Perovskite-inspired low-dimensional hybrid azetidinium bismuth halides: [(CH₂)₃NH₂]₃Bi₂X₉ (X = I, Br, Cl)

Abstract

Bi-based halide perovskites have been considered as alternatives to Pb-based perovskites with the intention of avoiding the use of lead in the field of photovoltaics. Over the last few years, novel Bi-based halide perovskites have shown potential in reaching good photovoltaic performance, as suggested by their similar electronic structure to Pb-based perovskites. Nevertheless, their lower dimensionality entails poor charge carrier transport. It has been consistently stated that the role of the A-site should be further studied. To explore this proposition, we have synthesized three different Bi-based halides with substitution on the A-site by azetidinium cations. In this contribution we report fundamental observations of azetidinium bismuth halides, [(CH₂)₃NH₂]₃Bi₂I₉, [(CH₂)₃NH₂]₃Bi₂Br₉, and [(CH₂)₃NH₂]₃Bi₂Cl₉ with prospects in optoelectronics and photovoltaics. These new materials exhibit 0D and 2D crystal structures at a molecular level and the optical feature of an excitonic band state.

---

Introduction

The halide perovskite system has given rise to new developments in optoelectronics, especially in photovoltaics.^1–3 In ABX₃ halides, research has concentrated on Pb-based iodide or halide mixtures due to their radical rise in power conversion efficiency (PCE). This is due to a large absorption coefficient, low exciton binding energy, and long charge carrier diffusion length compared to other divalent metal-based halides.^4–8 Despite this, these materials are not environmentally friendly as they contain Pb as a pollutant. Thus, there has been an effort to find alternatives for the B-site. Bi³⁺ has been considered a reasonable and less toxic B-site alternative. It has a similar atomic number, ionic radius, and electronic structure to Pb²⁺.^9–12 In general, it cannot be directly used to construct a conventional ABX₃ perovskite stoichiometry because of imbalanced charge neutrality. Therefore, Bi-based metal halides are considered perovskite-derivatives, where the chemical composition is A₃Bi₂X₉, which is ordered with one vacant Bi site in its unit cell.^13,14 A₃Bi₂X₉ can typically take two types of dimensionality: 0-dimensional (0D) or 2-dimensional (2D).^15–18 The 0D type forms [Bi₂X₉]-dimers that share the face of an octahedron and are isolated by an A-site, unlike the [BX₆] single octahedron in the normal 3D perovskite structure. This induces a large difference in electronic properties compared to 3D perovskites, such as charge carrier mobility, effective mass, and recombination channels.^17,18 Inorganic Cs₃Bi₂I₉ and organic–inorganic hybrid (MA)₃Bi₂I₉ (where MA is methylammonium, CH₃NH₃⁺) are 0D type. They have been applied in photovoltaics in single-junction cells. However, the PCE did not exceed 5%, due to the strong charge localisation in the dimers, the indirect nature of the bandgap, and deep defect states.^12,19–21 Nevertheless, the great stability and suitable bandgap can still prove potentially useful in photovoltaics. The 2D layered type forms octahedral networks with corner-sharing [BiX₆] sites with low density, especially in the case of the small radius of the A-site cation or a large halide anion: K₃Bi₂I₉, Rb₃Bi₂I₉ and (NH₄)₃Bi₂I₉ adopt 2D layered systems.^22,23 These systems commonly exhibit higher conductivity than 0D-systems and a direct bandgap. Therefore, the 2D type might be more suitable than the 0D type in terms of light absorbtion.^23,24 Although the 2D-systems exhibit better preconditions for performance in solar cells, the values of the PCEs have not been reported to be as high as those for the 0D type due to deep trap states.^12,25 In addition, they have not been employed in the fabrication of a photovoltaic device to date.

A-site compositional engineering for A₃Bi₂X₉ hybrid halides can be key to realizing optoelectronic performance, because the dipole of the organic cation at the A-site can greatly influence the electronic structure via a transformation in dimensionality.²⁶ Ünlü et al. made devices with replacement not only by alkali-metal cations but also by some small organic cations on the A-site. They reported photovoltaic performance with a good match to the thin-film properties of A₃Bi₂I₉.²⁷

We have used the azetidinium cation (Az⁺) on the cationic sublattice of A₃Bi₂X₉ (X = I, Br, Cl). Azetidinium, [(CH₂)₃NH₂]⁺, has a suitable effective ionic radius to replace MA and FA (formamidinium, CH(NH₂)₂⁺).^28,29 The molecular structure of Az⁺ is heterocyclic with one nitrogen and three carbons. The carbon–nitrogen ring structure is not flat, and it is polar (Fig. 1). Az⁺ has been investigated for its phase transition properties, especially in formate-based and cyanide-bridged perovskite systems.^30–32 (Az)PbI₃ has been synthesized and yields a PCE of 1.15% in a photovoltaic device.^33,34 Recently, our group confirmed the applicability of Az⁺ incorporation into a lead-free metal halide system with the synthesis of (Az)₂AgBiBr₆, displaying 1D chains of octahedra in the crystal lattice.³⁵

Fig. 1 Structure of the azetidinium molecular cation and chemical reactions for the synthesis of (Az)₃Bi₂X₉.

In this paper, we chose (Az)₃Bi₂X₉ (X = I, Br, Cl) for a detailed analysis, attempting to expand the range of potential absorber materials.

Experimental

Synthesis

Single crystals of (Az)₃Bi₂X₉ can be grown, but this is a time-intensive process, and we have not been able in all cases to produce single crystals which were large enough, not twinned, and not intergrown with other crystals. We thus chose to use polycrystalline samples for structural studies.

