Research ArticleMATERIALS SCIENCE

Magnetizing lead-free halide double perovskites

See allHide authors and affiliations

Science Advances  06 Nov 2020:
Vol. 6, no. 45, eabb5381
DOI: 10.1126/sciadv.abb5381

Abstract

Spintronics holds great potential for next-generation high-speed and low–power consumption information technology. Recently, lead halide perovskites (LHPs), which have gained great success in optoelectronics, also show interesting magnetic properties. However, the spin-related properties in LHPs originate from the spin-orbit coupling of Pb, limiting further development of these materials in spintronics. Here, we demonstrate a new generation of halide perovskites, by alloying magnetic elements into optoelectronic double perovskites, which provide rich chemical and structural diversities to host different magnetic elements. In our iron-alloyed double perovskite, Cs2Ag(Bi:Fe)Br6, Fe3+ replaces Bi3+ and forms FeBr6 clusters that homogenously distribute throughout the double perovskite crystals. We observe a strong temperature-dependent magnetic response at temperatures below 30 K, which is tentatively attributed to a weak ferromagnetic or antiferromagnetic response from localized regions. We anticipate that this work will stimulate future efforts in exploring this simple yet efficient approach to develop new spintronic materials based on lead-free double perovskites.

INTRODUCTION

As a new generation of solution-processed semiconductors, lead halide perovskites (LHPs) have taken a dominant position within the portfolio of compounds due to their outstanding optoelectronic properties, including high carrier mobility, long carrier lifetime, low trap density, and high absorption coefficient. These properties result in high-performance optoelectronic devices, including solar cells (1, 2), light-emitting diodes (3, 4), lasers (5), and photodetectors (6, 7). Very recently, LHPs also show interesting spin-related properties due to spin-orbit coupling (SOC) (8) of Pb. Thus, they exhibit spin-dependent optical selection rules (9, 10), large-scale Rashba and Dresselhaus splitting (1113), magneto-optical effects (14, 15), and polarized light-related effects (16). These observations make LHPs interesting for spintronic devices (17). However, the magnetic properties resulting from the SOC are weak, therefore limiting their further development in the field.

In addition to SOC enhancement, a straightforward yet more efficient approach to achieving spintronic materials is magnetic doping/alloying into nonmagnetic semiconductors. This approach has played an increasingly prominent role to revolutionize electronics and spintronics in conventional semiconductors and has also been very recently examined in LHPs, showing optically switched magnetism at 5 K (18). Since the magnetic properties in this approach mainly originate from magnetic dopants, it is more desired to turn to other optoelectronic lead-free halide perovskites (e.g., double perovskites), which provide much more room for doping/alloying (19), in addition to being environmentally friendly and stable.

Here, we focus on the benchmark double perovskite Cs2AgBiBr6, which is a promising lead-free and stable optoelectronic material. Through magnetic element iron alloying, we provide new strategies to develop a new generation of materials that can potentially couple optoelectronics with spintronics. Combining near edge x-ray absorption fine structure (NEXAFS) and solid-state nuclear magnetic resonance (ssNMR) measurements, we reveal that iron is in the form of Fe3+, replacing Bi3+ and forming FeBr6 clusters, which homogenously distribute throughout the double perovskite crystals. Our double perovskite alloy, Cs2Ag(Bi:Fe)Br6, exhibits a structural phase transition at ~120 K and a strongly temperature-dependent magnetic response at temperatures below 30 K.

RESULTS

Basic characterizations of double perovskite Cs2Ag(Bi:Fe)Br6

Double perovskites Cs2Ag(Bi:Fe)Br6 were prepared by using a procedure similar to that reported for Cs2AgBiBr6 (20, 21). CsBr, AgBr, BiBr3, and FeBr3 precursors were mixed into an HBr solution in a hydrothermal autoclave, which was heated at 120°C for 24 hours and then slowly cooled down, resulting in 2- to 5-mm black octahedral crystals as the final products. Figure 1A presents photographs of the Cs2AgBiBr6 and Cs2Ag(Bi:Fe)Br6 single crystals. The crystal color of Cs2Ag(Bi:Fe)Br6 changes from red to black after alloying, indicating enhanced absorption (fig. S1). Electron spin resonance (ESR) measurements of the Cs2Ag(Bi:Fe)Br6 crystal show a strong resonance located at g ~ 2.032 at room temperature, pointing out the existence of paramagnetic Fe (Fig. 1B). The concentration of Fe in Cs2Ag(Bi:Fe)Br6 is ~11.4%, as confirmed by both energy dispersive spectroscopy (EDS) and inductively coupled plasma optical emission spectroscopy (ICP-OES) measurements (fig. S2 and table S1).

