Layered-Defect Perovskite K₃Bi₂X₉ (X = I, Br, and Cl) Thin Films for CO₂ Photoreduction: An Analysis of Their Pseudocatalytic Behavior

Abstract

Lead-free layered-defect perovskite K₃Bi₂X₉ (X = I, Br, and Cl) films were proposed as efficient photocatalysts for the CO₂ reduction reaction (CO2RR) to obtain clean and sustainable formic acid (HCOOH), a widely used feedstock in the industry. The films exhibited high crystallinity, hexagonal morphologies, and visible light absorption, which were modified by proportionally increasing the diameter of the X anion. The obtained photocatalytic activities showed values of 299 µmol h⁻¹ (K₃Bi₂Br₉), 283 µmol h⁻¹ (K₃Bi₂I₉), and 91 µmol h⁻¹ (K₃Bi₂Cl₉). However, the stability of the films is an important parameter that must be solved; therefore, three strategies were implemented—one with an intrinsic approach (solvent engineering) and two others with an extrinsic focus (substrate modification and heterojunction engineering). These modifications favored yields of up to 738 µmol h⁻¹ and constant production over 6 h, demonstrating that the perovskite maintains continuous HCOOH generation. The analysis of the reaction medium showed the degradation of the material structure to BiOI and K⁺, which could have enhanced its affinity towards CO₂. In this manner, the degraded perovskite (K₃Bi₂I₉/BiOI) might still react with the CO₂ to generate HCOOH in an aqueous medium under visible light, showing pseudocatalytic behavior.

1. Introduction

The ever-increasing presence of CO₂ in the atmosphere has raised many sustainability concerns worldwide and has led to an extenuating evolution in climate politics, as this pollutant is the main component of the greenhouse effect gases that consequently cause climate change [1]. CO₂ is also the principal exhaust gas in processes that involve fossil fuel combustion. Its two high-energy C=O bonds (of approximately 750 kJ mol⁻¹) render this compound thermodynamically stable, and it is this high stability level that makes anthropogenic CO₂ mitigation such a difficult task [2]. Mainly, the sustainable policy-making has been focused on (i) environmental legislation to reduce the emission of greenhouse gases [3]; (ii) CO₂ absorption and storage strategies that prevent this pollutant from leaking to the atmosphere [4]; (iii) the utilization of alternative energy sources, which could help to diminish the need for fossil fuel combustion processes [5]; and (iv) the use of CO₂ as a chemical feedstock to obtain various industrial products [6]. Most of these strategies are partially effective at attenuating this global-scale pollution crisis. Regardless, the most promising solution strategies rely on sustainable technologies that can effectively terminate the presence of the most ubiquitous greenhouse gas in the environment. For this matter, the photocatalytic CO₂ reduction reaction (CO2RR) remains one of the most attractive mitigation strategies due to the possibility to utilize CO₂ as a source to produce clean and renewable fuels such as methane (CH₄), methanol (CH₃OH), carbon monoxide (CO), and formic acid (HCOOH) [7] from the conversion of this waste product [8,9]. These sustainable fuels can be produced with outstanding efficiency if the physical and chemical properties of the semiconductor materials (photocatalysts) are adequate for this purpose. For instance, the transport and separation of charge carriers can impact photocatalytic reactions and modulate the selectivity towards valuable products. Moreover, a high quantum yield is necessary for the emission of sufficient electrons for carrying out photocatalysis, and adequate energy band potentials are required for the reduction and oxidation reactions. Taking these factors into consideration, there are many suitable materials that have been extensively used, such as the layered double hydroxides (LDH) [10], g-C₃N₄ [11], and SrTiO₃ [12]. However, despite the good results obtained with these materials, there is a constant need for the exploration of new functional materials with beneficial properties to carry out an outstanding photocatalytic CO2RR at the industrial level, contemplating both environmental and economic sustainability. This is the case for perovskite-structured photocatalysts, which have been increasingly employed in successful photocatalytic systems in recent times.

The general formula for perovskites is ABX₃, where A and B are organic or inorganic cations (FA⁺, MA⁺, Pb²⁺, Ti⁴⁺, etc.) with different radii (A > B), while X comprises oxide (O₂⁻) or halide (I⁻, Br⁻, Cl⁻ or F⁻) anions. This variation results in the distortion of the original cubic lattice to diminished symmetries, such as orthorhombic, hexagonal, triclinic, or monoclinic [13]. Additionally, this diversity of lattices give rise to variations in the properties of the perovskites (e.g., electronic structure, dipole moment), which consequentially alter the generation and transport of charge carriers for the photocatalytic process [14].

