Synthesis and physical characteristics of narrow bandgap chalcogenide SnZrSe ₃

Version Changes

Revised. Amendments from Version 1

The new version of the publication entails: 1. Updated Figure 1 (XRD patterns) and complemented paragraph about solid state reaction and XRD analysis. More experimental evidence were provided on how to obtain higher yield of SnZrSe ₃ phase. 2. Update on experimental details covering new experiment parameters. 3. Where relevant, elaborated discussions regarding the SnZrSe ₃ optical properties with respect to its non-perovskite structure. 4. Added three new references.

Abstract

Background: The development of organic/inorganic metal halide perovskites has seen unprecedent growth since their first recognition for applications in optoelectronic devices. However, their thermodynamic stability and toxicity remains a challenge considering wide-scale deployment in the future. This spurred an interest in search of perovskite-inspired materials which are expected to retain the advantageous material characteristics of halide perovskites, but with high thermodynamic stability and composed of earth-abundant and low toxicity elements. ABX ₃ chalcogenides (A, B=metals, X=Se, S) have been identified as potential class of materials meeting the aforementioned criteria.

Methods: In this work, we focus on studying tin zirconium selenide (SnZrSe ₃) relevant physical properties with an aim to evaluate its prospects for application in optoelectronics. SnZrSe ₃ powder and monocrystals were synthesized via solid state reaction in 600 – 800 °C temperature range. Crystalline structure was determined using single crystal and powder X-ray diffraction methods. The bandgap was estimated from diffused reflectance measurements on powder samples and electrical properties of crystals were analysed from temperature dependent I-V measurements.

Results: We found that SnZrSe ₃ crystals have a needle-like structure (space group – Pnma) with following unit cell parameters: a=9.5862(4) Å, b=3.84427(10) Å, c=14.3959(5) Å. The origin of the low symmetry crystalline structure was associated with stereochemical active electron lone pair of Sn cation. Estimated bandgap was around 1.15 eV which was higher than measured previously and predicted theoretically. Additionally, it was found that resistivity and conductivity type depended on the compound chemical composition.

Conclusions: Absorption edge in the infrared region and bipolar dopability makes SnZrSe ₃ an interesting material candidate for application in earth-abundant and non-toxic single/multi-junction solar cells or other infrared based optoelectronic devices.

Plain language summary

The sun provides an enormous amount of energy to the planet and has the potential to meet all humanity power demands. Photovoltaic (PV) technologies, i.e. devices that convert sunlight directly to electricity, play an essential role in harvesting this solar energy. While the sun will shine and provide clean energy for many million years to come, the manufacture of PV technologies uses finite earth resources and depending on the type of technology can be energy demanding and therefore emit large amounts of CO ₂. For these reasons, scientists are constantly improving synthesis and fabrication processes to increase material utilization, improving device structure to increase power conversion coefficient and study stability dynamics to prolong device lifetime. In parallel, new materials composed of earth-abundant and non-toxic chemical elements are explored which could potentially replace or complement current technologies. In this work, we focus on investigating a promising compound composed of non-toxic and earth-abundant elements from ABX ₃ chalcogenide family materials – SnZrSe ₃. Using a solid state reaction method, we synthesized SnZrSe ₃ crystals and measured their optical and electrical properties. We found that a critical semiconductor feature – bandgap (which indicates what wavelength light will be absorbed) of SnZrSe ₃ is in the suitable range for application in solar cells. Additionally, electrical measurements indicated that SnZrSe ₃ can be engineered to behave as n- or p-type semiconductor which is very important for the formation of semiconductor devices. In summary, we believe that SnZrSe ₃ is an interesting candidate for development of greener PV technologies.

Introduction

Perovskite class of materials describe compounds sharing the same type of crystal structure (perovskite) and have specific stoichiometry – ABX ₃, where A and B are cations, and X is an anion. The unprecedented development of organic/inorganic lead halide perovskites (HP) for application in light-emitting diodes ¹ , X-ray detectors ² and photovoltaics ³ has been witnessed in the last decade. Despite their tremendous success, poor thermodynamic stability under ambient conditions ⁴ and the presence of a toxic chemical element (Pb) ⁵ question their wide-scale deployment in the future. This spurred an interest in search of materials with the same advantageous physical properties as HP, but with much higher chemical stability and environmentally benign compositions ⁶ .

The success of HP is thought to be caused by an unusual material property – defect tolerance. This means that despite having a high density of crystallographic defects in the semiconductor, charge carrier lifetime on the order of µs can be readily obtained ⁷ . However, there is no consensus on the origins of the defect tolerance and it has been associated with multiple material aspects that are characteristic to metal HP, such as: special bonding-antibonding character when antibonding states are at the top of the valence band ⁸ , coordination environment and ionic nature of chemical bonds ^{6, 9} , and with a high static dielectric constant effectively screening charged defects ^{6, 10} .

When searching for perovskite-inspired materials it is therefore important to know how defect tolerance is related to the structural characteristics of the compound. For instance, it has been proposed that materials comprised of cations with lone electron pairs can potentially have defect tolerance characteristics ¹⁰ . Alternatively, a more straightforward way is to explore compounds with perovskite structure and ABX ₃ composition. For this reason, and because of high thermodynamic stability, low toxicity and earth-abundant chemical composition, chalcogenide perovskites (X=Se, S) have gained a lot of interest in recent years ¹¹ .

