Supramolecular Approach for Fine-Tuning of the Bright Luminescence from Zero-Dimensional Antimony(III) Halides

Abstract

Halides of ns² metal ions have recently regained broad research interest as bright narrowband and broadband emitters. Sb(III) is particularly appealing for its oxidative stability (compared to Ge²⁺ and Sn²⁺) and low toxicity (compared to Pb²⁺). Square pyramidal SbX₅ anion had thus far been the most common structural motif for realizing high luminescence efficiency, typically when cocrystallized with an organic cation. Luminescent hybrid organic–inorganic halides with octahedral coordination of Sb(III) remain understudied, whereas fully inorganic compounds show very limited structural engineerability. We show that the host–guest complexation of alkali metal cations with crown ethers fosters the formation of zero-dimensional Sb(III) halides and allows for adjusting the coordination number (5 or 6). The obtained compounds exhibit bright photoluminescence with quantum yields of up to 89% originating from self-trapped excitons, with emission energies, Stokes shifts, and luminescence lifetimes finely-adjustable by structural engineering. A combination of environmental stability and strong, intrinsic temperature-dependence of the luminescence lifetimes in the nanosecond-to-microsecond range nominate these compounds as highly potent luminophores for remote thermometry and thermography owing to their sensitivity range of 200–450 K and high specific sensitivities of 0.04 °C^–1.

Organic–inorganic hybrid compounds with structurally and electronically isolated (0D) metal halide anions and organic cations are gaining increasing attention since the past few years.¹⁻⁵ The interest stems from their color-tunable emission from self-trapped excitons (STE), often with high photoluminescence quantum yield (PL QY) already at room temperature (RT). Bright STE emission is commonly reported for metal ions with ns² electron configuration, for example, Ge²⁺, Sn²⁺, Pb²⁺, Sb³⁺, Bi³⁺, Te⁴⁺, and Cu⁺ with halide ligands.⁶⁻²³ Since STE is the result of exciton interaction with lattice and large structural reorganization in the excited state, emission in prevailing majority of such materials is spectrally broad and exhibits large Stokes shift. Depending upon the depth of trapping, PL QY of up to 100% at RT is not uncommon and, unlike, for example, semiconductor quantum dots, PL QY is rather insensitive to the surface states and density of carriers and traps. At present, STE emissive halides are researched for applications in solid-state lighting,^24,25 remote thermography,²⁶ and X-ray scintillation.^8,27

Generally, the coordination of ns² ions by halides adopts diverse geometries, such as pyramidal, tetrahedral, disphenoidal, square pyramidal, and octahedral, which, in turn, determine the luminescence characteristics. The same metal ion may assume several of these coordinations and rational control of the preferred structure remains a formidable challenge. While adjusting the choice of bulky organic cations, typically quaternary ammonium or phosphonium salts, seems to offer a vast space for structure engineering, the outcome of such trials remains rather unpredictable with regard to the resulting coordination geometry. In this study, we sought of devising an alternative structure engineering strategy and showcase it for one specific ns²-ion case—Sb(III) halides. The latter are appealing for their environmental stability and low toxicity. We also note that hybrid Sb(III) halides, all comprising SbX₅^2– anion, are known to exhibit bright PL at RT.^16,28−32 On the other hand, fully inorganic Sb halides, such as A₃Sb₂X₉(where M = Rb, Cs, methylammonium, X = Cl, Br, I), do not yield discrete 0D anions, but rather condensed, non- or poorly luminescent structures (dimers, 1D-chains, 2D-layers).³³⁻³⁶ By adopting a well-known supramolecular concept of host-guest coordination of alkali and alkaline earth metal cations (Cs⁺, Rb⁺, Ba²⁺, etc.) with crown ethers, we could deliberately form either SbX₅^2– or SbX₆^3– ions. We have then crystallized a library of Sb-halides, with bright PL spanning the visible spectral range, and rationalized their PL properties considering the STE geometries in relation to their crystal structures. We also show that some of these halides, such as [C@Ba]₄[SbCl₆]₂[Sb₂Cl₈], [C@Ba]₄[SbBr₆]₂[Sb₂Br₈], [C@Cs]₂CsSbBr₆, [C@Rb]₂SbBr₅ (with C being 18-crown-6), are promising thermoluminophores for applications in remote thermometry and thermography.²⁶

