# Fileset

[HFI2023_Paper_Igarashi_240829.pdf](https://mdr.nims.go.jp/filesets/c21f4e5b-4fc1-40de-a10e-00a3dcb41ec3/download)

## Creator

Mutsuo Igarashi, [Tadashi Shimizu](https://orcid.org/0000-0003-1202-8185), [Atsushi Goto](https://orcid.org/0000-0002-9472-4098), [Kenjiro Hashi](https://orcid.org/0000-0002-0320-4768), Keiko Yamamichi, Takehito Nakano

## Rights

This version of the article has been accepted for publication, after peer review (when applicable) and is subject to Springer Nature’s AM terms of use, but is not the Version of Record and does not reflect post-acceptance improvements, or any corrections. The Version of Record is available online at: https://doi.org/10.1007/s10751-024-02040-7  [In Copyright](http://rightsstatements.org/vocab/InC/1.0/)

## Other metadata

[Slow dynamics observed on highly shifted spectral component of <sup>23</sup>Na NMR in low silica X zeolite loaded with potassium for saturating level](https://mdr.nims.go.jp/datasets/26f7eb9d-9a6a-40ec-b816-3339525c4c7f)

## Fulltext

Slow dynamics observed on highly shifted spectralcomponent of 23Na NMR in low silica X zeoliteloaded with potassium for saturating levelMutsuo Igarashi1*, Tadashi Shimizu2, Atsushi Goto2,Kenjiro Hashi2, Keiko Yamamichi1, Takehito Nakano31* National Institute of Technology, Gunma College, 580 Toribamachi,Maebashi, 371-8530, Gunma, Japan .2National Institute for Materials Science 3-13 Sakura, Tsukuba,305-0003, Ibaraki, Japan .3Institute of Quantum Beam Science, Graduate School of Science andEngineering, Ibaraki University, 2-1-1 Bunkyo, Mito, 310-8512, Ibaraki,Japan .*Corresponding author(s). E-mail(s): igarashi@gunma.kosen-ac.jp;Abstract23Na NMR spectra of Na-K form low silica X zeolite loaded with potassium forsaturating level, which corresponds to 9.0 atoms per unit, have been observed.Other than a centered grain of the spectral component, another separated peakcomponent with shift of somewhat larger than bulk Na has been detected. Suchthe extra peak component shows complex hysteresis on temperature and isaccompanied with quite slow dynamics of days-order. It is attributed to atomic re-configurations among sites occurs in the course of melting and re-crystallizationprocess among alloy of sodium and potassium, which possibly is formed on surfaceor outside of the zeolite crystals.Keywords: zeolite, sodium-potassium alloy, NMR, 23Na, Knight shift, slow dynamics11 IntroductionIt is known that alkali metal loading introduces metallic property into some kinds ofzeolites [1]. In some cases, it is accompanied by magnetic order. Na-K alloy clustersarrayed in zeolite low silica X (LSX) prepared by the alkali metal loading are oneof the systems whose conductivity and magnetism have been investigated in detail.The conductivity and magnetism change markedly depending on the Na content, suchas an insulating ferromagnetic state [2, 3], a metallic ferrimagnetic state [1, 4–6],and a metallic ferromagnetic state at high pressure [7]. In this study, we report theobservation of a strange NMR signal in a paramagnetic metal sample among thesemany Na-K alloy samples.LSX is a type of aluminosilicate zeolite whose framework structure type is FAU[8].Figure 1 shows the schematic illustration of the FAU structure. There are two typesof cages: the smaller one is called a β cage (inner diameter ≃ 7 Å) and the largerone is called a supercage (inner diameter ≃ 13 Å). Both are arranged in a diamondstructure [9]. Here we define the unit that contains a β cage and a supercage, which is1/8 of the unit cell of the diamond structure. The framework is formed by a covalentnetwork of Al, Si, and O, and the composition is Al12Si12O48 per unit. Since theframework is negatively charged by the number of Al atoms, cations are distributedin the cages. The typical cation sites are shown in Fig. 1. When the system containsmonovalent cation M, the chemical formula is written as M12Al12Si12O48 per unit.Such the pre-contained cation can be complex one and the formula is rewritten asNaxK12−xAl12Si12O48 for the case that both of Na and K are contained. We write it asNaxK12−x-LSX . x can vary from 0 to 12 by replacing the cation in aqueous solution.We here take a case of x = 4, where the formula being Na4K8-LSX. Na-K alloyclusters can be generated in the cages of it by loading of guest K atoms. Following aprevious report[5], we write the system as Kn/Na4K8-LSX. n means number of loadedK atoms per unit. This system shows ferrimagnetic transition for narrow range of naround n = 7.7[4]. Spectra of 23Na and 27Al NMR have been observed for n = 7.1[10],where