The first study on chalcogenide perovskite properties using first-principles calculations was introduced by Sun et al. ¹² (Note: that the study was limited to chalcogenide composition A ²⁺B ⁴⁺X ^2- ₃, where A=Ca, Sr, Ba; B=Ti, Zr; Hf, X=Se, S). It was shown that chalcogenide perovskites had photovoltaic-relevant properties and some even exceeded well-established materials used in solar cells, showing their great potential in photovoltaics. This sparked an interest in chalcogenide perovskites, of which BaZrS ₃ has drawn the most attention ^{13– 16} . It was, however, shown that among many ABX ₃ chalcogenides, under normal conditions, only very few exist as perovskite structures ¹⁷ : BaZrS ₃, BaHfS ₃, SrZrS ₃, SrHfS ₃, CaZrS ₃. All of them are wide bandgap (> 1.7 eV) semiconductors, and therefore have a relatively narrow application range in optoelectronics. Other ABX ₃ chalcogenides although in a non-perovskite ground phase have bandgaps spanning from near to mid-wavelength infrared regions ¹² . Additionally, because valence and conduction bands in edge(face)-sharing ABX ₃ chalcogenides are primarily determined by p and d orbitals of X and B elements, respectively, the absorption coefficient is expected to be high as well ¹⁸ . Note, however, that lower symmetry crystal structures such as needle-like or BaNiO ₃ type can lead to indirect optical transitions and therefore higher absorption onset as compared with ABX ₃ chalcogenides in perovskite-type structure.

In this work, we focus on examining SnZrSe ₃ which is predicted to have a narrow ~ 0.65 eV bandgap ¹⁹ . SnZrSe ₃ is composed of low-toxicity and earth-abundant chemical elements and is stable under ambient conditions. Additionally, SnZrSe ₃ and SnZrS ₃ are both stable in the needle-like phase suggesting a full chemical range of miscibility in SnZrSe _xS _3-x alloy and therefore a high tuneability of the bandgap. Experimentally estimated bandgaps for SnZrSe ₃ and SnZrS ₃ were 0.86 eV and 1.2 – 1.4 eV, respectively, however, the data are available only from one source ²⁰ . Therefore, herein we aim to synthesize and estimate the optical and electrical properties of SnZrSe ₃ for potential application in infrared-based optoelectronic devices.

SnZrSe ₃ was synthesized by solid state reaction in a powder form and as monocrystals. SnZrSe ₃ was found to crystallize in orthorhombic structure with space group Pnma as confirmed by single-crystal and powder diffraction methods. The absorption edge was estimated at around 1.15 eV indicating to the narrow bandgap nature of SnZrSe ₃, but considerably higher than reported before and predicted theoretically. Finally, depending on the chemical composition, SnZrSe ₃ was found to behave as an n- or p-type semiconductor highlighting bipolar dopability in SnZrSe ₃ compound.

Methods

SnZrSe ₃ samples were synthesized via a solid state reaction method. Elemental precursors comprising Sn (99.995%, AlfaAesar, -100 mesh), Zr (99.5% STREM Chemicals, -50 mesh), Se (99.999%, AlfaAesar, -200 mesh) and SnI ₂ (99.99%, SigmaAldrich, -10 mesh) were weighted in an Ar filled glove box. The total mass of the precursor before annealing was 0.5±0.005 g plus 0.01±0.005 g of SnI ₂. Then, the precursors were introduced to a quartz ampoule (inner diameter – 8 mm, outer – 10 mm, length – 150 mm) which was capped to protect the precursors from the ambient environment when taken outside the glovebox. Before sealing, all 16 ampoules were degassed for 30 – 60 min under approximately 2 Pa pressure. Four ampoules were immersed in an ultrasonic bath to remove precursors stuck to the quartz walls. Ampoules were sealed under vacuum using a flame from an oxygen and propane gas mixture and the final ampoule length was in the 60 – 80 mm range. First batch of ampoules had a carbonized inner wall to avoid quartz reaction with precursor materials during the synthesis step. However, we did not observe any traces of chemical reactions of precursors with a quartz ampoule and the carbonization step was not used for the rest of the samples.

Ampoules were introduced in the tube furnace, 13 of them in a horizontal and 3 of them in a vertical (placed in ceramic holder) positions at the centre of the heating zone. The heating process included either one or two steps: (i) temperature was raised to the top temperature within 3 – 7 h which was varied in the 600 – 800 °C range and held for 10 – 80 h; (ii) in the second step, the temperature was reduced to 100 – 600 °C within 7 – 60 h, and furnace power was switched off leaving to cool down naturally for a few hours. We found that larger crystals were formed when the temperature was reduced slowly in the second stage, instead of maintaining a high furnace temperature for a prolonged period of time.

Single crystal X-ray data of SnZrSe ₃ were obtained at 20 °C using an Xcalibur E diffractometer equipped with an Eos CCD space detector (Agilent Technologies) and a monochromatic source of MoK _α radiation (graphite monochromator, Oxford inst.). The data were collected and processed using the program CrysAlisPro (Oxford Diffraction Ltd., Version 1.171.37.35, provided with diffractometer) and were corrected for the Lorentz and polarization effects, and absorption. All calculations to solve the structures and to refine the proposed models were carried out with the SHELXS97 ²¹ and SHELXL2014 software packages ²² ( SHELX is free of charge for academic users).

The structure was refined by the full matrix least squares method on F2 with anisotropic displacement parameters using the program SHELXL ²¹ . Crystallographic data and structure refinement details and geometric parameters are given in CIF file (see Underlying data ²³ ). CIF files were deposited with the Cambridge Crystallographic Data Centre CCDC/ICSD, deposition number CSD 2166561, and can be accessed upon request ( https://www.ccdc.cam.ac.uk/).

Powder diffraction was measured using Rigaku diffractometer SmartLab with 9 kW rotating Cu anode in Bragg-Brentano geometry. Before measurements, crystals collected from the ampoule were grinded in an agate mortar. Diffractograms were recorded in the range of 10 to 60 2θ degrees with a scan step of 0.01 degree using linear D/tex ultra detector.

