Efficient Lone-Pair-Driven Luminescence: Structure–Property Relationships in Emissive 5s² Metal Halides

Abstract

Low-dimensional metal halides have been the focus of intense investigations in recent years following the success of hybrid lead halide perovskites as optoelectronic materials. In particular, the light emission of low-dimensional halides based on the 5s² cations Sn²⁺ and Sb³⁺ has found utility in a variety of applications complementary to those of the three-dimensional halide perovskites because of its unusual properties such as broadband character and highly temperature-dependent lifetime. These properties derive from the exceptional chemistry of the 5s² lone pair, but the terminology and explanations given for such emission vary widely, hampering efforts to build a cohesive understanding of these materials that would lead to the development of efficient optoelectronic devices. In this Perspective, we provide a structural overview of these materials with a focus on the dynamics driven by the stereoactivity of the 5s² lone pair to identify the structural features that enable strong emission. We unite the different theoretical models that have been able to explain the success of these bright 5s² emission centers into a cohesive framework, which is then applied to the array of compounds recently developed by our group and other researchers, demonstrating its utility and generating a holistic picture of the field from the point of view of a materials chemist. We highlight those state-of-the-art materials and applications that demonstrate the unique capabilities of these versatile emissive centers and identify promising future directions in the field of low-dimensional 5s² metal halides.

The rise of lead halide perovskites APbX₃ (with A⁺ being a large cation and X^– a halide) as exceptional optoelectronic compounds has kindled immense interest in related main-group metal halides over the past decade.¹⁻⁴ These are unique semiconductors because of the mix of covalent bonding in the inorganic [PbX_6/2]⁻ framework and ionic bonding to the A cation, yielding a dynamic structure that allows for local polar fluctuations,⁵ electron–phonon coupling,⁶ and the formation of large polarons to screen charge carriers.⁷ The remarkable optoelectronic features of these three-dimensional (3D) materials include long charge carrier diffusion lengths and lifetimes^8,9 as well as bright, narrow emission in nanocrystalline form,¹⁰ properties that are enhanced drastically by their defect-tolerant nature. This defect tolerance is based on the antibonding properties of the 6s² lone pair of the heavy metal Pb²⁺ coupled to halide p orbitals, which forces defects to be very shallow or even lie outside the band gap.^3,11

To replicate or improve upon the lead halide perovskites, the community turned to main-group metals with ns² lone pairs, primarily Sn²⁺, Sb³⁺, and Bi³⁺.¹² Spurred by solar cell research, which requires higher dimensionalities and the heavier halogens to maintain the low band gaps needed to provide efficient solar absorption, initial efforts began with CsSnI₃,¹³ Cs₃M₂I₉ (M = Sb, Bi),^14,15 and the quasi-two-dimensional (2D) Ruddlesden–Popper perovskites.¹⁶ The shift to lower dimensionality permitted a great deal of structural flexibility, and a plethora of metal halides soon followed, from 3D double perovskites A₂M⁺M³⁺X₆ (M⁺ = Ag, In, Tl; M³⁺ = Sb, In, Bi),¹⁷ to fully zero-dimensional (0D) hybrid systems with large cations completely isolating square-pyramidal¹⁸ or octahedral¹⁹ metal halide units. Even considering only octahedral metal halides, the myriad choices of compatible cations (organic and inorganic) generate an impressive array of structure types (Figure 1), described further below.

Figure 1.

Structural versatility of octahedral metal halides: dimensional reduction of the aristotypical AMX₃ 3D perovskite to low-dimensional metal halides.

The varied dimensionalities of metal halide perovskites give rise to a plethora of different properties deriving from the disruption of the 3D perovskite framework. For example, in 2D hybrid perovskites, the alternating organic–inorganic layers lead to anisotropic charge transport²⁰ and dramatic enhancements in the exciton binding energy due to dielectric confinement.²¹ Compositional tuning is similarly critical. For example, the importance of electronic dimensionality has been demonstrated in the double perovskites, which often behave similarly to 0D systems despite their 3D crystal structure.^22,23

The structural confinement found in low-dimensional metal halides generates vastly different properties compared with their 3D counterparts, in particular with regard to the mechanism for light emission. Low dimensionality tends to favor the formation of self-trapped excitons (STEs),^24,25 the recombination of which yields broadband photoluminescence (PL) with generally longer lifetimes.²⁵⁻²⁷ The emission of related compounds has also been interpreted using the frameworks of ion emission and atomic orbital theory.²⁸⁻³¹ Recent work has begun to realize the potential of these low-dimensional metal halides for a variety of applications complementary to the solar cells and narrow emitters of the 3D halide perovskites, namely, in solid-state lighting,^18,32 X-ray scintillation,^33,34 and remote thermometry.³⁵ Of these, the highest photoluminescence quantum yields (PLQYs) at room temperature (RT) have been based on Sn²⁺ or Sb³⁺ cations containing the active 5s² lone pair.^{19,34,36−40}

In this Perspective, we seek to provide insight into the structure–property relationships that dictate the performance of low-dimensional, emissive 5s² materials. While the focus here is on 0D compounds, there is a rich chemistry of these materials that has not been fully explored for light emission, and thus, we begin with a structural overview. In particular, we highlight the stereoactive effects of the 5s² lone pair, which generates the appropriate atomic environments for such emission, and unify the language used to describe this emission by considering the terminologies of ion emission, the atomic orbital approach, and the self-trapped exciton model. We believe that this burgeoning field will benefit from a common mechanistic understanding and a summary of the available structure types that exhibit broadband emission.

