# Fileset

[Dalton_Trans_2024_Igarashi240504.pdf](https://mdr.nims.go.jp/filesets/82464dee-805f-47b8-9105-fced45df5282/download)

## Creator

[Mutsuo Igarashi](https://orcid.org/0000-0003-2665-1031), [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](https://orcid.org/0000-0002-7111-8730)

## Rights

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

## Other metadata

[Hyperfine couplings between the paramagnetic moment and nuclei in the metallic phase of low silica X zeolite loaded with potassium](https://mdr.nims.go.jp/datasets/7e26cd02-a703-4216-9a2e-e8a4ede672cc)

## Fulltext

Journal NameHyperfine couplings between paramagnetic moment andnuclei in the metallic phase of low silica X zeolite loadedwith potassiumMutsuo Igarashi,∗a Tadashi Shimizu,b Atsushi Goto,b Kenjiro Hashi,b Keiko Yamamichi,aand Takehito NakanocTemperature dependences of NMR spectra have been observed for 23Na and 27Al in the metallicphase of Na-K form low silica X (LSX) zeolite loaded with potassium, where the condition of sat-uration is achieved with loading level of 9.0 atoms per supercage and the paramagnetic momentexists as contributing to the magnetism of the system beyond simple isolated spin. Separated twopeaks have been recognized for 23Na, where the shift values show quite linear relation with suscep-tibility and so-called K-χ plot works quite well to give values as 0.32 kOe/µB and 0.40 kOe/µBfor hyperfine coupling constants. Although no separated peak is seen on 27Al NMR spectrum, thespectral centroid deviates to positive side. The shoulder of the spectrum scales to susceptibility andK-χ plot also works well to give value as 0.15 kOe/µB for hyperfine coupling constant. Orbital ofpotassium-originated electron confined in the cage of LSX is understood as seeping out over theframe work of zeolite relatively wider than that of sodium-originated case.1 IntroductionIt is well known that alkali metal loading introduces metallicproperty into some kinds of zeolites1. Low silica X (LSX) isone of the representative zeolite substances on such the phe-nomenon. Metallic property of such the substrate with sodiumloading has most extensively studied ever2–4 than the one withother alkali species as potassium5, rubidium6, and so on. It hasbeen known that the metallic phase of the system is accompaniedwith paramagnetic moment observed through susceptibility mea-surement and the appearance of separated component on 23NaNMR spectrum3,7,8 has been detected. Hyperfine coupling con-stant between such the paramagnetic moment and 23Na has beenestimated for sodium loading case through so called Clogston-Jaccarino plot3. On the other hand for 27Al in this system, whereonly central component with temperature independent behavioron the NMR spectrum is observed, hyperfine coupling is at mosttakes quite less value than 23Na3. It is concluded that orbital ofthe electron comes from loaded atoms is well confined within thecage of LSX in the case of sodium loading.The skeletal structure type for LSX is FAU, which is the IUPAC∗ E-mail: igarashi@gunma-ct.ac.jpa National Institute of Technology, Gunma College, Maebashi, Gunma 371-8530,Japan.b National Institute for Materials Science, Tsukuba, Ibaraki 305-0003, Japan.c Institute of Quantum Beam Science, Graduate School of Science and Engineering,Ibaraki University, Mito, Ibaraki 310-8512, Japan.nomenclature. When the system contains a monovalent cation M,the chemical formula is written as M12Al12Si12O48 per supercage(or β -cage). This corresponds to 1/8 of the cubic unit cell. We willrefer to this as M-form LSX. Such a charge balance cation can be acomplex one and the formula is rewritten as NaxK12−xAl12Si12O48for the case where both Na and K are included. We write it asNaxK12−x-LSX and refer to NaK-form LSX. x can vary from 0 to 12by replacing the cation in aqueous solution.Here we take a case of x = 4, where the formula is Na4K8-LSX.Na-K alloy clusters can be generated in its cages by loading guestK atoms. Following a previous report9, we write the system asKn/Na4K8-LSX. n is the number