Raman measurements were performed using an inVia Raman microscope (Renishaw, Wotton-under Edge, UK) equipped with the 1800 lines/mm grating and thermoelectrically cooled (-70 °C) CCD camera at 532 nm wavelength excitation. Raman spectrum was taken using a 50x/0.75 NA (Leica) objective lens. Laser power was restricted to 0.05 mW and the integration time was set to 300 s. The Raman frequencies were calibrated using the silicon standard according to the line at 520.7 cm ^-1.

Diffuse reflectance was measured using Shimadzu UV-3600 two-beam spectrometer equipped with a multi-purpose compartment MPC-3100. Samples were placed in a 60 mm integrating sphere and a barium sulphate target was used for calibration.

Scanning Electron Microscope (SEM) images were taken using Helios Nanolab 650 equipped with a field emission gun. Chemical composition was recorded with an energy dispersive spectrometer (Oxford inst.) embedded in SEM. Before Energy Dispersive X-ray (EDX) measurements were made, calibration using a Cu plate was performed and other parameters such as working distance (5 mm), accelerating voltage (20 kV) and exposure time (30s) were kept constant.

Temperature dependent current-voltage measurements were recorded using a Keithley 6487 picoammeter/voltage source. The sample was placed in a closed helium cycle cryostat (Janis CCS-100/204) and the temperature was controlled via a digital temperature controller. Current-voltage measurements were made from 295 to 100 K with a 10 K temperature step size. Once the set temperature was reached on the thermocouple, it was waited for 3 min for temperature to equalize on the sample.

All data visualization and processing were performed using OriginPro 2021 version 9.8.0.200 ( https://www.originlab.com/index.aspx?go=Products/Origin) and files are accessible in Underlying data ²³ . Alternatively, data can be treated and analysed using freely available software Labplot ( https://labplot.kde.org/). Visualization of crystalline structure was done using freely available software Mercury ( https://www.ccdc.cam.ac.uk/solutions/csd-core/components/mercury/).

Results

SnZrSe ₃ solid state synthesis

SnZrSe ₃ was synthesized via solid state reaction from precursors in elemental form. Unless stated otherwise, a transport agent (SnI ₂) was used to accelerate the reaction and to grow single crystals. In other cases, no additives were introduced. In most of experiments, various shapes, sizes and forms of crystals were present inside an ampoule after annealing indicating formation of secondary phases. By analysing X-ray diffraction (XRD) data we found that in addition to SnZrSe ₃ phase, one or two of the following binary phases were present: ZrSe ₃ (ICDD# 03-065-2351), ZrSe ₂ (ICDD# 04-005-5128) and SnSe (ICDD# 04-009-2257). Based on multiple trials, we noticed that the following experimental parameters had a deciding factor on the phase composition: (i) spatial separation of precursor materials (ii) Se overpressure in the ampoule and (iii) annealing duration. When the ampoule was oriented vertically in the annealing zone, ZrSe ₃ and traces of SnSe phases were present despite the annealing temperature applied ( Figure 1a). Interestingly, the addition of transport agent in a form of SnI ₂ did not change the phase composition. This suggested that when precursors were in close proximity, the solid-state reaction was not accelerated or favoured for the formation of SnZrSe ₃ in the presence of a transport agent. When ampoules were oriented horizontally and precursor powder was spread over a certain length, phase composition strongly depended on the initial Se content. Using Se-deficient precursor, a large amount of ZrSe ₂ and absence of ZrSe ₃ phases were observed in the powder ( Figure 1b). Once stoichiometric or Se-rich precursors were used, little to no amount of ZrSe ₂ was found, which indicated that the formation of ZrSe ₃ phase was highly favoured under Se-rich atmosphere. Note that when the ampoules were oriented horizontally no secondary Sn-related phases were observed. Finally, the annealing duration at the top temperature had a significant effect on SnZrSe ₃ powder phase purity. In the initial experiments we used long annealing durations to facilitate a complete solid-state reaction. However, we noticed that shorter annealing durations yielded lower concentration of secondary phases ( Figure 1c). In the samples annealed under identical conditions but with different durations, we found that almost pure-phase powder was obtained annealing for 35 h instead of 60 h or more used previously ( Figure 1c). Longer or shorter than 35 h annealing durations resulted in the formation of ZrSe ₂ phase which concentration was much higher in the sample synthesized for a long period of time. This suggested that SnZrSe ₃ formed relatively quickly and under prolonged annealing time slowly decomposed into binary phases. Therefore, from these results we can infer that 700 °C is close to the decomposition temperature of SnZrSe ₃.

Figure 1. Phase composition analysis determined by X-ray diffraction method.

( a) Diffractograms of SnZrSe ₃ powder synthesized at various temperatures in a vertical position without and with transport agent. ( b) Diffractograms of SnZrSe ₃ powder samples synthesized in the same experiment at 740 °C temperature in a horizontal position with different initial Se-content in the precursor as indicated in the graph. ( c) Diffractograms of SnZrSe ₃ powder annealed for different durations. Top temperature – 700 °C, SnI ₂ transport agent, position – horizontal, precursor composition - stoichiometric. ( d) XRD pattern of collected and grinded crystals synthesized at 700 °C in horizontal position with a transport agent. Diffractogram of pure SnZrSe ₃ phase was simulated using CIF file ( underlying data ²³ ).