Polycrystalline (Az)₃Bi₂X₉ powders were synthesized by an evaporation method. A brief depiction of the chemical reaction for the synthesis of azetidinium bismuth halide is presented in Fig. 1. We used bismuth(III) iodide (99%) (BiI₃), bismuth(III) bromide (>98%) (BiBr₃) from Sigma-Aldrich, and bismuth(III) chloride (99.999%) (BiCl₃) from ACROS Organics. Azetidinium chloride (97%) (AzCl) was purchased from Sigma-Aldrich. Azetidinium bromide (AzBr) was synthesized in our laboratory by the reaction of hydrobromic acid (48%) (HBr) and azetidine (98%, Thermo Fisher Scientific). Azetidinium iodide (AzI) was synthesized by the reaction of hydroiodic acid (55–58%) (HI) and azetidine (98%). Powdered (Az)₃Bi₂I₉ was synthesized by the evaporation of a 0.2 M precursor at 65 °C, which consisted of AzI and BiI₃ dissolved in N,N-dimethylformamide. The polycrystalline powder of (Az)₃Bi₂Br₉ was synthesized by evaporation of a precursor made from AzBr and BiBr₃ dissolved in acetonitrile at 80 °C in a vacuum oven. The polycrystalline powder of (Az)₃Bi₂Cl₉ was synthesized by evaporation of a 0.3 M precursor containing AzCl and BiCl₃ dissolved in γ-butyrolactone. The precursor solution was filtered through a PTFE (polytetrafluorethylene) membrane filter of 0.45 μm pore size. The acquired crystals of (Az)₃Bi₂Cl₉ were crushed, and then cleaned in isopropanol.

Crystal structure

The crystal structures of the (Az)₃Bi₂X₉ materials were determined from powder X-ray diffraction (PXRD) data by the direct method using the program EXPO 2014.³⁶ It was an intricate matter to obtain the underlying crystal structure model from the powder diffraction data. The results are given in Fig. 2. Subsequently, the structural models were refined using the Rietveld method. The refinements converged to convincing chi² and R-values (see Table 1) confirming the correctness of the structural models. All materials could be obtained as single phases in specific synthesis runs and always as polycrystalline powders. In the case of (Az)₃Bi₂I₉, the structure was refined from the dataset of a sample that contained an additional, silver-containing, impurity phase (approx. 10%). The structure of (Az)₃Bi₂Cl₉ was refined from the dataset of a sample that contained a small amount of BiOCl (refined: 1.4%). The crystals in the latter two samples showed a much higher degree of structural order than the crystals of the corresponding monophasic samples.

Fig. 2 Refined crystal structure diagrams of (Az)₃Bi₂X₉ (X = I, Br, Cl). The azetidinium cations have specific sites in the crystal lattice of (Az)₃Bi₂X₉ but a random orientation of the ring. The orientations of the azetidinium ring do not coincide with Rietveld refinement results but represent a snapshot of the disordered cation. (Edited by VESTA software.).

Table 1 Crystallographic parameters for the structural refinements of (Az)₃Bi₂X₉

Unit cell content	C18H48N6Bi4I18	C36H96N12Bi8Br36	C18H48N6Bi4Cl18
Diffractometer	STOE STADI MP with Mythen 1 K position sensitive detector	PANalytical X'pert PRO with PIXcel3D-Medipix3 1D scanning line detector
Wavelength	1.54059 Å (Cu Kα1 Radiation)	1.5418 Å (Cu Kα1/2 Radiation)
Sample holder	Flat zero scattering foil	Flat Si circular plate
2θ range of data used [°]	6.0–86.9	7. 0–94.9	7. 0–95.0
Step size [°2θ]	0.01500	0.01313	0.01313
No. contributing reflections	301	929	944
No. geometric restraints	6	26	20
No. structural parameters	23	54	39
No. profile parameters	8	15	16
FWHM at ca. 25°2θ [°2θ]	0.08–0.09	0.12–0.13	0.06–0.07
R F	0.034	0.046	0.048
R wp	0.110	0.130	0.136
χ2	1.83	2.19	2.31
Space group	P63mc (No. 186)	Cmc21 (No. 36)	P31c (No. 159)
Crystal lattice	Hexagonal	Orthorhombic	Trigonal
a [Å]	9.1537(1)	8.4576(3)	8.4764(1)
b [Å]	9.1537(1)	16.2886(6)	8.4764(1)
c [Å]	22.2111(4)	20.8689(5)	20.5819(5)
V UC [Å^3]	1611.73(4)	2874.94(8)	1280.69(4)
Density (calc.) [g cm^−3]	3.574	3.035	2.363

---

For further details, please see the ESI.†

Results and discussion

(Az)₃Bi₂I₉ exhibits hexagonal symmetry, with space group P6₃mc (No. 186). The crystal structure contains 4 trivalent Bi³⁺ cations and 18 I⁻ anions per unit cell, forming four [BiI₆] octahedra. Two of these octahedra always share common faces, thereby generating two [Bi₂I₉] dimers (Fig. 2). The Az⁺ cations occupy specific positions surrounding the [Bi₂I₉] dimers per unit cell. As the octahedra are isolated within the unit cell, they form an effectively 0D octahedral structure. The space group for (Az)₃Bi₂I₉ is the same as those for (MA)₃Bi₂I₉ and (GA)₃Bi₂I₉. Methylammonium (MA) and guanidinium (GA) have similar effective ionic radii to Az⁺.^37,38 It is expected that its structural and electrical properties will be close. In the crystal lattice, Az⁺ cations have a largely random orientation with no specific location for the N atom. We assume that Az⁺ is dynamically disordered at room temperature. This is the result of XRD. However, it should be considered that it is quite difficult to locate C and N atoms in the vicinity of heavy scatterers like I and Bi, from XRD powder diffraction data.

(Az)₃Bi₂Br₉ adopts an orthorhombic lattice with the space group, Cmc2₁ (No. 36), which also contains 0D isolated [Bi₂Br₉] dimers surrounded by Az⁺ cations that are highly disordered with respect to the carbon and nitrogen positions. The structure of (Az)₃Bi₂Br₉ is pseudo-hexagonal and very similar to that of (Az)₃Bi₂I₉, but the actual symmetry is lower than that of (Az)₃Bi₂I₉.

(Az)₃Bi₂Cl₉ is distinctive, as it has a low-density 2D structure with corner-sharing [BiCl₆] interconnecting octahedra. Az⁺ is also highly disordered and intercalated between [Bi₂Cl₉] layers. It assumes space group P31c (No. 159) with trigonal symmetry. The corner-sharing only corresponds to 3 Cl⁻ anions in a one-octahedron framework, in which the vacant site between the [BiCl₆] octahedra is occupied by disordered Az⁺ cations.

