# Fileset

[65_MT-Y2024005.pdf](https://mdr.nims.go.jp/filesets/74a4ca4e-b85f-41b1-a370-13ed49d4d7ef/download)

## Creator

[Kiyoshi Kobayashi](https://orcid.org/0000-0001-9644-1879), [Shogo Miyoshi](https://orcid.org/0000-0003-0375-1187), [Tohru S. Suzuki](https://orcid.org/0000-0001-9458-6863)

## Rights

[In Copyright](http://rightsstatements.org/vocab/InC/1.0/)

## Other metadata

[Production of Solidified Body from a Melt and Its Electrical Conductivity of CsSnBr3 Using Precursor Prepared by Mechanochemical Reaction Process](https://mdr.nims.go.jp/datasets/349e309f-33b7-4d15-b1ab-9e7219e8a293)

## Fulltext

Production of Solidified Body from a Melt and Its Electrical Conductivity of CsSnBr3 Using Precursor Prepared by Mechanochemical Reaction ProcessProduction of Solidified Body from a Melt and Its Electrical Conductivity of CsSnBr3Using Precursor Prepared by Mechanochemical Reaction Process+1Kiyoshi Kobayashi+2, Shogo Miyoshi and Tohru S. SuzukiNational Institute for Materials Science, Tsukuba 305-0047, JapanCesium-tin-bromide perovskite (CsSnBr3) has focused on as a candidate material for all-inorganic perovskite solar cell and thermoelectricenergy converter because of its optical, electrical, and thermal properties. On the other hand, the electrical properties have not been clarified yetbecause of several inconsistent reports. In this paper, we produced CsSnBr3 bulk from a melt using the precursor powder prepared by amechanochemical process. From powder X-ray diffraction analysis, main phase of the precursor was CsSnBr3 perovskite, and minor impuritieswere Cs2SnBr4 and CsSn2Br5. Although main phase of the bulk produced from a melt was CsSnBr3, small amount impurity, CsSn2Br5 wasconfirmed. From the electrical conductivity (σt) measurement, irreversible temperature dependence of σt was observed at first time increasingtemperature. The conductivity measured from room temperature to 443K at the first time showed metallic behavior. On the other hand, thetemperature dependence is changed into opposite with decreasing temperature. At this operation, σt decrease with decreasing temperature. Thissemiconductor like behavior was found to be reversible after first increasing temperature operation. These results indicate that post anneal as wellas production process of CsSnBr3 would be important to control its electric property. [doi:10.2320/matertrans.MT-Y2024005](Received June 11, 2024; Accepted August 19, 2024; Published October 25, 2024)Keywords: CsSnBr3, cesium tin bromide, perovskite, mechanochemical synthesis, electrical conductivity1. IntroductionThe cubic cesium tin bromide perovskite (CsSnBr3) is apotential material for all-inorganic perovskite solar cells andthermoelectric energy conversion materials because of itsexcellent optical, electrical, and thermal properties [1, 2].Nevertheless, the synthesis difficulty was understood fromthe phase relationship in the CsBr-SnBr2 system (Fig. 1) [3,4]. Although the melting point of CsSnBr3 is approximately723K, the eutectic point exists around 10mol% CsBr-90mol% SnBr2 at 477K. Therefore, when a mixed powderwith 50mol% CsBr-50mol% SnBr2 is heated directly, aliquid with a composition around 10mol% CsBr-90mol%SnBr2 is initially formed at 477K; then, the liquid isprecipitates at the bottom of the container. The precipitationof the liquid around the 10mol% CsBr-90mol% SnBr2composition continues on further heating. As the results,even though the mixed powder is changed into liquid above723K, it is difficult to obtain a homogeneous liquid.Therefore, in order to obtain the homogeneous CsSnBr3solidified body from the melt, it is necessary to use a specialinstrument for stirring the liquid above 723K.In oxide systems with similar phase relationship to theCsBr-SnBr2 system, the inhomogeneity can be prevented byemploying low-temperature synthesis processes, such ascoprecipitation and sol-gel methods [5–8]. However, thewet process using solvent-like coprecipitation and sol-gelmethods is difficult to apply to a non-oxide system. A notabledry process suitable for low-temperature synthesis is themechanochemical method. In the mechanochemical