ferrimagnetic transition occurs at 7 K. It possesses paramagnetic moment above7 K and multiple components shift through hyperfine interaction between paramag-netic moment and 23Na nuclei have been recognized. As has been reported[6], thissystem shows metallic property for n ≳ 6. Studies with variation on n are expected togive hints to resolve the nature of the electron system.We have studied these materials using NMR. In the 23Na NMR spectrum for n =9.0, where paramagnetic moment exists all through the temperature above 2 K andmetallic property is recognized, we found several components. The main component iswell coupled to the paramagnetic moments over the wide temperature range between25 and 300 K, allowing us to estimate the hyperfine coupling constants [11]. Besidessuch a main component, we found another component around room temperature,which is less coupled with the paramagnetic moment of the sample. This componenthas hysteresis in its temperature dependence and shows quite slow dynamics. We focuson the strange behavior of this component in this paper.22 ExperimentalSample preparation has been done with the same procedure as before[4]. Vapor ofK atom is thermally absorbed in vacuum environment up to the saturation level.The loading level was estimated as 9.0 K atoms per unit using chemical analysis. Weexpress, by similar manner with the previous report[5], the sample as K9.0/Na4K8-LSX, where 9.0 means number of loaded K atoms per unit. X-ray diffraction analysishas not been done for it but, taking into account an analysis on the case of Nan/Na12-LSX[12], it is quite natural to consider that there are several variations of sites forcations. No magnetic transition is observed above 2 K in spite that previously reportedsample, K7.1/Na4K8-LSX, shows ferrimagnetic transition at 7 K.NMR spectra of 23Na have been measured at 6.3 T at several points of tempera-ture below 300 K. Another magnetic field of 9.4 T has been utilized for part of theobservation such as long time one. References for the frequency shift are taken as theresonance frequency of 23Na in NaCl aqueous solution for both magnetic fields. NMRspectrum is obtained by Fourier transformation of the spin echo signal. In order toestablish thermal equilibrium state of the nuclear magnetization, waiting time betweeneach pulse was set effectively long enough. Typical value of it has been several 10seconds.3 Results and Discussion23Na spectrum of K9.0/Na4K8-LSX at 300 K is shown in Fig. 2. In the region aroundzero shift there is a structured component similar to the one which is previouslyreported for K7.1/Na4K8-LSX[10]. In addition to it, a separated component with some-what large shift is seen. The shift value of this component is ∼1400 ppm. The reasonwhy the shape of such the component looks as out of phase is that one shot pulse withnarrow Fourier component around zero shift has been adopted for this observation.Such the sharp component with large shift is not seen for n = 7.1. There shouldbe some certain reasons for an appearance of this component. It looked coming frommetallic nature of the sample itself, since shifted component in metallic materials oftencorresponds to Knight shift and, moreover, the value of the shift is different fromthat of bulk metal sodium, ∼1100 ppm[13]. Therefore trial efforts had been made formeasurements of the spin-lattice relaxation time T1 down to lower temperature fromroom temperature to obtain additional information about the metallic property of thesample. Several measurements, however, have shown that as the waiting time of the RFpulse is extended, the tail part of the magnetization recovery becomes progressivelylonger up to several tens of minutes.Then we decided to follow temperature variation of the spectrum with taking longpause between each measurement. An example of the spectral variation on temperatureis shown in Fig. 3 (a). Cooling steps had been taken continuously in the first sequenceof the measurements from 290 K until 215 K. Heating steps had successively beentaken in the second sequence of the measurements from 215 K until 270 K. Betweeneach step of data acquisition, long time pause more than one hour had been taken.As seen in the figure, the process has hysteresis. Such the behavior is observed notonly for the shift value of the peak but also for the signal intensity. Peak height is3plotted in the bottom of Fig. 3 (b). The peak height decreases from 290 K to 215 K,in spite of cooling process. If the spectrum is given by nuclei with steady environment,nuclear paramagnetic moment obeys Curie law and