Typical 0D hybrids feature SbX₅^2– square pyramidal anions (Figure 1A).^{16,28−31,37,38} Discrete SbX₆^3– octahedra (Figure 1B) are very seldom.³⁸⁻⁴¹ Indeed, common organic cations are single charged (Ph₄P⁺, Et₄N⁺, etc.) and, hence, favor the lower charge on the halometalate anion. Our supramolecular approach readily serves both needs—controlled bulkiness and adjustable charge of the cation—as 18-crown-6 ether readily coordinates alkali metal (K⁺, Rb⁺, Cs⁺) or alkaline earth metal (Ba²⁺, Figure 1C). Consequently, the coordination number around Sb(III) could be adjusted in the obtained compounds (Figure 2), concomitantly with the tailoring of the emission color. The approach also allows for combining differently charged inorganic cations in one structure, such as in [C@Rb][C@Ba][SbBr₆] and [C@Cs][C@Ba][SbBr₆] (Figure 2). In the following, we correlate the structure of the obtained compounds with the PL characteristics. In this context, we would like to emphasize that such host-guest chemistry engineering is interesting not only for controlling formability of specific halometalate ions, as in this study, but also because it can alter the optical and other properties in non-obvious ways. An example of the latter is [K(dibenzo-18-crown-6)]₂MnX₄ (X = Cl, Br), which exhibits circularly polarized emission owing to the difference in the symmetry of the crystal structure.⁴² We would also note that the supramolecular approach, despite integration of crown molecules, still leads to a rather compact packing of Sb-halide anions. For instance, the molar volume of [C@Rb]₂SbCl₅ is just 70% of that of (Ph₄P)₂SbCl₅ (457 vs 655 cm³ mol^–1, respectively), while both compounds exhibit high PL QY.

Figure 1.

(A) Square pyramidal coordination of Sb³⁺ ion with C_4v point symmetry. (B) Octahedral coordination of Sb³⁺ ion with O_h point symmetry. (C) General chemical formula of the 18-crown-6 complex with alkali or alkaline-earth metal ions. (D) Photo of several obtained 0D Sb(III) halide compounds with a complex crown-ether/metal cation, namely, [C@Rb]₂SbBr₅, [C@Cs]₂CsSbBr₆, [C@Ba]₄[SbBr₆]₂[Sb₂Br₈], and [C@Ba]₄[SbCl₆]₂[Sb₂Cl₈] under UV (365 nm) excitation.

Figure 2.

(A) Synthetic approach towards obtaining compounds with SbX₅^2– anionic species that results in various crystal structures motives (half-filled atoms represent 50% static disorder). (B) Several synthetic approaches that result in SbX₆^3– anionic species and local geometry of the obtained compounds.

Two equivalents of single-charged complex cations result solely in square-pyramidal SbX₅^2– (Figure 2A). The SbBr₅^2– anion is “sandwiched” between two crown-ether metal complexes, yielding short distances between halide ions and inorganic cations. The crystal structure usually contains also solvent molecules, which is typical also for many 0D organic–inorganic hybrids. There are also several modes for solvent molecule coordination, of which one could highlight the appearance of an oxygen atom from DMF molecule in the Sb coordination sphere ([C@Cs]₂SbBr₅·DMF).

