DSpace at VNU: Insulator-to-metal Transition and Magnetism of Potassium Metals Loaded into Regular Cages of Zeolite LSX

Journal of the Korean Physical Society, Vol. 63, No. 3, August 2013, pp. 512∼516

Insulator-to-metal Transition and Magnetism of Potassium Metals Loaded
into Regular Cages of Zeolite LSX
Takehito Nakano, Duong Thi Hanh, Akihiro Owaki and Yasuo Nozue∗
Department of Physics, Graduate School of Science, Osaka University, Osaka 560-0043, Japan

Nguyen Hoang Nam
Center for Materials Science, Faculty of Physics,
Hanoi University of Science, Vietnam National University, Hanoi, Viet Nam

Shingo Araki
Graduate School of Natural Science and Technology, Okayama University, Okayama 700-0082, Japan
(Received 6 June 2012)
Zeolite LSX (low-silica X) crystals have an aluminosilicate framework with regular supercages
and β-cages. They are arrayed in a double diamond structure. The loading density of guest K
atoms per supercage (or β cage), n, can be controlled from 0 to ≈ 9. At n < 2, samples are nearly
nonmagnetic and insulating. The Curie constant has a clear peak at n = 3, and the electrical
resistivity suddenly decreases simultaneously. The electrical resistivity suddenly decreases again at
n = 6 and shows metallic phase at n > 6. These properties are explained by the polaron effect
including the electron correlation. Ferrimagnetic properties are observed at n ≈ 9. A remarkable
increase in the resistivity is observed at very low temperatures at n ≈ 9, and is discussed in terms
of the hypothesis of a Kondo insulator.
PACS numbers: 75.50.Gg, 75.50.Ee, 75.30.Mb, 71.38.-k, 75.20.Hr, 82.75.Vx
Keywords: Alkali metal, Cluster, Ferromagnetism, Ferrimagnetism, Polaron, Kondo lattice, Zeolite
DOI: 10.3938/jkps.63.512

I. INTRODUCTION

LSX [5] and sodalite [6–8], respectively.
Zeolite LSX has the FAU framework structure with

Si/Al = 1 as shown in Fig. 1(a). The chemical formula
of framework is given as Al12 Si12 O48 per supercage (or
β-cage). In LSX, β cages having the inside diameter of
≈7˚
A are arrayed in a diamond structure by the sharing
of double 6-memberred-rings (D6MRs). Among them,
the supercages of FAU with the inside diameter of ≈
13 ˚
A are formed and arrayed in a diamond structure by
the sharing of 12-membered-rings (12MRs) having the
inside diameter of ≈ 8 ˚
A. The distance between adjacent
supercages is 10.8 ˚
A which is shorter than the inside
diameter of supercage. Each β cage shares 6-memberedrings (6MRs) with adjacent four supercages. The chemical formula of zeolite LSX containing K cations is given
as K12 Al12 Si12 O48 per supercage (or β-cage) and is abbreviated as K12 -LSX, here. In the present paper, we
load guest nK-atoms into K12 -LSX. The total chemical
formula is given as K12+n Al12 Si12 O48 (abbreviated as
Kn /K12 -LSX, here).
When potassium metal is heavily loaded into Na4 K8 LSX, N´eel’s N-type ferrimagnetism has been observed
and is explained by assuming two non-equivalent magnetic sublattices of clusters in β-cages and supercages

Alkali metals loaded into the regular nanospace of zeolites exhibit exotic electronic properties that depend on
the structure of zeolites, the loading density, and the
alkali metals. The aluminosilicate frameworks of zeolite crystals provide different types of regular arrays of
nanocages, such as the double-diamond structure of β
cages and supercages, the CsCl structure of α and β
cages, and the body centered cubic structure of β cages in
zeolites LSX (low-silica X), A and sodalite, respectively.
The aluminosilicate framework has negative charges by

the number of Al atoms. Exchangeable cations (positive
ions), such as K+ , are distributed in the space of the
framework for the charge neutrality. The s-electrons of
guest alkali metals are shared with the zeolite cations
to form cationic clusters and are confined in the space
of cages of the framework. These s-electrons exhibit exotic magnetisms, although bulk alkali metals are nonmagnetic [1,2]. Ferromagnetism, ferrimagnetism and antiferromagnetism have been observed in zeolites A [2–4],
∗ E-mail:



-512-


Insulator-to-metal Transition and Magnetism of Potassium Metals · · · – Takehito Nakano et al.

