through the carbon dioxide–bicarbonate equilibrium. Liu et al. (2015) observed the dependence of CO₂ reduction rate on HCO₃⁻ concentration, with current density increasing linearly with the solution molarity. More recently, Hursán and Janáky (2018) have demonstrated by means of isotopically labelled ¹³CO₂ and H¹³CO₃⁻ that bicarbonate does not directly contribute to the generation of CO₂RR products but it rather behaves as a CO₂ reserve by regularly supplying CO_2(aq) dissolved at the electrode surface.

3.2 Hetero-doped (N, S, B) C-nanomaterials in CO₂RR

3.2.1 The 2 electrons reduction process to carbon monoxide (CO)

Carbon monoxide (CO) represents the most commonly observed reduction product with carbon-based CO₂RR electrocatalysts. Besides the above-mentioned contribution by Kumar et al. (2013) where N-doped carbon nanofibers were used to promote the CO₂RR to CO in IL with high Faradaic efficiency (98%), many other N-doped materials have been prepared and successfully applied to the selective carbon monoxide production. Wu et al. (2015) and Sharma et al. (2015) prepared N-doped carbon nanotubes (N-CNTs) by chemical vapor deposition (CVD) with interesting electrocatalytic activity in the CO production using an aqueous bicarbonate solution as electrolyte, with a maximum Faradaic efficiency of 80% already at low potential values (− 0.78 V vs. SHE). The combination of experimental results with XPS analyses and DFT calculations led the authors to identify both N-pyridinic and N-graphitic (quaternary) as the active sites for the CO₂ activation. On the other hand, pyrrolic groups were supposed to have no impact on CO₂RR. Later, Xu et al. (2016) synthesized N-CNTs by calcination of a mixture of poly(diallyldimethylammonium chloride) and oxidized CNTs to be tested as CO₂RR electrocatalysts. CO was found to be the main reaction product with FE up to 90%. DFT calculation recognized quaternary nitrogen sites as the main players for the observed activity due to their ability to stabilize CO ^·−₂ intermediate by reducing the free energy associated with the first electron-transfer process. Other than monodimensional (1D) N-CNT systems, several 2D and 3D N-doped composites have been developed and exploited in the process. Lu et al. (2016) prepared a graphitic carbon nitride (g-C₃N₄)/MWCNTs composite by means of a high-temperature polycondensation of dicyandiamide with oxidized CNTs. Authors attributed the catalyst ability to selectively reduce CO₂ to CO (FE = 60%) to a synergistic effect between carbon nitride and MWCNTs. In particular, they invoked the formation of active carbon–nitrogen bonds along with the high specific surface area and material conductivity as the key factors for the observed catalytic outcomes. An impressive CO₂RR activity was reported later by Jhong et al. (2017) that prepared a g-C₃N₄/MWCNTs composite by depositing a thin layer of nitride on a pure MWCNTs surface. With FE as high as 98% in the CO₂RR to CO and a maximum current density of 90 mA/cm², this composite material ranks among the most efficient metal-free carbon-based catalysts for carbon monoxide production reported so far. Most recently, a 2D ultrathin g-C₃N₄ layer has been developed by a hydrothermal–thermal exfoliation method and reported to boost CO₂RR to CO as the main product (FE = 80%) (Zhang et al. 2018b). The polarized surface induced by the ultrathin nanostructure was claimed by authors to play a key role in enhancing CO₂RR activity. Looking finally at three-dimensional carbon materials, 3D hierarchical porous N-doped carbons prepared by pyrolysis of biomasses (Li et al. 2017a) and 3D graphene foam grown by CVD on a template (Wu et al. 2016a) have been reported for the selective CO₂RR to CO with comparable selectivity (FE = 85%). In both cases, authors have claimed pyridinic nitrogen defects as essential for the CO₂ activation and conversion to CO.