the intensity simply follows inverseof temperature. On the other hand, heating process from 215 K until 250 K doesnot give recovering to the signal intensities of the previous cooling process as beforebut, after branched behavior around 260 K, partial recover happens at 270 K. Suchthe observation suggests clearly that there is a somewhat strange hysteresis in thetemperature range of 220 K to 260 K. Moreover the behavior such as time durationfor an appearance of peak split depends on the temperature history of the sample.Such the fact may corresponds to first order transition. Multiple phases may coexistin the course of the change.In order to grasp the nature of this behavior, a trial observation through long termhas been done. The result is shown in Fig. 4 (a). At first to prepare a uniform thermalequilibrium condition for nuclear magnetization, the temperature of the sample hadbeen kept constant at 319 K for 1 hour, where no hysteresis is recognized. After suchthe preparation stage, rapid cooling down to 225 K had been done. Just after suchthe cooling treatment, the signal intensity is higher than that at 319 K. This behaviorqualitatively follows Curie law for nuclear magnetism of 23Na. Other than such thebasic feature, small deviation to lower side of the shift has been recognized. Keepingconstant temperature after such the rapid cooling, the spectrum had repeatedly beenrecorded many times with intervals around one hour. Shift of the peak gradually goesto the higher side as time passing and a shoulder like structure appears on the less shiftside of the peak. We did not make a precise examination for an appearance of suchthe shoulder, since we rather focused on overall behavior. Thick curves in the figureamong many spectra are representative ones. In order to present more the behavior,the intensity at each peak is plotted in the bottom of Fig. 4 (b). Through more than afew days, it continuously decreases. Although behavior after keeping the temperaturefor 100 hours is unknown, we terminated this experiment. We think it is not naturalthat regrowth of the spectrum up to the one similar to starting time occurs, even ifwe had continued the measurement. Disappeared signal must exist somewhere on thespectrum but we did not search for it.After 23 hours it can be seen by eye that a shoulder component appears to theleft of the main peak in the spectrum. Its peak intensities are estimated by eye andplotted in Fig. 4 (b). If the spectrum is fitted with two Lorentzian components, sucha small component can be picked up for the period prior to 23 hours. However, sincewe are not aiming for a clear separation of these two components, but simply hopingto describe the phenomenon, we have not performed such a fitting. To indicate thepossibility of a left shoulder component before 17 hours, a dotted line is drawn inFig. 4 (b) as a guide.The above behaviors must be certainly an appearance of some thermal dynamicsother than phenomena of electromagnetism, which in usual does not contribute toquite slow dynamics. Taking into account the fact that the shift value ∼ 1400 ppm issimilar but different to that of metallic sodium ∼ 1100 ppm[13], it may be attributedto Na-K alloy. Melting point of eutectic compound of Na an K is 260 K[14], whichjust corresponds to the upper limit temperature where the strange hysteresis starts4in our data. The starting point temperature 260 K of the hysteresis is well preserved.This correspondence is not accidental. Since, in general, melting point is expected todecrease when such the alloy is formed in quite small pore of supercage whose diameteris ∼ 13 Å. Although Na-K alloy nanoclusters are formed in LSX, they are not expectedto have exactly the same melting point as the bulk alloy.There is small amount of excess K atoms in the purpose to guarantee the conditionof saturation for loading K atoms fully into the cages. Therefore Na-K alloy in the out-side space of zeolite particles can be provided with such the K atoms. It is consistedwith the fact that a shifted component similar to the case of n = 9.0 does not appearon the spectrum of n = 7.1. Excess K atoms correspond to Na-K alloy located outsideof zeolite particle. As described in the experimental section, however, the mother sub-stance has not ever been exposed to Na vapor. Na atoms must come from somewhere.Unique source to provide with Na to the environment of zeolite particle is restrictedwithin the original substance of zeolite itself, where Na cations are included as thecharge compensation ions. Although there is no way to confirm the pass to give