-513-

at n ≈ 3. A sudden decrease in the electrical resistivity is
observed again at n ≈ 6, and a metallic phase is observed
at n > 6. Ferrimagnetic properties are observed at n ≈
9. In addition, a remarkable increase in the electrical
resistivity is observed at very low temperatures in the
metallic phase at n ≈ 9, and a Kondo insulator model is
discussed.

II. EXPERIMENTAL PROCEDURES
Fig. 1. (Color online) (a) Schematic illustration of the aluminosilicate framework of zeolite LSX having the FAU framework structure. β-cages are arrayed in a diamond structure
by the sharing of double 6-membered rings. Among them, supercages are formed. (b) Illustration of alkali-metal clusters
stabilized in β-cages and supercages of LSX zeolite.

[8–10]. When nNa atoms are loaded into Na12 -LSX
(Nan /Na12 -LSX), the optical spectrum shows an insulating phase up to n ≈ 10 and suddenly changes to a
metallic spectrum at n ≈ 12 [11]. The electrical resistivity dramatically decreases by several orders of magnitude
with increasing n from 11 to 12 [12]. Many paramagnetic
moments are thermally excited at n ≈ 12 [12]. The insulating and non-magnetic phase at n < 11 is explained
by the polaron effect as follows: An s-electron has a finite interaction with the displacement of cations, which
is called the electron-phonon interaction. A small polaron, which is a self-trapped state of an electron, can be
stabilized when the electron-phonon interaction is large
enough to trap the electron at the local lattice deformation induced by the electron itself [13]. If the electronphonon interaction is weak, a large polaron is stabilized
and moves freely. The small polaron is immobile because of a large lattice distortion. Two electrons can
be self-trapped by the strong electron-phonon interaction, and the small bipolaron in the spin-singlet state
is stabilized. If the electron-phonon interaction is large
enough to combine bipolarons, small multiple-bipolarons
can be stabilized. They are the case at n < 11. Large
polarons, however, are stabilized at n > 11 in the metallic state, if multiple-bipolarons become unstable due to
the increase in the Coulomb repulsion among electrons.
The thermal excitation of the paramagnetic susceptibility has been observed in the metallic state and is assigned
to paramagnetic moments of thermally excited small polarons. The anomalous paramagnetic behavior has been
observed in NMR study of 23 Na [14]. This insulatorto-metal transition and the thermal excitation of paramagnetic moments are explained by both the electron
correlation and the electron-phonon interaction in the
deformable structure of cations [13].
In the present research, we have studied the magnetic
property and the electrical resistivity in Kn /K12 -LSX. A
remarkable increase in the paramagnetic moments and a
sudden decrease in the electrical resistivity are observed

We used synthetic zeolite powder of Na12 -LSX which
were checked in terms of the chemical analysis for Si/Al
ratio and the X-ray analysis for structural quality and
purity. We exchanged Na cations with K ones in KCl

aqueous solution many times in order to prepare K12 LSX. The complete dehydration of the zeolite powder
was made by heating at 500 ◦ C for one day under high
vacuum. Distilled K metal and dehydrated zeolite powder were sealed in a glass tube, and K metal was adsorbed
into zeolite powder at 150 ◦ C through the vapor phase
as well as the direct contact with the zeolite powder. In
order to improve the homogeneity of loading density of K
metal, we performed the heat treatment of zeolite powder for two weeks. Finally, we obtained a homogeneous
K-loading. The average loading density of nK atoms per
supercage (or β cage) was controlled by adjusting the
weight ratio of K metal and zeolite. No residual K metal
was seen in either the optical spectrum or the optical
microscope image.
Samples for magnetic measurement were sealed in
quartz glass tubes. The DC magnetization was measured
by using a SQUID magnetometer (MPMS-XL, Quantum
Design) in the temperature range 1.8 - 300 K. For the
electrical resistivity measurements, the sample powder
was put between two gold electrodes, and an adequate
compression force ≈ 1 MPa was applied during the measurements. The electrical resistivity of the sample was
obtained by multiplying the measured resistance by the
dimensional factor (area/thickness) of compressed powder. Due to the constriction resistance between powder
particles, the observed electrical resistivity is about oneorder of magnitude larger than the true value. The relative values, however, can be compared with each other,
because of the constant compression force. Because of
the extreme air-sensitivity of the sample, the sample
powder was kept in a handmade airproof cell. These procedures were completed in a glovebox filled with pure He
gas containing less than 1 ppm O2 and H2 O. Then, the
cell was set in the sample chamber of Physical Property
Measurement System (PPMS, Quantum Design). The
sample temperature was controlled between 300 and 2 K.
Impedance measurements on the cell were made by the