The polycrystalline powder of (Az)₃Bi₂I₉ has a red colour (Fig. 3). The most intensive reflection is the (101) plane at 2θ = 11.77° that was confirmed by its PXRD pattern. The largest d-spacing in the lattice is in the (002) plane at 2θ = 7.97°, followed by (100), (101) and (102). Every reflection has a relatively large full-width-at-half-maximum (FWHM) in its PXRD pattern; thus, comparatively large strains exist in the crystal lattice. The refined polycrystalline powder of (Az)₃Bi₂Br₉ is yellow. The PXRD of (Az)₃Bi₂Br₉ has a main reflection of the (111) plane at 2θ = 12.56°, which is not identical to (Az)₃Bi₂I₉ owing to its different crystal symmetry. (Az)₃Bi₂Cl₉ is an almost white powder. It generally appeared in quite highly crystalline form for every sample we synthesized, with much higher intensity values than (Az)₃Bi₂I₉. The (102) reflection at 2θ = 14.76° has the highest peak intensity. Its better degree of crystallinity may be due to the 2D [BiCl₆] octahedral network being more rigid than isolated dimers.

Fig. 3 Powder X-ray diffraction patterns (PXRD) of (Az)₃Bi₂X₉ phases with approximately randomly oriented crystals and photographs of the powdered materials. Simulated patterns based on the refined structures are shown for comparison.

Thin films

To investigate the thin-film properties, we deposited (Az)₃Bi₂X₉ thin-film layers with one-step spin coating as an initial step. According to their XRD patterns all (Az)₃Bi₂X₉ strongly exhibit specific grain growth. To determine the features of substrate-dependent growth, we prepared three different (transparent) substrates and coated the thin films onto those: purified normal glass, fluorine-doped tin oxide (FTO), and indium tin oxide (ITO). On a glass substrate, the thin films of (Az)₃Bi₂I₉ and (Az)₃Bi₂Br₉ show a preference for (00l) growth orientation with large intensities of (002), (004), and (006). (Az)₃Bi₂Cl₉ presented growth with (002) and (004) reflections (Fig. 4a). In the case of (Az)₃Bi₂I₉, the greatest intensity is the (006) reflection at 2θ = 24.39°, well exceeding the intensities of the (002) and (004) reflections at 2θ = 8.15° and 16.24°, respectively. (0010), and (0012) reflections with relatively small intensity at 2θ = 41.12° and 49.84°, respectively, were also detected. The diffraction pattern of (Az)₃Bi₂Br₉ on glass had the greatest intensity for the (002) reflection at 2θ = 8.60° followed by (004) at 2θ = 17.12° and then (006) at 2θ = 25.75°. There were lower but distinctive reflections of (0010) at 2θ = 43.51° and (0012) at 2θ = 52.81°. At (006), (0010), and (0012) reflections peak splitting was detected. In the diffraction pattern of (Az)₃Bi₂Cl₉ on glass, we confirmed that the intensities of the reflections of (002) at 2θ = 8.65° and (004) at 2θ = 17.30° are extremely high compared to the corresponding peaks of (Az)₃Bi₂I₉ and (Az)₃Bi₂Br₉. Remarkable peak splitting happens for the (004) reflection.

Fig. 4 (a) X-ray diffraction patterns of (Az)₃Bi₂X₉ thin films on glass substrate with pictures of the samples (the green dotted line represents the powder X-ray diffraction), (b) SEM morphological images of (Az)₃Bi₂X₉ on different transparent substrates.

Compared to the film coated on glass, the XRD pattern of (Az)₃Bi₂I₉ thin film on FTO shows no growth preference, so the pattern is close to its PXRD (Fig. S3, ESI†). The FTO substrate has a tetragonal structure (rutile), inducing a significant mismatch with the crystal growth of (Az)₃Bi₂I₉. The (Az)₃Bi₂I₉ thin film on ITO is similar to that on glass, but the (002) reflection is as dominant as the (006) reflection. The XRD pattern of (Az)₃Bi₂Br₉ on FTO presents only some very low peaks at around 2θ = 12.5° and 17.5°, corresponding to a secondary, nearly amorphous phase and some reflections of FTO, but no reflection that indicates the occurrence of (Az)₃Bi₂Br₉ crystals. We assume that the orthorhombic structure of (Az)₃Bi₂Br₉ yields an intermediate mismatch, generating increased disorder in the thin film, seen as an amorphous phase. The XRD pattern of (Az)₃Bi₂Br₉ on ITO presents the same pattern as on glass. Although the intensities of the reflections are much lower than those on glass. Similarly, in the XRD patterns of (Az)₃Bi₂Cl₉ on FTO and ITO, it was confirmed that the intensities of the (002) and (004) reflections are much lower than those on glass, although those peaks are sharp. Unlike the (Az)₃Bi₂I₉ and (Az)₃Bi₂Br₉ thin films on FTO, (Az)₃Bi₂Cl₉ on FTO resulted in highly textured growth because there is a lattice match with the substrate. The identification of other peaks of (Az)₃Bi₂Cl₉ was not simple due to the relatively high and sharp intensity of its (002) and (004) reflections. We focused on the very low-intensity region of the XRD pattern of (Az)₃Bi₂Cl₉ on every substrate (Fig. S3d, ESI†). The resulting patterns seemingly presented not only (00l) reflections but also small broad peaks, coinciding with (100), (200), and (300) orientations in all patterns.

The morphology of the (Az)₃Bi₂X₉ thin film was visualized in SEM images for comparison with the XRD results (Fig. 4b). The grains of (Az)₃Bi₂I₉ on glass are mostly platelets, which are not coplanar with the substrate. This generates a very rough film. The morphology of (Az)₃Bi₂I₉ on FTO is different from that of the film on glass. A homogenous distribution of needle-like grains is displayed, but no hexagonal platelets. (Az)₃Bi₂I₉ on ITO has nearly the same morphology as on glass. As with its XRD pattern, the preferred (000l) orientation is dominant. (Az)₃Bi₂Br₉ on glass is less rough. The crystallites seem to have grown in star-like patterns, which have merged to form a film. The morphological state of (Az)₃Bi₂Br₉ on ITO is nearly identical to that of (Az)₃Bi₂Br₉ coated on glass. (Az)₃Bi₂Br₉ on FTO has smaller grains. It seems that the crystals hardly grow, with only a low density of coverage. In addition, the shape of the grains is completely different from those on glass or ITO. The XRD pattern differs for the reflections with low intensity in Fig S3 (ESI†). (Az)₃Bi₂Cl₉ on glass has large hexagonal grains with small lumps that seem to be in an intermediate stage of crystal growth (Fig. 4b). However, the crystal layer does not completely cover the glass surface. This phenomenon is also shown in the films coated on FTO and ITO. The size of the grains of (Az)₃Bi₂Cl₉ is usually larger than those of (Az)₃Bi₂I₉ or (Az)₃Bi₂Br₉. The XRD pattern shows that the texture of the large grains corresponds to the (00l) reflection.