Dimensionality and Emission in Octahedral Metal Halides

Here a brief introduction to dimensional reduction of metal halides is given through the series of perovskite-derivative structures based largely on corner-connected octahedral metal cations (Figure 1). We must mention that a variety of other non-perovskite polytypes exist, such as the face-shared octahedra of hexagonal perovskites⁴¹ or entirely non-octahedral structures, with similar diversity of dimensionalities and properties.^36,42−45 The aristotypical 3D perovskite exhibits a cubic structure, with lower-symmetry structures derived from octahedral tilts⁴⁶ in cases with a non-ideal Goldschmidt tolerance factor.¹ Halide perovskites exhibit band-edge emission at RT with high internal PLQYs and efficient photon recycling,^47,48 but self-absorption leads to weak PLQYs in bulk, so isolation as nanocrystals is needed to achieve emission with high external PLQYs.^4,10

The shift to 2D structures already leads to a variety of different structure types, the properties of which are dictated by the direction along which the perovskite framework is cut.^49,50 ⟨100⟩ 2D perovskites consist of flat slabs of octahedral layers interspersed with layers of large organic cations, yielding the Ruddlesden–Popper or Dion–Jacobsen perovskites (Figure 1).^2,16,51 The shape of the perovskite framework is maintained with such flat sheets, so the ⟨100⟩ 2D materials somewhat resemble their 3D counterparts with high absorption coefficients, dispersive band structures within the inorganic layer, and good solar cell efficiencies.^20,52,53 However, the introduction of large interlayer organic cations interrupts the covalent perovskite framework, generating a quantum well structure with anisotropic optical and electronic properties and extremely high exciton binding energies due to dielectric confinement.²¹ These large binding energies stabilize free excitons at room temperature, causing bulk ⟨100⟩ 2D perovskites to emit more efficiently at room temperature.⁵⁴ Broad emission from STEs is common at low temperatures but requires structural distortion of the Pb–X–Pb angle to manifest at room temperature,²⁶ so free exciton emission is most common in the ⟨100⟩ 2D perovskites.

The ⟨110⟩-type 2D perovskites possess further anisotropy, as corner-sharing occurs in a straight line along one direction while the other direction has zig-zag chains that disrupt charge transport within the inorganic layer, providing further confinement.^55,56 In some sense, the ⟨110⟩ 2D perovskites share more similarities to one-dimensional (1D) perovskites than to their ⟨100⟩ or 3D counterparts, and accordingly, the emission properties of these materials are drastically different, with the manifestation of RT STE emission in ⟨110⟩ perovskites.^26,57,58 This class represents an edge case in this regard, as it includes several ⟨110⟩ compositions exhibiting both free exciton and STE emission.^26,56

2D defect perovskites formed from trivalent cations (e.g., Sb³⁺ or Bi³⁺) must balance the excess charge on the M site with one-third occupancy of vacancies; this yields the formula A₃M₂□X₉ (where □ represents a vacancy) and forms bilayers cut along the ⟨111⟩ direction of the 3D perovskite (Figure 1).^59,60 It should be noted that the ⟨111⟩ direction of the 3D perovskite corresponds to the R symmetry point of the Brillioun zone, where the band edge states lie, and thus, the 2D defect perovskite most effectively breaks up the 3D perovskite. Furthermore, while this structure maintains 2D hexagonal symmetry within the octahedral layer, it is further removed from the 3D perovskite than the ⟨100⟩ or ⟨110⟩ 2D perovskites because no straight lines connect more than two octahedra.⁶¹ Accordingly, these compounds exhibit STEs and relatively poor charge transport properties compared with either 3D or ⟨100⟩ 2D perovskites.^14,61,62 For further information on 2D perovskites, we refer the reader to recent reviews that describe these compounds in greater detail.^50,51

Further reducing the dimensionality to 1D compounds begets wholly different properties compared with the 3D perovskites because of the limitation of orbital overlap to a single crystallographic axis. While there are a variety of corner-connected octahedral structures ranging from straight trans-connected chains to zigzag cis-connected chains (Figure 1)⁶³ and corrugated double chains,⁶⁴ their optical properties are relatively underexplored. The corrugated compounds Cs₃M₂Cl₉ (M = Sb, Bi) are known to be emissive at low temperatures only,⁶⁵ whereas efficient STE emission from 1D metal halides has been obtained from edge-shared double chains.⁶⁶

0D metal halides possess fully isolated octahedral units, which can be bridged by inorganic cations as in Cs₄SnBr₆ or large organic cations as in (C₄N₂H₁₄Br)₄SnBr₆.^19,37 While corner-connected octahedra form many higher-dimensional frameworks, corner-connected 0D clusters are quite rare: no divalent cations exhibit such anions, while only seven compounds are known to exhibit the dimeric [M₂X₁₁]^5– anion.⁶⁷

Instead, 0D octahedral clusters tend toward edge-sharing or face-sharing octahedra to help reduce the total charge of the anion. For example, in Sb-based dimers the total charge on the anion decreases from −5 in corner-shared Sb₂Br₁₁ to −4 in edge-shared Sb₂X₁₀ and −3 in face-shared Sb₂X₉ (Figure 1). Such polynuclear clusters can contain as many as eight or 18 octahedra for trivalent and bivalent cations, respectively.^44,68 However, such clusters have not been extensively explored for their optical properties, and hence, the focus here is on the fully isolated 0D structures built of mononuclear emissive metal centers. The emission of fully 0D metal halides is based wholly on STEs and is most efficient for the fully isolated octahedra, especially with large organic cations.^19,37

0D Units and Chemistry of the 5s² Lone Pair

In addition to octahedral units, ns² metal halides can form structures based on pyramidal MX₃ units, disphenoidal MX₄ units, and square-pyramidal MX₅ coordination (Figure 2),^31,43 which can be built into polymeric frameworks and chains just as in the case of the octahedral metal halides.^43,69 The increasing structural diversity with reduced dimensionality is illustrated in Table 1 for 5s²-based metal halides (Sn²⁺, Sb³⁺), with archetypes known to exhibit emission at room temperature marked in green. The coordinative flexibility of 0D units permits a wide range of structures, many of which possess emission at room temperature. It should be noted that the table is limited to dimeric units for clarity and that a variety of polynuclear 0D compounds exist beyond those listed.^43,44,87

Figure 2.

Atomic environments of 0D ns² metal halides, with the lone pair visualized in orange.

Table 1. Anionic 5s² Metal Halide Structural Units.