Fig. 1 Basic characterizations of the double perovskite Cs2Ag(Bi:Fe)Br6.

(A) Photographs and crystal structures of double perovskites Cs2AgBiBr6 and Cs2Ag(Bi:Fe)Br6. (B) ESR signals from Cs2AgBiBr6 and Cs2Ag(Bi:Fe)Br6 at room temperature (RT). a.u., arbitrary units. (C) Powder XRD patterns for Cs2AgBiBr6 and Cs2Ag(Bi:Fe)Br6, and the simulated XRD pattern of Cs2AgBiBr6 as reference, an expansion of highly intense reflections (220) (D) and (400) (E), illustrating shift toward higher 2θ angles and the symmetric to asymmetric transformation after Fe3+ alloying. (F) NEXAFS Fe 2p3/2 edge spectrum for Cs2Ag(Bi:Fe)Br6, in which the baseline has been corrected.

Both single-crystal x-ray diffraction (SC-XRD) and power XRD (PXRD) measurements indicate that Cs2Ag(Bi:Fe)Br6 retains the cubic phase of Cs2AgBiBr6. On the basis of the SC-XRD measurements, Cs2AgBiBr6 crystallizes in the cubic Fm-3m space group with the lattice parameter of a = 11.27 Å (table S2), manifesting a three-dimensional classic perovskite structure (Fig. 1A, bottom). When 11.4% Fe is alloyed into Cs2AgBiBr6, the structure retains its original cubic phase with lattice parameter slightly decreasing to a = 11.25 Å (table S2). This lattice shrink phenomenon results from smaller radius of Fe3+ (0.65 Å)/Fe2+ (0.78 Å) relative to Bi3+ (1.03 Å)/Ag+ (1.15 Å) (22), consistent with the Vegard’s law (23). The PXRD patterns further confirm the cubic double perovskite structure of Cs2Ag(Bi:Fe)Br6, since all the diffraction peaks are almost the same as the simulated and experimental Cs2AgBiBr6 (Fig. 1C). The highly intense reflections (220) and (400) shift toward high 2θ angles together with symmetric to asymmetric transformation after Fe alloying (Fig. 1, D and E), which is consistent with the lattice shrink observed from SC-XRD.

We additionally notice that Fe exhibits multiple valence states in compounds (e.g., FeO, Fe2O3, and Fe3O4) and subsequently perform NEXAFS, x-ray photoelectron spectroscopy (XPS), and synchrotron radiation–XPS (SR-XPS) measurements so as to understand the valence state and surface distribution of Fe in our alloyed double perovskites. The NEXAFS Fe 2p3/2 edge spectrum for Cs2Ag(Bi:Fe)Br6 (Fig. 1F) shows a double peak in the range of 705 to 713 eV. Peak deconvolution indicates an intensity ratio of approximately 1:3 between the Fe 2p3/2 signal at 708.1 eV and that at 709.9 eV, pointing out that the Fe is in the state of trivalence (24). The valence state of Fe is further confirmed by the XPS Fe 2p3/2 core level spectrum (fig. S3), as evidenced by the small atomic shake-up satellite as a high binding energy shoulder at 712.5 eV on the major Fe 2p3/2 peak at 707.5 eV that is characteristic of the trivalent state (25). It is important to note that SR-XPS of the Fe 2p level at a low photoelectron kinetic energy of 50 eV (electrons escaping from a shallow depth of one monolayer) does not reveal the presence of Fe in the sampled region, while laboratory XPS using deeper originating Fe 2p photoelectrons at a kinetic energy of 777 eV (compared with an Auger electron energy of 650 eV in the NEXAFS measurement, a comparable kinetic energy and photoelectron inelastic mean free path) gives a clearly discernable Fe 2p XPS signal. The laboratory XPS, SR-XPS, and NEXAFS results confirm that the alloyed Fe is not at the surface of the crystal but is instead buried beneath the surface monolayer.