method,the driving force of chemical reactions is impact force;therefore, heating is unnecessary. Hence, the liquid formationat the eutectic point, as shown in Fig. 1, and the precipitationof the liquid do not occur. Thus, a homogeneous liquid with aCsSnBr3 composition can be prepared.Although the CsSnBr3 perovskite is expected to be asuitable material for all-solid perovskite solar cells andthermoelectric devices, its electrical property is not clarifiedyet. Only a few studies have reported on the electricalconductivity (σt) of the CsSnBr3 bulk. Moreover, thetemperature dependence of σt shows large discrepancyamong different reports [2, 9–11]; certain studies concludethat the temperature dependence of σt exhibits a metal-likebehavior, whereas others report that σt shows semiconductor-like behavior.In this study, we employed a mechanochemical reactionprocess to prepare a precursor for producing the homoge-neous CsSnBr3 melt [12, 13]. Further, we produced theCsSnBr3 bulk from the melt and investigated its electricalconductivity.973873773673573473T (K)100806040200mol % SnBr2Cs4SnBr 6CsSnBr3CsSn2Br5Fig. 1 Phase diagram in CsBr-SnBr2 system [3, 4].+1This Paper was Originally Published in Japanese in J. Japan Soc. PowderPowder Metallurgy 70 (2023) 427–431.+2Corresponding author, E-mail: kobayashi.kiyoshi@nims.go.jpMaterials Transactions, Vol. 65, No. 11 (2024) pp. 1397 to 1401©2024 Japan Society of Powder and Powder Metallurgyhttps://doi.org/10.2320/matertrans.MT-Y20240052. Experiments2.1 Mechanochemical synthesisCesium bromide (CsBr, 99.9%, Wako Pure ChemicalIndustries, Ltd., Japan) and tin (II) bromide anhydrous(SnBr2, 99.9%, Kojundo Chemical Lab. Ltd., Japan) wereused as raw materials. In a nitrogen-filled glove box, thesame molar amount of CsBr and SnBr2 were weighed to atotal amount of 20 g. The weighed CsBr and SnBr2 were setin a calcia-stabilized zirconia (CSZ) pot. Then, approximately100 g of yttria-stabilized zirconia (YSZ) balls were set in theCSZ pot. The diameter of the ball was 10mm. A cap with aTeflonμ ring was put on the pot. Subsequently, the pot andcap set was placed in an enclosed stainless steel container.The container was taken out from the glove box and set in aplanetary ball mill instrument (PULVERETTE 6, Fritsche,German). The mechanochemical reaction was conducted at400 rpm for 3 h. After the milling, the mechanochemicallyreacted sample was scraped off from the inside wall of the potin the glove box. The scraped sample and YSZ balls wereplaced in the pot and placed in the enclosed stainless-steelcontainer. The sample was mechanochemically reacted againunder the same conditions as the first time. The reactedsample was scraped off from the pot and pulverized intopowder using a zirconia mortar and pestle in the glove box,and the powder was used as a precursor to produce thesolidified body from the melt. The formed phase of theprecursor was confirmed by using X-ray diffraction (XRD)instrument (TTR-III, Rigaku Co. Ltd., Japan).2.2 Production of solidified body from the meltThe precursor powder was enclosed in a vacuum quartzglass tube of diameter 10mm. The glass tube was heated at823K in a vertically placed electric furnace. It was graduallycooled by moving it down from the center to the bottom ofthe furnace. The moving rate was 0.01mm/s. Hence, thesolidified body with a diameter of approximately 10mm anda length of 60mm was obtained. The solidified body wasremoved from the quartz tube at room temperature; then, adisk shape sample of 1.7mm thickness was cut from thebody using a low-speed diamond blade saw. Herein, thedisk is denoted as a solidified disk. The formed phase at thesurface of the cut disk was confirmed by XRD analysis.A part of the solidified body was fractured to conductcomposition analysis using an energy dispersive X-rayspectrum meter (Element EDS, EDAX, USA). The spectrummeter was attached to a scanning electron microscope(FlexSEM 1000 II, Hitachi High-Tech. Corp., Japan).To confirm the validity of the precursor, a solidified bodyfrom a melt was produced using a CsBr-SnBr2 mixedpowder. The production condition was the same to the bodymade from the