Excited states play a critical role in the optoelectronic properties of metal halides, and in these 0D complexes the geometry is dictated by the outer-shell ns² lone pair electrons.³¹ Hence, it is helpful to visualize the stereochemistry of the lone pair in each coordination environment, shown in orange in Figure 2. The lone pair induces a statically expressed distortion in these units, from the expected trigonal-planar to trigonal-pyramidal in MX₃ anions, tetragonal to disphenoidal in MX₄ anions, and trigonal-bipyramidal to square-pyramidal in MX₅.³¹

In contrast, the octahedral MX₆ anion can be found in either inert crystallographic environments, where the site symmetry is essentially cubic,^37,88 or in environments where the lone pair is expressed through a trigonal or tetragonal deformation.^19,85,89 Intriguingly, even the inert MX₆ octahedron exhibits some tendency toward off-centering, with signs of dynamic excited-state structural fluctuations observed in tellurium halides^88,90,91 and Cs₄MX₆.^37,92 The importance of such features in the optical performance will be discussed for the case of 5s² metal halides based on Sb³⁺ and Sn²⁺.

The 5s² lone pair has an exceptionally rich chemistry because of the combination of stereoactivity and structural flexibility compared with neighboring metals. Following the inert pair effect, the oxidative stability of ns² cations increases moving down the periodic table, with 6s² cations such as Pb²⁺ and Bi³⁺ having the most chemically inert lone pairs.^93,94 Lone pair electrons have strong stereochemical effects on the metal–ligand coordination environment; however, this lone pair expression is distinct from the chemical stability of the “inert” lone pair. The same larger 6s² cations offer more space for the lone pair and reduced surface charge density on the cation than their lighter neighbors, increasing the polarizability and coordination number. These features permit structurally inert lone pairs such as the highly symmetric structures of PbI₂ and BiI₃, which have centered metal cations with no obvious lone pair expression.⁶⁹

The 3D halide perovskites AMX₃ based on Ge²⁺, Sn²⁺, and Pb²⁺ are an instructive example to explore the differences between the 4s², 5s², and 6s² lone pairs. The small size of the Ge²⁺ cation does not leave sufficient space for the lone pair electrons, and thus, AGeX₃ structures possess significant off-centering of the Ge atom toward a face of the would-be octahedron, leading to a trigonal-pyramidal coordination.⁹⁵ Sn²⁺- and Pb²⁺-based AMX₃ compounds exhibit perovskite-related structures with crystallographically inert lone pairs (with the exception of the highly distorted low-temperature phases of MASnBr₃),⁹⁶ but this belies a dynamic nature that leads to polar fluctuations at room temperature⁵ and exceptional thermal expansion cofficients.⁹⁷ Stereochemical off-centering at high temperatures was first observed in CsSnBr₃, with the Sn atom exhibiting a trigonal shift toward the face of the octahedron with space group R3̅m,⁹⁸ exactly the same space group and off-centering found in the crystal structure of AGeI₃.⁹⁵ Subsequent studies have shown that this dynamic nature is more pronounced in the Sn²⁺ and Br^– analogues than in the respective Pb²⁺ or I^– compounds.⁹⁹

This highlights that lone pair expression is also heavily influenced by the halogen size. For example, the binary halide SnBr₂ exhibits a different structure than SnCl₂, SnI₂, PbCl₂, and PbBr₂ because of the combination of stereochemical influence of its lone pair electrons with the appropriate size disparity between the metal and halogen, despite the fact that SnCl₂ exhibits the strongest repulsion of the lone pair.¹⁰⁰ Similarly, while BiI₃ adopts a layered structure with perfectly symmetric [BiI₆]^3– octahedra, BiCl₃ exhibits trigonal distortions that lead to pyramidal coordination,¹⁰¹ requiring additional crystal forces to return to the high-symmetry octahedra of Cs₃BiCl₆.⁶⁴ The smaller size of Sb causes all three binary halides SbX₃ (X = Cl, Br, I) to exhibit such off-centering,¹⁰¹ but the templating influence of other cations can enforce high-symmetry crystallographic sites such as the [SbCl₆]^3– octahedra in Rb₇Sb₃Cl₁₆⁷⁹ and Cs₂NaSbCl₆.¹⁰² The case of Cs₂NaSbCl₆ is especially interesting, as the enforced octahedral symmetry on a normally expressed SbCl₃ lone pair leads to a large lattice constant of 10.7780 Å for Cs₂NaSbCl₆ while that of the isostructural Cs₂NaInCl₆ is 10.5313 Å,¹⁰² despite the similar ionic radii of Sb³⁺ and In³⁺ (0.76 vs 0.80 Å).¹⁰³ This points to a dynamic distortion of the [SbCl₆]^3– octahedron in which the stereochemistry of the lone pair is expressed dynamically while the cubic long-range symmetry is maintained.²⁹ However, this does not occur in the isostructural Cs₂NaSbBr₆, in which the ionic radius is consistent with the expected value,²⁹ highlighting the importance of metal–halogen overlap in determining the dynamics of the lone pair.

The contrast of lone pair stability and expression makes the 5s² lone pair unique, as it lies on the cusp between the dynamically expressed lone pairs of 6s² cations and the statically expressed lone pairs of the 4s² cations. This interplay of dynamic and static off-centering is at its peak in the 0D 5s² metal halides, permitting structural distortions that can yield efficient luminescence in materials where similar 4s² or 6s² structures have nonexistent or weak emission. For example, while Cs₄PbBr₆ does not emit at room temperature,^104,105 Cs₄SnBr₆ and (C₄N₂H₁₄Br)₄SnBr₆ have bright emission with PLQYs of 15%³⁷ and 95%,¹⁹ respectively. This can be understood as a function of the enhanced hybridization between the metal ns² lone pair and the halide np orbitals, with recent work showing that the higher energy of the Sn 5s level leads to the trend Cs₄SnBr₆ > Cs₄SnI₆ > Cs₄PbBr₆ favoring efficient STE emission in the former while the latter only emits at low temperatures.⁹² Similarly, in the disphenoidal metal bromides (bmpip)₂MBr₄, (bmpip)₂SnBr₄ has a PLQY of 75%, while that of (bmpip)₂PbBr₄ is only 24% and that of (bmpip)₂GeBr₄ is below 1%.³⁴ To better understand the broad, bright luminescence driven by the 0D 5s² lone pair, we bridge together the various models that have been utilized to characterize this emission.