Based on the obtained results we propose a simplified reaction mechanism described below. During the temperature ramp in the synthesis process, it is suggested that SnSe is formed first, which crystallizes already at 300 °C ²⁴ :

$$Sn(l)+\frac{1}{x}S{e}_x(g)\to SnSe(s)(1)$$

where l, g and s indicate liquid, gaseous and solid states of matter; x denotes Se molecule size, which can be from 2 to 8 depending on the temperature ²⁵ . In parallel, Zr reacts with Se vapour and ZrSe ₂ is formed as the most stable form of Zr – Se system ^{26, 27} :

$$Zr(s)+\frac{1}{x}S{e}_x(g)\to ZrS{e}_2(s)(2)$$

Above 600 °C, SnSe enters a gaseous phase and starts to react with zirconium diselenide:

$$SnSe(g)+ZrS{e}_2(s)\to SnZrS{e}_3(s)(3)$$

When there is a surplus of Se in the atmosphere, a competing reaction occurs:

$$ZrS{e}_2(s)+\frac{1}{x}S{e}_x(g)\to ZrS{e}_3(s)(4)$$

Therefore, we believe that the thermodynamic balance between reactions (3) and (4) which depends on spatial separation of materials, partial Se pressure and annealing temperature determine the final product composition. While inspecting the ampoule annealed sequentially first at 700 °C and then at 800 °C, it was noticed that after the second annealing large SnZrSe ₃ crystals deteriorated, and small fine needle-like crystals typical for ZrSe ₃ were found (Figure S1 in Extended data ²³ ). This indicated that SnZrSe ₃ decomposes at higher temperatures. This observation was in agreement with the results obtained from time-dependent annealing series, where prolonged synthesis duration at 700 °C led to decomposition of SnZrSe ₃. Therefore, to synthesize SnZrSe ₃ phase with high yield, it is essential to keep balance between top temperature and annealing duration. With our experimental setup it was around 35 h at 700 °C. However, if lower temperature is applied, longer duration may be required and vice versa.

In this study, to obtain a pure single-phase SnZrSe ₃ powder, a part of large needle-shaped crystals was collected from the ampoule’s walls and was grinded in a mortar. Other crystals were used as-grown for single crystal XRD, Raman scattering and electrical measurements. The XRD pattern of thus obtained powder is presented in Figure 1d. A very good match was found between the experimental XRD pattern and the one simulated from single crystal XRD data. No traces of secondary phases were detected. Note that intensity of XRD peaks of (h0l) planes was higher than in the simulated pattern ( Figure 1d). This shows that a preferred orientation of (h0l) planes was present even in the powder sample because needle-like shaped grains tend to orient with their long axis parallel to the surface. Single-phase powder was later used for optical measurements.

Determination of the SnZrSe ₃ structure

To determine the crystalline structure of SnZrSe ₃, needle shaped crystal with dimensions of 0.25 × 0.02 × 0.015 mm and synthesized at 700 °C was selected for single crystal X-ray diffraction measurements. The summary of crystallographic information is presented in Table 1. More details on the refinement procedure and parameters can be found in the experimental description and crystallographic information file ( Underlying data ²³ ) attached to the publication.

Table 1. Summary of SnZrSe ₃ single crystal crystallographic information.

Crystallographic information was collected using single crystal XRD method and based on SnZrSe ₃ crystal synthesised at 700 °C temperature. The key information was selected.

Compound	Crystal system	Space group	Unit cell parameters
			a, Å	b, Å	c, Å
SnZrSe 3	Orthorhombic	Pnma	9.5862(4)	3.84427(10)	14.3959(5)
Atomic coordinates
Atom	WP	x	y	x	U(eq)
Sn(1)	4c	4556(1)	7500	6664(1)	22(1)
Zr(1)	4c	1608(1)	2500	5498(1)	12(1)
Se(1)	4c	2697(1)	2500	7191(1)	16(1)
Se(2)	4c	3318(1)	7500	4876(1)	13(1)
Se(3)	4c	-155(1)	7500	6104(1)	11(1)
Geometric parameters
Within ribbon (along b)		Intra ribbon (along a)		Inter ribbon (along c)
Bond	Length, Å	Bond	Length, Å	Bond	Length, Å
Sn(1)-Se(1)	2.7291(5)	Se(2)-Se(2)	3.771	Sn(1)-Se(3)	3.225
Sn(1)-Se(2)	2.8352(7)	Sn(1)-Se(2)	3.573	Se(1)-Se(3)	3.736
Zr(1)-Se(1)	2.6514(7)
Zr(1)-Se(2)	2.6798(5)
Zr(1)-Se(3)	2.7041(5)

It was determined that SnZrSe ₃ crystal is orthorhombic and belongs to space group Pnma ( Figure 2a). The calculated unit cell parameters were a = 9.5862(4) Å, b = 3.84427(10) Å, c = 14.3959(5) Å and unit cell volume V = 530.52(3) Å ³. The needle axis of the crystal habit corresponded to the shortest crystallographic axis b. The X-ray structural analysis showed that the crystalline structure of SnZrSe ₃ compound was isomorphous and isostructural to the sister compound SnZrS ₃ ²⁸ . The fundamental building block of the crystalline structure was a ribbon ( Figure 2b) which comprised of double edge-sharing Zr octahedra extending along b direction indefinitely. Within a ribbon and along b direction atoms are held by strong bonds as evidenced by the short interatomic distances of < 3 Å ( Table 1). On the contrary, ribbons themselves are held together via van der Waals forces because of the longer interatomic distances (> 3 Å) found along directions a and c ( Table 1). Within a unit cell cations occupy two non-equivalent sites Sn(1) and Zr(1), whereas anions have three distinct positions ( Figure 2b). Sn(1) is coordinated with three Se atoms forming a trigonal pyramidal geometry, whereas Zr is coordinated with six Se atoms forming a distorted octahedra geometry. Additionally, the bonding environment around Sn cation in terms of bond length was found to be anisotropic ( Figure 2c). Sn-Se bonds which are shorter than 3 Å are part of the ribbon structure, whereas in other directions they are longer than 3 Å giving a clear spatial separation between ribbons.