There are two major subgroups into which perovskites can be classified—metal oxide perovskites (MOPs) and metal halide perovskites (MHPs). Therefore, due to the perovskites’ remarkable structure–property relationship, it is to be expected that there will be significant changes between the characteristics of these subgroups. MOPs are formed when the X anion in the perovskite’s ABX₃ structure is occupied by an O₂⁻ ion, while the A and B cations have oxidation state values that sum a total of 6+ (being either 2+ and 4+ or 1+ and 5+ for A and B, respectively). A notable aspect of these materials is that they have previously shown photocatalytic activity towards the CO2RR. For instance, Sr-doped NaTaO₃ was proposed to produce 352 µmol of CO g⁻¹ h⁻¹ with 2% Ag as a cocatalyst and 0.1 M NaHCO₃ as a hole scavenger [15]. Moreover, SrTiO₃ had a production rate of 369 µmol g⁻¹ h⁻¹ using Au and Rh in a heterojunction [16], while C-doped LaCoO₃ generated 95 µmol g⁻¹ h⁻¹ with Na₂CO₃ under UV–visible light [17]. As for MHPs, these materials might show additional tilting in their structure (compared to MOPs) due to variations in the X anion. However, when the Pb²⁺ cation occupies the B site, the obtention of stable cubic-structured MHPs becomes theoretically feasible [18]. There are reports that showcase the good properties of lead-based MHPs that result in remarkable performances in the photocatalytic CO2RR [19]. Most of these reports are focused specifically on the CsPbBr₃ perovskite, due to its adequate reduction potentials [20]. However, the highly volatile ionic interaction of MHPs also favors their transformation into other phases [21], which can be beneficial for the exploration of new materials for photocatalysis, resulting in the efforts made to render these materials more stable and to resolve the toxicity issues derived from the development of the most promising MHPs—lead halide perovskites [22]. Among the attempts to replace lead in the structure of halide perovskites for this matter, the most notable reconfigurations rely on the use of elements of similar ionic radii to produce minimum alterations to its lattice, thereby retaining the valuable properties of lead halide perovskites without the toxicity and stability disadvantages. These alternative materials are commonly referred to as lead-free halide perovskites (LFHPs). These can be formed with different elements that achieve low toxicity while maintaining low costs of production. In this sense, the use of trivalent atoms (e.g., Bi³⁺, As³⁺, Sb³⁺) can accomplish this requirement, as they can be used to replace the divalent Pb²⁺ in the B site of the ABX₃ formula, resulting in a different configuration—A₃B₂X₉. Of these trivalent atoms, bismuth has been shown to improve visible light absorption [23], which has led to good results in the CO2RR [24]. Additionally, the use of bismuth-based LFHPs for this application has not yet been extensively explored, and of these materials Cs₃Bi₂X₉ (X = I, Br, Cl) has been the most utilized, as the Cs⁺ cation has previously shown good integration in the lattice of this configuration, as opposed to organic cations [25]. The highest generation of solar fuels using this perovskite has been obtained with a Cs₃Bi₂I₉/CeO₂ heterojunction, reaching production rates as high as 135 µmol h⁻¹ [26]. Another cation with the possibility to form the A₃Bi₂X₉ configuration is potassium (K). This is a widely abundant element with relatively insignificant effects on the environment. However, the use of K₃Bi₂I₉ for the CO2RR has not been investigated due to its low resistance to environmental factors, such as moisture, high temperatures, or the presence of atmospheric oxygen [27]. For this reason, the K₃Bi₂I₉ perovskite has only been tested for solar cells [28]. Therefore, in this study, the use of the K₃Bi₂X₉ (X = I, Br, Cl) LFHPs for the photocatalytic CO2RR with different stability strategies was investigated.

Theoretically, K₃Bi₂I₉ crystallizes in the monoclinic C2/c space group (Figure 1) [27]. The atomic arrangement of the potassium cations and the iodide anions can be described as a sequence of densely packed, slightly distorted KI₃⁻² layers. There are two inequivalent K⁺ sites and five inequivalent I⁻ sites. Meanwhile, Bi³⁺ is bonded to six I⁻ atoms to form corrugated incomplete face-sharing BiI₆ octahedra, which only allow for an oversized irregular gap for the K⁺ site in the center, altering the symmetry of this material.