From the literature overview outlined above, it is clear that the debate on the structure/composition/reactivity relationship at the origin of the catalytic performance of these metal-free systems in CO₂RR is still open. Despite many research efforts both on experimental and theoretical ground, the structural complexity of doped nanocarbons along with their variable N-composition (coexistence of different N-configurations as a function of the material synthesis) hampers the unambiguous identification of CO₂RR active sites. On this ground, a rational bottom–up synthesis based on the exohedral chemical functionalization of plain C-nanomaterials with defined N-containing functionalities has been recently developed by some of us (Tuci et al. 2018). Such a non-conventional approach to the N-decoration of carbon nanostructures has allowed a fine-tuning of N-containing functional groups at the nanomaterial outer surface, thus offering unique hints to the comprehension of their role in the underlying CO₂RR mechanism. In particular, basic NH-aziridine groups have been proven to promote the selective CO₂ reduction to CO with markedly higher performance compared to that of classic pyridine groups (often claimed in the literature as the active species for the process). The higher activity of the former is basically ascribed to their inherent ability to chemisorb CO₂ through the formal aziridine-carbonation reaction. Such an effect is claimed to drive the catalytic process by changing CO₂ geometry from linear to bent (in the carbonation product) and thus favoring the kinetically sluggish first electron transfer process for the generation of the CO ^·−₂ species. Finally, the latter is stabilized by the positively charged carbon atoms neighboring the N site before undergoing a further 1 electron reduction step with the participation of two H⁺ (Fig. 6).

Fig. 6

Reproduced from Tuci et al. (2018) by permission of The Royal Society of Chemistry

CO₂ electroreduction process via CO₂(aq) chemisorption to NH-aziridine groups (amine carbonation). N 1s core regions and relative fits for the high-resolution XPS spectrum of catalyst before (left) and after (right) the electrocatalytic run are shown in the upper part of the scheme. N 1s components at higher binding energies (401.3 and 401.0 eV) are attributed to the formation of amine carbonate whereas components at 399.1 eV belong to free aziridine groups.

Beyond nitrogen as dopant, just a few examples of carbon nanomaterials doped with other heteroatoms have been reported so far for promoting CO₂RR to CO. Sulfur was chosen as a carbon dopant with very modest result. Indeed, only 2% of FE towards CO was found with S-doped nanoporous carbon; whereas, a slight increase (up to 11%) was reported for N and S co-doped materials at work in the process (Li et al. 2016b). Better results have been achieved with F- and P-doped materials. As for the former, a F-doped sample was obtained by a simple pyrolysis of a commercial carbon with PTFE as the fluorine source (Xie et al. 2018). The catalyst exhibited high FE to CO (up to 90%) with the positive charged carbon atoms identified by authors as the active sites for the process. Indeed, due to the high electronegativity of fluorine nuclei, a relevant positive charge density on the neighboring carbon atoms was supposed to strongly bind the CO₂ reduction intermediates, ultimately favoring CO₂RR and suppressing the competitive HER. Finally phosphorous, as a heteroatom possessing the same valence electrons as nitrogen, was explored as dopant in C-nanostructures. On this ground, P-doped onion-like carbons have been reported by Liu et al. (2018b) as efficient catalysts for CO₂ reduction to CO with FE up to 81%. Authors point out P–C functionalities as more active than P–O ones in stabilizing CO₂ reduction intermediates and in enhancing the electronic transfer. Overall, P-doped carbons demonstrate to be valuable alternatives to N-doped nanocarbons offering comparable performance for CO production.

3.2.2 The 2 electrons reduction process to formate (HCO₂ ⁻) and formaldehyde (HCOH)

Formate (HCO₂⁻) represents an important 2 electrons reduction product of CO₂. Beyond the above-mentioned paper by the Zhang et al. (2014), there are just a few examples of N-doped nanocarbons employed in the CO₂RR to give formate as the main reduction product. Wang et al. (2016) and Wang et al. (2017) have described the use of N-doped graphene and N-doped nanoporous carbons/carbon-nanotube composite membranes for the CO₂RR to HCO₂⁻ with FE up to 73 and 81%, respectively. Both papers agree on the central role played by carbon atoms neighboring pyridinic sites for the CO₂ coordination without any direct involvement of N atoms. Even if authors did not give an explanation on why their N-doped systems produce formate instead of the most common CO obtained with similar materials (see Sect. 3.2.1), they invoke a reaction mechanism where CO₂ coordinates to the catalyst (and in particular to the carbon atom adjacent to N-pyridine) through one of its oxygen atoms (Fig. 7).