of loaded K atoms per supercage(or β cage). This system shows a ferrimagnetic transition for anarrow range of n around n = 7.710. Spectra of 23Na and 27AlNMR have been observed for n = 7.17 where ferrimagnetic tran-sition occurs at 7 K. It has a paramagnetic moment above 7 K,and several components have been detected that have shifts dueto hyperfine interactions between the paramagnetic moment and23Na nuclei. As has been reported5, this system shows metallicproperty for n >∼ 6. Studies with variation on n are expected togive hints to resolve the nature of the electron system. In this pa-per, we report NMR property of n = 9.0 case to explore how thebehavior of such the component changes through increasing n.2 ExperimentalIn the framework structure of LSX, β cages and supercages con-struct a double diamond structure11. Typical sites for cationsJournal Name, [year], [vol.],1–6 | 1SiAlOSupercage of FAUβ-cage++++ cation sites10.8 ÅI++I’IIIIIIII’Fig. 1 Schematic illustration of the framework of low silica X zeolite.Possible cation sites based on symmetry are shown with small circles.called site I, II and III are shown by small circles in Fig. 1. Thesites for Al and Si atoms correspond to crossing points on theframework in the figure. The Si and Al atoms are alternatively or-dered in the framework through the Si-O-Al bonds. Our sampleis made of such the substance by loading of external K into thosecages.Sample preparation has been done with the same procedureas before10. Vapor of K atom is thermally absorbed in vacuumenvironment. The loading level was estimated as 9.0 K atomsper supercage using chemical analysis. Therefore the sample forthe measurement is signed as n = 9.0. X-ray diffraction analysishas not been done for it but, taking into account the result ofan analysis on the case of Nan/Na12-LSX12, it is quite natural toconsider that there are several kinds of sites for cations. No occur-rence of magnetic transition has been detected with susceptibilitymeasurement above 2 K, although previously reported sample ofn = 7.1 shows ferrimagnetic transition at 7 K.For NMR measurements, approximately 0.3 g of the powdersample was sealed in a quartz glass cell. The probe cell was cylin-drical, with an outer diameter of 10 mm, an inner diameter of 8mm, and an overall sample width of about 15 mm. NMR spectraof 23Na and 27Al have been measured at 63 kOe at several pointsof temperature below 300 K. References for the frequency shiftsare taken with those of 23Na and 27Al in aqueous solution of NaCland AlCl3, respectively. Durations of first (π/2) and second (π)pulse to obtain spin echo have typically been taken as 6 µs and12 µs, respectively. Interpulse spacing for echo has typically beentaken as 100 µs to obtain spin echo signal. NMR parameters forNMR measurement at representative temperatures are tabulatedin Table 1. T1e is an effective T1 determined as a time that recov-ered magnetization becomes 1− 1/e (≃ 0.7) of thermal equilib-rium one. It is estimated in the zero shift region of the spectrum,at where T1 is relatively longer than tail part of the spectrum.Recycle delays between NMR measurements have been taken assufficiently long to escape from saturation of the pulse excitation.However the case of 4 K, where magnetization recovery especiallyin the region close to zero shift has long time constant, violatessuch the constraint. Therefore quality of the data at 4 K is lessthan others. NMR spectrum is obtained by Fourier transforma-tion of the spin echo signal. When the spectrum is wider thanTable 1 Measurement parameters for 23Na NMR spectra.Temperature Recycle delay Estimated T1e(K) (s) (s)300 2 < 0.190 20 ∼ 0.425 60 ∼ 0.74 60 -Fourier component of the rf pulse, multiply obtained spectra ateach frequency have been summed over for obtaining wide rangespectrum. Susceptibility and optical spectra have been taken inthe similar way in the reference5.3 Results and Discussion3.1 23Na NMR23Na NMR spectrum of Kn/Na4K8-LSX at representative temper-atures are shown in Fig. 2 for n = 9.0. Previously reported onefor