precursor powder. The formed phase wasanalyzed by the XRD pattern from the cut surface similarto the surface prepared from the disk surface made from theprecursor powder.2.3 Electrical conductivity measurementCarbon disks were used as electrodes. The solidified diskwas sandwiched by two carbon disks. Platinum wires werewound around each carbon disk as lead wires. Theconductivity was measured under nitrogen flow in thetemperature range between room temperature and 448K.The measurements were carried out while repeatedlyincreasing and decreasing the temperature cycle betweenthe room temperature and 448K. Electrical conductivitymeasurements were carried out based on the two-probeimpedance method using a potentiostat/galvanostat (Refer-ence 600, Gamry, USA). No clear spectra were observedby impedance measurements due to the relaxation process;therefore, the spectra were explained only by the resistor. Thetotal conductivity was calculated from the resistance, surfacearea, and thickness of the solidified disk.3. Results and Discussions3.1 Mechanochemical reactionMacroscopic pictures of the 50mol% CsBr-50mol%SnBr2 mixed powder and the precursor obtained by themechanochemical reaction are shown in Fig. 2. The rawmaterial mixture (50mol% CsBr-50mol% SnBr2) is whitebecause CsBr and SnBr2 are colorless. Conversely, theprecursor is black; therefore, the chemical reaction is visuallyfound to proceed by the impact force at planetary ballmilling. The main peak positions in the XRD patterncollected from the precursor powder is found to show goodagreement with the peak positions of CsSnBr3 [14] (Fig. 3).The positions of other small peaks showed good agreementwith those of Cs4SnBr6 and CsSn2Br5 [3, 15]. These formedphases are compounds having lower SnBr2 compositioncompared to the eutectic one (90mol% SnBr2) shown inFig. 1. It was found that formation of the inhomogeneousliquid at 477K due to the eutectic reaction as shown in Fig. 1was prevented by the mechanochemical reaction even thoughthe mechanochemical reaction was essentially a solid-statereaction process.3.2 Solidified body from a meltThe XRD pattern collected from the cut surface of the  (a)(b)Fig. 2 Overview of (a) mixed powder of raw materials, CsBr and SnBr2,and (b) precursor obtained by mechanochemical reaction.K. Kobayashi, S. Miyoshi and T.S. Suzuki1398solidified body from the melt is shown in Fig. 4(a). Exceptfor the XRD pattern of an adhesive to fix the solidified disk(Fig. 4(b)), the main peaks were found to show goodagreement with CsSnBr3 peaks. On comparing the peaksbetween the precursor powder and the cut surface of thesolidified disk, the CsSn2Br5 peak intensities were relativelylow. In addition, the peaks of Cs4SnBr6 were not present inthe solidified disk. Based on these results, the amount ofimpurity segregation decreased in the case of the solidifieddisk.After comparing the reported XRD peaks of CsSnBr3, thepeak intensities at 111 and 222 that were collected from thesolidified disk surface were relatively high. Therefore, thecut surface of the solidified disk was oriented to the [111]direction in which large grains existed. In this study, thesolidified body was produced by slow cooling from themelt of the precursor enclosed in a vacuum quartz glass tube.From this method, single crystal and/or large grains werepreferably grown from seeds formed during slow cooling.Large grains oriented to the [111] direction were containedbecause a tendency of the crystal orientation was observed bythe XRD pattern.The XRD pattern collected from the cut surface of thesolidified body produced from the 50mol% CsBr-50mol%SnBr2 powder is shown in Fig. 4(f ). From this pattern, nopeaks from CsSnBr3 were confirmed; further, phaseidentification was not possible. The reason for the formationof such solidified bodies was suggested to be the lack ofhomogeneity of the liquid composition as explained in theintroduction.3.3 Electrical conductivityFigure 5 shows an Arrhenius plot of σt measured usingthe solidified disk. At the first increase of the temperatureprocess from room temperature to 448K, σt decreasedwith an increasing temperature, i.e., metal-like