Chemical characterization

Azetidinium cations were characterized using Fourier transform infrared (FT-IR) transmittance spectra of powder samples at room temperature (Fig. 5a). The results exhibit the assignment of transmission peaks of the azetidinium cation in the crystal lattice.^34,39–41 The broad band at around 3359 cm⁻¹ is considered to be the N–H stretching vibrational mode of the Az⁺.⁴⁰ This band is clear in the spectrum of (Az)₃Bi₂I₉, but the intensity is lower in the spectra of (Az)₃Bi₂Br₉ and (Az)₃Bi₂Cl₉. This might be due to weak hydrogen bonds in these two compounds. C₂–N⁺–H₂ stretching can be assigned to 3190–3150 cm⁻¹ and C₂–N⁺–H₂ deformations can be assigned to 1570–1530 cm⁻¹ and 1445–1430 cm⁻¹.⁴¹ The two transmission peaks of all materials at 3090–3080 cm⁻¹ and 2990–2960 cm⁻¹ can indicate –CH₂ asymmetric/symmetric stretching.^40,41 An obvious peak at 690–670 cm⁻¹ is detected in all materials, which indicates azetidine ring deformation, according to the calculations and data of H. Nielsen and N. Gajhede [1989].⁴⁰

Fig. 5 (a) FT-IR spectra of (Az)₃Bi₂X₉ powders and (b) Raman spectra of (Az)₃Bi₂X₉ powders and thin films on glass (laser used: 532 nm).

We performed Raman spectroscopy for both powder and thin films (on glass) using a 532 nm laser from 4000 cm⁻¹ to 100 cm⁻¹ at room temperature (Fig. 5b). A common characteristic is observed where there are obvious Raman peaks only below 400 cm⁻¹. These Raman bands clearly indicate the vibrational mode of the [BiX₆]³⁻ octahedra, in good agreement with compounds reported in the literature.^42–46

The Raman spectrum of (Az)₃Bi₂I₉ powder can be explained with 2 peaks: a low-intensity one at ∼118 cm⁻¹ caused by the vibrational mode of a [BiI₆]³⁻ singular octahedron in the [Bi₂I₉]³⁻ dimers, and a high-intensity one at ∼137 cm⁻¹ caused by the Bi–I bonds of the internal [BiI₆]³⁻ octahedron.^42,43 The spectrum of the (Az)₃Bi₂I₉ thin film has the same characteristics as that of the powder. The Raman spectrum of (Az)₃Bi₂Br₉ powder presents 3 peaks. There is a strong peak at ∼187 cm⁻¹ and a weaker one at ∼160 cm⁻¹ assigned to the Bi–Br stretching vibrational mode of the [BiBr₆]³⁻ octahedron.⁴⁴ The other separate band at ∼114 cm⁻¹ can be assigned to the axial stretching vibrational mode of the Br–Bi–Br bridge in the [BiBr₆]³⁻ octahedron.⁴⁴ The spectrum of the thin film has a weaker and unclear band compared to that of the powder. This might be because the optimized film is quite thin, leading to less Raman scattering.

In the case of the Raman spectrum of (Az)₃Bi₂Cl₉, the two large bands at ∼293 cm⁻¹ and at ∼253 cm⁻¹ can be assigned to Bi–Cl stretching, and the other small band at ∼149 cm⁻¹ can be assigned to δ (Cl–Bi–Cl) bending vibrational modes in the anionic crystal sub-lattice [BiCl₆]³⁻.^45,46 A slight shift is observed in the Bi–Cl stretching bands (around 2 cm⁻¹) in the spectrum of the thin film, but the δ (Cl–Bi–Cl) bending band is shifted a long way to ∼126 cm⁻¹. As we suggested above, our thin film of (Az)₃Bi₂Cl₉ is in an unusual state, which does not give complete coverage, while its crystallinity is very high. This factor might have a great influence on the Raman shift of the thin-film state.

Optical properties

The Tauc plot method with Kubelka–Munk transformation of diffuse optical reflectance spectra was used, extending up to 4 eV.^47,48 (Fig. 6a and b). (F(R)hν)^1/2 is plotted against photon energy hν, reflecting a potential indirect band transition; see Fig. 6a. By linear extrapolation, an indirect bandgap transition of (Az)₃Bi₂I₉ is approximately assessed as 1.97 eV; however, it seems that sub-band transitions occur at 2.52, 2.99, and 3.32 eV, although these band states are not clear in the graph. In the case of (Az)₃Bi₂Br₉, the dominant band transition is estimated as 2.58 eV and that of (Az)₃Bi₂Cl₉ is estimated as 3.12 eV in the indirect band transition plot. Noteworthy is the weak broad-band state at (Az)₃Bi₂Cl₉ below its dominant band transition. For comparison, (F(R)hν)² is also plotted against photon energy hν, where it presents a direct band transition in Fig. 6b. A direct band transition value of (Az)₃Bi₂I₉ is assessed as 2.09 eV, and it presents a sub-band transition at 2.51 eV followed by 2.99, and 3.35 eV, similar to its indirect band transition plot. Absorption edges in the direct band transition plots of (Az)₃Bi₂Br₉ and (Az)₃Bi₂Cl₉ are extracted as 2.67, and 3.17 eV, respectively. Additionally, (Az)₃Bi₂Br₉ presents a second band transition at 3.65 eV in the higher region of the band edge, which is not estimated in its indirect band transition plot. For (Az)₃Bi₂Cl₉, the sub-band states are unclear, but the broad-band state is not observed below its band edge.

Fig. 6 Tauc plots of the polycrystalline powders of (Az)₃Bi₂X₉ through Kubelka–Munk transformation and bandgap estimation with (a) an indirect transition that describes photon energy vs. (F(R)hν)^1/2 and (b) a direct transition that describes photon energy vs. (F(R)hν)², and absorbance spectra of (c) (Az)₃Bi₂I₉, (d) (Az)₃Bi₂Br₉, and (e) (Az)₃Bi₂Cl₉ in accordance with different substrates (glass, FTO, and ITO).