Unifying Theory of 5s² Emission

Emission from the isolated center can manifest itself in many different forms, e.g., a dopant in a host matrix or pure material, as well as under different conditions, e.g., cryogenic or room temperature. Researchers from different fields of spectroscopy and materials science have contributed to isolated center emission studies. Because of this, different theoretical models are circulating in the literature. While these models share many features, the absence of a unified model (or the discussion of such a model) leads to contradicting and confusing terminology. Here we attempt to bring together the important theoretical concepts into a unified framework.

Two main models are presently utilized, arising from either (1) treating 0D centers as isolated ions dressed in the halide ligand field or (2) presenting these as the limiting case of complete electronic and structural isolation in the framework of an extended solid. Overall, both approaches lead to the same theoretical picture of the isolated MX_z centers, as illustrated in Figure 3A. Unification of the two models provides a framework that can explain and predict many properties of emissive 0D metal halides.

Figure 3.

(A) Schematic representation of the relationship between the atomic orbital model, the molecular orbital model, and the semiconductor model of an extended solid. (B) Energy band diagram associated with the free ns² ion. (C) Energy band diagram of the metal–halide molecular orbitals. (D) Configurational coordinate diagram of the simplified STE model in 0D 5s² metal halides. (E) Unified model with the configurational coordinate diagram, in which the ground and excited states are described using their atomic character as derived from the active ns² metal ion.

The first theoretical description is derived from the simple two-electron Seitz model (Figure 3B). In this model, a free ion with the ns² configuration has the singlet electronic ground state, denoted with atomic term symbol ¹S₀. The splitting of the nsnp excited states into ¹P (singlet) and ³P (triplet) states occurs as a result of the Coulomb (F) and exchange (G) interactions. Further, the ³P states are split into non-degenerate states ³P₂, ³P₁, and ³P₀ by the spin–orbit coupling interaction. Additionally, the overall energy of all of the described states is non-uniformly altered by the ligand field (in this case by the coordinating halides). This interaction with the ligand field for ns² ions includes a large degree of hybridization between the atomic orbitals of the central metal and ligands, which is described by molecular orbital (MO) theory (Figure 3C). In this case, the ground and excited states are localized on both central ions and ligands. Atomic term symbols in this description are substituted with molecular term symbols (i.e., ¹S₀ becomes a_1g). The relative energy of the latter depends directly on the molecular geometry (i.e., octahedral, square-pyramidal, disphenoidal) and has been summarized by Vogler and Nikol for different possible molecular geometries.³¹ However, it remains very common in the literature to use atomic term symbols even for systems with a high degree of hybridization and denote the resulting molecular states as ¹P₁-derived, ³P₀-derived, etc.^38,106,107 In this way, the model still serves its main purpose: to explain the transitions that occur between the states in terms of their allowed or forbidden nature, which can be experimentally reflected in the measured emission decay time (i.e., longer radiative lifetimes for partially forbidden transitions and shorter for allowed ones) or the intensity of the corresponding excitation band.^30,108

For example, the dramatic reduction of the radiative lifetime of Cs₂NaSbCl₆ from 1.4 ms at 5 K to 0.15 μs at higher temperatures with no change in the emission intensity until ∼200 K is derived entirely from changes in the character of the radiative excited state, from the forbidden ³P₀ → ¹S₀ transition at low temperatures to the ³P₁ → ¹S₀ transition by ∼25 K.²⁹ On the other hand, the MO approach becomes useful for comparing optical properties for different geometries as well as correctly predicting the ground- and excited-state geometries. This is especially crucial for these 0D systems in which the halide bonds are not shared, and thus, the metal halide anion and its lone pair are free to distort in the excited state, and the excited-state geometry is dramatically different than the ground-state geometry.

Historically, the Seitz model was widely used to interpret the optical properties of ns² ion impurities in binary alkali metal halides¹⁰⁹ or in various other oxide and halide hosts.^29,30 The STE model for interpretation of the optical properties of 0D metal halides became widespread only recently because of the research focus on ns² metal halide semiconductors. While free excitons are dominant in the prototypical 3D AMX₃ semiconductors, the structural versatility of metal halides yields many compounds where STEs dominate.⁵⁰

The typical theoretical description of phonon-assisted STE recombination involves a single configurational coordinate (SCC) diagram, in which the abscissa represents a change in molecular geometry upon excitation (Q) and the ordinate represents a change in energy (Figure 3D).¹¹⁰ Ground and excited states in this model are described as parabolic potential energy functions, which are harmonic in the first approximation. It should be noted that in this SCC diagram, the free exciton excited state would be represented as a parabola centered at the ground state with higher energy;⁵⁷ however, free excitons have not been observed in 0D 5s² metal halides, and hence, our model considers the localized self-trapped exciton as the only stable excited state. This is consistent with the description of Ueta et al. demonstrating that 1D or lower-dimensional systems have unstable free exciton states that localize upon any perturbation,²⁴ which means that the self-trapped exciton is the only relevant state in 0D 5s² metal halides. This simplifies the STE model to an SCC diagram essentially identical to the generalized vibronic transitions described by the Franck–Condon principle. Here we maintain the use of the term “self-trapped exciton” to be consistent with the literature on higher-dimensional metal halides.

The minimum of each parabola represents the equilibrium geometry of the ground or excited state, while the horizontal difference between these geometries, ΔQ, is interpreted as the structural change that the metal halide center undergoes upon excitation. The difference between the excited- and ground-state geometries is reflected in the large Stokes shift of 0D 5s² metal halide emission. Both absorption and emission transitions can occur from different vibronic states (denoted as flat lines in the parabola), thus yielding an amalgam of individual transitions with emission probabilities following a Poisson distribution. The convolution of these transitions results in a broadened emission peak, often with a tail toward lower energy.⁵⁹ The magnitude of the electron–phonon coupling is described by the Huang–Rhys parameter, which is larger for stronger coupling and defines the degree of broadening associated with STE emission.^59,110

The STE model’s configurational coordinate diagram is a simplified projection that represents a complex energy surface as a two-dimensional plot and therefore has limitations, especially in complex materials with multiple emissive centers. However, it has proven quite useful in characterizing STE emitters in recent years,^59,111,112 and in conjunction with the Toyozawa model,²⁴ important features such as the Huang–Rhys parameter and effective phonon energy can be derived. Importantly, this model also rationalizes a key property of STE emitters: temperature-dependent emission quenching.^24,108 The crossing of the ground- and excited-state parabolas provides a nonradiative relaxation pathway from the excited-state minimum to the crossing point purely through phonon excitation, known as hot nonradiative recombination.²⁴ The energy difference (height) between the intercept and the excited-state minimum defines the energy barrier (and thereby the temperature, as described further below) required to quench the radiative process (Figure 3D).