Discrete SbX₆^3– octahedra were successfully obtained in several ways, namely by having (i) three equivalents of A⁺ cation, (ii) one equivalent of A²⁺, and (iii) one equivalent of each A²⁺ and A⁺ (see Supporting Information, SI, for details). The first scenario results in a compound [C@Cs]₂CsSbBr₆, where only two Cs atoms form complex with the crown ether. This also leads to the packing of [C@Cs]₂CsSbBr₆ units in 1D fashion along the b direction (Figure S1). This compound can be viewed as the first antimony-based hybrid in between fully inorganic (e.g., Cs₃Sb₂Br₉) and compounds with organic cations. Scenario 2 yields compounds [C@Ba]₄[SbX₆]₂[Sb₂X₈] (X = Cl or Br, isostructural compounds, Figure S2D) that features two 0D Sb(III) halide species (Figure S3, partially shown in Figure 2). Apart from octahedral species (Figure S3B), the overall charge neutrality is maintained by the dimer [Sb₂Br₈]^2– species, which are built from two edge-sharing square pyramids (Figure S3C). Such edge-sharing SbX₅ pyramids are known from the literature; however, they display no emission at RT.⁴³ The scenario (iii) yields compounds [C@Rb][C@Ba][SbBr₆] and [C@Cs][C@Ba][SbBr₆], crystallizing in a high symmetry space group, R3a. This is due to the fact that the cumulative positive charge of 3+ from two cations is matching the negative charge of the anionic species. Crystal data and structure refinement details for all crystallized compounds as well as photos of crystalline samples can be found in Tables S1–9 and Figure S4.

Having a library of novel 0D Sb(III)-halides allows for correlating the anion geometry and other structural details with the PL characteristics (see optical properties on Figures 3 and S5–S7 and in Table S10). All obtained compounds are luminescent at RT, exhibiting broad emission bands (full width at half maxima, fwhm, of 350-510 meV) and large Stokes shifts for both square pyramidal (Figure 3A) or octahedral (Figure 3B) anions. While PL bands are located at 550–750 nm, PL excitation (PLE) spectra appears below 450 nm and contains several bands. The optical properties of isolated Sb³⁺ ions, as well as any ns² ion, can be described in terms of transitions between the states derived from Sb 5s and 5p states hybridized with ligand p states. Despite significant contributions from the ligand atomic orbitals, some features of the PLE and PL can still be simply interpreted using Seitz model.^44,45 In this model, the ground state (5s²) is a singlet denoted with atomic term ¹S₀ and the excited states (5s¹5p¹) consist of triplets and a singlet with corresponding atomic terms ³P₀, ³P₁, ³P₂ and ¹P₁, respectively.

Figure 3.

(A) Normalized PLE and PL spectra of the compounds with square pyramidal coordination, [C@A]₂SbX₅, where A = Rb, Cs and X = Cl, Br. (B) Normalized PLE and PL spectra of the compounds with octahedral coordination, [C@Ba]₄[SbBr₆]₂[Sb₂Br₈], [C@Rb][C@Ba][SbBr₆], [C@Cs][C@Ba][SbBr₆], and [C@Cs]₂CsSbBr₆. (C) Single-coordinate diagram depicting the excitation and emission processes from of the localized states. (D) PL peak, (E) Stokes shift, and (F) fwhm trends across different compounds at RT.

For 0D ns² group 4 metal halides, it has been recently demonstrated that various RT emission bands and their components can be ascribed to transitions between these states.⁸ A similar study has also been conducted on (Et₄N)₂SbCl₅ proving that the model is valid in case of Sb.²⁹ Seitz model can also be extended with molecular orbital theory to account for the ligand field influence.⁴⁶ Upon hybridization with molecular orbitals of the ligands, resulting molecular states are derived from respective atomic states (³P₀, ³P₁, ³P₂, and ¹P₁) and preserve multiplicity (singlet or triplet). Increasingly common in the recent literature is to outline the optical properties of the STE in ns² metal halides using a simple configurational diagram (Figure 3C). In this diagram, the ground and excited states are represented by potential energy curves. The abscissa axis in this diagram represents the change in geometry: zero point is taken as the ground state equilibrium geometry. The further on this coordinate is the excited state minimum, the bigger is the difference between the ground and excited-state geometries. This depiction readily visualizes the large PL Stokes shift. SbX₅^2– ions display emission shifted to the red region of the visible spectrum, unlike to emission from SbX₆^3– occurring at higher energies. Likewise, also the PLE onset for SbX₅^2– is found at lower energies. This suggests that higher coordination number leads to larger HOMO-LUMO gaps. For SbX₅^2–, the Stokes shift is much larger for chlorides than bromides (Figure 3E), which results from greater structural rearrangement in the excited state for chlorides. Interestingly, RT broadening has two contributors: intrinsic, related to the variation in the local environment of different emitting centers and extrinsic, caused by the energy states broadening at higher temperatures due to activation of vibrational states. The trends obtained from experimental data evidence that square pyramidal coordination results in narrower emission as compared to octahedral. The fwhm in SbX₅^2– compounds scale with the average Sb–X bond length for the same halide (Figure S8), which could be explained by stronger hybridization between the central ion and ligands causing larger thermal broadening. At the same time, no such correlation with the average bond length could be seen for SbX₆^3–, suggesting a dynamic effect discussed below.