Fig. 7

Reproduced from Wang et al. (2017) with permission from Wiley–VCH

Proposed mechanism of the CO₂RR in 0.1 M KHCO₃ aqueous solution by HNCM/CNT membrane.

This reactivity accounts for a radically different coordination mode of CO₂ respect to that postulated for N-doped nanocarbons engaged in the CO₂RR to CO, where the substrate (CO₂) was supposed to interact with the catalyst through its carbon atom (see Sect. 3.2.1 and Fig. 6) (Sharma et al. 2015; Tuci et al. 2018; Li et al. 2016a). Therefore, the different CO₂ coordination mode can be invoked as a key factor to control the process selectivity towards CO or formate. This hypothesis would also explain CO₂RR performance observed with differently doped systems, such as with B-doped carbon nanomaterials, where the electron-poor nature of the heterodopants is supposed to foster its coordination with one of the more electron-rich oxygen atoms of CO₂.

To date, only a few examples of B-doped nanocarbons have been exploited for CO₂ electroreduction. Among them, B-doped graphene prepared by pyrolysis of GO with boric acid has been reported to produce HCO₂⁻ with a FE of 66% (Sreekanth et al. 2015). Two research groups have independently demonstrated that B-doped nanodiamonds (B-NDs) were electrocatalytically active materials for CO₂RR to formaldehyde and formate (main by-product) with FEs of 74 and 26% or 54% and 14%, respectively (Nakata et al. 2014; Liu et al. 2017b). Taking into account that formaldehyde (HCOH) derives directly from the formate reduction (Nakata et al. 2014), it is evident that B-doped samples foster the HCO₂⁻ conversion once again through the coordination of its oxygen atom to the electron-poor B-sites. To the best of our knowledge, B-doped nanodiamonds are unique materials for the CO₂RR to HCOH, thus unveiling that B-doping of sp³ hybridized nanocarbons are somehow suitable systems for promoting the further HCO₂⁻ reduction to formaldehyde (Nakata et al. 2014). On the other hand, there are no examples of N-doped nanocarbons reported to date as active electrocatalysts for the generation of HCOH as the main species.

Therefore, it can be inferred that the distinctive behavior of B-doped C-nanomaterials in CO₂RR mirrors with the different coordination mode of the substrate (CO₂) at the catalyst active sites. Boron is an electron-poor element respect to carbon and it prefers coordination to one oxygen atom of CO₂. On the other hand, the electron-rich N sites in N-doped C-nanocarbons will bind CO₂ through its carbon atom, preferentially. Even if any definitive conclusion on the underpinning CO₂RR mechanism is formally hampered by the complexity of the electrocatalytic materials, it seems reasonable to postulate that different coordination modes of CO₂ are responsible of the observed process selectivity towards CO or formate.

3.2.3 CO₂RR to C2 products

An important class of CO₂ reduction products is represented by C2 compounds such as acetate (CH₃COO⁻), ethanol (CH₃CH₂OH) or light hydrocarbons. C2 products are obviously highly attractive targets although their selective synthesis from CO₂ remains rather difficult to be addressed because the process requires C–C coupling step in addition to the classical CO₂ reduction process. The following sections summarize the literature achievements in this challenging research area.