Figure 2. Crystalline structure of needle-like SnZrSe ₃ phase.

Single crystal XRD results confirmed that SnZrSe ₃ crystal is orthorhombic ( a) The main building block of SnZrSe ₃ structure is ribbon which is presented in ( b) with labelled atom sites. ( c) The projection of SnZrSe ₃ crystal structure on ( a, c) plane highlight the anisotropy around Sn cation. Dashed lines indicate > 3 Å interatomic distances in Sn coordination environment.

Optical characterisation of bandgap

For the optical characterization of SnZrSe ₃, single-phase powder as determined from XRD measurements was selected. To examine if there were any impurity phases present on the surface of the powder, Raman scattering measurements were conducted. As seen before, XRD results showed no evidence of secondary phases ( Figure 1c). For the Raman scattering study, additional samples were measured – a SnZrSe ₃ monocrystal which served as a reference case. Raman spectra contained vibrational bands located at 71, 119, 133, 160, 196 and 243 cm ^-1 ( Figure 3a). Because there is no reference Raman spectrum of SnZrSe ₃ in the literature, we first examined if there were secondary phases that had been observed in XRD patterns. Positions of the main and the most intensive Raman bands of ZrSe ₂ ²⁹ , SnSe ³⁰ and ZrSe ₃ ³¹ are shown as dotted lines ( Figure 3a). No evidence of ZrSe ₃ phase was found. However, one Raman band of SnSe (71 cm ^-1) and three bands of ZrSe ₂ (134 cm ^-1, 196 cm ^-1, 242 cm ^-1) overlap very well with some of the SnZrSe ₃ bands. Despite a good match, we believe it is not a response from secondary phases, but because of structural similarities between SnZrSe ₃ and ZrSe ₂ (SnSe) that give rise to similar vibrational bands. The main structural element in ZrSe ₂ is ZrX ₆ octahedral which is also the case in SnZrSe ₃ ( Figure 2b). Coordination environment of Sn in SnSe and SnZrSe ₃ also share similar structural features therefore rendering alike Raman bands. In addition, the Raman spectrum of SnZrSe ₃ monocrystal perfectly matched with Raman spectrum of the SnZrSe ₃ powder sample. This strongly supports that SnZrSe ₃ powder was free of secondary phases and overlapping bands were a result of similar structural characteristics between SnZrSe ₃ and ZrSe ₂ (SnSe).

Figure 3. Raman spectra and optical properties of SnZrSe ₃ powder.

( a) Raman scattering spectra of single phase SnZrSe ₃ powder and monocrystal synthesized at 700°C. Excitation wavelength - 532 nm. ( b) Diffuse reflectance spectrum and thereof calculated apparent absorption using eq. 5 of single-phase SnZrSe ₃ powder. Inset – Tauc plot of SnZrSe ₃ around absorption edge. ( c) Second derivative of measured apparent absorption as a function of energy of single-phase SnZrSe ₃ powder.

To estimate the absorption and to calculate optical bandgap and other critical points in the electronic band, diffuse reflectance was measured of the single-phase powder sample ( Figure 3b). According to the Schuster-Kubelka-Munk formulation, apparent absorption and diffuse reflection are related as follows ³² :

$$F(R_{\infty })=\frac{{(1-R_{\infty })}^2}{2R_{\infty }}=\frac{K}{S}(5)$$

Where, R _∞- diffuse reflectance, K – apparent absorption, S – reflection coefficient.

Assuming S did not change considerably over the measured energy range, F(R _∞) was taken to reflect the SnZrSe ₃ absorption coefficient. Diffused reflectance together with the calculated F(R _∞) are depicted in Figure 3b. To estimate the bandgap, we first applied the common Tauc method ³³ . In brief, material’s absorption coefficient (for photon energy above the bandgap) is proportional to the material’s bandgap as follows: α E _ph ≈C( E _ph - E _g ) ⁿ, where α is absorption coefficient, C – constant reflecting joint density of states in the bands, E _ph – is a photon energy, E _g – material’s bandgap and n -exponent depending on the optical transition nature: for direct transition n=2, for indirect n=0.5. Then, to find the bandgap value, ( αE _ph ) ⁿ is plotted versus E _ph and fitted linearly. A linear fitting region is extrapolated until it crosses x axis where α≈0 and the crossing point is defined as material’s bandgap. Direct transition was considered for SnZrSe ₃ therefore (F(R _∞)· E) ² vs E plot was used ( Figure 3b, inset). The fitted linear region near the absorption edge resulted in a bandgap value of around 1.16 eV. In the second approach, to estimate the position of critical points (CP) in the band structure, the second derivative of F(R _∞) was calculated ( Figure 3c). To reduce background noise and highlight CP features, data points were smoothed using Savitzky-Golay methods with 3 ^rd order polynomial and a 100 point window. Note, that we did not fit the derivative with the Aspnes’ function, because to obtain correct fitting results, high-accuracy measurements of material dielectric function are required. Nevertheless, the position of CP can be still estimated as the inflexion point of the CP feature, which typically has one positive and one negative extrema ( Figure 3c, indicated by arrows). In addition, to validate the certainty of CPs identified in the spectrum, we calculated second derivative under various smoothing conditions for the same sample and measured the diffuse reflectance on SnZrSe ₃ sample made from another batch. In all cases, inflexion points were located at the same positions (Figure S2 in Extended data ²³ ). First, clear CP features below the SnZrSe ₃ absorption edge were observed at 0.65 and 0.87 eV, respectively. The position of these CPs was in very close agreement with H ₂O absorption bands which are located at 1940 nm (~0.64 eV) and 1450 nm (~0.85 eV) ³⁴ . This indicated that H ₂O was present in BaSO ₄ which was used as a white reference plate in the diffuse reflectance measurements. Such a case is quite common when BaSO ₄ is used as a reference plate. Other CPs were located at 1.14 and 1.46 eV ( Figure 3c). The low energy CP was assigned to the SnZrSe ₃ bandgap because it coincided well with the value estimated from the Tauc plot ( Figure 3b). However, all other higher energy CPs cannot be assigned to a specific optical transition and is beyond the scope of this work. Based on the diffuse reflectance results, SnZrSe ₃ bandgap was around 1.15 eV which is almost twice as large as predicted from first-principles ¹⁹ and is substantially higher than measured by Richard ²⁰ . First-principles calculations using hybrid potentials can predict bandgaps consistent with experimental values therefore such large discrepancy we relate to one of the two reasons: (i) SnZrSe ₃ is an indirect bandgap semiconductor and using diffuse reflectance method we were not able to clearly detect weak indirect optical transitions which would be expected to be lower in energy than direct (ii) due to low density of states at the band edges, strong optical absorption starts with an high onset leading significant absorption only at higher photon energies ¹⁹ . The value measured by Richard was very close to the water absorption band located at 1450 nm. Although we consistently found a bandgap value of about 1.15 eV, more samples such as thin films, and other bandgap measurement methods would be useful to consolidate the real bandgap value of SnZrSe ₃ and nature of transition (direct/indirect).