For clarification of the bandgap study, absorption spectra of thin films of (Az)₃Bi₂X₉ on variable substrates (glass, FTO, and ITO) were measured using UV-vis spectroscopy as a subsequent step (Fig. 6c–e). Overall, the absorption spectra of the thin films present a slightly wider bandgap compared to those of the powders. There are a variety of factors for the band-shift between powders and thin films. This sort of deviation was often observed in Bi-based halides.⁴⁹ For example, for thin-film deposition of the analogue (MA)₃Bi₂I₉, large deviations of the bandgap appear for different fabrication methods, where the bandgap ranges from 1.8 eV to 2.2 eV.^50–52

The absorption spectrum of the (Az)₃Bi₂I₉ thin film presents a steep dominant peak centered at 2.49 eV, a wide absorption band, and sub-band transitions at 3.11 eV, and 3.61 eV (Fig. 6c). This spectral structure shows similarities to those of Cs₃Bi₂I₉ and (MA)₃Bi₂I₉.^53,54 As those analogues have the same dimensionality as our (Az)₃Bi₂I₉, the dominant peak at 2.49 eV is assumed to be an excitonic band state. In the case of the electronic structure of (MA)₃Bi₂I₉, it is considered to consist of electron transitions from the ground ¹S₀ state to the triplet excited states ³P₂, ³P₁, and ³P₀ of Bi³⁺ in the [Bi₂I₉] dimers.⁵⁴ Therefore, it can be assumed that the sub-band transitions of (Az)₃Bi₂I₉ are induced in the [Bi₂I₉] electron transitions. Recently, Klein et al. suggested that the shape of the absorption edge changes when the Tauc plot is used if the absorption edge is dominated by an excitonic transition, so the Tauc plot cannot be fitted.⁵⁵ To clarify the excitonic region, we fitted the absorbance of its thin film on glass with a Gaussian function, giving an FWHM of 285 meV and an absorption peak at 2.502 eV (Fig. S4, ESI†). The fitting of the graph shows that the dominant absorption at the band edge is an excitonic band state. It is considered that this excitonic band is caused by the localization of exciton transitions due to the electronic structure of isolated [Bi₂I₉] dimers. Depending on substrates, no evident shifts of the absorption band edge were detected; only the increase in absolute absorption was presented. Compared to the thin film on glass, the excitonic absorption peaks for the thin films on ITO and on FTO are intense. The crystallinity of the thin film on ITO might tend to be lower according to our XRD dataset, but this would not be the sole factor determining an increase in the absolute value of absorption. There is no observation of preferred growth for the thin film on FTO, as shown in Fig S3 (ESI†). This indicates that the texture of the thin film decreases the dominant excitonic absorption peak of (Az)₃Bi₂I₉.

The absorption spectrum of (Az)₃Bi₂Br₉ thin film presents a strong and broad absorption centered at around 3.20 eV (Fig. 6d). This strong band can be estimated as an exciton transition, as our structural determination of (Az)₃Bi₂Br₉ adopts an 0D face-sharing octahedral structure like (Az)₃Bi₂I₉, predicting that it will have a similar electronic structure. However, it features a sub-gap observed at 2.90 eV that is partially overlapped at the exciton transition. This sub-gap is weakly distinctive; therefore, it is difficult to estimate whether it is due to an excitonic transition or a deep trap-state. Due to this sub-gap, it was difficult to ascertain the fitting of the plot for the excitonic peak with either Lorentzian or Gaussian functions. In addition, this sub-gap is not observed in the Tauc plot of the powder reflectance. The sequential band is detected at around 3.59 eV. The thin film on ITO appears to show relatively weak excitonic absorption compared to those on glass or FTO, but not that critical shift. We confirmed that the excitonic peak at around 3.19 eV of the thin film on FTO is slightly different from the others.

In the case of the thin film of (Az)₃Bi₂Cl₉, the absorption spectrum presents a dominant band edge at 3.35 eV followed by a sequential band at 3.52 eV (Fig. 6e). The excitonic peak near the band edge is not as obvious as those of (Az)₃Bi₂I₉ or (Az)₃Bi₂Br₉. This is considered to be due to an expansion of dimensionality in comparison with (Az)₃Bi₂I₉ and (Az)₃Bi₂Br₉, resulting in weakened excitonic confinement.²⁴ The thin films of (Az)₃Bi₂Cl₉ on glass and FTO show characteristics of linearly increasing absorption up to 3.30 eV, but that on ITO has a weak and broad absorption band up to 3.28 eV. As mentioned above, the thin film of (Az)₃Bi₂Cl₉ does not completely cover the substrates, regardless of the kind of substrate; therefore, it should be normal to have different defect states from (Az)₃Bi₂I₉ and (Az)₃Bi₂Br₉. In other words, this weak and broad absorption is expected to come from a deep defect state induced by the very low density of the thin-film layer. Therefore, further research into trap states or emission spectroscopy should be conducted. Meanwhile, weakening of the excitonic peak is observed with a thin film of (Az)₃Bi₂Cl₉ on glass compared with that on FTO. Observing that the grain size of the thin film on glass is larger than that on FTO, we assume that this is a typical phenomenon caused by its high crystallinity, resulting in a large distribution of 2-dimensional grains. The absorption spectrum of the thin film on ITO largely presents a weakened excitonic peak. Moreover, the sub-band at 3.52 eV is not distinctive. This should be re-assessed after optimization with completely filled layers.

The remarkable thing is that the excitonic peak does not exist in the powder reflectance spectra, while it is clearly visible in the thin-film absorbance spectra. There could be multiple reasons for this. One obvious option is that surface defects act as quenching sites for the excitons on the large powder surface.^56–58 Crystallite formation in powders may also include more point defects than films, but the underlying mechanisms are too complex to be resolved here.

Conclusions

It has been proven that (Az)₃Bi₂X₉, where the Az⁺ cation is incorporated into the A-site of Bi-based halides, represent specific low-dimensional perovskite-derivatives. (Az)₃Bi₂I₉ and (Az)₃Bi₂Br₉ adopt an 0D-isolated octahedral structure, while (Az)₃Bi₂Cl₉ adopts a 2D corrugated layered octahedral structure at the molecular level with randomly oriented Az⁺ cations in the crystal lattice. (Az)₃Bi₂I₉ is in a different space group to (Az)₃Bi₂Br₉. Due to the peculiar molecular structure and random orientation of the Az⁺ cation, there is further interest in studying these systems at low temperature to determine their molecular dynamics.