behavior.Then, the temperature dependence of σt was changed to asemiconductor-like behavior when σt decreased with adecreasing temperature. Subsequently, σt showed a reversibletemperature dependency by repeating the temperature cycleas shown in Fig. 5, i.e., semiconductor-like behavior. Theactivation energy calculated from the slope of the Arrheniusplot was 0.02 eV above 333K and 0.3 eV below 333K. Thecritical temperature of 333K is different from the reportedone of the semimetal - semiconductor phase transition(393K) [11] and structural phase transition (286K) [14]. Inaddition, this our results are different from the reported onethat the σt spike appeared at the semimetal - semiconductorFig. 3 X-ray diffraction pattern of (a) the precursor obtained bymechanochemical reaction. Calculated patterns of (b) CsSnBr3 [14],(c) Cs4SnBr6 [3], and (d) CsSn2Br5 [15] are plotted for comparison.Intensity / Counts6050403020102θ / degree (CuKα)100110111200210 211220310311222321CsSnBr3Cs4SnBr6CsSn2Br5(a)(b)(c)(d)(e)(f)Adhesive to fix sampleFig. 4 (a) X-ray diffraction pattern collected from a cut surface of CsSnBr3solidified body from a melt using a precursor by mechanochemicalreaction. X-ray diffraction patterns of (b) adhesive used to fix thesample, and calculated patterns of (c) CsSnBr3 [14], (d) Cs4SnBr6 [3], and(e) CsSn2Br6 [15] are plotted for comparison. X-ray diffraction patterncollected from a cut surface CsSnBr3 solidified body prepared from50mol% CsBr-50mol% SnBr2 mixture.Fig. 5 Arrhenius plot of CsSnBr3 solidified body using mechanochemicalprecursor. Conductivity data from (1) Ref. [10], (2) Ref. [2], (3) Ref. [9],and (4) Ref. [11] are plotted for comparison.Conductivity of CsSnBr3 Prepared by Mechanochemical Reaction 1399phase transition [11]. The origin of the activation energychange has not been clarified yet.The reported logσt values shown in Fig. 5 show a widescattering range from ¹2 to ¹6. The reported results ontemperature dependence showed a large discrepancy becauseof the metal-like behavior [2, 9, 10], semiconductor-likebehavior [11], and existence (or not) of logσt spike due tothe phase transition. The log σt values obtained in this studyfall in the range of the existing reported data [2, 9–11]. Thelog σt with semiconductor-like behavior was reversible for thetemperature cycle. However, no log σt spike appeared.The observed differences were attributed to variations inthe sample production process. The sample showing a log σtspike due to the semimetal - semiconductor transition wasa single crystal [11, 16] grown from heated ethylene glycoldissolved CsBr and SnBr2. The single crystal was grown byslowly cooling the solution. The sample production processwas different from our study. Regarding the conductivitymeasurement, σt of the single crystal was reported to bemeasured after several temperature cycles in order to appearreversible temperature dependence of σt to remove aninfluence on thermal stress and so on [11]. The fact thattemperature dependence of σt became reversible after severaltemperature cycles was similar to our results.The bulk and single crystal samples showing metal-likebehavior is produced by slow cooling from a melt [2, 9, 10].The melt was produced from a mixture of CsBr and SnBr2.With regards to the temperature reversibility of σt, therewas no explanation; therefore, the same phenomenon of thetemperature behavior of σt observed at the first temperatureincrease of our sample possibly occurred.The anisotropy of σt might not appear for CsSnBr3 becauseof its cubic structure. Therefore, the difference between ourσt results and the reported data is not due to the 111 crystalorientation as explained in Fig. 4.Based on these results, CsSnBr3 probably has severalmetastable states. The solidified bulk and single crystal arefrozen at the metallic metastable state that may betransformed into a stable semiconductor phase by increasingthe temperature. The stability of this semiconductor phasewas high; therefore, the reversible temperature dependenceon σt appeared after the first temperature increase.The reason that reported σt values are scattered inapproximately a 4-order