n = 7.1 are also shown7. In the region around zero shift thespectrum of this time data for n = 9.0 has a structured componentsimilar to the one for n = 7.1.In addition, a separated component labelled ’#’ accompaniedwith a somewhat large shift is seen at 300 K. The shift value ofthis component is ∼1400 ppm, which is different from that for23Na in the bulk metal sodium, ∼1100 ppm13. Such a sharp com-ponent with a large shift is not seen for n = 7.1. It was thoughtin the initial stage of the study that the component ’#’ was dueto the metallic nature of the sample itself. But, the quite strangebehavior of this component has changed such the recognition8.It disappears from the spectrum below between 220 K and 260 Kthrough a complex hysteresis. It almost certainly comes from Na-K alloy distributed outside the particle of LSX crystal. Fortunatelya contribution from it on the whole system is considered as quiteminor. Based on signal intensity, the thickness of the Na-K alloydistributed on the surface of the zeolite crystal was estimated tobe less than ∼ 30 nm, which is quite less than the average size ofLSX crystal, ∼ 2 µm. Such a thin film can be assumed to have noinfluence on the optical spectra used to estimate the propertiesof the K clusters in the cages. In view of our aim, i.e. to dis-cuss the properties of LSX with loaded K, we omit this separatedcomponent from the latter discussion.In the structured component located in the center part, at leasttwo shifted peaks are present. They are labeled P1 and P2, asshown in the inset of Fig. 2. The shifts of P1 and P2 become strongespecially below 25 K. Accompanying such the behavior, width ofthe slopes in both side from the center part becomes large. Sincewe are interested in the shifts of the peaks, measurement hasbeen omitted for negative shift side from the point at the symbol’*’. Even for P1 and P2 the location of them become ambiguousbelow 15 K. They are superimposed to the long tail in positiveside.Moreover, a weak but observable component with a tempera-ture independent positive shift is observed, as indicated by ’@’ inthe figure. The shift of this signal ’@’ has a shift of ∼ 1200 ppm.2 | 1–6Journal Name, [year], [vol.],-0.1 0.0 0.1 0.2-0.5 0.0 0.5*23NaKn / Na4K8-LSX63Cuin coiln=7.1Intensity (arb. units)Shift (%)4K25K90K300K P2P1 @xxxn=7.1 Shift (%)25 Kn=9.0Intensity (arb. units)-0.5 0.0 0.5*@ 63Cuin coil# n=9.0Fig. 2 23Na NMR spectra of Kn/Na4K8-LSX at representative tempera-tures in a field of 63 kOe for n = 7.1 and n = 9.0.It is similar to that of 23Na in the metallic bulk sodium13. Fora similar reason as for the component labelled ’#’, we exclude itfrom the analysis. Although no additional experimental analy-sis was carried out for it, it could be a signal from deposited Naformed on the surface of the LSX crystal at the time of samplepreparation.We observed clear evidence on metallic property for the samplesuch as plasma edge on optical spectrum and expected tempera-ture independent Knight shifts on NMR spectrum as an evidencefor metallic nature. However, the shifts of the two peaks P1 andP2 depend on temperature. They can be manifestations of hyper-fine coupling between the magnetic electrons and the nuclei viathe Fermi contact interaction. Therefore, we need to confirm arelationship between Knight shift and susceptibility. The Knightshift is given by the non-zero probability of existence at the nu-clear sites for electrons with the Fermi energy EF . It is formallywritten as14,K =∆HH0=8π3< |uk(0)|2 >EF χ ≡ Aχ, (1)where H0 is the strength of the external magnetic field. ∆H is thedeviation of the effective internal field from the external field atthe nuclear site. Knight shift K is defined with both H0 and ∆H.uk(0) is the wave function at the nuclear site and < |uk(0)|2 >EF isan average of |uk(0)|2 at EF . χ is the susceptibility of the electronsystem. A is the hyperfine coupling constant. In the case of a sim-ple metal, χ is almost temperature independent and the Knightshift is essentially constant.We made so called Clogston-Jaccarino plot (K-χ plot) to esti-mate A15. The peak positions have ambiguity on the spectra forlower temperatures as described above, the plot has been doneabove 15 K. It is shown in Fig. 3. Each data follows a straightline. Thus, we can evaluate A as the slope, K/χ, which are shownin the third column of Table 2. As instantly recognized, the com-ponent ’@’ is independent of χ. This is consistent with our as-sumption, where the component ’@’ comes from Na located out-side of LSX crystals. Although K/χ is a dimensionless quantity inthe CGS unit system†, the value of A is reported in many tradi-tional papers with the unit Oe/µB, which allows us to comparethe values between different substances. The unit Oe/µB meansthat we observe an effective field A Oe when a Bohr magnetonµB = 9.27× 10−21 erg/Oe is located at the focused site. Then itis required to assume how much amount of the magnetic mo-ment distributes at where it is. We are not currently able to rig-orously estimate the distribution of the electron spins, some as-sumption must be included. In many cases of Kn/NaxK12−x-LSX,the Curie-Weiss law is observed in the temperature dependenceof the magnetic susceptibility; the Curie constant is close to thevalue assuming a spin-1/2 localized magnetic moment in each βcage1,5,9,10,16. The ferrimagnetic and ferromagnetic orders ob-served at 0 ≤ x ≤ 4 and x ≃ 7, respectively, at low temperaturescannot be explained without the presence of localized magneticmoments in β cages. Moreover, the presence of electrons confinedin β cages is clearly observed in the optical spectra16. Therefore,we will estimate the value of the hyperfine coupling constant as-suming a magnetic moment of one Bohr magneton per β cage.The cubic unit cell of this system with a lattice constant of 2.5 nmcontains eight of β cages. Then one β cage corresponds to a vol-ume V0 = 1.95×10−21 cm3. Then, A is given by (K/χ) × (µB/V0)in the unit of Oe/µB. The obtained values are shown in the fourthcolumn of Table 2. For 23Na, P1 and P2 are thought to originatefrom two different crystallographic sites. A of P1 is 80% of thatof P2. These values are directly proportional to the probability ofthe existence of electron spins on the nucleus at those sites, as isclear from eq. (1).We attempted to fit the spectrum to characterize the three shiftcomponents, P1, P2 and ’@’17. The fit results have two main char-acteristics: the P2 component is three or more times broader thanthe other components, and the integrated intensity of the P2 com-ponent is dominant, accounting for about 80% of the total, whilethose of P1 and ’@’ are about 9% and 11%, respectively. Referringan analysis on the cation distribution for the hydrated NaxK12−x-† K is clearly a dimensionless quantity by definition. χ is expressed in units ofemu/cm3 in this paper, which is also a dimensionless quantity in the CGS unit sys-tem.Journal Name, [year], [vol.],1–6 | 3Table 2 Values of K/χ by K-χ plot of the peaks on 23Na and 27Al NMRspectra.Component Nuclei K/χ (K/χ) × (µB/V0)Name (cm3/emu) (kOe/µB)@ 23Na 1.6±1.6 0.008±0.008P123Na 69±2 0.33±0.01P223Na 86±2 0.41±0.01D 27Al 32±3 0.15±0.01LSX by X-ray diffraction18, the cation sites I and I’ are most likelyoccupied by Na cations at x = 4. As shown in Fig. 1, the site Iis located at the center of doubled six-membered ring (D6R) andthe site I’ is located in the β cage on one side of D6R. Althoughthe occupancy of Na cation sites in K-loaded samples is not com-pletely known, we speculate the following from the NMR data.Since site I’ is more inside the β cage than site I, we can expecta higher weight of the wave function of the electron confined inthe β cage. Thus, P2 with a larger A may originate from site I’and P1 with a smaller A may originate from site I. This assump-tion is also supported by the spectral widths. Because 23Na withI = 3/2 has an electric quadrupole moment, the electric field gra-dient (EFG) dominates the spectral width. The site I in the centerof D6R should have higher symmetry and have less EFG, result-ing in the narrow spectrum. On the other hand, the