Temperature-dependent structure of double perovskite Cs2Ag(Bi:Fe)Br6

We further investigate the effect of Fe alloying on the crystal structures at low temperatures. We perform specific heat capacity (Cp) measurement to examine the phase transition behavior of Cs2Ag(Bi:Fe)Br6 (Fig. 2A). A thermal anomaly is observed at ~120 K, which is probably attributed to a weak first-order or second-order phase transition. To deeply understand the thermal anomaly observed from the Cp measurements, we characterize the crystal structures of Cs2Ag(Bi:Fe)Br6 by using temperature-dependent synchrotron powder diffraction (SPD). Figure 2B and fig. S4 show the SPD patterns of Cs2Ag(Bi:Fe)Br6 in 300 to 30 K, where the cubic Bragg reflection (4 0 0) split into two peaks at ~120 K, indicating that the phase transition at this temperature is a symmetric breaking structural phase transition. In contrast, the reflection peaks of (4 4 4) only regularly shift toward the high-angle side, possibly due to lattice shrink during the cooling process.

Fig. 2 Temperature-dependent specific heat capacity and SPD.

Specific heat capacity (A) and SPD patterns (B) of Cs2Ag(Bi:Fe)Br6 at different temperatures. The lattice constant (C) and rotation angle (D) in 293 to 30 K for Cs2Ag(Bi:Fe)Br6. The crystal structures of Cs2Ag(Bi:Fe)Br6 at high temperature (HT, 300 K) (E) and low temperature (LT, 30 K) (F), red lines representing unit cells. Inset plot in (A) represents a magnification of the Cp peak at 120 K.

To determine the low-temperature crystal structure of Cs2Ag(Bi:Fe)Br6 accurately, we use the JANA2006 soft package (26) to do the Rietveld refinements. The Rietveld refinement shows a perfect single-phased material; thus, Cs2Ag(Bi:Fe)Br6 sample is free of CsBr, BiBr3, AgBr, and Fe clusters or any other impurity (fig. S5). Table S3 lists the lattice parameters at different temperatures. A cubic to tetragonal (c > a) first-order structural transition is observed at 120 K. This temperature agrees well with thermal anomaly temperature (Tc) observed in the temperature-dependent Cp (Fig. 2A). With decreasing temperature, Cs2Ag(Bi:Fe)Br6 experiences a symmetry breaking phase transition with the space group changing from Fm-3m to I4/m. Meanwhile, an abnormal change is observed from the temperature-dependent lattice constant and rotation angle between AgBr6 and (Bi:Fe)Br6 octahedron (Fig. 2, C and D), further confirming the structural phase transition at 120 K. As the temperature gradually decreases, the crystal phase retains I4/m with the lattice volume slightly decreasing at 30 K, which correlates well with the temperature dependence of Cp. As for pristine Cs2AgBiBr6, the structural behaviors are almost the same as Cs2Ag(Bi:Fe)Br6, indicating that the structural phase transition is intrinsic, without any influence from Fe3+ alloying (27). The lattice parameters of Cs2Ag(Bi:Fe)Br6 are observed to be smaller than those of Cs2AgBiBr6, consistent with SC-XRD results (22).

ssNMR of double perovskites Cs2AgBiBr6 and Cs2Ag(Bi:Fe)Br6

Next, we proceed to investigate how Fe ions distribute within the Cs2AgBiBr6 matrix. We analyze miscellaneous 133Cs and 209Bi ssNMR [magic angle spinning (MAS) ssNMR)] experiments to gain deep insight in the structure for both Cs2AgBiBr6 and Cs2Ag(Bi:Fe)Br6 systems. As shown in Fig. 3A, only one symmetric peak located at δiso = 80.7 ± 0.5 parts per million (ppm) in 133Cs MAS NMR spectra of Cs2AgBiBr6, confirming the existence of one crystallographic position of the Cs+ ions in pristine Cs2AgBiBr6. In contrast, besides the peak at δiso = 80.7 ± 0.5 ppm, an additional weak peak appears at δiso = 83.4 ± 0.5 ppm (Fig. 3B) in Cs2Ag(Bi:Fe)Br6, indicating the presence of paramagnetic ions (Fe3+) in close proximity to Cs+ ions (2830). The presence of paramagnetic metal ions (e.g., Fe3+ and Cu2+) usually causes extremely rapid longitudinal and transverse relaxation of the nearby nuclei due to electron spin couplings. This behavior allows us to probe a more precise distribution of the paramagnetic species in the perovskite matrix by using 133Cs NMR T1 relaxation measurements (31).