The success of the synthesis of the polycrystalline powders and thin films implies that they are potential materials for optoelectronics and photovoltaics as new light absorbers. In particular, it will be worth trying to do deep research on their optical features in relation to their electronic structure. The excitonic peak that was detected in the absorbance of the thin films is not clear in the Tauc plot from the reflectance spectra of the powders. We did not identify whether they have an obviously direct or indirect transition at the forbidden band. This should be further characterized through various theoretical and experimental methods. Accordingly, we believe that this study can provide insight into developing the scientific scope for light absorbing materials in the future.

Author contributions

Y. U. J. organized the research and the experiment, synthesized the polycrystalline powders, and successfully deposited the films. B. M. solved the crystal structure of the materials. B. M. and A. N. S. performed the structure refinements. M. E. C. helped with synthesis, FT-IR, Raman and NMR analysis. N. B. and D. C. L. helped with data interpretation, text and funding.

Data availability

Crystallographic data for (Az)₃Bi₂I₉ has been deposited at the CCDC crystallographic database under 2333118.† Crystallographic data for (Az)₃Bi₂Br₉ has been deposited at the CCDC crystallographic database under 2333116.† Crystallographic data for (Az)₃Bi₂Cl₉ has been deposited at the CCDC crystallographic database under 2333115.† No primary research results, software or code have been included.

Conflicts of interest

There are no conflicts to declare.

Acknowledgements

A. N. S acknowledges the support of the project CICECO-Aveiro Institute of Materials, UIDB/50011/2020, UIDP/50011/2020 & LA/P/0006/2020, financed by national funds through the FCT/MEC (PID-DAC). D. C. L., N. B., and Y. U. J acknowledge funding through the German Science Foundation (Deutsche Forschungsgemeinschaft, DFG) under project number 424708448. Fruitful discussions with Vladimir V. Shvartsman are highly acknowledged. Felix Niemeyer is acknowledged for his NMR measurements. Matthias Epple and Ivanna Kostina are acknowledged for the help of FT-IR measurement. Ulrich Hagemann is acknowledged for the help of Raman spectroscopy.