site I’ has alower symmetry and has stronger EFG, resulting in the broaderspectrum. As mentioned above, P2 accounts for about 80% of thepeak integrated intensity. This means that the occupancy of siteI’ is much higher than that of site I. It is quite possible that thedistribution of Na cations in the dehydrated and K-loaded sam-ple of Kn/NaxK12−x-LSX is different from that in the hydrated andnon-loaded sample of NaxK12−x-LSX. Precise tracking of the vari-ation of shifted peaks on NMR spectrum by changing the loadingdensity is to be done to examine such occurrence. At lower tem-peratures, the spectrum becomes very broad due to the increasein paramagnetic magnetization of the electron spins. As seen inFig. 2, the P1 and/or P2 component may be present around the0.3% shift value at 4 K, but the peak positions are not clear, mak-ing analysis difficult. As mentioned earlier, the ’@’ componentis independent of the bulk magnetic susceptibility. Therefore, itmay originate from Na ions located on the surface or outside ofthe LSX crystal8.As seen in the inset of Fig. 2, there are several shifted peakspointed with symbols of ’x’ also for n = 7.1. Although an analysiswith K-χ plot has not yet been done for them, we are sure thatthose peaks also scales with χ. We gained a feeling that A seemsto vary with n, although experiments with varying n has not beenachieved yet. n variation of A may be one of a good index onelectronic state in this system. We observed susceptibility of bothn = 7.1 and n = 9.0 in the field of 63 kOe and obtained quite sim-ilar value. It is, for example, around 0.9× 10−5 emu/cm3 at 25K. Increasing n makes effective local fields from such the similarmagnitude of magnetic moment in the cage strengthen. AlthoughK-χ plot has not ever been done for sample of n = 7.1, A is ex-pected to be smaller than n = 9.0 case.0 1x10-5 2x10-505001000150010100@(b)DP1P2K (ppm) [emu/cm3](a)K9.0 / Na4K8-LSX  T (K) Fig. 3 Figures for the estimation of hyperfine coupling constants. (a)The susceptibility of K9.0/Na4K8-LSX plotted versus temperature. (b)Clogston-Jaccarino plot (K-χ plot) of the shifts. Triangles with upper(purple) and lower (green) directions, labeled 23P1 and 23P2 in the figure,correspond to the peaks P1 and P2 on the 23Na NMR spectra in the insetof Fig.2. Pentagons marked with the symbol ’@’ are also plotted in asimilar manner. Open circles (red) are the shift values of the uppermostGaussian component D in the four Gaussian fits for the shoulder of the27Al NMR spectrum in Fig.4. See text for the treatment of the curvefitting. Fitted to each data, the linear lines in (b) are drawn.3.2 27Al NMRThe 27Al spectrum of n = 9.0 case at representative tempera-tures is shown in Fig. 4 together with the previously reported dataof n = 7.17. The observation at 300 K for n = 9.0 was not madebecause of its minor importance. Although there are spectral sig-nals outside the area from the places marked with ’*’, we haveomitted the observation of them by the similar reason. Based onour experience on various zeolites, the wider tailed area comesfrom satellite transition among Zeeman levels of 27Al. The spin-spin relaxation time T2 of such the transition is too short to givean observable spin echo signal at higher temperatures, but be-comes slow at lower temperatures, allowing the spin echo signalto be observed. It is similar to a phenomenon reported for zeoliteLTA19, where T2 of the satellite component tends to decrease byraising of temperature. Shortening T2 causes a decrease in signal4 | 1–6Journal Name, [year], [vol.],intensity when observing the tailed components with constant τ,which is a separation time between rf pulses. Spectral width ofthe satellite powder pattern itself does not change with temper-ature. Since we are interested in the magnetic property and themostly wider component can effectively be separated from thecentered component around the peak, we omit the satellite com-ponents from our discussion.Then we tried to fit the spectrum only for the centered compo-nent. As seen in the inset of Fig. 