Fig. 3 ssNMR of Cs2AgBiBr6 and Cs2Ag(Bi:Fe)Br6.

133Cs-133Cs SD/MAS NMR and 133Cs Hahn-echo MAS NMR spectra of Cs2AgBiBr6 (A) and Cs2Ag(Bi:Fe)Br6 (B). 133Cs T1 saturation recovery buildup curves of Cs2AgBiBr6 (C) and Cs2Ag(Bi:Fe)Br6 (D). (E) 209Bi ssNMR of Cs2AgBiBr6 and Cs2Ag(Bi:Fe)Br6 conducted at static conditions. (F) Schematic representation of possible scenarios for Fe3+ distribution inside the perovskite lattice: parent perovskite lattice, isolated Fe3+ ions, small Fe3+ clusters, and large Fe3+ clusters. The percentages are the ideally calculated values for Cs+ exhibiting fast relaxation in different scenarios.

The detected 133Cs NMR signal contains two partially overlapped peaks (components), and determination of their relaxation parameters (T1 relaxation times and amounts of individual components) is performed in two steps. First, the saturation recovery buildup curves of the detected 133Cs NMR signal are analyzed using a double-exponential saturation recovery function and preliminary values of T1(133Cs) relaxation times, and the corresponding fractions of individual components are obtained. Second, each 133Cs NMR spectrum recorded in T1(133Cs) saturation recovery experiment is fitted. Then, the buildup curve of individual components is separately analyzed by using the single-exponential saturation recovery functions (Fig. 3, C and D). This way, both peaks are analyzed, and the values of T1 relaxation times and the amounts of individual phases are refined. Figure 3 (C and D) shows the resulting 133Cs T1 relaxation curves for Cs2AgBiBr6 and Cs2Ag(Bi:Fe)Br6. We obtain a slow T1 relaxation time of 306 s in Cs2AgBiBr6 and one slow- and one fast-relaxing phase corresponding to signals at δiso = 80.7 ± 0.5 ppm (31 s) and δiso = 83.4 ± 0.5 ppm (4 s) in Cs2Ag(Bi:Fe)Br6. The shortened 133Cs T1 relaxation times indicate that the paramagnetic ions (Fe3+) are alloyed into the perovskite framework. Since we now know the amount of Fe3+ and also the amount of Cs+, which is in close proximity to Fe3+, simple calculations lead us to conclude that Fe3+ is homogenously distributed as small FeBr6 clusters throughout the lattice (more details in the Supplementary Materials) (31).

The remaining doubt is whether Fe3+ ions replace Bi3+. Hence, we perform the 209Bi ssNMR experiment on both Cs2AgBiBr6 and Cs2Ag(Bi:Fe)Br6 systems. As shown in Fig. 3E, a very broad signal locates at about δiso = 5500 ppm for both Cs2AgBiBr6 and Cs2Ag(Bi:Fe)Br6. The 209Bi ssNMR spectrum of Cs2AgBiBr6 provides relatively symmetric signal with evident J-coupling (JX−Bi = 2002 Hz, where X corresponds to Br). In the case of Cs2Ag(Bi:Fe)Br6, the 209Bi ssNMR spectrum shows broadened spectral line with J-coupling completely missing. This phenomenon is caused by the fast T1 relaxation times of 209Bi nuclei induced by the paramagnetic Fe3+, which is homogenously distributed in the matrix, replacing Bi3+ ions.