Notes and references

1. A. Kojima, K. Teshima, Y. Shirai and T. Miyasaka, Organometal halide perovskites as visible-light sensitizers for photovoltaic cells, J. Am. Chem. Soc., 2009, 131(17), 6050–6051.
2. J. H. Im, C. R. Lee, J. W. Lee, S. W. Park and N. G. Park, 6.5% efficient perovskite quantum-dot-sensitized solar cell, Nanoscale, 2011, 3, 4088–4093.
3. M. M. Lee, J. Teuscher, T. Miyasaka, T. N. Murakami and H. J. Snaith, Efficient hybrid solar cells based on meso-superstructured organometal halide perovskites, Science, 2012, 338, 643–647.
4. National Renewable Energy Laboratory, Best research-cell efficiency chart, accessed, 01, 2024, https://www.nrel.gov/pv/assets/images/efficiency-chart.png.
5. S. D. Stranks, G. E. Eperon, G. Grancini, C. Menelaou, M. J. P. Alcocer, T. Leijtens, L. M. Herz, A. Petrozza and H. J. Snaith, Electron-hole diffusion lengths exceeding 1 micrometer in an organometal trihalide perovskite absorber, Science, 2013, 342(6156), 341–344.
6. M. Liu, M. B. Johnston and H. J. Snaith, Efficient planar heterojunction perovskite solar cells by vapour deposition, Nature, 2013, 501, 395–398.
7. A. K. Jena, A. Kulkarni and T. Miyasaka, Halide perovskite photovoltaics: background, status, and future prospects, Chem. Rev., 2019, 119, 3036–3103.
8. L. N. Quan, B. P. Rand, R. H. Friend, S. G. Mhaisalkar, T. Lee and E. H. Sargent, Perovskites for next-generation optical sources, Chem. Rev., 2019, 119, 7444–7477.
9. M. Lyu, J. Yun, M. Cai, Y. Jiao, P. V. Bernhardt, M. Zhang, Q. Wang, A. Du, H. Wang, G. Liu and L. Wang, Organic–inorganic bismuth (III)-based material: A lead-free, air-stable and solution-processable light-absorber beyond organolead perovskites, Nano Res., 2016, 9(3), 692–702.
10. N. Cates and M. Bernechea, Research Update: Bismuth based materials for photovoltaics, APL Mater., 2018, 6, 084503.
11. S. Attique, N. Ali, S. Ali, R. Khatoon, N. Li, A. Khesro, S. Rauf, S. Yang and H. Wu, A potential checkmate to lead: bismuth in organometal halide perovskites, structure, properties, and applications, Adv. Sci., 2020, 7, 1903143.
12. X. Chen, M. Jia, W. Xu, G. Pan, J. Zhu, Y. Tian, D. Wu, X. Li and Z. Shi, Recent progress and challenges of bismuth-based halide perovskites for emerging optoelectronic applications, Adv. Opt. Mater., 2023, 11, 2202153.
13. K. M. McCall, C. C. Stoumpos, S. S. Kostina, M. G. Kanatzidis and B. W. Wessels, Strong Electron–Phonon Coupling and Self-Trapped Excitons in the Defect Halide Perovskites A₃M₂I₉ (A = Cs, Rb; M = Bi, Sb), Chem. Mater., 2017, 29, 4129–4145.
14. Y. E. Ajjouri, V. S. Chirvony, N. Vassilyeva, M. Sessolo, F. Palazon and H. J. Bolink, Low-dimensional non-toxic A₃Bi₂X₉ compounds synthesized by a dry mechanochemical route with tunable visible photoluminescence at room temperature, J. Mater. Chem. C, 2019, 7, 6236–6240.
15. B. Chabot and E. Parthé, Cs₃Sb₂I₉ and Cs₃Bi₂I₉ with the hexagonal Cs₃Cr₂Cl₉ structure type, Acta Crystallogr., 1978, B34, 645–648.
16. V. I. Sidey, Y. V. Voroshilov, S. V. Kun and E. Y. Peresh, Crystal growth and X-ray structure determination of Rb₃Bi₂I₉, J. Alloys Compd., 2000, 296, 53–58.
17. A. J. Lehner, D. H. Fabini, H. A. Evans, C. Hebert, S. R. Smock, J. Hu, H. Wang, J. W. Zwanziger, M. L. Chabinyc and R. Seshadri, Crystal and Electronic Structures of Complex Bismuth Iodides A₃Bi₂I₉ (A = K, Rb, Cs) Related to Perovskite: Aiding the Rational Design of Photovoltaics, Chem. Mater., 2015, 27, 7137–7148.
18. J. K. Pious, M. L. Lekshmi, C. Muthu, R. B. Rakhi and C. Vijayakumar, Zero-dimensional methylammonium bismuth iodide-based lead-free perovskite capacitor, ACS Omega, 2017, 2, 5798–5802.
19. K. Hong, J. Kim, L. Debbichi, H. Kim and S. H. Im, Band Gap Engineering of Cs₃Bi₂I₉ Perovskites with Trivalent Atoms Using a Dual Metal Cation, J. Phys. Chem. C, 2017, 121, 969–974.
20. B. Ghosh, S. Chakraborty, H. Wei, C. Guet, S. Li, S. Mhaisalkar and N. Mathews, Poor Photovoltaic Performance of Cs₃Bi₂I₉: An Insight through First-Principles Calculations, J. Phys. Chem. C, 2017, 121, 17062–17067.
21. M. Shi, G. Li, W. Tian, S. Jin, X. Tao, Y. Jiang, E. A. Pidko, R. Li and C. Li, Understanding the effect of crystalline structural transformation for lead-free inorganic halide perovskites, Adv. Mater., 2020, 32, 2002137.
22. A. J. Lehner, D. H. Fabini, H. A. Evans, C. A. Hébert, S. R. Smock, J. Hu, H. Wang, J. W. Zwanziger, M. L. Chabinyc and R. Seshadri, Crystal and Electronic Structures of Complex Bismuth Iodides A₃Bi₂I₉ (A = K, Rb, Cs) Related to Perovskite: Aiding the Rational Design of Photovoltaics, Chem. Mater., 2015, 27, 7137–7148.
23. S. Sun, S. Tominaka, J. Lee, F. Xie, P. D. Bristowe and A. K. Cheetham, Synthesis, crystal structure, and properties of a perovskite-related bismuth phase, (NH₄)₃Bi₂I₉, APL Mater., 2016, 4, 031101.
24. K. M. McCall, C. C. Stoumpos, O. Y. Kontsevoi, G. C. B. Alexander, B. W. Wessels and M. G. Kanatzidis, From 0D Cs₃Bi₂I₉ to 2D Cs₃Bi₂I₆Cl₃: Dimensional Expansion Induces a Direct Band Gap but Enhances Electron–Phonon Coupling, Chem. Mater., 2019, 31, 2644–2650.
25. K. Ahmad, P. Kumar, H. Kim and S. M. Mobin, Optoelectronic and Photovoltaic Properties of (NH₄)₃Bi₂I₉: A Perovskite-like Energy Material for Pb-free Perovskite Solar Cells, ChemNanoMat, 2022, 8, e20220006.
26. M. Pazoki, M. B. Johansson, H. Zhu, P. Broqvist, T. Edvinsson, G. Boschloo and E. M. J. Johansson, Bismuth iodide perovskite materials for solar cell applications: electronic structure, optical transitions, and directional charge transport, J. Phys. Chem. C, 2016, 120(51), 29039–29046.
27. F. Ünlü, A. Kulkarni, K. Lê, C. Bohr, A. Bliesener, S. D. Öz, A. K. Jena, Y. Ando, T. Miyasaka, T. Kirchartz and S. Mathur, Single- or double A-site cations in A₃Bi₂I₉ bismuth perovskites: What is the suitable choice?, J. Mater. Res., 2021, 36, 1794–1804.
28. G. Kieslich, S. Sun and A. K. Cheetham, Solid-state principles applied to organic–inorganic perovskites: new tricks for an old dog, Chem. Sci., 2014, 5, 4712–4715.
29. G. Kieslich, S. Sun and A. K. Cheetham, An extended tolerance factor approach for organic–inorganic perovskites, Chem. Sci., 2015, 6, 3430–3433.
30. M. Mączka, T. A. da Silva, W. Paraguassu, M. Ptak and K. Hermanowicz, Raman and IR Studies of Pressure- and Temperature-Induced Phase Transitions in [(CH₂)₃NH₂][Zn(HCOO)₃], Inorg. Chem., 2014, 53, 12650–12657.