3.2.3.1 CO₂RR to acetate ions

The first example of a C2 compound synthesized by electrochemical CO₂ reduction has been reported by Liu et al. (2015). They designed a N-doped nanodiamond/Si rod array enabling CO₂ conversion to acetate (FE = 77.6%) together with a minor amount of formate (FE = 14.6%). Although further studies are needed to better understand the origin of acetate selectivity in the process, authors postulated that N-Csp³ functionalities in N-NDs make the kinetic for a new C–C bond formation faster than that of the protonation of the CO ^·−₂ intermediate to give formate. Two years later, the same group reported on analogous N-NDs for acetate production with a FE of 62% (Liu et al. 2017a). Interestingly, these authors also demonstrated the radically different behavior of the boron-doped counterparts (B-NDs). Indeed, the latter allowed CO₂ reduction to C1 products only (i.e., formaldehyde and formate, see also Sect. 3.2.2). Accordingly, they inferred that N doping in NDs was somehow responsible for the occurrence of C–C coupling in the process, thus paving the way to the direct CO₂ electrochemical conversion to C2 compounds.

3.2.3.2 CO₂RR to ethanol

One of the more desired CO₂ reduction products is represented by ethanol, an important raw material and a promising renewable fuel. To date, the highest FE towards EtOH (93.2%) has been reported by Liu et al. (2017b) with B–N co-doped NDs. Their material has been found to efficiently promote the C–C coupling to EtOH, while limiting both C1 products formation and HER. With the support of DFT calculations, authors demonstrated that their catalyst superior performance for EtOH production originates from the synergistic effect of B–N co-doping that promote the coupling of two neighboring CO* species to give C2 products. Their conclusions confirm once again the supposed pivotal role of N-doping in NDs for promoting the C–C coupling (see Sect. 3.2.3.1).

In the same period, metal-free cylindrical mesoporous N-doped carbons (c-NC) synthesized by Song et al. (2017) using a soft-template method have been proven to convert CO₂ into ethanol with FE as high as 77% at low potential values (0.56 V vs. RHE). Noteworthy, the only by-product observed is represented by H₂ due to HER, with no other reduction products coming from CO₂. Again, the key feature for the selective production of ethanol lies on the occurrence of a C–C coupling, while limiting the competitive CO and/or other C1 products formation. Authors have assumed that pyridinic nitrogens are the preferential activation sites for CO₂ to give CO* intermediates stabilized by the material electron-rich cylindrical surface, which supplies the electrons for CO* dimerization to ethanol while avoiding the undesired release of CO as reduction product. To demonstrate the superiority of such a cylindrical mesoporous structure, the inverse mesoporous N-doped carbon (denoted as i-NC) that preserves similar pore parameters and N-content, was also prepared for comparison (Fig. 8). Over i-NC, the maximum ethanol FE reached only 44% with formation also of CO and H₂ as by-products.

Fig. 8

Reproduced from Song et al. (2017) with permission from Wiley–VCH

A Illustration of c-NC and i-NC for CO₂ electroreduction. B TEM images of (a1, a2) c-NC and (b1, b2) i-NC viewed along (a1, b1) [110] and (a2, b2) [100] directions.

3.2.3.3 CO₂RR to light hydrocarbons

Besides methane (CH₄) as the simplest hydrocarbon prepared from CO₂RR in 2016 (see Sect. 3.1) (Sun et al. 2016), only one example of C2 hydrocarbon has been reported to date with appreciable FE values. Wu et al. (2016b) have demonstrated that N-doped graphene quantum dots (NGQDs) electrocatalysts lead to ethylene (C₂H₆) as the major CO₂RR product, although with moderately low FE (31%). Indeed, products distribution appears relatively broad with the formation of EtOH and acetate other than C1 products (CO and formate) and C3 compounds such as n-propanol. The authors indicate, among the various N-configurations available in their material, N-pyridinic as the most active site for the process to occur. In particular, they postulated N-pyridinic edge sites as those engaged in the new C–C bond formation. As a result, the higher the amount of pyridinic edge sites in NGQDs, the higher the yield of C2 products.