4, the higher shift side of thespectrum for n = 9.0 clearly has a shoulder-like structure at 25K. The spectral centroid apparently has a positive shift. Since westill don’t have precise knowledge on environment of nucleus anddo not know appropriate functions to fit the experimental data,we tried to fit the whole spectrum with several Gaussians. At leastfour functions are needed to give a good coincidence. A typicalexample of the fitting result is shown in the inset of Fig. 4 by thincurves A, B, C and D located below the data points. Ratio of theintegral intensity of the component D, IDIA+IB+IC+ID, is 3×101 %,where IA, IB, IC and ID are intensities of the components A, B, Cand D, respectively.A similar analysis worked well at lower temperatures, 4 K and15 K17. The A and B components are understood to originatefrom the Al sites, where there is no Fermi contact with the elec-tron wave function, since the spectral centroid does not changemuch with temperature. The spectral shape of the componentwhose centroid is shifted is not known analytically, but it can bereproduced by two components, C and D, when decomposed bythe Gaussian function. These signals originate from Al sites withFermi contact interactions with the electronic wavefunction.As shown in Fig. 3(b), we created a K-χ plot for component Dto evaluate the upper bound on the coupling constant. This plotfits quite well with a linear line. The hyperfine coupling constanthas been estimated with the manner similar to the case of 23Naand is also shown in Table 2. This non zero value means thatthe orbital of the electron confined in the cage seeps out from thecage into the site of 27Al on the zeolite framework. The A of 27Alis smaller than that of 23Na. The probability of existence of theelectron wave function on 27Al is about 38–47% of that on 23Na,estimated from the ratio of A.Referring such the property, one notices little tilt on the highershift side shoulder around the place marked with ’S’ for the n= 7.1sample as shown in the inset of Fig. 4. This may be a result of theshift caused by the hyperfine interaction. However, the shift is toosmall to estimate the coupling constant. Although it is difficult todiscuss quantitatively, we can result that the difference of the shiftis not given by difference of magnetic moment but is given bydifferent coupling constant as is discussed above in the section of23Na. It is reasonable that higher loading of K into the cage leadsthe tail of the electron’s orbital to seep out into the framework ofLSX. It can also be said that the increase of the Fermi energy withthe loading density causes a slight hybridization of the electronicstates confined in the cage with those of the framework.In Na-loaded sodalite, which is a type of zeolite, A = 1.05kOe/µB was observed in the 27Al NMR20. This value is muchlarger than that observed in this study in K9.0/Na4K8-LSX. So-dalite has a much more compact crystal structure than LSX, withthe β cages arranged in a body-centered cubic structure with alattice constant of about 9 Å. The Na-loaded sodalite is an anti-ferromagnetic insulator with one unpaired s-electron occupyingeach cage. The high density of the structure is thought to cause alarge overlap of the s-electron wavefunction with the Al positionof the framework, resulting in a larger value of A compared tothat in K9.0/Na4K8-LSX.4 Conclusion23Na and 27Al NMR spectra between 300 K and 4 K of low silica Xzeolite loaded with potassium for saturation level have been ob-served. For 23Na, two peaks are found with susceptibility-scaledshift as similar to less-loaded sample, in which 7.1 potassiumatoms are loaded per supercage. The magnitude of the shiftsare several times larger than that of the less-loaded sample. Themain cause is that the hyperfine coupling constants have becomestronger. For 27Al, a deformed spectrum is found with a shoulder-like shape on the higher frequency side of the peak, and the cen-troid of the entire spectrum has a temperature-variance-positiveshift. The whole spectrum was analyzed with multiple sum