Further insight into the structure-property relation was obtained with the aid of Density Functional Theory (DFT) calculations as implemented in the cp2k code (see SI for details). We have performed both ground state and excited state geometry optimization. Generally, the electronic structure of all ns² metal halides is similar in terms of atomic contributions. Overall, the electronic structure is highly localized, and the typical band structure diagram features flat non-dispersed bands (Figure S9). Because of this, we only concentrate on the analysis of the density of states (DOS). The highest occupied molecular orbital (HOMO) has an antibonding character and is derived from Sb-p and X-p states, while the lowest unoccupied molecular orbital (LUMO) also has an antibonding character and is derived from Sb-p and X-p states. A significant difference between square pyramidal and octahedral coordination is the stereoactivity of the lone pair (Figure 4A). In the case of SbX₅, the lone pair occupies a vacant axial position, stabilizing the square pyramidal geometry. There is no apparent evidence of the lone pair stereoactivity in SbX₆^3– from the structure determination (systematic axial bond elongation or Sb atom off-centering for stabilizing HOMO, e.g., first and second-order Janh-Teller distortions).⁴⁶ A less obvious player could be a dynamic distortion,^47,48 which may be a hidden cause for a broader emission from SbX₆ compounds. Figures 4B and S10 displays typical element- and orbital-decomposed DOS plots for SbX₅ and SbX₆.

Figure 4.

(A) Atomically projected molecular orbitals for SbX₅ (top) and SbBr₆ (bottom) molecular geometries. The HOMO in both cases has antibonding σ* nature, whereas LUMO has antibonding π* nature. (B) Cumulative DOS of [C@Rb]₂SbBr₅(top) and [C@Cs][C@Ba][SbBr₆] (bottom); other atomic contributions (O, C, H, etc.) are omitted for clarity. (C) The distortion of the ground state molecular geometry in the excited state for SbX₅ (top) and SbX₆ (bottom) geometries showcasing an axial metal halide bond extension.

In accordance with experimental PLE data, calculated HOMO–LUMO gap is smaller for square pyramidal geometry. SbX₅ LUMO has two nondegenerate energy states as expected for C_4v symmetry, whereas three excited states in SbX₆ LUMO are all degenerate due to O_h symmetry. In agreement, SbCl₅ compounds typically exhibit one additional emission band in the blue region, unlike SbCl₆^3–.^29,39 Typically for an STE, a large structural change occurs once electron is promoted to LUMO. This structural change has been previously accessed with DFT for various 0D Sn and Sb emitters.^{8,10−12,16,49} We have simulated the structural change by relaxing the triplet geometry in unrestricted Kohn-Sham method as implemented in the cp2k code (details can be found in SI). Figure 4C depicts the geometry change in the excited state. In both SbX₅ and SbX₆, the axial Sb–X bonds are elongating by the significant value of about 0.5 Å, while the equatorial Sb–X bonds are shortened by about 0.1 Å. In addition, in SbX₅, there is a unique mode of distortion: out-of-plane movement of Sb atom above the axial halides plane (Figure S11). While the relative change in bond length is comparable for Cl and Br ligands, the Sb displacement is stronger for Cl, explaining a larger difference in Stokes shift for SbBr₅ and SbCl₅(from 1.4 to 1.8 eV) in comparison to SbBr₆ and SbCl₆(1.45 to 1.4–1.3 eV). The energy states of both isolated octahedral and dimer units in [C@Ba]₄[SbBr₆]₂[Sb₂Br₈] lie close together in DOS (Figure S12). The HOMO states that originate from SbBr₆^3– are at 0.59 eV in the DOS, while LUMO is at 3.95 eV. The HOMO–LUMO difference equates to 3.36 eV (370 nm). For the Sb₂Br₈ dimeric unit, HOMO–LUMO gap is 2.95 eV (420 nm).