4 Conclusions

This review aims at describing the state-of-the-art carbon-based nanomaterials as metal-free catalysts for CO₂ photo-electrochemical (PEC) and electrochemical reduction (CO₂RR). Despite a certain number of interesting reports appeared in the literature up to date, one of the main issues still pending within these reduction technologies deals with the development of efficient and robust catalytic materials capable to operate under aqueous environments for the highly selective CO₂ reduction to molecules of added value with high productivity while limiting the occurrence of side reactions (i.e., HER).

Conducting polymers and g-C₃N₄ represent attractive materials for PEC applications due to their low cost, abundance and low toxicity. However, some issues such as low photocurrent, charges recombination and films inhomogeneity need to be overcome to get highly efficient materials for CO₂ activation and conversion. For example, to reduce the production costs and increase the sustainability of the overall process, additive manufacturing technology should be applied while producing organic photoelectrodes. The development and optimization of these processes also passes through the use of ink-jet or spray-coating preparation methods. Doping processes and low-cost polymerization methods (such as electropolymerization) should be used for the realization of highly efficient and stable photoelectrodes; while PPy and PEDOT should be considered in more details also for carbon dioxide conversion. In fact, PEDOT possesses a higher potentiality for carbon dioxide reduction by PEC technology due to the presence of sulfur atoms on its polymeric chain. The main limitations on the application of g-C₃N₄ are the fast recombination of the photogenerated charges and poor electronic conductivity as well as the limited absorption of visible light. These issues can be overcome building highly conductive nanohybrid materials with a metallic nanoparticles coating without reducing the band gap energy of g-C₃N₄. At the same time, the lower photocurrent produced should be improved by doping processes, realization of new interfaces and using a co-catalyst material. Also the deposition process is crucial and a new technique should be created for the film deposition directly on the conducting substrate.

As for CO₂RR, one main issue is represented by selectivity towards a target product and on this ground, some general trends can be summarized on the basis of the most recent literature outcomes. CO is the most common reduction product when hetero-elements such as N, P or F are used as dopants in C-based nanostructures. On the other hand, when the electropositive B atom is introduced in the C-network, formate represents the preferred 2-electrons reduction product. No B-doped nanocarbons have been reported to date as suitable electrocatalysts for CO₂RR to CO, while formaldehyde has been observed as the main CO₂RR product with B-doped nanodiamonds as electrocatalysts only, unveiling the pivotal role played by the B/C-sp³ combination to give HCOH selectively. No carbon-based metal-free catalysts have been reported to date for methanol synthesis with appreciable FEs, while methane has been obtained with N-doped nanocarbons only conducting CO₂RR in IL electrolytes that are supposed to play a non-trivial role in driving selectivity towards CH₄.

C2 compounds are certainly the more attractive reduction products in CO₂RR, although just a few examples of metal-free catalysts have been reported to date to foster the transformation efficiently. Even if the reaction mechanism is still far from being fully explained, one of the main challenges to be addressed to obtain C2 products is to prepare electrocatalytic materials featuring an ideal binding strength towards CO₂ reduction intermediates and products. Taking into account CO as the most common 2 electrons reduction product, electrocatalysts that bind CO weakly will produce it as the major product because it quickly moves away for the catalyst active sites. On the other hand, catalysts that bind CO too strongly will result in poor catalytic activity due to poisoning effects; this is not usually encountered with metal-free carbon-based materials but it is a well-known problem encountered with noble metal-based electrocatalysts (e.g., Pt). The key point is, therefore, to bind CO with the “just right” strength to avoid its fast release from active sites and allow its further evolution into C2 compounds. This key issue has to be accomplished with a tricky combination of selected dopant species and tailored chemico-physical/morphological properties of nanocarbon materials. It is evident that additional research efforts are needed for the rational design of robust and efficient catalysts with targeted selectivity towards a desired product. To this aim, one main step forward is represented by a deep comprehension of the mechanisms at work in the CO₂RR along with the elucidation of the structure/composition/reactivity relationships between doped nanocarbons and CO₂. A rational bottom–up synthesis of hetero-doped carbon nanomaterials combined with DFT calculations can give an important boost to get new progresses in the field and to accomplish many of the ambitious goals of this technology.