Electrical properties

Conductivity type is a very important factor when considering the formation of semiconductor heterojunctions. Without intentional doping, conductivity usually depends on the dominating intrinsic point defects in the material, which in turn are related to the compound stoichiometry. To study conductivity properties of SnZrSe ₃, we first tested the largest needle-like crystals with a hot-point-probe method. This method allows us to identify the conductivity type by observing current/voltage sign upon increase in temperature gradient between probes. Before measurements, the system was calibrated with well-known n-type commercial fluorine doped SnO ₂ (SigmaAldrich) sample and boron doped commercial p-type (100) Si wafer. It turned out that some as-grown crystals showed a constant positive voltage change upon temperature gradient increase and some – negative ( Figure 4a). This indicated that SnZrSe ₃ can exhibit both n and p-type conductivity. Note that n-type behaviour was much more pronounced suggesting higher carrier concentration was present in n-type than in p-type crystals or much higher mobility of electrons. This was also in-line with calculated resistivity which was two orders of magnitude higher for p-type sample ( Table 2). To find if there was a relation between off-stoichiometry and conductivity type, we measured the chemical composition of crystals showing n and p-type behaviour, respectively. The average composition of the crystals measured over more than 5 points are summarized in Table 2. In both samples, the cationic ratio A/B was identical, whereas the n-type crystal was slightly more Se-rich than p-type, but both were Se-deficient with respect to stoichiometric ratio of [X]/([A]+[B]) which is 1.5. At this point, it is difficult to confirm if the different Se quantity was the origin of respective conductivity type and would require more samples to be synthesized and tested. In addition, there was quite a large compositional variation as seen from the high standard deviation. Important to note that since iodine was used as a transport agent to facilitate the solid state reaction during synthesis, it could be inadvertently introduced in the lattice giving rise to extrinsic doping. Iodine could not be detected by EDX, but we acknowledge that very small amounts (beyond the EDX detection limit) can have a significant contribution to the conductivity behaviour. In summary, although the origin of doping is not clear in SnZrSe ₃, the bipolar conductivity behaviour is a very desirable feature for semiconductor-based technologies ³⁵ .

Figure 4. Electrical properties of SnZrSe ₃ crystals.

( a) Voltage as a function of a temperature difference between probes measured on SnZrSe ₃ crystals using hot-point-probe method. Commercial FTO (fluorine doped SnO ₂) and boron doped Si wafer were used to calibrated system. ( b) Electrical conductivity as a function of reciprocal temperature in three single-crystal SnZrSe ₃ samples.

Table 2. Chemical composition and electrical parameters of SnZrSe ₃ crystals from F1 and K1 batches.

Two batches (F1 and K1) of crystals were studied which showed pronounced n- and p-type conductivity behaviour. Chemical composition clearly showed correlation with composition in uniform crystal samples.

Sample	Sn, at.%	Zr, at.%	Se, at.%	Sn/Zr	Se/(Sn+Zr)	ρ s, Ω·cm	E a1, eV	E a2, eV
n-F1 (HHP)	21.5±1.1	20.9±0.4	57.6±1.3	1.03	1.36
p-K1 (HHP)	22.1±1.5	21.4±0.9	56.5±1.0	1.03	1.30
F1-1	22.0±0.12	21.0±0.13	57.0±0.13	1.05	1.33	1.5·10 3	0.28	0.22
F1-2	21.0±0.52	21.5±0.12	57.5±0.41	0.98	1.35	3.9·10 4	0.24	0.20
K1-1	19.4±0.09	22.4±0.2	58.2±0.14	0.86	1.39	1.7·10 5	0.39	-