The onset of the experimental PLE spectrum is at 3.22 eV, rather close to the HOMO–LUMO gap of SbBr₆. In the literature, Sb₂Br₈^2– units were found to luminesce at cryogenic temperatures.⁴³ Taken together, these considerations suggest that the RT emission in [C@Ba]₄[SbBr₆]₂[Sb₂Br₈] can be assigned to SbBr₆^3– centers. DFT analysis also points to insignificant electronic hybridization with the DMF solvent molecule, commonly found to coordinate SbBr₅ anion in one of the obtained compounds (Figure S13).

We have recently demonstrated that STE-emissive ns² metal halides are highly promising luminophores for remote thermometry and thermography, as demonstrated with Cs₄SnX₆ and Sn(II) hybrids.²⁶ Luminescent Sb(III) halides make for a highly appealing alternative due to their much higher environmental stability (Sn²⁺ readily oxidizes). A candidate STE emitter must exhibit the following PL characteristics: strong temperature-dependence of the PL lifetime in the desired (for sensing) temperature range and variation of this lifetime in the nanosecond-to-microsecond range. The measured PL decays (Figure 5A and B; see also Table S11) can be fitted with either a monoexponential function or a biexponential with a shorter component. The corresponding radiative and nonradiative processes can be outlined with the configuration coordinate diagram (Figure 5C). The slower, radiative decay in this model (k_r) occurs from the center of the excited state parabola and the ground state. A competing, faster nonradiative decay path (k_nr) is a phonon-assisted de-trapping of the STE. This process has an activation barrier ΔE and, hence, accelerates at higher temperatures. Furthermore, such an increase in k_nr starts earlier (i.e., at lower temperatures) for compounds with larger Stokes shift (see a general correlation trend between the Stokes shifts in Table S10 and PL lifetimes at RT in Table S11). Consequently, at sufficiently low temperatures, STE-emitters shall ideally exhibit PL QY of 100% (in the absence of additional absorptive processes like the presence of impurities, etc., at the excitation wavelengths). At higher temperatures, there is a temperature range in which the PL QY and measured PL lifetime must rapidly drop. Remote measurement/mapping of the PL lifetime by, for instance, TCSPC or time-of-flight techniques, is the physical basis for remote thermometry/thermography. Structure-engineering of STE-emitters allows tuning of ΔE and hence the temperature range of PL lifetime sensitivity. Already RT values of PL QY and PL lifetime provide initial guidance as to the PL lifetime sensitivity range. For instance, moderate PL QY (20-60%), along with a PL lifetime of 0.1–1 μs, may indicate a partially quenched state at RT, and hence, such a compound may be suited as a thermoluminophore in a temperature range of approximately RT ± 50 °C. A material with a near-unity RT PL QY and a long PL lifetime (1–10 μs) is likely sensitive to the elevated temperatures (within its thermal durability window). Conversely, a compound with RT PL QY of just a few% or less and PL lifetime of few tens of ns is expected to exhibit a sensitivity range upon cooling or even at cryogenic temperatures. For Sb–X bond series, the iodides have the lowest vibrational frequency and are activated already at very low temperatures, thereby explaining why SbI₅^2– or SbI₆^3– compounds usually not emissive at RT.^41,50 On the contrary, chlorides have the highest vibrational frequency (Figure S14), which means that RT modes, responsible for emission quenching, are not sufficiently populated and this results in almost unity PL QY at RT.

Figure 5.

RT emission decay curves (355 nm laser excitation) for the selected compounds with (A) square pyramidal and (B) octahedral coordinations. (C) Single coordinate diagram model to account for the non-radiative (k_nr) exciton recombination pathway, showing excited states with different Stokes shift, corresponding to different temperatures of PL quenching. (D) Temperature-dependent average radiative decay time τ dependencies for the selected compounds: [C@Ba]₄[SbCl₆]₂[Sb₂Cl₈] (green), [C@Ba]₄[SbBr₆]₂[Sb₂Br₈] (yellow), [C@Cs]₂CsSbBr₆ (orange), and [C@Rb]₂SbBr₅ (red).