suchthe component to the environment, diffusion process of the cation during the samplepreparation can be supposed as an origin. Heating above 423 K is necessary there topromote diffusion of the cations in the time of loading external K atoms. Since suchthe process is performed in much higher temperature than the melting point of Na-Kalloy ∼ 260 K [14], Na-K alloy liquid possibly distributes on the surface of LSX micro-crystal of ∼ 2 µm. Although there are some reports on 23Na NMR of Na-K alloy[14],such the slow dynamics has not been reported so far. The environment of the zeolitesurface possibly gives a special condition to Na-K alloy to have the slow dynamics.Here we estimate thickness of Na-K alloy distributed on the surface of zeolitecrystal. Integral intensity of this shifted component is estimated to be at most 10% ofthe whole spectrum. If we suppose that each particle of zeolite is just sphere, thicknessof such the Na-K alloy is estimated as less than ∼ 30 nm. Such the thin film canbe supposed to have no influence on optical spectra, which are adopted to estimateproperties of K clusters in the cages. When analyzing the NMR signal associated withthe Na-K clusters in the cages of LSX, the ∼ 1400 ppm shift component, which is thefocus of this paper, can be ignored.Despite the above discussion, the possibility still remains that the focused spectralcomponent is given by 23Na nuclei in the cage of LSX. Whether outside or insidethe crystal is more likely to be determined by observing the precise loading leveldependence on the external K atom just below the saturation condition.4 Conclusion23Na NMR spectra for zeolite low silica X loaded with K for saturating level hasbeen observed. Other than grain of spectrum around the zero shift, a separated peakat ∼ 1400 ppm has been recognized. It shows strong hysteresis between 260 K and220 K and slow dynamics with the time scale up to days-order. Possible candidatefor the source of such the component is Na-K alloy formed on the surface of zeolitemicro-crystals.5Acknowledgments. The authors are grateful to Y. Nozue for his profound support.M. I. thanks D. Arcon and P. Jeglic for fruitful discussions. M. I. is especially grate-ful to A. Krajnc for helpful assistance. This study was partially supported by JSPSKAKENHI (Grant Numbers JP15540353 and JP16K05462) and MEXT KAKENHI(Grant Number JP19051009).Declarations• Funding: This work was financially supported by JSPS KAKENHI (Grant Num-bers JP15540353 and JP16K05462) and MEXT KAKENHI (Grant NumberJP19051009).• Ethics approval: not applicable• Availability of data and materials: not applicable6References[1] Nakano, T., Nozue, Y.: Electrons of alkali metals in regular nanospaces of zeolites.Adv. Phys.: X 2, 254–280 (2017) https://doi.org/10.1080/23746149.2017.1280415[2] Kien, L.M., Goto, T., Hanh, D.T., Nakano, T., Nozue, Y.: Ferromagnetism ofNa–K alloy clusters incorporated in zeolite low-silica X. J. Phys. Soc. Jpn. 84,064718 (2015) https://doi.org/10.7566/JPSJ.84.064718[3] Kien, L.M., Watanabe, I., Nakano, T.: µ+SR study on ferromagnetism of Na-Kalloy clusters in zeolite low-silica X. Interactions 29, 245 (2024) https://doi.org/10.1007/s10751-024-01862-9[4] Nakano, T., Goto, K., Watanabe, I., Pratt, F.L., Ikemoto, Y., Nozue, Y.: µSRstudy on ferrimagnetic properties of potassium clusters incorporated into low sil-ica X zeolite. Physica B 374-375, 21–25 (2006) https://doi.org/10.1016/j.physb.2005.11.007[5] Hanh, D.T., Nakano, T., Nozue, Y.: Strong dependence of ferrimagnetic prop-erties on Na concentration in Na–K alloy clusters incorporated in low-silica Xzeolite. J. Phys. Chemi. Solids 71, 677–680 (2010) https://doi.org/10.1016/j.jpcs.2009.12.064[6] Nakano, T., Hanh, D.T., Owaki, A., Nozue, Y., Nam, N.H., Araki, S.: Insulator-to-metal transition and magnetism of potassium metals loaded into regular cagesof zeolite LSX. Journal of the Korean Physical Society 63, 512–516 (2013) https://doi.org/10.3938/jkps.63.512[7] Araki, S., Hoang Nam, N., Shimodo, K., Nakano, T., Nozue, Y.: Ferromagnetismof potassium metal under pressure loading into zeolite low-silica X. Phys. Rev. B99, 094403 (2019) https://doi.org/10.1103/PhysRevB.99.094403[8] The three-letter code is the framework type code assigned by the StructureCommission of the International Zeolite Association (IZA-SC). InternationalUnion of Pure and Applied Chemistry (IUPAC) recommends it as the nomencla-ture.