ofGaussian functions, and the decomposed positively shifted com-ponent scales well with the susceptibility. The positively shiftedcomponent is concluded to be given by sites of 27Al nuclei, whichare located at the framework of zeolite, with hyperfine interac-tion from the atomic clusters in the cages. The less-loaded sampleshows a symmetrically widened spectrum and gives at most quitelittle shift.Author ContributionsM.I. designed the research. T.N. synthesized and characterizedthe sample. M.I., T.S., A.G., K.H., and K.Y. performed the NMRexperiments. M.I. and K.Y. analyzed the NMR data. M.I. and T.N.wrote the main manuscript. M.I. finalized the manuscript. Allauthors reviewed the manuscript.Conflicts of interestThere are no conflicts to declare.AcknowledgementsThe authors are grateful to Y. Nozue for his support. We alsothank T. Kodaira and T. Ikeda for providing high-quality zeo-lite crystals, and S. Tamiya for the chemical analysis. This workwas supported by JSPS KAKENHI (Grant Numbers JP15540353,JP15KK0165, JP16K05462, JP19K03738) and MEXT KAKENHI(Grant Number JP19051009).Notes and references1 T. Nakano and Y. Nozue, Adv. Phys.: X, 2017, 2, 254–280.2 T. Nakano, T. Mizukane and Y. Nozu, J. Phys. Chem. Solids,2010, 71, 650–653.3 M. Igarashi, T. Nakano, P. T. Thi, Y. Nozue, A. Goto, K. Hashi,S. Ohki, T. Shimizu, A. Krajnc, P. Jeglič and D. Arčon, Phys.Rev. B, 2013, 87, 075138–1–075138–7.4 M. Igarashi, P. Jeglič, A. Krajnc, R. Žitko, T. Nakano, Y. Nozueand D. Arčon, Sci. Rep., 2016, 6, 18682–1–18682–8.Journal Name, [year], [vol.],1–6 | 50.00 0.05-0.5 0.0 0.527AlKn / Na4K8-LSXn=7.1Intensity (arb. units)Shift (%)4K25K90K300KS  DCBAn=7.1Shift (%)25 Kn=9.0Intensity (arb. units)-0.5 0.0 0.5* ** * n=9.0Fig. 4 27Al NMR spectra of Kn/Na4K8-LSX at representative temper-atures in a field of 63 kOe for n = 7.1 and n = 9.0. The observation at300 K for n = 9.0 was not made because of its minor importance. TheGaussian curves in the inset labeled A (brown), B (magenta), C (green),and D (blue) are extracted components by fitting the data, which cor-responds to the thick-dashed line (red) on the experimental data (gray).The shift value of the peak of the component D is used in Fig.3 for theClogston-Jaccarino plot. Measurements have been skipped for 25 K and4 K on the spectral parts in the outer sides from the place signed with’*’, because of less importance of them.5 T. Nakano, D. T. Hanh, A. Owaki, Y. Nozue, N. H. Nam andS. Araki, Journal of the Korean Physical Society, 2013, 63,512–516.6 P. Jeglič, T. Nakano, T. Mežnaršič, D. Arčon and M. Igarashi,J. Phys. Soc. Jpn., 2020, 89, 073706–1–073706–4.7 M. Igarashi, T. Nakano, T. Shinizu, A. Goto, K. Hashi, K. Goto,K. Yamamichi and Y. Nozue, J. Magn. Magn. Mater., 2007,310, e307–e309.8 M. Igarashi, T. Shimizu, A. Goto, K. Hashi, K. Yamamichi andT. Nakano, submitted.9 D. T. Hanh, T. Nakano and Y. Nozue, J. Phys. Chemi. Solids,2010, 71, 677–680.10 T. Nakano, K. Goto, I. Watanabe, F. L. Pratt, Y. Ikemoto andY. Nozue, Physica B, 2006, 374–375, 21–25.11 H. A. M. Verhulst, W. J. J. Welters, G. Vorbeck, L. J. M. van deVen, V. H. J. de Beer, R. A. van Santen and J. W. de Haan, J.Phys. Chem., 1994, 98, 7056–7062.12 T. Ikeda, T. Nakano and Y. Nozue, J. Phys. Chem. C, 2014,118, 23202–23211.13 Metallic Shifts in NMR, ed. G. C. Carter, L. H. Bennett and D. J.Kahan, Pergamon Press, Oxford, 1977.14 C. P. Slichter, Principles of Magnetic Resonance, Springer-Verlag, Berlin, 1989.15 A. M. Clogston, V. Jaccarino and Y. Yafet, Phys. Rev., 1964,134, A650–A661.16 L. M. Kien, T. Goto, D. T. Hanh, T. Nakano and Y. Nozue, J.Phys. Soc. Jpn., 2015, 84, 064718–1–064718–9.17 See Electric Supplementary Information.18 Y. Lee, S. W. Carr and J. B. Parise, Chem. Mater., 1998, 10,2561–2570.19 M. Igarashi, T. Kodaira, T. Shimizu, A. Goto, K. Hashi,T. Nakano and Y. Nozue, Chem. Phys. Lett., 2007, 436, 80–83.20 I. Heinmaa, S. Vija and E. Lippmaa, Chemical Physics Letters,2000, 327, 131–136.6 | 1–6Journal Name, [year], [vol.], Introduction Experimental Results and Discussion 23Na NMR 27Al NMR Conclusion