range is probably a difference inimpurity composition. From the chemical analysis from thefractured surface of the solidified rod used in this research, anegligible amount of oxygen was detected (Fig. 6). However,we could not be achieved a quantitative estimation becauseof the large error. When oxygen was substituted at thebromide ion site, a hole was formed due to the differencein their valence numbers. Further investigation is necessaryto investigate the oxygen contamination observed becausethere has been no report on this so far. In addition, a defectchemical analysis involving the doping of aliovalent ions isalso necessary because of the absence of studies on CsSnBr3that investigate its defect chemistry.Another possible contaminant is zirconia because weemployed the mechanochemical reaction using a zirconiapot and balls. The Zr-Lα peak should appear at 2.04 keV inthe energy dispersive X-ray spectrum (EDS). By confirma-tion of Fig. 6, no peak is found at 2.04 keV; therefore, theconcentration of Zr in our sample is lower than the detectionlimit of the spectrum meter (1 atom%). Hence, thecontamination of Zr is ignorable.The electric properties must be controlled for applicationsto perovskite solar cells and thermoelectric devices; inparticular, the stability of electric properties is important.As explained above, not only the absolute value but also thetemperature dependence of σt for CsSnBr3 strongly dependson the production process. The temperature reversibility ofσt results in stable electric properties. Hence, post annealingis necessary to produce CsSnBr3 samples with stable electricproperties.eZAF Quant Result - Analysis Uncertainty: 16.37 %Error %Atom%Weight%Element12.6341.739.17C K10.9110.383.04O K5.9426.9539.42Br L5.669.8121.31Sn L7.4411.1327.06Cs LMag. Live Time Time Const. ResolutionDetectorFig. 6 Results of composition analysis on a fracture surface of CsSnBr3 solidified body from a melt using mechanochemical precursor.K. Kobayashi, S. Miyoshi and T.S. Suzuki14004. ConclusionIn this study, a CsSnBr3 precursor powder was prepared bythe mechanochemical reaction to produce the solidified bodyfrom the melt. In addition, the solidified body was producedusing the precursor; then, the electrical conductivity wasmeasured using the solidified disk cut from the body.The colorless mixed powder of CsBr and SnBr2 waschanged into a black powder by a mechanochemical reaction.Although the main phase of the blackened precursor wasCsSnBr3, small peaks from CsSn2Br6 were observed. Fromthe XRD analysis of the solidified body produced from the50mol% CsBr-50mol% SnBr2 powder, the peaks fromCsSnBr3 were not observed and phase identification wasnot feasible because of the formation of inhomogeneousliquid. Therefore, we confirmed that the CsSnBr3 precursorpreparation process was valid for the production of theCsSnBr3 solidified body from the melt.The total electrical conductivity (σt) measured at the firsttemperature increase showed metallic behavior, i.e., σtdecreased as the temperature increased. On the contrary, σtchanged into a semiconductor-like behavior because σtdecreased as the temperature decreased in the temperaturerange from 448K to room temperature. Thus, the σt valueswere different when σt was measured at the first temperatureincrease and the subsequent temperature decline. After thefirst temperature cycle, σt demonstrated a reversible temper-ature change with a semiconductor-like behavior. Based onthese results, we concluded that the produced CsSnBr3solidified bulk was in a metastable state. The Arrhenius plotof the reversible σt showed a bend at 333K, which was indisagreement with the reported phase transition temperatures.Moreover, a large variation of σt values of CsSnBr3 has beenconfirmed in previous studies. Thus, further research isnecessary to clarify the electrical properties of CsSnBr3.AcknowledgmentsThis work was supported by the JSPS KAKENHI GrantNumber 17H01317.REFERENCES[1] M. Roknuzzaman, K. Ostrikov, H. Wang, A. Du and T. Tesfamichael:Towards lead-free perovskite photovoltaics and optoelectronics byab-initio simulations, Sci. Rep. 7 (2017) 14025.