Similar to other halide perovskites (ABiX₃; where A = organic and inorganic cations), K₃Bi₂I₉ may have low resistance to various environmental factors (e.g., moisture, oxygen, irradiation, etc.), mainly due to the ionic interactions that connect the atoms of MHPs, which lead to a facilitated interaction of water or oxygen to disrupt their lattice and spontaneously form vacancies upon irradiation [29]. There are two different approaches for tackling these issues—intrinsic and extrinsic [30]. The intrinsic approach relies on the modification of the internal structure of materials to strengthen its resistance to degradation. Some examples of these modifications are surface passivation [31], recrystallization [32], and cation exchange [33]. On the other hand, the extrinsic approach focuses on the interface of a material with its reaction medium. For this, some of the most common strategies are heterojunction engineering [34] and surface encapsulation [35]. However, it is not clear how these strategies impact the structural and optical properties after several evaluation cycles. Considering this, here, layered-defect LFHP perovskite K₃Bi₂X₉ (X = I, Br, Cl) films were synthesized through a reproducible method and evaluated for the photocatalytic CO2RR. The stability of the perovskites was investigated through the implementation of three strategies following intrinsic and extrinsic approaches, with the aim of guaranteeing their long-term implementation. The results could serve as the basis for the development of low-cost, stable, and efficient LFHP photocatalysts with inconsequential environmental impact for sustainable applications.

2. Materials and Methods

2.1. Materials

The BiI₃ (99%), KBr (99%), BiBr₃ (98%), BiCl₃ (98%), N,N-Dimethylformamide (DMF, 99.8%), and dimethylsulphoxide (DMSO, 99.5%) were purchased from Sigma-Aldrich (St. Louis, MO, USA). The KI (99%) and KCl (99%), were purchased from Fermont (Quebec, QC, Canada). The isopropyl alcohol (IPA; 99%), dietylether (DE; 99.7%), H₂SO₄ (96%), and H₂O₂ (30%) were purchased from DEQ (Monterrey, Mexico). All chemicals were used as received, without further purification. The water used for the experiments was purified using a bi-distillation method.

2.2. Synthesis of K₃Bi₂X₉ (X = I, Br, and Cl)

LFHP perovskites with a formula of K₃Bi₂X₉ (X = I, Br, Cl) were synthesized by mixing 1.5 M KX and 1 M BiX₃ (as shown in Equation (1)) in 1 mL of a mixture of N,N-dimethylformamide (DMF) and dimethyl sulfoxide (DMSO) in a ratio of 7:3 v/v. The integration of the halide reactants generated the following reaction in the precursor solution:

3KX + 2BiX₃ → K₃Bi₂X₉

(1)

The solutions were sonicated for 15 min and stirred in a magnetic stirrer at 70 °C for 2 h. Subsequently, the mixture was cooled to room temperature and then filtered using a PTFE syringe (pore size = 0.2 mm) to obtain a light red (I), greenish-yellow (Br), or white (Cl) solution. The glass substrates used for the deposition of the solution were previously cleaned by immersing them in a piranha solution, which consisted of concentrated H₂SO₄ and H₂O₂ in a 4:1 ratio. The substrates were then rinsed with distilled water and dried at room temperature. Prior to the spin-coating deposition step, the filtered solution and the glass substrates were heated to 70 °C for 10 min. As a final step, the solution was deposited on the substrates at 2000 rpm for 30 s on the spin-coater and then annealed on a hot plate to 100 °C for 2 min.

2.3. Application of Stability Strategies

It is essential to develop stability enhancements to the structure of perovskites to guarantee their sustainable long-term implementation. As previously stated, different intrinsic and extrinsic stability strategies were implemented in this study, with the aim of ensuring a consistent performance during several photocatalytic cycles. These methodologies were carried out in the following manner.

(a)

Intrinsic approach

To eliminate impurities in the film and strengthen the structure of the synthesized materials, a recrystallization method was performed. For the correct elimination of impurities, the solvent must be miscible with the precursor solution’s solvent (and non-miscible with the crystallized material) to preserve the deposited perovskite film. For this matter, the solvent must also possess low polarity and viscosity. The properties of the chosen solvents for the recrystallization of the K₃Bi₂X₉ perovskites are shown in

Table 1. Miscibility and properties of various commonly used solvents.

Properties	Diethyl Ether (DE)	Isopropyl Alcohol (IPA)
Polarity index	2.8	3.9
Boiling point (°C)	35	82
Viscosity (cP)	0.32	2.3
Miscibility with DMF	Miscible	Miscible

. Isopropyl alcohol and diethyl ether have useful characteristics for recrystallization, as they are both miscible with DMF, the main component of the perovskite precursor solution. These solvents also have low polarity rates and adequate viscosity. Therefore, in a final spin-coating step (under the same conditions of 2000 rpm and 30 s), the K₃Bi₂I₉ films were washed with IPA and DE separately. The films were then heated on the griddle for 20 min and stored. The annealing times were reduced based on our observations.

(b)

Extrinsic approach

In this study, two stability strategies with an extrinsic approach were considered to assure good stability in the perovskite film. The first one consisted of changing the film’s substrate and growing the K₃Bi₂I₉ layers on flexible earth-abundant mica substrates via spin-coating deposition. Conversely, the second one involved the design of a heterostructure with BiOI, which would grant the perovskite with higher stability and photocatalytic activity in visible light. It is worth mentioning that the determination of an adequate number of layers of the deposit was carried out on the film prior to this modification, with the purpose of maximizing the photocatalytic activity of K₃Bi₂I₉ before adding the effect of any cocatalyst.