Figure 5D exemplifies PL lifetime vs T dependencies, showing, for instance, that [C@Cs]₂CsSbBr₆ can measure temperature upon cooling, whereas [C@Ba]₄[SbCl₆]₂[Sb₂Cl₈] is operational at higher temperatures. On the basis of these PL lifetime vs temperature curves, specific sensitivity α can be calculated (Figure S15), which is, for instance, up to 0.04 °C^–1 for [C@Cs]₂CsSbBr₆. This value is comparable with 0.05 °C^–1 for (C₄N₂H₁₄I)₄SnI₆ and Cs₄SnBr₆ compounds in our earlier study²⁶ and higher than that of commonly reported thermoluminophores (0.005–0.030).⁵¹

In conclusion, we have employed supramolecular chemistry strategy to selectively access a specific molecular geometry of antimony halide 0D centers, SbX₅ or SbX₆. This has enabled us to reveal the trends in compounds with a broad, largely Stokes shifted and highly efficient emission. The PL of synthesized compounds covers a major part of the visible spectrum (550–750 nm), with large Stokes shifts from 1 to 2 eV, while RT PL QY varies from 11 to 89%. Understanding the optical properties trends allows for the choice of the emitter, aiming for certain PL QY, Stokes shift, or spectral color of the emission. Furthermore, we have assessed the potential of such stable Sb-based emitters for a thermography application, revealing that by tuning the coordination and composition, thermoluminophores covering a broad temperature range (from 200 K to above 450 K) with good specific sensitivity (up to 0.04 °C^–1) were designed and synthesized.

Supporting Information Available

The Supporting Information is available free of charge at https://pubs.acs.org/doi/10.1021/acsmaterialslett.0c00174.

• Materials, synthesis, characterization, computational details, packing in [C@Cs]₂CsSbBr₆, experimental and simulated PXRD patterns, packing of the [C@Ba]₄[SbBr₆]₂[Sb₂Br₈], optical microscope images of crystalline samples, PL and PLE of the [C@Ba]₄[SbCl₆]₂[Sb₂Cl₈], comparison of PLE and Kubelka–Munk spectra, 2D PL/PLE maps, optical properties (PL peak position, fwhm, PLE peak, Stokes shift, PLQY) for the obtained antimony(III) halide hybrids, PL fwhm dependence on the average Sb–X bond, orbital-projected band structure plots, orbital-decomposed cumulative DOS, structural change from ground to the excited state, DOS of [C@Ba]₄[SbBr₆]₂[Sb₂Br₈] and [C@Cs]₂SbBr₅, results of mono- or biexponential fitting of the RT decay times, Raman spectra of [C@Ba]₄[SbCl₆]₂[Sb₂Cl₈] and [C@Ba]₄[SbBr₆]₂[Sb₂Br₈], specific sensitivity, and photostability test (PDF)

• Crystallographic data for [C@K]₂SbBr₅, [C@Ba]₄[SbBr₆]₂[Sb₂Br₈], [C@Cs][C@Ba][SbBr₆], [C@Rb]₂SbCl₅, [C@Cs]₂SbCl₅, [C@Rb]₂SbBr₅, [C@Cs]₂SbBr₅, [C@Cs]₂CsSbBr₆, and [C@Rb][C@Ba][SbBr₆] (CCDC 1997955) (ZIP)

The authors declare no competing financial interest.

Notes

Experimental and computational methods and crystallographic data (CIF files) for [C@Cs]₂SbBr₅, [C@Cs]₂SbCl₅, [C@Rb]₂SbBr₅, [C@Rb]₂SbCl₅, [C@Ba]₄[SbBr₆]₂[Sb₂Br₈], [C@Rb][C@Ba][SbBr₆], [C@Cs][C@Ba][SbBr₆], [C@Cs]₂CsSbBr₆ and [C@K]₂SbBr₅ are deposited in the Cambridge Crystallographic Data Centre database (CCDC) with the following deposition numbers: 1997940, 1998078, 1997957, 1998034, 1998064, 1997955, 1998033, 1997953, 1998071.