[2] H. Xie, S. Hao, J. Bao, T.J. Slade, G.J. Snyder, C. Wolverton andM.G. Kanatzidis: All-inorganic halide perovskites as potentialthermoelectric materials: Dynamic cation off-centering induces ultra-low thermal conductivity, J. Am. Chem. Soc. 142 (2020) 9553–9563.[3] B.M. Benin, D.N. Dirin, V. Morad, M. Wörle, S. Yakunin, G. Rainó,O. Nazarenko, M. Fischer, I. Infante and M.V. Kovalenko: Highlyemissive self-trapped excitons in fully inorganic zero-dimensional tinhalides, Angew. Chem. Int. Ed. 57 (2018) 11329–11333.[4] R.H. Andrews, S.J. Clark, J.D. Donaldson, J.C. Dewan and J. Silver:Solid-state properties of materials of the type-Cs4MX6 (where M = Snor Pb and X = Cl or Br), J. Phys. Chem. Dalton Trans. (1983) 767–770.[5] S. Kitajima, K. Kobayashi, T. Higuchi and Y. Sakka: Developmentof new synthesis route of lanthanum germanate oxyapatite fromhomogeneous aqueous solution, Ceram. Trans. 243 (2014) 103–108.[6] Y. Igarashi, S. Kitajima, K. Kobayashi, T. Higuchi and Y. Sakka: Low-temperature synthesis process of lanthanum germanate oxyapatite bycitrate combustion method, J. Jpn. Soc. Powder Powder Metallurgy61 (2014) 582–586.[7] K. Kobayashi, S. Kitajima, Y. Igarashi, T. Higuchi and Y. Sakka:Research and development of the coprecipitation process forlanthanum germanate oxyapatite, J. Am. Ceram. Soc. 98 (2015) 66–70.[8] K. Kobayashi, Y. Igarashi, N. Saito, T. Higuchi, T. Sakka and T.S.Suzuki: Stabilization of the high-temperature phase and totalconductivity of yttrium-doped lanthanum germanate oxyapatite, J.Ceram. Soc. Jpn. 126 (2018) 91–98.[9] S. Clark, C.D. Flint and J.D. Donaldson: Luminescence and electricalconductivity of CsSnBr3 and related phases, J. Phys. Chem. Solids 42(1981) 133–135.[10] K. Yamada, H. Kawaguchi, T. Matsui, T. Okuda and S. Ichiba:Structural phase-transition and electrical-conductivity of the perovskiteCH3NH3Sn1¹xPbxBr3 and CsSnBr3, Bull. Chem. Soc. Jpn. 63 (1990)2521–2525.[11] R.L. Narayan and S.V. Suryanarayana: Transport-properties of theperovskite-type halides, Mater. Lett. 11 (1991) 305–308.[12] Z. Hong, D. Tan, R.A. John, T.D.E. Tay, Y.K.T. Ho, X. Zhao, Z.C.Sum, N. Mathews, F. Garcia and H.S. Soo: Completely solvent-freeprotocols to access phase-pure, metastable metal halide perovskitesand functional photodetectors from the precursor salts, iScience 16(2019) 312–325.[13] Y.E. Ajjouri, F. Locardi, M.C. Gélvez-Rueda, M. Prato, M. Sessolo,M. Ferretti, F.C. Grozema, F. Palazon and H.J. Bolink: Mechano-chemical synthesis of Sn(II) and Sn(IV) iodide perovskites and studyof their structural, chemical, thermal, optical, and electrical properties,Energy Tech. 8 (2019) 1900788.[14] D.H. Fabini, G. Laurita, J.S. Bechtel, C.C. Stoumpos, H.A. Evans,A.G. Kontos, Y.S. Raptis, P. Falaras, A. van der Ven, M.G. Kanatzidisand R. Seshadri: Dynamic stereochemical activity of the Sn2+ lone pairin perovskite CsSnBr3, J. Am. Chem. Soc. 138 (2016) 11820–11832.[15] D. Becker and H.P. Beck: High pressure study of NH4Pb2Br5 typecompounds. I Structural parameters and their evolution under highpressure, Z. Anorg. Allg. Chem. 630 (2004) 1924–1932.[16] M. Mori and H. Saito: An X-ray study of successive phase-transition inCsSnBr3, J. Phys. C 19 (1986) 2391–2401.Conductivity of CsSnBr3 Prepared by Mechanochemical Reaction 1401https://doi.org/10.1038/s41598-017-13172-yhttps://doi.org/10.1021/jacs.0c03427https://doi.org/10.1002/anie.201806452https://doi.org/10.1039/dt9830000767https://doi.org/10.1039/dt9830000767https://doi.org/10.1002/9781118771464.ch10https://doi.org/10.2497/jjspm.61.582https://doi.org/10.2497/jjspm.61.582https://doi.org/10.1111/jace.13302https://doi.org/10.1111/jace.13302https://doi.org/10.2109/jcersj2.17198https://doi.org/10.2109/jcersj2.17198https://doi.org/10.1016/0022-3697(81)90072-Xhttps://doi.org/10.1016/0022-3697(81)90072-Xhttps://doi.org/10.1246/bcsj.63.2521https://doi.org/10.1246/bcsj.63.2521https://doi.org/10.1016/0167-577X(91)90208-Nhttps://doi.org/10.1016/j.isci.2019.05.042https://doi.org/10.1016/j.isci.2019.05.042https://doi.org/10.1002/ente.201900788https://doi.org/10.1021/jacs.6b06287https://doi.org/10.1002/zaac.200400207https://doi.org/10.1088/0022-3719/19/14/005