• Growth of K₃Bi₂I₉ in earth-abundant mica substrates: The adherence and stability of the films was improved via the substitution of the substrates using flexible (and natural) mica. These substrates were cut to the same size as the previously used glass substrates (7.6–1.3 cm), and their frames were covered with mildly adhesive tape to prevent leakage through the edges. After these preparation steps, the spin-coating method was performed as stated above (2000 rpm; 30 s).
• Formation of m-BiOI/K₃Bi₂I₉ heterostructures: Based on the hypothesis that a heterostructure of BiOI with the K₃Bi₂I₉ perovskite would cause self-assembled co-sharing of the I⁻ ions in both materials (thereby enhancing the film adhesion), a microwave-assisted methodology was carried out. For this purpose, the BiOI film was grown on mica substrates using a microwave–hydrothermal method, as previously reported by our research group [36]. The utilized equipment was a Mars 6^® synthesis research microwave into which a Teflon reactor was introduced. This reactor contained a single mica film, BiI₃ and Bi(NO₃)₃ as reactants, and a solvent mixture of isopropyl alcohol and glycerol. The solvents and reactants were added separately in a 2:1 v/v ratio to each other. The reaction conditions were 600 W of power, a temperature of 180 °C, a reaction time of 30 min, and a heating step of 10 °C/min. Finally, the K₃Bi₂I₉ spin-coating deposition procedure was performed on this film, as described previously.

A summary of these experiments is schematically represented in Figure 2.

2.4. CO₂ Photoreduction Experiments

The photocatalytic activity of the materials was evaluated at room temperature utilizing the experimental arrangement shown in Figure 3. In short, a Pyrex glass reactor was employed, into which the perovskite film was introduced along with 100 mL of water stirred at 200 rpm. In addition, an LED lamp (µ > 400 nm) placed on the side of the reactor was used as a source of visible light irradiation. Prior to each experiment, the reactor was saturated with a continuous flow of CO₂ to eliminate the presence of environmental air. This gaseous flow was maintained throughout the experimental runs.

2.5. Characterization Techniques

For X-ray diffraction (XRD) characterization, the employed equipment was a Malvern Panalytical Empyrear Dy1681 model (Malvern, Worcestershire, UK) with Cu Kα radiation (4 kV, 100 mA) and an angular resolution of 0.26° (FWHM on LaB6). The measurements were carried out over a range of 2θ = 5–70°, with a step size of 0.02° and a time of 0.3 s for each step. The morphological characterization was achieved using scanning electron microscopy (SEM) in a JEOL JSM-6490LV microscope (JEOL USA, Peabody, MA, USA) coupled with an Oxford model energy-dispersive detector (EDS). On the other hand, diffuse reflectance spectroscopy (DRS) was employed to calculate the bandgap energy (Eg) of the materials through optical characterization with a wavelength range of 200–800 nm using an Agilent Technologies UV-Vis-NIR Cary 5000 spectrophotometer equipped with an integration sphere (Agilent Technologies, Santa Clara, CA, USA). The material was placed in a quartz sample holder and a barium sulfate (BaSO₄) standard was used as the reference. A tangent line was drawn in the region where a change in absorbance was appreciated. The gaseous reaction products were measured using GC gas chromatography using a Thermo Scientific Trace 1310 model (Thermo Fisher Scientific, Waltham, MA, USA), while the liquid products were examined through HPLC by utilizing a Shimadzu Prominence-i LC-2030C model (Shimadzu Scientific Instruments, Columbus, MD, USA).

3. Results and Discussion

3.1. Characterization of K₃Bi₂X₉ (X = I, Br, and Cl)

The crystalline phases of the perovskite materials were investigated through XRD, and the results are shown in Figure 4. All samples presented intense and defined reflections, which can be attributed to their high crystallinity. The K₃Bi₂I₉ film showed similarities with the reflections of the theoretical card of K₃Bi₂I₉ (mp-1120792). The conventionally synthesized sample showed reflections at 2θ = 12°, 26°, 39°, and 53°. Regarding the K₃Bi₂Br₉ and K₃Bi₂Cl₉ films, there were analogous reflections to K₃Bi₂I₉ with a slight shift towards higher diffraction angles, which is consistent with previous reports that state that the size of the X anion in the A₃B₂X₉ crystal structure is inversely proportional to the angles of the XRD reflections [37]. Likewise, the three films presented similar reflections to each other, although with a slight displacement, which was proportional to the increase in the ionic radius of the structure’s anion. Moreover, based on Scherrer’s equation [38], the crystallite size of these materials was calculated, showing values of 108 nm (K₃Bi₂I₉), 100 nm (K₃Bi₂Br₉), and 83 nm (K₃Bi₂Cl₉).