To estimate the ionization potential (E _a) of defects contributing to the conductivity, temperature dependent I-V curves were measured. To avoid grain boundary effects, as monolithic as possible SnZrSe ₃ samples were selected (Figure S3 in Extended data ²³ ). However, due to their small size, their conductivity type could not be measured directly by HPP and was assumed from the measurements of larger crystals from the same batch. We found that the resistivity of samples varied in 10 ³ – 10 ⁵ Ω·cm range highlighting the insulating nature of SnZrSe ₃. Calculated E _a of defects are summarized in Table 2 and Figure 4b. We see that for F1 series samples two E _a were calculated, but because conductivity is the product of carrier concentration and carrier mobility, it cannot be ruled out that reduction in E _a was because of the change of carrier scattering mechanisms, and therefore increased mobility upon temperature decrease. Overall, all defects that were found to contribute to the conductivity can be considered as deep defects, because E _a » 0.025 eV, which is a thermal excitation energy at room temperature (kT), k is Boltzmann constant and T – temperature (295 K). Chemical composition of the crystals was also measured ( Table 2) to link composition with conductivity behaviour. These crystals were much more homogenous as evidenced by a small standard deviation. Notably, the specific resistance was found to correlate with the Sn/Zr ratio: the more Sn-rich sample was, the smaller resistivity. In addition, it is likely that Sn-rich composition also led to n-type conductivity, whereas Sn-deficiency – to the p-type. This would also explain why in the Sn-deficient sample (K1-1) we observed very different E _a than in the other two cases. E _a in F1 samples was related to the donor defects, whereas in K1 – to the acceptor type defects. This was not obvious when the composition of large crystals was measured because of their chemical inhomogeneity. Resistivity can also be related to the gradual change of Se content as well. However, to rely on the Se ratio to the cations is less accurate, because we found small amounts of oxygen present in the crystals, especially if exposed for a long time in the ambient environment. The oxygen is most likely adsorbed from the air or in some amorphous state because phase composition determined by XRD of the powder sample that was stored for more than half a year under ambient conditions (T=20-25°C, RH=30-60%) did not change (Figure S4 in Extended data ²³ ). This also shows high thermodynamic stability of the SnZrSe ₃.

Discussion

A characteristic structural feature of perovskite compounds is that cation B forms an octahedra geometry with sharing corners ³⁶ . In fact, the stability of perovskite structures are predicted by estimating Goldschmidt’s tolerance factor, t, which reflects an ability to squeeze octahedra in a cubic sub-lattice ³⁷ . However, ABX ₃ chalcogenides do not follow predictions based on t only. None of the ABX ₃ chalcogenides exist in a perfect cubic structure even when t=1.0 (which is a golden ratio for cubic perovskite). Because chalcogen ionic radius is much larger than that of oxygen, the octahedral factor ( µ) must also be accounted for. Then, t is plotted against µ, it can be clearly seen that only a few ABX ₃ chalcogenides fall in the region of having a perovskite structure ¹⁷ . For SnZrSe ₃ calculated t and µ values are 0.79 and 0.36, which falls outside the perovskite region ( t > 0.85, µ > 0.4).

Instead, ABX ₃ chalcogenides are found to exist in distorted perovskite (model structure – GdFeO ₃), needle-like (model structure NH ₄CdCl ₃) and hexagonal (model structure BaNiO ₃) structures. Other crystal structures also exist in ABX ₃ chalcogenides, for example, CeTmS ₃-type in A ³⁺B ³⁺X ^2- ₃ ³⁸ , CuTaS ₃-type in A ¹⁺B ⁵⁺X ^2- ₃ ³⁹ and there are other possible variations although more rare. In needle-like and hexagonal structures, B cation octahedra is sharing edges (faces) instead of corners as is the case in SnZrSe ₃ ( Figure 2c). This will give rise to anisotropic carrier transport because low effective mass is expected in the direction of edge-sharing octahedra compared to other directions. Indeed, high anisotropy in electronic band dispersion was shown in SnZrS ₃ and SnZrSe ₃ (needle-like phase) using the first-principles calculations ¹⁹ .

Additionally, it is important to note the difference in bonding environment between SnZrSe ₃ and other needle-like ABX ₃ chalcogenides containing alkali or alkaline earth cations, for instance, SrZrSe ₃ or RbCdCl ₃. In SrZrSe ₃, cation A is positioned almost equidistantly from the nearest neighbouring atoms ⁴⁰ , whereas in SnZrSe ₃ as shown before there is a clear anisotropy in terms of interatomic distances ( Figure 2c). The origin of the ribbon-like structure in SnZrSe ₃ could be related to the stereochemically active lone electron pair. On theoretical grounds, it has been shown that stereochemically active lone pairs lead to distorted low symmetry crystal structures ⁴¹ . Sn in SnZrSe ₃ is in a +2 valence state which leads to two unpaired 5s electrons. For the binary compounds, if there is a strong interaction between cation s states and anion p states, electronic stabilization is achieved through lattice distortion and lone electron pair is ejected outwards forming a structural void. That leads to asymmetric bonding around lone electron pair containing cation. Many chalcogenides containing cation with lone electron pair such as Sb ₂Se ₃, Sb ₂S ₃, Bi ₂S ₃, SnHfS ₃, SnZrS ₃, PbHfS ₃ and PbZrS ₃ have a ribbon-like low symmetry crystal structure. Because of these structural similarities at least binary compounds also share some electronic and optical characteristics, for instance indirect bandgap with a small difference between direct and indirect gaps and anisotropic carrier transport properties ⁴² . Based on first-principles calculations, SnZrSe ₃ is predicted to have a small (< 0.1 eV) difference between indirect and direct gaps as well ¹⁹ . Such characteristic is highly desired in absorber materials for photovoltaics because high absorption coefficient and long carrier lifetime can be realised simultaneously ⁴² . On one hand absorption coefficient (α) depends on the density of states in the bands. If states at direct gap are dispersive it can lead to slowly increasing α resulting in significant absorption onset. Based on optical properties calculated for SnZrSe ₃, α > 10 ⁴ cm ^-1 is reached for photon energy > 1.5 eV which renders an onset of around 0.9 eV ¹⁹ . Another theoretical work also shows that among ABX ₃ chalcogenides those in needle-like phase have a larger absorption onset than those in perovskite structure ¹² , and this was measured for SrZrS ₃ case experimentally ⁴³ . On the hand, in materials with dispersive bands, a high carrier mobility can be expected. Consequently, for practical application, for instance solar cells, an optimum absorber thickness will be required to balance between charge carrier mobility, lifetime and absorption coefficient.