Magnetic properties characterization of double perovskite Cs2Ag(Bi:Fe)Br6

Encouraged by the fact that Fe3+ ions are incorporated into the lattice, we perform temperature-dependent magnetic susceptibility for Cs2Ag(Bi:Fe)Br6 by superconducting quantum interference device (SQUID). The results exhibit diamagnetism at room temperature and weak temperature dependence of magnetic susceptibility above 30 K (Fig. 4A), indicating that diamagnetism from core electrons in this compound dominates over paramagnetic contribution from the alloyed magnetic ion Fe3+. The magnetic susceptibility increases rapidly below 20 K, indicating that a certain kind of magnetic response appears. The magnetic susceptibilities under zero-field–cooled (ZFC) and field-cooled (FC) procedures basically overlap with each other in the whole temperature range (Fig. 4A) and exhibit very tiny difference below 30 K (fig. S7). Field-dependent magnetization M(H) is linear with a negative slope above 50 K, which is consistent with temperature-dependent magnetic susceptibility (Fig. 4B). M(H) at 2 K has a positive component at low fields and saturates to ~0.15 μB per Fe after subtracting the diamagnetic component. The saturated magnetic moment is much smaller than both high spin (5 μB per Fe) and low spin (1 μB per Fe) state of Fe3+. In addition, there is no obvious magnetic hysteresis in the M(H) loop in the low fields (see the inset of Fig. 4B).

Fig. 4 Magnetic characterizations of Cs2Ag(Bi:Fe)Br6.

(A) Temperature-dependent magnetic susceptibility of Cs2Ag(Bi:Fe)Br6 under zero-field-cooled (ZFC) and field-cooled (FC) procedures with a magnetic field H = 1000 Oe. (B) Isothermal field-dependent magnetization of Cs2Ag(Bi:Fe)Br6. The inset plot presents the low–magnetic field magnetization from −0.1 to 0.1 T. (C) ESR spectrum from Cs2AgBiBr6 (the gray curves) and Cs2Ag(Bi:Fe)Br6 (the red curves) crystal powder, measured at 7.2 and 40 K, respectively. The simulated ESR spectra (the blue curves) are obtained from a spin Hamiltonian analysis, which consist of the contributions from Fe3+, with S = 5/2 and g = 2.032, and a defect center of unknown origin, with S = 1/2 and g = 2.032. The discrepancy between the simulated and experimental ESR spectra indicates the existence of a third ESR signal—a broad background signal, which becomes more pronounced at low temperatures. (D) Temperature dependence of total microwave absorption (the black spheres), together with the contributions from the three individual ESR components revealed from the spin Hamiltonian analysis.

To gain further insights into the observed temperature dependence of the magnetic response probed by SQUID, we perform ESR measurements. ESR is well-known to be capable of providing microscopic information associated with the chemical origins of the spin (magnetic) species, being hence probed. In our case, this has enabled us to identify the paramagnetic centers or possible magnetic resonance signals arising from magnetic coupling such as ferromagnetic or antiferromagnetic resonance (FMR or AFMR) signals that are involved in the total ESR spectra as well as their contributions to the overall magnetic response evolving with temperature. Figure 4C shows ESR spectra from Cs2Ag(Bi:Fe)Br6 crystal powder (the red curves) obtained at 7.2 and 40 K, respectively. Strong multiline ESR signals are observed throughout the temperature range of 5 to 300 K: These are introduced upon the incorporation of Fe in Cs2AgBiBr6, as they are absent in the ESR spectra from the pristine Cs2AgBiBr6 (shown as the gray curves in Fig. 4C). As shown in Fig. 4C (right), the ESR spectrum at 40 K from the Cs2Ag(Bi:Fe)Br6 crystal powder is best described by a sum of two paramagnetic species: (i) isolated Fe3+ of tetragonal symmetry with S = 5/2 and g = 2.032 and (ii) a defect center of unknown origin with S = 1/2 and g = 2.032. The assignment is supported by the good agreement between the simulated ESR spectra of these two centers by the spin Hamiltonian and the experimental spectrum (the spin Hamiltonian analysis is described in the Supplementary Materials). Upon cooling down below 35 K, in addition to the two aforementioned paramagnetic centers, an extra, broad component emerges and becomes dominant in the ESR spectrum at 7.2 K.

To find a correlation with the temperature-dependent magnetic response from the SQUID and additionally identify the relative importance of various spin species, we carefully investigate and analyze the ESR signal intensity (directly proportional to χ) as a function of temperature. The results are summarized in Fig. 4D, where the temperature dependences of the ESR intensities of the individual ESR components and the total ESR signal intensity are displayed. Evidently, the total ESR signal intensity exhibits a similar temperature-dependent trend as the SQIUD data shown in Fig. 4A, confirming a direct correlation in measuring χ by the two techniques. Using the protocol of spectral deconvolution of the three different ESR contributions, the intensities of the isolated Fe3+ and the S = 1/2 center are found to gradually and monotonically increase with deceasing temperature, as expected according to the paramagnetic centers. Our analysis proves that the sharp rising of the magnetic response at T < 35 K is not related to these two paramagnetic species but rather due to the third and broad component that becomes dominant at T < 35 K, as shown in Fig. 4D.