Thus, the Seitz model mainly focuses on the nature of the electronic states and, with the MO extension, on their degeneracy in the electric field of the ligand, while the SCC model describes phonon–exciton interactions. The successful interpretation of observed phenomena in the optical properties of 0D materials requires the aid of all three theoretical concepts. The unified model is depicted in Figure 3E. It is based on the SCC diagram, where the excited state can be split into several nondegenerate states, each reflecting the nature of the contributing atomic state. As we will show, the splitting and degeneracy of these states vary with the coordination geometry, but this model is flexible enough to describe the observed optical features.

Properties of Emissive 5s² 0D Materials

Among the various emissive 5s² 0D materials that have been reported, the simplest case with which to study such structure–property relationships comes from Te⁴⁺. The high charge of the metal center necessitates a greater number of halide ligands, restricting the known Te⁴⁺ halides to materials containing octahedrally coordinated Te⁴⁺ regardless of the nature of the countercation.^87,113 The most commonly adopted structure is that of the vacancy-ordered double perovskite A₂TeX₆ with regular octahedra.⁸⁷ The isostructural nature of these compounds has encouraged excellent fundamental studies; for example, the role of the halide in luminescence quenching has been studied in the series Cs₂TeX₆¹¹⁴ and A₂TeX₆,¹¹³ with chlorides showing the brightest emission at RT. The dynamic nature of these octahedra has also been extensively studied, with many reports on the dynamic Jahn–Teller effect-driven tetragonal distortions in the excited state.^{86,88,90,113,115} The apparent simplicity of these materials, however, has not been manifested in numerous luminescent examples, as the quenching temperatures are relatively low in these compounds,⁸⁶ and those that do emit at RT are plagued by low PLQYs. This appears to be due to an apparent concentration quenching effect⁸⁶ that is less pronounced in other 5s² 0D metal halides, but the root cause deserves further investigation given the exceptional performance of Sn²⁺ and Sb³⁺ emissive materials and Te⁴⁺’s distinct advantages of stability and its tendency to adopt a single coordination that encourages rational design.

At the other end of the spectrum, Sn²⁺ with its low charge exhibits the greatest structural diversity of 0D compounds, with examples of luminescent materials containing octahedra, disphenoids, and trigonal pyramids (Table 2 and Figure 4A). Additionally, a 0D structure containing square pyramids (Rb₃SnCl₃I₂) was also recently discovered; however, its potential for luminescence was left unexplored.¹¹⁶ Though Sn²⁺ is a versatile ion for materials discovery, its tendency to oxidize to Sn⁴⁺ increases the synthetic difficulty of discovering new materials.¹¹⁷ This may be one of the reasons why so far only seven room-temperature luminescent 0D structures have been reported. Of these, the most prominent examples consist primarily of disphenoids and octahedra (Figure 4).

Table 2. Summary of the Optical Properties of Highly Emissive 0D 5s² Metal Halide Compounds.

compound	coordination unit	Eexc/abs (nm)	Eem (nm)	Stokes shift (eV)	FWHM (eV)	PLQY (%)	lifetime (μs)	ref
Cs4SnBr6	MX6	340	540	1.28	0.51	15	0.54	(37)
HDM3SnBr8	MX6	350	600	1.60	0.44	86	3.12	(120)
(C4N2H14Br)4SnBr6	MX6	355	570	1.32	0.40	95	2.2	(19)
(PEA)6SnBr8[CCl2H2]2	MX6	340	566	1.46	0.44	89.5	2.7	(121)
(C4N2H14I)4SnI6	MX6	410	620	1.02	0.38	75	1.1	(19)
(bmpip)2SnBr4	MX4	350	658	1.66	0.37	75	4	(34)
(C6H22N4Cl3)SnCl3	MX3	280	638	2.48	0.42		7.6	(71)
Bzmim3SbCl6	MX6	342	525	1.26		87.5	2.4	(85)
Bzmim2SbCl5	MX5	375	600	1.24		22.3	2.6	(85)
Bmim2SbCl5	MX5	370	583	1.22		86.3	4.26	(18)
(C9NH20)2SbCl5	MX5	380	590	1.16	0.43	98	4.2	(19)
TTA2SbCl5	MX5	370	625	1.37	0.45	86	7.5	(38)
TEBA2SbCl5	MX5	360	590	1.34	0.48	98	7.7	(38)
(Ph4P)2SbCl5	MX5	375	648	1.39	0.41	87	4.6	(122)
(PPN)2SbCl5	MX5	410	635	1.07	0.44	98.1	4.1	(119)
Rb7Sb3Cl16	MX6, M2X10	365	560	1.16	0.53	3.8	0.2	(79)

Figure 4.

(A) 0D structures with statically expressed lone pairs (SnX₄, SbX₅) and the corresponding energy band diagram. (B) Dynamically distorted octahedral MX₆ and the corresponding band diagram. (C) Excitation-dependent PL of (bmpip)₂SnBr₄ showing singlet and triplet emission.³⁴ (D) PL and PL excitation (PLE) spectra of (TTA)₂SbCl₅ showing the separate excitation of the triplet and singlet states (reproduced using the previously reported synthesis³⁸ and characterized using an optical setup described elsewhere³⁴). (E) PL and PLE spectra of Cs₄SnBr₆,³⁷ (C₄N₂H₁₄Br)₄SnBr₆,¹⁹ and [18-crown-6]₂Cs₃SbBr₆.¹¹⁸

Intermediate in 0D structural diversity is trivalent Sb³⁺. Most of the luminescent examples in the literature are hybrid organic–inorganic materials with square-pyramidal [SbCl₅]^2– units isolated by large organic cations (Figure 4A); the remainder are octahedral or polynuclear octahedral units obtained in cation/alkali halide-rich compositions, e.g., (bzmim)₃SbCl₆ and Rb₇Sb₃Cl₁₆ (Figure 4 and Table 2).