The absorption spectra of the K₃Bi₂X₉ materials were obtained using DRS (Figure 5). These showed similarities to each other, although K₃Bi₂I₉ displayed additional absorption bands in the visible spectrum from 500 to 600 nm, which were attributed to the excitonic peak of these perovskites. These results are consistent with previous reports stating that this compound could have multiple gap states within its bandgap energy range [39]. The energy values in all K₃Bi₂X₉ perovskites displayed the following tendency: 1.8 eV (K₃Bi₂I₉) < 2.7 eV (K₃Bi₂Br₉) < 3.2 eV (K₃Bi₂Cl₉). As can be seen, the bandgap energy values apparently increase proportionally to the size of the X-anion in the K₃Bi₂X₉ structure. The electronegativity difference between the anions and cations of these perovskites might be a defining factor in these phenomena, since the electron sharing involved in compound electronegativity has been shown to be related to the bandgap energy required for the electron transfer process from the VB to the CB [40]. The obtained bandgap energy values were comparable to those previously reported for these compounds, apart from K₃Bi₂I₉, which presented a lower energy value (1.8 eV).

The morphologies of the perovskite samples were characterized through SEM, as shown in Figure 6. All three samples showed distinct morphologies. It was observed that the K₃Bi₂I₉ film presented a heterogeneous morphology, composed of elongated particles larger than 5 µm (Figure 6a). On the other hand, K₃Bi₂Cl₉ exhibited an amorphous layer composed of clusters of particles with indistinct shapes, whereby a mildly hexagonal morphology was visible in the larger grains (<7.5 µm) and several sharp nanoflakes were present throughout (Figure 6b). In the case of K₃Bi₂Br₉, an amorphous clustered layer was also visible (Figure 6c). Nonetheless, there was a higher degree of homogeneity in the semi-hexagonal particles of smaller sizes (<3.8 µm) that composed its morphology.

3.2. Preliminary Study of the Photocatalytic Activity of K₃Bi₂X₉ (X = I, Br, and Cl)

Once the films were characterized, they were evaluated as photocatalysts in the photocatalytic reduction of CO₂. The main product obtained was formic acid, which is one of the solar fuels with a greater value in the market (1767–3135 €/ton), as shown in Figure 7. As can be seen, there were comparable HCOOH production rates between K₃Bi₂I₉ (283 µmol h⁻¹) and K₃Bi₂Br₉ (299 µmol h⁻¹), while K₃Bi₂Cl₉ presented a lower production rate for this sustainable solar fuel (91 µmol h⁻¹).

The photocatalytic activities of each perovskite may have been influenced by their visible light absorption. The DRS spectrum of K₃Bi₂Br₉ shows absorbance bands in the short range of the visible spectrum (380–450 nm). Hence, the lower HCOOH production observed in perovskite K₃Bi₂Cl₉ could be related to its lower activation in the visible range. The high performance of K₃Bi₂Br₉ can also be explained by the relatively higher uniformity of its thin film, which has been linked to higher photocatalytic activity [41] and electric conductivity rates [42]. In addition, according to the XRD analyses, this perovskite presented intense reflections related to its high crystallinity, which could have improved its photocatalytic performance. Even though the perovskites K₃Bi₂I₉ and K₃Bi₂Br₉ displayed similar performances, the following stages of this study were carried out by utilizing K₃Bi₂I₉ films, since this material also had a higher absorption rate in the visible light spectrum.

The formic acid production rate obtained with the K₃Bi₂X₉ perovskites was much higher than with other reported photocatalytic materials [43,44,45,46]. For instance, K₃Bi₂Br₉ (the most efficient perovskite in this study) had a 4.4-fold higher production rate than Cu-doped BiYO₃ (68 µmol h⁻¹) [43]. K₃Bi₂Br₉ also had a better production rate than ZnV₂O₆ (10-fold higher; 28.9 µmol h⁻¹), with NaOH acting as a sacrificing agent [44]. Furthermore, in contrast with other perovskites evaluated in the same reaction system, K₃Bi₂Br₉ had a 1.5-fold higher production rate than NaTaO₃ (199 µmol h⁻¹), which was tested in seawater to assess the effects of dissolved salts (e.g., NaCl, MgCl₂, MgSO₄, CaSO₄, among others) as hole-scavengers [45].