Combining the results from the ESR and SQUID studies, the following conclusions can be drawn. First, judging from the fact that the Fe3+ ions remain magnetically isolated throughout the entire temperature range of 5 to 300 K, the Cs2Ag(Bi:Fe)Br6 crystal is not globally magnetic. Otherwise, the isolated Fe3+ ESR signal would have decreased due to their strong magnetic coupling occurring below the magnetic phase transition temperature, which would have directly correlated with a concomitant increase of an FMR- or AFMR-like signal. Second, although the origin of the broad ESR component that is responsible for the steep rising of magnetic response at low temperatures is unknown, it is induced by the Fe incorporation in the crystal and it resembles that of an FMR or AFMR signal. As no obvious shift in the resonance field of the broad ESR signal can be resolved, which should otherwise reflect the internal magnetic field induced by magnetization, it is not possible at present to single out whether or not the broad ESR signal corresponds to an FMR or AFMR signal. If it is induced by FMR or AFMR coupling, the corresponding magnetic response should then be (i) local, only representing a small part of the total sample volume, and (ii) rather weak, to account for the negligible reduction of the isolated Fe3+ ESR signal intensity and the small saturated magnetic moment detected in SQUID. A possible origin of the corresponding magnetic response is ferromagnetic coupling from localized regions with high Fe3+ concentrations due to a strong composition fluctuation. If this is true, then it could hint that it may be possible to make the Cs2Ag(Bi:Fe)Br6 crystal a magnetic semiconductor provided that a higher Fe composition could be achieved. Thus, this work points toward some possible directions for further studies aiming to explore lead-free double perovskites for future spintronic applications. We cannot either rule out the possibility of magnetic inclusions of phase separated Fe clusters/composites beyond the detection limit of our characterization techniques in this work. The origin of this magnetic response needs further investigations on more samples with different iron concentrations, which is beyond the scope of this study at this stage.

DISCUSSION

In summary, we alloy magnetic element Fe3+ into optoelectronic double perovskite Cs2AgBiBr6, providing a new class of halide double perovskite alloys, which can be potentially used for spintronics. The magnetic Fe3+ ions homogenously distribute with small FeBr6 clusters in the crystal lattice. The temperature-dependent specific heat and SPD of Cs2Ag(Bi:Fe)Br6 demonstrate a cubic-tetragonal structural transition at ~120 K. Combined SQUID and ESR measurements reveal a strongly temperature-dependent magnetic response at temperatures below 30 K, which is tentatively attributed a weak ferromagnetic or antiferromagnetic response from localized regions containing, e.g., phase separated magnetic clusters/composites or high Fe compositions due to alloy fluctuations. We believe that halide double perovskites hold great potential for a next generation of spintronic devices that have been widely studied in oxide-based perovskites and merit further study to realize their full potential.

MATERIALS AND METHODS

Preparation of Cs2AgBiBr6 crystals

Solid CsBr (213 mg, 1.00 mmol) and BiBr3 (225 mg, 0.5 mmol) were dissolved in 3 ml of 47% HBr. Solid AgBr (94 mg, 0.5 mmol) was then added to the solution, and the mixture was transferred into a 25 cm3 Teflon-lined digestion autoclave. The autoclave was sealed and placed in the oven where it was heated at 120°C for 24 hours. After being slowly cooled to room temperature, red octahedral single crystals were achieved (yield ca. 79% based on Ag).

Preparation of Cs2Ag(Bi:Fe)Br6 crystals

Solid CsBr (213 mg, 1.00 mmol), BiBr3 (135 mg, 0.3 mmol), and FeBr3 (59.1 mg, 0.2 mmol) were dissolved in 4 ml of 47% HBr. Solid AgBr (94 mg, 0.5 mmol) was then added to the solution, and the mixture was transferred into a 25-cm3 Teflon-lined digestion autoclave. The autoclave was sealed and placed in the oven where it was heated at 120°C for 24 hours. After being slowly cooled to room temperature, black octahedral single crystals were achieved (yield ca. 67% based on Ag).