From a purely structural perspective, too few structures are known at this point to predict the structure-directing effect that a specific cation may have. However, some empirical observations lead us to conclude that coordinatively unsaturated metal centers such as SnX₄ or MX₅ (M = Sn, Sb) can be obtained only through the use of (a) organic and sufficiently bulky countercations (e.g., bmpip, PPh₄, bzmim, etc.; Table 2) or (b) a combination of halides with very different sizes (e.g., mixed Cl^–, I^–).¹¹⁶ Smaller inorganic countercations can enable 0D structures in alkali halide-rich environments that sufficiently isolate the metal halide complex with high coordination numbers (e.g., Cs₄SnBr₆), but less alkali halide favors the formation of higher-dimensional frameworks (e.g., CsSnBr₃), as the small cations cannot fully isolate the 0D units (Table 1).

On the basis of these various Sn²⁺- and Sb³⁺-based materials, several examples can be compared to see how the geometry and the metal center affect the resulting optical properties. To examine these differences, the SCC diagrams for 0D Sn and Sb units can be redrawn on the basis of their geometry and the stereoactivity of the lone pair (Figure 4A,B). The “static” lone pair is expressed and found in coordinatively unsaturated geometries such as disphenoids (SnX₄) and square pyramids (SbX₅) (Figure 4A). In both cases, the reduced symmetry imposed by these motifs results in nondegenerate excited states, which may allow for additional transitions if the energy levels are sufficiently separated, although the separation may be quite small (Figure 4A).³¹ This has been demonstrated by Morad et al., with the observation of both ³P₁ → ¹S₀ and ¹P₁ → ¹S₀ transitions at RT in disphenoidal (bmpip)₂SnBr₄ (Figure 4C).³⁴ This also highlights the fact that the triplet STE state tends to be the lowest-energy excited state for such 5s² metals. This same model extends to square-pyramidal Sb (i.e., MX₅ in Table 2), for which Wang et al. first observed two distinct emission bands from (bmim)₂SbCl₅,¹⁸ and subsequently several other singlet-emitting hybrid antimony chlorides have been found.^19,119 One such compound is TTA₂SbCl₅, which further demonstrates the excitation-wavelength-dependent nature of the emission (Figure 4D).³⁸ High-energy excitation succeeds in populating the high-energy singlet ¹P₁ state, resulting in emission directly from this state as well as emission from the thermodynamically favored lower-energy ³P₁ triplet state.

This situation is simplified in octahedrally coordinated systems (Figure 4B). The larger coordination number reduces the static expression of the lone pair and therefore off-centering and distortion, often leading to high-symmetry sites and similar bond lengths. The regularity (or near-regularity) of such sites obscures the optical properties present in the disphenoidal and square-pyramidal coordinations, as the dynamic distortions of these octahedra lead to stronger mixing of the excited states. This is best illustrated by comparing three 5s² bromides: Cs₄SnBr₆, (C₄N₂H₁₄Br)₄SnBr₆, and [18-crown-6]₂Cs₃SbBr₆ (Figure 4E).^19,37,118 Although these three materials have different metal centers and different degrees of distortion or lone pair expression, the PL spectra appear to be triplet (³P_0,1,2)-dominated with one broad, featureless emission peak. This can be understood in terms of the higher degeneracy of the regular octahedral environment: while the trigonally distorted octahedron would have fully distinct singlet and triplet states, the regular octahedron does not fully separate these levels, and therefore, the singlet remains unobserved.³¹ This also correlates with the reduced Stokes shift and emission line width in octahedral compounds relative to their more distorted counterparts; e.g., Vogler and Nikol compared the absorption and emission of disphenoidal SbCl₄^– and octahedral SbCl₆^3– in solution.³¹

Given the enormous diversity of structures offered by these 5s² systems and their recent (re)discovery as optical materials, it is apparent that the community has only scratched the surface concerning their properties and how to utilize them. We highlight two rising applications that take advantage of the unique features of these 0D materials: X-ray scintillation and remote thermometry and thermography.

X-ray scintillation utilizes two features shared by these 0D 5s² metal halide materials: (1) efficient relaxation processes for high-energy absorption to the STE triplet state and (2) highly Stokes-shifted emission such that the compounds are transparent to their emission. The first such study demonstrated that (bmpip)₂SnBr₄ could function as an excellent X-ray scintillator (Figure 5A).³⁴ A very recent work found that square-pyramidal (PPN)₂SbCl₅ also exhibits efficient X-ray scintillation,¹¹⁹ and it is likely that many of these materials are radioluminescent. An open question is whether the long emission lifetimes (Table 2) are problematic for X-ray detection. Although these materials have low overall atomic numbers, they still perform comparably to commercially available NaI:Tl or CsI:Tl scintillators and seem to outperform organic scintillators. This class of materials deserves much more investigation in this context, and we hope to see collaborations between traditional scintillator groups and materials chemists to find effective scintillators among the 5s² metal halides.

Figure 5.

(A) Radioluminescence spectra of (bmpip)₂SnBr₄ and NaI:Tl under 50 kV Ag tube X-ray irradiation, with images of the NaI:Tl commercial scintillator and a (bmpip)₂SnBr₄ pellet under ambient light and X-ray irradiation. Adapted from ref (34). Copyright 2019 American Chemical Society. (B) Configurational coordinate diagram showing the change in energy barrier with different shifts of the excited state, with red and blue representing high and low ΔE (and thus quenching temperatures), respectively. (C) PL lifetime vs temperature for Rb₇Sb₃Cl₁₆,⁷⁹ Cs₄SnBr₆,³⁵ and (C₄N₂H₁₄I)₄SnI₆.³⁵

The second application, thermometry and thermography, is based on the steep temperature dependence of the PL lifetimes of these materials, as first demonstrated by our group for Cs₄SnBr₆, (C₄N₂H₁₄I)₄SnI₆, and [C(NH₂)₃]₂SnBr₄.³⁵ The PL lifetimes of these compounds vary by up to 2 orders of magnitude over a temperature range of 100 °C, giving rise to some of the highest specific sensitivities among known thermoluminophores. Temperature sensitivity is also observed in the PLQY of many materials, including these, but the dramatic changes in lifetime of 0D 5s² metal halides impart a much stronger sensitivity than simply utilizing the emission intensity could provide.³⁵ This property is rationalized by the unified model presented earlier, which is simplified in Figure 5B to highlight the effect of shifting the excited-state parabola. The intersection of two parabolas (colored and black curves) represents a nonradiative pathway back to the ground state. As the Stokes shift increases and the material’s PL red-shifts, the upper excited-state parabola shifts further from the ground-state parabola, lowering the energy barrier ΔE for the phonon-assisted “hot” nonradiative recombination.