However, according to the literature, there are concerns about the stability of these materials [47]. For this purpose, the stability of K₃Bi₂I₉ was analyzed after three consecutive tests, the results of which are shown in Figure S1. These results indicated that after the first photocatalytic evaluation, the film exhibited significant decay (up to 20 times) in its activity to reduce CO₂. This result could be explained by its unfavorable estimated formation energy (−113.95 kJ mol⁻¹) [48], which also promotes dissociation to its components (i.e., metal halides KX and BiX₃) under external environmental factors (e.g., light, moisture, oxygen) due to the perovskites’ thermodynamic instability [49]. However, in order to provide solutions to this instability issue, three strategies were implemented, the results of which are discussed in the following sections.

3.3. Implementation of Intrinsic and Extrinsic Strategies for the K₃Bi₂I₉ Perovskite

As shown, the K₃Bi₂I₉ films gave an outstanding performance in the photoreduction of CO₂. However, to ensure their economic sustainability, it was mandatory to improve their stability, so efforts were made to amend this aspect. Two different approaches were used—extrinsic and intrinsic. In this study, the growth of K₃Bi₂I₉ on mica substrates was the extrinsic methodology of choice. This adjustment provided more flexibility and greater adhesive properties, as well as allowing the simplification of the film deposition method, as these substrates do not require previous treatments (as opposed to glass). Parallel to this technique, an intrinsic recrystallization method was also implemented, with the aim of eliminating the presence of impurities and enhancing the stability of the K₃Bi₂I₉ films. This strategy consisted of depositing DE and IPA as anti-solvents using a final spin-coating step following the deposition of K₃Bi₂I₉. The films were then characterized using XRD, SEM, and DRS techniques based on their crystal structure, morphology, and optical properties.

Strategies 1 and 2: Recrystallization and Substrate Modification

Figure S2 reveals that the diffractogram of the recrystallized films grown on mica substrates show similar reflections to the mica by itself. According to the theoretical K₃Bi₂I₉ reference, the prominent reflections of this substrate could be overlapped with most of the perovskite reflections, except for 2θ = 12°, for which slightly different intensity levels can be observed for the two types of anti-solvents used for recrystallization—IPA and DE. These results differ from as-synthesized perovskites on glass, since this substrate is amorphous and t is not visible in a diffractogram, as opposed to mica.

The recrystallized three-layered K₃Bi₂I₉ mica film was also characterized via DRS to determine its optical properties. As is shown in Figure S3, there was an increased absorbance intensity in the recrystallized materials’ spectra compared to the as-synthesized film. This was possibly caused by the better adhesion of the film on the mica substrate, as well as improved formation of the films during the recrystallization process with the anti-solvents. The spectra also show decreased absorbance in the 200–300 nm band in the recrystallized films, which could be associated with a lower number of impurities in these samples, especially when using IPA as an anti-solvent. Moreover, the bandgap energy values of these samples were calculated from the DRS spectra. The as-synthesized K₃Bi₂I₉ mica film had an energy value of 1.67 eV, whereas the two recrystallized films showed a decreased value (1.49 eV).

The morphology of the recrystallized perovskite was analyzed using scanning electron microscopy (Figure 8). Regarding the as-synthesized film in the mica substrate (without recrystallization; Figure 8a,b), there was an apparent formation of large grains, forming a heterogeneous layer with incomplete coverage of the substrate. After the implementation of a recrystallization technique using DE as an anti-solvent (Figure 8c,d), it was possible to achieve better coverage of the substrate, in addition to the formation of a better-defined lamellar morphology with elongated shapes. However, the recrystallization process also resulted in the conglomeration of the deposition layers on the substrate (Figure 8c). On the contrary, by using IPA in this process (Figure 8e,f), it was possible to achieve high-quality crystallization of the film, obtaining rectangular morphologies.

Based on the observations obtained from the characterization process, IPA was chosen as the anti-solvent, with which the experiments resumed. The effect of the number of deposit layers was investigated, with the purpose of determining the optimal quantity of perovskite photocatalyst present on the mica substrate. Firstly, XRD characterization was performed on K₃Bi₂I₉ films with 3, 5, and 10 layers of deposit (Figure S4). By depositing 3 layers, a prominent reflection was detected at 2θ = 8°, which was associated with the presence of the perovskite. As the number of layers increased to 5, this reflection showed a higher intensity due to the greater amount of film deposited on the substrate. However, the intensity decreased when the number of layers was increased to 10, which could be associated with a greater thickness on the film that probably led to greater compaction and lower crystallinity.