Basic characterizations of Cs2AgBiBr6 and Cs2Ag(Bi:Fe)Br6 crystals

The optical absorption measurements were carried out using a PE Lambda 950 ultraviolet-visible spectrophotometer. XRD patterns of the products were recorded with a X’Pert PRO x-ray diffractometer using Cu Kα1 irradiation (λ = 1.5406 Å). Energy-dispersive x-ray spectroscopy analysis was performed using an LEO 1550 scanning electron microscope operated at 20-kV accelerating voltage, with an Oxford Instruments X-Max 80-mm2 Silicon Drift Detector. ICP-OES measurements were carried out using a PerkinElmer Avio 500 spectrometer with the Cs2Ag(Bi:Fe)Br6 crystal powder dissolved in aqua regia. XPS of Cs2Ag(Bi:Fe)Br6 crystals was performed using a Scienta ESCA200 spectrometer with a base pressure of 2 × 10−10 mbar (ultrahigh vacuum) and monochromatized Al (Kα) radiation (hν = 1486 eV). NEXAFS spectroscopy and SR-XPS measurements were performed at the Elettra synchrotron radiation facility in Trieste, Italy. Specific heat (Cp) measurement was performed using physical property measurement system (PPMS-9 T) Quantum Design Co. with Cs2Ag(Bi:Fe)Br6 crystals.

Temperature-dependent SPD

The SPD experiments were carried out at beamline BL02B2 at SPring-8, Japan (32). The Cs2AgBiBr6 and Cs2Ag(Bi:Fe)Br6 crystal samples were crushed into a fine powder and filled into a borosilicate glass capillary with a diameter of 0.1 mm. The incident x-ray beam was monochromatized to 24.8 keV (0.5 Å) using Si(111) double crystals. The detail of the wavelength was 0.50018 Å determined by using standard reference materials CeO2.The temperature of the powder sample was controlled using a nitrogen and helium gas blower. Rietveld refinements were performed using the JANA2006 software package (26).

Solid-state nuclear magnetic resonance

The ssNMR spectra were recorded at 11.7 T using a Bruker Avance III HD spectrometer. The 4-mm cross-polarization MAS (CP/MAS) probe was used for 133Cs and 209Bi experiments at Larmor frequency of ν(133Cs) = 65.611 MHz and ν(209Bi) = 80.858 MHz, respectively. 133Cs MAS NMR experiments were collected at 10-kHz spinning speed without 1H decoupling. The recycle delay was 4 s for all ssNMR experiments. The 133Cs chemical shift was calibrated using solid CsCl (133Cs, 228.1 ppm) (33). The pulse length was set to 2.4 μs at 100 W for maximal signal intensity. The 133Cs rotor-synchronized (1 loop) spin-echo MAS NMR experiments (90°-nτR-180°-nτR-acq.) (34) were performed. The 133Cs-133Cs correlation MAS NMR spectra were recorded using nuclear Overhauser effect spectroscopy–type three-pulse sequence. Duration of the spin-exchange period between the second and third pulse was 10 μs. Spectral width in both frequency dimensions was rotor-synchronized to be 10 kHz. The indirect detection period t1 consisted of 128 increments each made of 48 scans. The 209Bi chemical shift was calibrated using a saturated solution of Bi(NO3)3·5H2O in concentrated HNO3 (209Bi, 0.0 ppm) (35). The pulse length was 9.0 μs at 100 W at maximal signal intensity. 209Bi NMR experiments were collected at static conditions using spin-echo NMR experiments (90°-τ-180°-acq.) (34), and the delay between pulses was 20 μs. To compensate frictional heating of the spinning samples, all NMR experiments were measured under active cooling. The sample temperature was maintained at 298 K, and the temperature calibration was performed on Pb(NO3)2 using a calibration procedure described in the literature (36). Dried sample was packed into ZrO2 rotors and subsequently stored at room temperature. All NMR spectra were processed using the TopSpin 3.5 pl2 software package.

Magnetic characterization

Magnetic susceptibility measurements were performed in the commercial Quantum Design magnetic property measurement system (MPMS3). Lots of small Cs2Ag(Bi:Fe)Br6 crystals with a total mass of 29.9 mg were loaded into a polycrystalline sample holder, which was fixed in a brass sample rod. Both ZFC and FC procedures were used to check possible magnetic transition in this compound. ESR was performed with a Bruker Elexsys E500 spectrometer operating at about 9.3 GHz. ESR spectra were recorded in dark. Cs2Ag(Bi:Fe)Br6 and Cs2AgBiBr6 crystal powder samples were prepared via collecting many randomly oriented small crystals into sealed and evacuated quartz tubes and placed in an He-flow cryostat.