Unlike conventional semiconductors, the PL lifetime in 0D metal halides is intrinsically robust, i.e., without noticeable effects of the structural defects and surfaces. This robust nature is a consequence of the strongly localized excited states of these 0D 5s² metal halides, which lead to excited-state energy bands with high Stokes shifts and momentum offsets that inhibit nonradiative recombination from defect states and favor hot recombination. The only other known source of nonradiative quenching comes from parasitic impurity phases such as CsSnBr₃, which absorbs in the emissive region of Cs₄SnBr₆,³⁷ but such quenching is only known to affect the PLQY.

Thus, hot nonradiative recombination dominates in 0D 5s² metal halides, dictated by the energy barrier ΔE that governs the quenching temperature of the luminescence, T_quench. ΔE, along with the average or effective phonon energy, E_ph, directly influences the emission lifetime and PLQY through the temperature-dependent competition between the radiative and nonradiative relaxation rates (k_r vs k_nr, respectively). The onset of quenching, with corresponding changes in lifetime, defines the temperature-sensitive regime. For example, in Figure 5C this regime is found at the higher temperatures, where k_r and k_nr compete effectively against one another and k_r remains non-negligible. In contrast, the flat region at low temperatures (<325 K for (C₄N₂H₁₄I)₄SnI₆; <250 K for Cs₄SnBr₆ and Rb₇Sb₃Cl₁₆) involves a purely radiative process, where k_nr is negligible, and the PL lifetime remains rather unaltered and the PLQY is saturated. This trend appears to be general not only for 0D Sn materials but also for 0D Sb materials (Figure 5C).⁷⁹ This robust thermal sensitivity, which can be tuned by structural engineering, and the wide temperature and lifetime ranges that it covers make 5s² metal halides a compelling class of novel thermoluminophores.

Cryogenic temperatures can present exceptions to this behavior that may also prove suitable for thermometric applications, as below 77 K in Rb₇Sb₃Cl_l6 a second temperature-sensitive regime can be observed without any change in the PLQY.⁷⁹ This purely radiative yet thermally sensitive regime is likely driven by changes in the character of the excited-state levels, as described above for Cs₂NaSbCl₆.^29,108 The unchanging intensity with concurrent temperature sensitivity ensures that the signal from the thermometric probe will remain strong throughout the entire sensitivity regime. This unique behavior may enable high thermometric sensitivities at extremely low temperatures (below 10–15 K), where accurate temperature values can be difficult to obtain through other means without altering the temperature of the system and hence a simple optical probe may be quite valuable.

Polynuclear 0D 5s² Compounds

Beyond these isolated 0D units, a plethora of polynuclear 0D cluster compounds based on 5s² metal cations exist.^43,78,87,123 Their optical potential remains to be thoroughly explored. RT PL has been reported for [Sb₂I₉]^3– dimers in Cs₃Sb₂I₉, but with rather weak emission intensity.⁵⁹ This may be due to the asymmetry in these polynuclear systems, which provides stable lone pair off-centering in coupled units that may enhance nonradiative recombination (Figure 6). However, the discovery of Rb₇Sb₃Cl₁₆ and its emissive distorted [Sb₂Cl₁₀]^4– dimer with a PLQY of 3.8% demonstrates that there are more optically active centers to be discovered.⁷⁹ Recent reports on a bright [Pb₃Cl₁₁]^5– cluster with near-unity PLQY provide further encouragement,¹²⁴ and we expect emissive clusters to be a rich avenue of exploration moving forward.

Figure 6.

Atomic environments of polynuclear 0D 5s² metal halides derived from square pyramids and octahedra, with the lone pairs visualized in orange.

Other Emissive Metal Centers

We also wish to highlight a recent series of metal halides based on ns⁰ (d¹⁰) cations that have exhibited remarkably efficient and broadband luminescence similar to that observed in the 5s² metal halides. Examples of nd¹⁰ emitters have been known for many years among the Cu⁺ and Ag⁺ halides,^125−127 but this field has been reinvigorated with a flurry of investigations in the last 5 years showcasing the utility of these materials in scintillators and light-emitting diodes (LEDs).^128−131 This group has been further bolstered by several new broadband-emissive compounds based on isoelectronic Zn²⁺ and Cd²⁺.^132−135 Mixing of nd¹⁰ns² systems has also led in some cases to enhanced emission,¹³⁶ and we expect that there is much more synergy to be found among such mixed-metal systems. In particular, combining the efficient excitation of low-band-gap ns²-based materials with highly emissive nd¹⁰ centers could make for attractive white-light phosphors.

In the case of In³⁺, the reported emission characteristics are extraordinarily similar to those of Sb³⁺-doped indium halide materials.^39,137−139 A new report found that undoped and Sb³⁺-doped Cs₂InCl₅·H₂O had essentially identical features and temperature-dependent intensities,³⁹ while recent work in our group showed that pure Cs₂NaInCl₆ and Cs₂InCl₅·H₂O do not emit.¹⁴⁰ In our view, trace Sb³⁺ dopants in the precursors may be responsible for these emissions, similar to the miniscule levels of Sn impurities that were found by Mitzi et al. to generate STE emission in PEA₂PbBr₄.¹⁴¹ Careful experimental work would be required to demonstrate this, perhaps utilizing InX monohalides as indium precursors that would be chemically less likely to contain trace Sb³⁺ impurities. A very recent report further validates this view by demonstrating that pure Cs₂InCl₅·H₂O does not luminesce without Sb³⁺ doping.¹⁰⁷

Summary and Outlook

The future is bright for emissive lone pair materials, and the structural versatility of 5s² metal halides yields a wide compositional space for the design of emitters with unique properties. Low-dimensional emissive 5s² metal halides offer several advantages: quantum efficiencies near unity; earth-abundant, nontoxic metal cations; intrinsic emission that does not require doping, electronic passivation or particle size control to achieve high performance; and large Stokes shifts that minimize self-absorption. Applications of such compounds include X-ray scintillation, phosphors for white-light emission, and high-resolution thermography, and we expect great strides in each of these fields in the coming years. That being said, challenges persist that will need to be overcome to enable the development of successful devices that fulfill the promise of these applications.