The effect of the number of layers was also investigated via a DRS analysis (Figure S5). The results showed that when modifying the number of layers of the photocatalyst, the bandgap energy values were 1.49 and 1.56 eV for the three and five-layered films, respectively. It should be noted that samples with 5 and 10 layers showed no significant differences in their bandgap values. Recrystallized K₃Bi₂I₉ was evaluated in the photocatalytic CO₂ reduction to determine the effect of employing different amounts of the photocatalyst by modifying the number of deposited layers (Figure 9). As expected, it was found that a greater number of deposited layers allowed us to obtain higher production rates of HCOOH, reaching a maximum production rate of 210 µmol h⁻¹ with 10 layers of deposit. Taking the superficial area into consideration, this production resulted in a 2.2-fold increment with respect to the initial K₃Bi₂I₉ sample on glass substrates (Figure S6), owing to improved superficial adhesion and the removal of impurities due to the substrate exchange and recrystallization, respectively. In contrast, when 15 and 20 layers were tested, the generation of HCOOH decreased, possibly because the irradiated light was not able to be absorbed by the film during the photocatalytic evaluations. Therefore, 10-layered films of K₃Bi₂I₉ were used for the photocatalytic evaluations that followed in this study.

3.4. Stability

To determine the stability of the 10-layered K₃Bi₂I₉ film recrystallized with IPA, its photocatalytic activity was evaluated in three consecutive runs, in which two different scenarios were used: (i) a test with the same reaction liquid (Figure 10a) and (ii) another one where the liquid was changed (Figure 10b). This helped determine whether the photocatalyst could still be activated after its desorption from the substrate. As seen in Figure 10a, under the first scenario, the generation of HCOOH was drastically lowered after the first run, which was likely due to the dissociation of the perovskite to its components (KI and BiI₃) upon contact with water. This is consistent with previous reports, which stated that metal halide perovskites are prone to having low resistance to environmental factors, such as water. On the other hand, Figure 10b shows that by retaining the reaction liquid (the second scenario), there is a stark difference in the performance of this perovskite, as it retains its high HCOOH production rate (<500 µmol). Thus, according to the results, the liquid reaction medium appears to hold the photocatalyst when diluted in the water, which remains activated by the irradiated light in the reactor.

3.5. Discussion of the Photocatalytic Mechanism

According to the results presented in Figure 10, the contact of the K₃Bi₂I₉ perovskite with water seems to activate a hydrolyzation reaction in which this material is dissociated into its former reactants, as seen in Equation (2) [50]. Subsequently, the formation of BiOI (another visible light photocatalyst) seems to occur from the interaction of BiI₃ with water (Equation (3)), which can explain the fact that the HCOOH generation remains constant within the photocatalytic reaction after several cycles without changing the reaction medium:

K₃Bi₂I₉ → 3KI + 2BiI₃

(2)

BiI₃ + H₂O → BiOI + 2HI

(3)

To corroborate this hypothesis, the presence of K⁺ ions in the liquid medium was confirmed via ion exchange chromatography (Figure 11b). It is evident that the concentration of potassium rises after only one cycle of reaction, which suggests that the perovskite starts to decompose and its components (i.e., metal ions) dissolve in the reaction medium. This process is shown in Figure 11(a3) for the case of potassium. The interaction of potassium iodide with water generates potassium hydroxide (KOH), and hydrogen iodide (HI) can be also produced. Then, KOH can dissociate in the presence of water to hydroxyls and potassium ions, which could explain the chromatography results we obtained.

Strategy 3: Heterojunction Engineering

Based on the assumption that K₃Bi₂I₉ is ultimately converted to BiOI after its interaction with water during the reaction (Equation (3)), it is reasonable to infer that the synthesis of this compound as a cocatalyst would lead to a beneficial interaction in the case of the degradation of the perovskite. Regarding bismuth-based compound heterojunctions, a co-sharing of the Bi³⁺ ion between layers has been previously reported for Cs₃Bi₂I₉/Bi₂WO₆ [51]. It was found that such co-sharing can promote intimate contact and effective electron coupling, which is advantageous for interfacial charge transfer. The implementation of a heterojunction would, thus, strengthen the bond between the perovskite and BiOI. Therefore, for the enhancement of the stability of the perovskite film, a single layer BiOI was deposited as a cocatalyst on a mica substrate, after which 10 layers of K₃Bi₂I₉ film were deposited. A representation of this procedure is shown in Figure 12. After photocatalytic evaluations, the BiOI layer is expected to thicken as a consequence of the degradation of K₃Bi₂I₉ (as previously shown in Figure 11(a2)). The mechanism with which this reaction takes place in the perovskite can be described as “pseudocatalysis” [52], since this material effectively facilitates the photocatalytic reduction of CO₂, although it simultaneously converts into other products (contradicting the principle of a catalyst). Additionally, one of these products, BiOI, is the material expected to continue with the reduction process, resulting in uninterrupted CO2RR product generation, even when the perovskite is subject to instability.