[9] Verhulst, H.A.M., Welters, W.J.J., Vorbeck, G., Ven, L.J.M., Beer, V.H.J., San-ten, R.A., Haan, J.W.: New Assignment of the Signals in 23Na DOR NMR tosodium sites in Dehydrated NaY zeolite. J. Phys. Chem. 98, 7056–7062 (1994)https://doi.org/10.1021/j100079a027[10] Igarashi, M., Nakano, T., Shimizu, T., Goto, A., Hashi, K., Goto, K., Yamamichi,K., Nozue, Y.: NMR property of low silica X zeolite with incorporated potassium.J. Magn. Magn. Mater. 310, 307–309 (2007) https://doi.org/10.1016/j.jmmm.2006.10.2437https://doi.org/10.1080/23746149.2017.1280415https://doi.org/10.7566/JPSJ.84.064718https://doi.org/10.1007/s10751-024-01862-9https://doi.org/10.1007/s10751-024-01862-9https://doi.org/10.1016/j.physb.2005.11.007https://doi.org/10.1016/j.physb.2005.11.007https://doi.org/10.1016/j.jpcs.2009.12.064https://doi.org/10.1016/j.jpcs.2009.12.064https://doi.org/10.3938/jkps.63.512https://doi.org/10.3938/jkps.63.512https://doi.org/10.1103/PhysRevB.99.094403https://doi.org/10.1021/j100079a027https://doi.org/10.1016/j.jmmm.2006.10.243https://doi.org/10.1016/j.jmmm.2006.10.243[11] Igarashi, M., Shimizu, T., Goto, A., Hashi, K., Yamamichi, K., Nakano, T.:Hyperfine couplings between the paramagnetic moment and nuclei in the metal-lic phase of low silica X zeolite loaded with potassium. Dalton Transactions 53,9838–9843 (2024) https://doi.org/10.1039/D4DT00599F[12] Ikeda, T., Nakano, T., Nozue, Y.: Crystal Structures of Heavily Na-Loaded Low-Silica X (LSX) Zeolites in Insulating and Metallic States. J. Phys. Chem. C 118,23202–23211 (2014) https://doi.org/10.1021/jp507894u[13] Carter, G.C., Bennett, L.H., Kahan, D.J. (eds.): Metallic Shifts in NMR.Pergamon Press, Oxford (1977)[14] Charnaya, E.V., Lee, M.K., Tien, C., Chang, L.J., Wu, Z.-J., Kumzerov, Y.A.,Bugaev, A.S.: Continuous melting and thermal-history-dependent freezing in theconfined Na-K eutectic alloy. Phys. Rev. B 87, 155401–15 (2013) https://doi.org/10.1103/PhysRevB.87.1554018https://doi.org/10.1039/D4DT00599Fhttps://doi.org/10.1021/jp507894uhttps://doi.org/10.1103/PhysRevB.87.155401https://doi.org/10.1103/PhysRevB.87.155401Fig. 1 Schematic illustration of the framework of zeolite low silica X (LSX), which has the FAUstructure [8]. The sites for Al and Si atoms correspond to crossing points on the framework in thefigure. The Si and Al atoms are alternatively ordered in the framework through the Si-O-Al bonds.Typical cation sites I, I’, II, II’, and III are shown with the small circles.9-1000 0 1000 2000Kn/Na4K8-LSXn=9.0n=7.1Intensity (arb. units)Shift (ppm)Fig. 2 23Na NMR spectra of Kn/Na4K8-LSX at 300 K in a field of 6.3 T for n = 7.1 and n = 9.0.101380 1400 1420 1440 1460 148023NaK9.0/Na4K8-LSX(b)270K265K260K260K250K225K215K215K220K220K230K230K  Intensity (arb. unit)Shift (ppm)290K(a)1400141014201430Temperature (K)Shift (ppm)200 220 240 260 280 30001 Peak Height (arb unit)Fig. 3 An example of temperature variation of the separated component on 23Na NMR spectraof K9.0/Na4K8-LSX taken in a field of 9.4 T. (a) Spectra observed every one hour in the course ofcooling from 290 K down to 215 K (process of the blue-colored dash-dotted arrow) and successiveheating from 215 K up to 270 K (process of the red-colored dash-doubledotted arrow). (b) The shiftand the height of each peak versus temperature. The height is normalized to the value at the startingtemperature, 290 K.111380 1400 1420 1440 146023Na K9.0/Na4K8-LSX(a)97 h75 h47 h23 h319K9 h  Amplitude (arb. unit)Shift (ppm)225K0 h(b)140014201440 Shift (ppm)0 50 10001 Peak Height (arb. unit)Time (h)Fig. 4 Long term observation of the separated component on 23Na NMR spectra of K9.0/Na4K8-LSX with constant temperature. The sample had been held at 319 K for one hour to prepare an initialthermal equilibrium condition and forced to be cooled until 225 K as rapid as possible. Observationof the spectrum at 225 K started with such the condition. (a) Spectra at each time. Representativeones are emphasized with thick lines. (b) The shift and the height of each peak versus temperature.The height is normalized to the value at the starting time of measurement at 225 K. Trends of theirbehavior are shown with dashed lines. Two points are plotted at each time point after 23 hours,since it can be confirmed by eye that the spectrum is composed of two parts. On the other hand,for the period prior to 23 hours, where it looks like one component, plots were made only for thepeaked component. Although fitting the spectrum can possibly extract two components from such asingle peaked shape, we did not perform such a fitting. To indicate the possibility of a left shouldercomponent for such a period, dotted lines are drawn as a guide.12 Introduction Experimental Results and Discussion Conclusion Acknowledgments