Many open questions deserve further exploration to understand the structure–property relationships and unveil the practical potential of these compounds. At present, it is difficult to predict the coordination environment generated by different organic cations, and finding design principles will require many more examples to tease out the effects of cation size, polarity, bonding characteristics, etc. Even in the absence of such design principles, several important trends can be observed on the basis of the available literature. For example, nearly all examples of highly emissive compounds in Table 2 can be grouped into two kinds, namely, Sn(II) bromides and Sb(III) chlorides, indicating that the orbital overlap and/or phonon energies in these combinations are particularly suited for highly efficient emission. In the Cs₂TeX₆ system, the RT emission intensity increases for lighter halides, similar to Sb(III) halides but contrary to Sn(II) bromides. Further studies, such as the computational comparison of Cs₄SnBr₆, Cs₄SnI₆, and Cs₄PbBr₆ that identified the importance of the Sn 5s²–Br 4p overlap,⁹² are needed to determine whether the energy levels of Sn(II) and Br are ideal for efficient emission or whether the lower stability of Sn(II) chlorides is the key to the observed differences. The structural and photophysical aspects of the excited state are rather poorly understood. While we have elaborated on some intriguing trends in the structure–property relationships of these compounds and the connection to the 5s² lone pair, a true picture of the dynamics of these systems is currently lacking. Besides commonplace transient absorption spectroscopy, excited states of these materials can be probed by ultrafast measurements that combine optical excitation and structure probing (ultrafast electron or X-ray scattering).^{50,57,99,142−144}

We also observe that the hybrid organic–inorganic materials routinely exhibit higher PLQYs than their fully inorganic counterparts, which may indicate more effective 0D isolation in the former materials with large and soft organic cations (Table 2). This likely derives from the lower-energy phonons and their enhanced coupling in inorganic materials, whereas the high-energy phonons of the organics improve the structural isolation of the lone-pair-driven vibrational modes in these 0D materials and push the temperature region for thermal quenching to higher temperatures (Figure 5C). However, the inorganic compositions benefit from enhanced stability and higher melting points, as manifested in the ability to prepare nanocrystals (NCs) of Cs₄SnBr₆,^145,146 which has not yet been demonstrated for any hybrid organic–inorganic 0D compositions.

The properties that allow them to excel as thermosensitive phosphors may hinder the utility of metal halides as solid-state lighting phosphors. During LED operation, the excitation source typically heats the phosphor, with local surface temperatures potentially exceeding 70 °C.^147,148 While hybrid organic–inorganic metal halides generally exhibit higher PLQYs at ambient and elevated temperatures, their chemical stability becomes a limiting factor. Furthermore, the majority of these materials exhibit excitation peaks below 400 nm, which are unsuitable for use even with purple LEDs (405 nm) and require them to be paired with UV LEDs (for which an inexpensive and efficient UV light source is unavailable at present) for use as downconversion phosphors in white-light LEDs.¹⁴⁹ The combination of these two limitations necessitates further structural exploration of 5s²-based materials suitable for white-light LEDs.

With regard to application in scintillators, a pertinent challenge is their intrinsically slow radiative recombination, in the low microsecond range (Table 2). Time-resolved radioluminescence studies are needed to determine the decay time under X-ray excitation and the possibility of afterglow. These compounds may therefore not be suitable for applications that require photon counting and/or spectroscopic X-ray energy resolution capabilities, such as positron emission tomography and nuclear security. However, static X-ray imaging such as used in dental and physician practices would benefit from the low-cost nature of these materials and the high reported sensitivity,¹¹⁹ which enables a low dose rate to the patient.

So far, we have found that emission characteristics and durability of known 5s² metal halides neatly match the theoretical and practical requirements of thermometric applications. The intrinsic nature of the STE emission makes the emission characteristics rather insensitive to the particle size, surface, and common structural defects and hence the synthesis method. The temperature range of high sensitivity is exceptional and broadly tunable by chemical design, while the respective PL lifetimes lie in convenient ranges for various detection schemes (ns to μs). Future work should also focus on the design of nanoscopic thermoluminescent probes. This requires the synthesis of size-tunable and environmentally stable nanoparticles of 0D metal halides. For high-resolution thermal imaging of biological systems, these NCs must be water-compatible and water-dispersible.

Biographies

Kyle M. McCall received his Ph.D. in Applied Physics from Northwestern University in 2019 under the joint supervision of Professors Mercouri Kanatzidis and Bruce Wessels. He is now working as a postdoctoral researcher under Professor Maksym V. Kovalenko in the Functional Inorganic Materials laboratory at ETH Zürich. His research interests focus on halide materials for radiation detection and light emission.

Viktoriia Morad received her M.S. in Chemistry from ETH Zürich in 2018 and since then has been pursuing a Ph.D. under the supervision of Prof. Maksym V. Kovalenko at ETH Zürich. Her research is focused on synthesis and applications of metal halides for light emission.

Bogdan M. Benin received his M.S. in Chemistry from Kent State University in 2016 and is now pursuing his Ph.D. under the supervision of Prof. Dr. Maksym V. Kovalenko at ETH Zürich. His research currently focuses on the discovery and application of luminescent zero-dimensional lead-free metal halides.

Maksym V. Kovalenko received his Ph.D. from Johannes Kepler University Linz in 2007 and was a postdoctoral researcher at the University of Chicago in 2008–2011. Since 2011 he has led the Functional Inorganic Materials group at ETH Zürich and Empa. He has been a full professor since 2020. The research activities of his group focus on discovery and engineering of novel inorganic materials and their applications for light emission and detection as well as in rechargeable batteries.

The authors declare no competing financial interest.