Regarding the characterization of this heterojunction, its optical properties show a discernible change compared to pure BiOI (Figure 13a), as the formation of an excitonic peak with an additional band takes place (450–600 nm), similarly to the K₃Bi₂I₉ film (Figure 5). There is also another band at 250–350 nm, which corresponds to this perovskite. The BiOI film appears to have a band at 450–600 nm. It is worth noting that a band was observed at 200 nm in all samples, corresponding to the mica substrate. From these spectra, the bandgap energy values were obtained, which were 1.79 eV (K₃Bi₂I₉/BiOI) and 1.61 eV (BiOI). These values were higher than those of recrystallized K₃Bi₂I₉ (1.49 eV). The K₃Bi₂I₉/BiOI heterojunction films were analyzed using SEM (Figure 13b–e). The formation of a lamellar morphology was apparent in these films (Figure 13b), consistent with SEM images for the slightly compressed perovskite film recrystallized with DE. Likewise, a slight compaction could have ensued after 10 coatings of K₃Bi₂I₉ during recrystallization with IPA. In the heterojunction, the presence of BiOI is evident due to the characteristic nanosheet morphology that can be seen in the gaps of the perovskite layer (Figure 13c). The exposure of BiOI could lead to the conversion of CO₂ on the surfaces of both the perovskite and this photocatalyst.

The K₃Bi₂I₉/BiOI heterojunction was evaluated in the CO2RR over 6 h of the reaction, as depicted in Figure 14a, along with a schematic representation of the basis of this heterojunction and its co-sharing of I⁻ ions (Figure 14b). Compared to the production rate displayed by the K₃Bi₂I₉ film in the same conditions (Figure 10), the heterojunction displayed an important increase in its photocatalytic activity (38%; 738 µmol h⁻¹ HCOOH), which may have been caused by an improved photogenerated charge transfer (due to adequate band alignment; Figure 14c), as well as better retention of the catalyst in the mica substrate during the reaction (as is visible in Figure 14d). The BiOI nanosheet morphology is present in the heterojunction after one photocatalytic cycle (previously shown in Figure 13d,e), along with nanoflowers that cover the substrate in a thin layer, caused by the migration of the material to the reaction medium (i.e., water). The outstanding efficiency with which the CO2RR takes place in the K₃Bi₂I₉/BiOI heterojunction confirms the aforementioned pseudocatalytic behavior, which may also have been enhanced by its morphology, since there are reports that nanoflower morphologies allow for a greater number of adsorption sites to be obtained, in which CO₂ could be converted to HCOOH [53]. Furthermore, an improved affinity for CO₂ could have been achieved, owing to the forceful interaction between the K⁺ ions and CO₂ [54]. This was also evident in the remarkable production of HCOOH observed in the perovskite when potassium ions were not removed from the reaction liquid, as previously displayed in Figure 10. The strength of this interaction is such that it has previously led to the development of many potassium-based materials for CO₂ adsorption [55]. Adequate interfacial contact with CO₂ has also been reported for Bi³⁺ ions [56], which are readily available on the surface of BiOI and could have been the determinant in the obtained photocatalytic activity.

4. Conclusions

Lead-free bismuth halide perovskites of the K₃Bi₂X₉ family (X = I, Br, Cl) were synthesized via a reproducible one-pot method at room temperature. These materials displayed an outstanding performance (up to 299 µmol h⁻¹ using K₃Bi₂Br₉) in photocatalytic CO₂ reduction for the obtention of formic acid, a clean alternative fuel. The anions’ electronegativity seems to influence these materials in terms of their morphologies, light absorption rates, and bandgap energy values. Additionally, a variety of modification methods (intrinsic and extrinsic) were carried out on the K₃Bi₂I₉ perovskite to ensure a reliable performance. The implementation of these strategies led to the obtention of more efficient functional films, with enhanced adhesion and flexibility, well-defined morphologies, and improved photocatalytic activity (2.2-fold; 535 µmol h⁻¹ during 12 h of reaction). An analysis of the reaction medium and the characterization of the sample suggested a pseudocatalytic behavior in the K₃Bi₂I₉ perovskite, where newly formed BiOI and K⁺ enhanced the photocatalytic processes via more efficient charge transfer, improved light absorption, and an increase in the CO₂ affinity of the K⁺ ions. Thus, it is in this manner that this material can provide reliable performance in the CO2RR, with the possibility of sustainable industrial applications. For its part, the implementation of a heterojunction engineering approach allowed for remarkable efficiency rates (738 µmol h⁻¹) that were up to 38% higher than the improved K₃Bi₂I₉ perovskite during 6 h of constant photocatalytic evaluation. This was probably caused by the junction of the energy bands of these catalysts, which resulted in an improved photogenerated charge transfer, as well as better retention of the film in the substrate due to the co-sharing I⁻ ion binding. The obtained results demonstrated that the design and implementation stability strategies promoted an efficiency enhancement for the CO2RR in the lead-free K₃Bi₂I₉ perovskite, providing sustainable alternatives to scale this product up at the following engineering levels.