In this work, we found that SnZrSe ₃ showed p as well as n type conductivity behaviour. Ambipolar behaviour was also recently realised in BaZrS3 thin films ⁴⁴ . This is in contrast to other multicomponent well-known Cu-based photovoltaic materials such as Cu(In,Ga)Se ₂, Cu ₂ZnSn(Se,S) ₄, CuSb(Se,S) ₂ and Cu ₂Sn(Se,S) ₃ where usually one type carrier is dominant. Because of low formation energy of Cu vacancy defect (acceptor type), these materials are intrinsically p-type ^{45– 48} . Inability to alter the conductivity type and magnitude on demand, puts constrains on the device structure and requires formation of heterojunctions. On the contrary, if a semiconductor can be tuned to behave as p or n type, it opens up wider possibilities to design device structure, for example employing homojunctions and there is also a wider choice of partner layers for formation of heterojunctions. Bipolar dopability is therefore desired in the material because it facilitates the optimisation of device design for targeted application.

SnZrSe ₃ is a promising material candidate for photovoltaic application. Nonetheless, to really highlight the potential of this material, the deposition of SnZrSe ₃ in thin film form should be demonstrated and absolute values of optical absorption measured. This has not been done or reported thus far. Based on the experience in synthesis of other ABX ₃ chalcogenide thin films (BaZrS ₃) ¹⁶ , deposition process of SnZrSe ₃ could be challenging. Because of the large difference in vapour pressure of constituting elements, conventional chalcogenide thin film synthesis methods can be unsuitable. Therefore, likely alternative synthesis approaches should be explored to synthesize SnZrSe ₃ thin films.

Conclusions

In this work, we studied the properties of SnZrSe ₃ intending to explore ABX ₃ chalcogenide materials beyond the perovskite structure. We confirmed that the ground phase of SnZrSe ₃ is needle-like (s.g. Pnma) where the main building block was a ribbon forming a quasi-one-dimensional crystal structure. Coordination anisotropy around cation A was observed in SnZrSe ₃ which was a sign of a stereochemically active electron lone pair of Sn. The bandgap of SnZrSe ₃ was found to be 1.15 eV which is much smaller than in perovskite chalcogenides, therefore, broadening application range of ABX ₃ chalcogenides. In addition, we found that as-grown SnZrSe ₃ crystals were insulating (ρ _s = 10 ³-10 ⁵ Ω·cm), showed bipolar dopability and deep intrinsic defects. In terms of ribbon-like crystal structure and optical bandgap, SnZrSe ₃ has similar properties as Sb ₂X ₃ – which is one of the most perspective materials for earth-abundant and non-toxic photovoltaics, but SnZrSe ₃ offers a wider range of tunability in terms of doping and bandgap. However, the next important step in the validation of SnZrSe ₃ prospects is to find a synthesis approach for thin film deposition, which could be not as straightforward as evidenced from experience with perovskite chalcogenides.

Ethics and consent

Ethical approval and consent were not required.

Funding Statement

This project has received funding from the European Union’s Horizon 2020 research and innovation programme under the Marie Sklodowska-Curie grant agreement No 895046 (project Chalcopyrite-perovskites for infrared photovoltaics [IRPV]). This research was also supported by projects National Agency for Research and Development (ANCD) 20.80009.5007.15 and TK141 supported by European Union through the European Regional Development Fund, and by the Estonian Research Council under grant agreement No PRG1023.

Data availability

Underlying data

Zenodo: Dataset for publication “Synthesis and physical characteristics of narrow bandgap chalcogenide SnZrSe3”. https://doi.org/10.5281/zenodo.7867349 ²³ .

This project contains the following underlying data:

• –Sample description.txt (synthesis conditions of the samples presented in the publication).
• –F1IS_V_6244.cif (crystallographic information file about SnZrSe ₃ structure as determined by single-crystal XRD method; CIF can be opened with free of charge available software, e.g. VESTA ( https://jp-minerals.org/vesta/en/), Mercury ( https://www.ccdc.cam.ac.uk/solutions/csd-core/components/mercury/)).
• –XRD.zip (raw XRD patterns of powder samples presented in the publication in .ras and .raw formats; ras file can be read and plotted using open-source platform Labplot; raw/ras files can be opened and analysed in Profex (free of charge, https://www.profex-xrd.org/)).
• –Raman.zip (raw Raman spectra in .txt format and OriginPro project file where data was plotted; alternatively, Raman spectra can also be read and plotted using open-source platform Labplot ( https://labplot.kde.org/)).
• –Optics.zip (raw diffuse reflectance data in .txt format and OriginPro project file where data was processed and plotted; diffuse reflectance spectra can also be read, analysed and plotted using open-source platform Labplot ( https://labplot.kde.org/)).
• –JV-T.zip (I-V curves at specific temperature in .txt format and OriginPro project files where data was processed and plotted; I-V data files can be read, analysed and plotted using open-source platform Labplot ( https://labplot.kde.org/)).
• –Images.zip (optical photographs of the sample and untreated SEM images of crystals).

CIF files were deposited with the Cambridge Crystallographic Data Centre CCDC/ICSD, deposition number CSD 2166561, and can be accessed upon request ( https://www.ccdc.cam.ac.uk/structures/).

Extended data

Zenodo: Dataset for publication “Synthesis and physical characteristics of narrow bandgap chalcogenide SnZrSe3”. https://doi.org/10.5281/zenodo.7867349 ²³ .

This project contains the following extended data:

• –Extended data.pdf (additional information supporting claims in the publication with direct link to the main text, such as figures).

Data are available under the terms of the Creative Commons Attribution 4.0 International license (CC-BY 4.0).