X-ray single crystallography

The SC-XRD data for Cs2Ag(Bi:Fe)Br6 and Cs2AgBiBr6 were collected at 298 K with graphite monochromated Mo Kα (λ = 0.71073 Å) on a charge-coupled device area detector (Bruker D8 VENTURE and Bruker-SMART), respectively. Data reductions and absorption corrections were performed with the APEX3 suite, SAINT and SADABS software packages, respectively. Structures were solved by a direct method using the SHELXL-97 software package. The nonhydrogen atoms were anisotropically refined using the full-matrix least-squares method on F2. The details about data collection, structure refinement, and crystallography are summarized in table S2.

SUPPLEMENTARY MATERIALS

Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/6/45/eabb5381/DC1

https://creativecommons.org/licenses/by-nc/4.0/

This is an open-access article distributed under the terms of the Creative Commons Attribution-NonCommercial license, which permits use, distribution, and reproduction in any medium, so long as the resultant use is not for commercial advantage and provided the original work is properly cited.

REFERENCES AND NOTES

Acknowledgments: We thank J. Klarbring, S. I. Simak, I. A. Abrikosov, and A. Wildes for the helpful discussions. Funding: This work was financially supported by Knut and Alice Wallenberg Foundation (Dnr KAW 2019.0082), the Swedish Energy Agency (2018-004357 and P43288-1), the Swedish Government Strategic Research Area in Materials Science on Functional Materials at Linköping University (faculty grant SFO-Mat-LiU no. 2009-00971), and the Grant Agency of the Czech Republic (grant GA19-05259S). We acknowledge the Diamond Light Source for access to beamline I19 (MT20805). We also thank L. Saunders and D. Allan for the technical support. L.W. acknowledges the China Scholarship Council (CSC) for the financial support. Work at Argonne (magnetic susceptibility measurements) was supported by the U.S. Department of Energy, Office of Science, Basic Energy Sciences, Materials Sciences and Engineering Division. Synchrotron radiation experiment was performed at the BL02B2 of SPring-8 with approval of the Japan Synchrotron Radiation Research Institute (JASRI) (proposal no. 2017B0074). We are grateful for the support of the European Community’s Seventh Framework Programme (FP7/2007–2013, grant agreement no. 312284) for the research at the Materials Science Beamline at the Elettra Synchrotron and the CERIC-ERIC Consortium for access to experimental facilities and financial support of the Czech Ministry of Education (LM2015057). G.A.C., M.C., and R.D.M. pay special thanks to N. Tsud and K. C. Prince at Elettra Synchrotron for assistance with the experiments. We also thank J. Bradley for assistance with measurements at the Elettra Synchrotron. M.C. and G.A.C. acknowledge the CERIC users’ grant for travel funding to visit the Elettra Synchrotron, while R.D.M. thanks the International Synchrotron Access Program of the Australian Synchrotron for travel funding to attend the experiment at the Elettra Synchrotron. F.G. is a Wallenberg Academy Fellow. Author contributions: W.N. and F.G. conceived the idea for the manuscript and designed the experiments. W.N. and F.J. developed the synthesis procedures and performed the basic chemical and physical characterization. J.B. performed the SQUID measurements and analyzed the results under the supervision of D.Y.C. and M.G.K. Y.P. and F.M. measured and analyzed the ESR results under the supervision of W.M.C. and I.A.B. L.K., S.A., and J.B. measured and analyzed the ssNMR results. S.S., S.K., H.I., and Y.K. helped with the specific heat and SPD measurements. L.W. performed the SC-XRD measurement and analysis under the supervision of L.S. M.C., R.D.M., and G.A.C. performed the NEXAFS measurements. W.N. wrote the paper with the contributions from the coauthors. F.G. supervised the project. All authors commented on the manuscript. Competing interests: The authors declare that they have no competing interests. Data and materials availability: All data needed to evaluate the conclusions in the paper are present in the paper and/or the Supplementary Materials. Additional data related to this paper may be requested from the authors.

Stay Connected to Science Advances

Navigate This Article