31. T. Asaji, Y. Ito, H. Fujimori and B. Zhou, Ring-Puckering Motion of Azetidinium Cations in a Metal–Organic Perovskite [(CH₂)₃NH₂][M(HCOO)₃] (M = Zn, Mg)—A Thermal and 1H NMR Relaxation Study, J. Phys. Chem. C, 2019, 123, 4291–4298.
32. M. Rok, M. Moskwa, J. Hetmańczyk, Ł. Hetmańczyk and G. Bator, Switchable dielectric constant, structural and vibrational studies of double perovskite organic–inorganic hybrids: (azetidinium)₂[KCr(CN)₆] and (azetidinium)₂[KFe(CN)₆], CrystEngComm, 2022, 24, 4932–4939.
33. S. R. Pering, W. Deng, J. R. Troughton, P. S. Kubiak, D. Ghosh, R. G. Niemann, F. Brivio, F. E. Jeffrey, A. B. Walker, M. S. Islam, T. M. Watson, P. R. Raithby, A. L. Johnson, S. E. Lewis and P. J. Cameron, Azetidinium lead iodide for perovskite solar cells, J. Mater. Chem. A, 2017, 5, 20658–20665.
34. R. Panetta, G. Righini, M. Colapietro, L. Barba, D. Tedeschi, A. Polimeni, A. Ciccioli and A. Latini, Azetidinium lead iodide: synthesis, structural and physico-chemical characterization, J. Mater. Chem. A, 2018, 6, 10135–10148.
35. Y. U. Jin, B. Marler, A. D. Karabanov, K. Winkler, I. C. J. Yap, A. Dubey, L. Spee, M. E. Castillo, F. Muckel, A. N. Salak, N. Benson and D. C. Lupascu, Lead-free organic–inorganic azetidinium alternating metal cation bromide:[(CH₂)₃NH₂]₂AgBiBr₆, a perovskite-related absorber, RSC Adv., 2023, 13, 36079–36087.
36. A. Altomare, C. Cuocci, C. Giacovazzo, A. Moliterni, R. Rizzi, N. Corriero and A. Falcicchio, EXPO2013: a kit of tools for phasing crystal structures from powder data, J. Appl. Cryst., 2013, 46, 1231–1235.
37. R. Hoye, R. E. Brandt, A. Osherov, V. Stevanovic, S. D. Stranks, M. W. B. Wilson, H. Kim, A. J. Akey, R. C. Kurchin, J. R. Poindexter, E. N. Wang, M. G. Bawendi, V. Bulovic and T. Buonassisi, Methylammonium bismuth iodide as a lead-free, stable hybrid organic–inorganic solar absorber, Chem. – Eur. J., 2016, 22, 2605–2610.
38. P. Szklarz, A. Pietraszko, R. Jakubas, G. Bator, P. Zielinski and M. Gałazka, Structure, phase transitions and molecular dynamics of [C(NH₂)₃]₃[M₂I₉], M= Sb, Bi, J. Phys.: Condens. Matter, 2008, 20, 255221.
39. H. Günter, G. Schrem and H. Oberhammer, The gas-phase structure of azetidine: Microwave spectroscopy, and electron diffraction and normal coordinate analysis, J. Mol. Spectrosc., 1984, 104, 152–164.
40. P. H. Nielsen and M. Gajhede, Reassignment of the fundamental vibrations of azetidine from ab initio calculations, J. Phys. Org. Chem., 1989, 2, 183–186.
41. H. G. O. Becker, G. Domschke, E. Fanghänel, M. Fischer, K. Gewald, R. Mayer, D. Pavel, H. Schmidt and K. Schwetlick, Organikum, 1990, A.3.5, 86–88.
42. A. Nila, M. Baibarac, A. Matea, R. Mitran and I. Baltog, Exciton–phonon interactions in the Cs₃Bi₂I₉ crystal structure revealed by Raman spectroscopic studies, Phys. Status Solidi B, 2017, 254(No. 4), 1552805.
43. G. M. Paternò, N. Mishra, A. J. Barker, Z. Dang, G. Lanzani, L. Manna and A. Petrozza, Broadband Defects Emission and Enhanced Ligand Raman Scattering in 0D Cs₃Bi₂I₉ Colloidal Nanocrystals, Adv. Funct. Mater., 2019, 29, 1805299.
44. A. Miniewicz, R. Jakubas, C. Ecolivet and A. Girard, Raman scattering in ferroelectric (CH₃NH₃)₃Bi₂Br₉ single crystals, J. Raman Spectrosc., 1994, 25, 371–375.
45. L. El-Adel, A. Ouasri, A. Rhandour and L. Hajji, Raman-Infrared spectroscopy, thermal behaviour, dielectric, and UV-fluorescence studies of [C₆H₅NH₃]₃BiCl₆·3H₂O, Solid State Commun., 2021, 340, 114541.
46. A. Ouasri, F. Lambarki, R. Fakherddine, A. Aatiq and A. Rhandour, Structural characterisation, BFDH morphology, DSC, infrared and Raman studies of the disordered tetramethylammonium nonachlorodibismuthate [(CH₃)₄N]₃Bi₂Cl₉, Polyhedron, 2024, 251, 116875.
47. P. Kubelka and F. Munk, An article on optics of paint layers, Z. Tech. Phys., 1931, 12, 593–601.
48. P. Makuła, M. Pacia and W. Macyk, How to correctly determine the band gap energy of modified semiconductor photocatalysts based on UV-vis spectra, J. Phys. Chem. Lett., 2018, 9, 6814–6817.
49. F. Ünlü, M. Deo, S. Mathur, T. Kirchartz and A. Kulkarni, Bismuth-based halide perovskite and perovskite-inspired light absorbing materials for photovoltaics, J. Phys. D: Appl. Phys., 2022, 55, 113002.
50. X. Chen, Y. Myung, A. Thind, Z. Gao, B. Yin, M. Shen, S. B. Cho, P. Cheng, B. Sadtler, R. Mishra and P. Banerjee, Atmospheric pressure chemical vapor deposition of methylammonium bismuth iodide thin films, J. Mater. Chem. A, 2017, 5, 24728.
51. B. Park, B. Philippe, X. Zhang, H. Rensmo, G. Boschloo and E. M. J. Johansson, Bismuth Based Hybrid Perovskites A₃Bi₂I₉ (A: Methylammonium or Cesium) for Solar Cell Application, Adv. Mater., 2015, 27, 6806–6813.
52. C. Wu, Q. Zhang, G. Liu, Z. Zhang, D. Wang, B. Qu, Z. Chen and L. Xiao, From Pb to Bi: A promising family of Pb-free optoelectronic materials and devices, Adv. Energy Mater., 2020, 10, 1902496.
53. G. M. Paternò, N. Mishra, A. J. Barker, Z. Dang, G. Lanzani, L. Manna and A. Petrozza, Broadband Defects Emission and Enhanced Ligand Raman Scattering in 0D C_s3Bi₂I₉ Colloidal Nanocrystals, Adv. Funct. Mater., 2019, 29, 1805299.
54. T. Kawai, A. Ishii, T. Kitamura, S. Shimanuki, M. Iwata and Y. Ishibashi, Optical Absorption in Band-Edge Region of (CH₃NH₃)₃Bi₂I₉ Single Crystals, J. Phys. Soc. Jpn., 1996, 65(5), 1464–1468.
55. J. Klein, L. Kampermann, B. Mockenhaupt, M. Behrens, J. Strunk and G. Bacher, Limitations of the Tauc plot method, Adv. Funct. Mater., 2023, 33, 2304523.
56. D. Han, H. Shi, W. Ming, C. Zhou, B. Ma, B. Saparov, Y. Ma, S. Chen and M. Du, Unraveling luminescence mechanisms in zero-dimensional halide perovskites, J. Mater. Chem. C, 2018, 6, 6398–6405.
57. I. Rörich, Q. Niu, B. van der Zee, E. del Pino Rosendo, N. I. Crăciun, C. Ramanan and P. W. M. Blom, Exciton Quenching due to Hole Trap Formation in Aged Polymer Light-Emitting Diodes, Adv. Electron. Mater., 2020, 6, 1700643.
58. S. Athanasopoulos, E. Hennebicq, D. Beljonne and A. B. Walker, Trap Limited Exciton Transport in Conjugated Polymers, J. Phys. Chem. C, 2008, 112, 11532–11538.

---

Footnote

† Electronic supplementary information (ESI) available. CCDC 2333115, 2333116 and 2333118. For ESI and crystallographic data in CIF or other electronic format see DOI: https://doi.org/10.1039/d4qm00878b

---