4-terminal measurement method by using Agilent 4824A
LCR meter in the frequency range from 20 Hz to 2 MHz
and DC. We analyzed the frequency dependence of the
complex impedance by the Cole-Cole plot and checked


-514-

Journal of the Korean Physical Society, Vol. 63, No. 3, August 2013

Curie constant at n ≤ 2 indicates that about 20% of supercages have magnetic moments of spin 1/2. Electrons
in β cages are not observed in the optical spectra at low
loading densities [11]. In Fig. 2(a), the Curie constant
has a clear peak at n ≈ 3 and quickly decreases at n ≈
4. The peak value at n ≈ 3 amounts to ≈ 100% distribution of magnetic moments with spin 1/2. The Curie
constant gradually increases for n > 4, and has the large
value corresponding to ≈ 100% distribution of magnetic
moments at n ≈ 9.
The Weiss temperature (TW ) estimated from the
Curie-Weiss law is plotted in Fig. 2(b). It shows small
negative values up to n ≈ 8.5, and quickly decreases
down to –10 K at n ≈ 9. Spontaneous magnetization is
clearly observed at n ≈ 9. The extrapolated Curie temperature (TC ) is plotted in the same figure. From the
negative value of the Weiss temperature, the existence
of an antiferromagnetic interaction is very clear. Hence,
the observed spontaneous magnetization is assigned to
the ferrimagnetism, where two non-equivalent magnetic
sublattices, possibly clusters in supercage- and β-cagenetworks, have an antiferromagnetic interaction through
6MRs, likely, N´eel’s N-type ferrimagnetism observed in
Kn /Na4 K8 -LSX [8–10].

2. Electrical resistivity

Fig. 2. (Color online) (a) Loading density dependence of
the Curie constant in Kn /K12 -LSX, and (b) that of the Curie
(TC ) and the Weiss (TW ) temperatures.

the reliability of the resistivity at < 109 Ωcm. A very
small background resistivity, originating from the electric
circuit inside the cell, was included at the order of 0.1
Ωcm. This background was subtracted from the value.

III. EXPERIMENTAL RESULTS AND
DISCUSSION

The electrical resistivity at 300 K is quite n-dependent
as shown in Fig. 3(a). The resistivity at n ≤ 2 is very
high, as expected from the optical spectrum [15], but
suddenly decreases at n > 2 in Fig. 3(a). The resistivity
gradually increases up to n = 6. However, the resistivity
suddenly decreases again at n ≈ 6 and shows very small
values at n > 6. As shown in Fig. 3(b), the resistivity
at n = 6.2 is very low even at low temperatures. This
result implies that some amounts of carriers exist at low
temperatures, indicating that a nearly metallic phase is
realized at n > 6. With the increase in n, the resistivity decreases at higher temperatures (T > 20 K), but
quickly increases at very low temperatures (T < 20 K).
At n = 9.0, the value at the lowest temperature is more
than 100 times of those at higher temperatures. This result clearly indicates that a very small gap, such as ≈ 1
meV, exists at the Fermi energy. Samples showing these

strange temperature dependences exhibit ferrimagnetic
properties as well.

1. Magnetic properties

The Curie-Weiss behavior is observed in the temperature dependence of the magnetic susceptibility of
Kn /K12 -LSX. The loading density dependence of the
Curie constant is estimated from the Curie-Weiss law
and is plotted in Fig. 2(a). If each supercage (or β-cage)
has the magnetic moment of spin 1/2, the Curie constant
is expected to be 3.21 × 104 Kemu/cm3 . The observed

3. Polaron effects

In order to explain the high Curie constant and the
low resistivity at n ≈ 3 found in Figs. 2(a) and 3(a), respectively, we propose the polaron effect for s-electrons
in zeolite. According to the theory of self-trapping of an
electron in the deformable lattice [13], the self-trapped


Insulator-to-metal Transition and Magnetism of Potassium Metals · · · – Takehito Nakano et al.

Fig. 3. (Color online) (a) Loading density dependence of
the electrical resistivity at 300 K in Kn /K12 -LSX, and (b)
the temperature dependence of the electrical resistivity at n
= 6.2, 8.4, 9.0.

Fig. 4. (Color online) Schematic illustration of adiabatic
potentials for polarons expected at n < 6 and n > 6 in
Kn /K12 -LSX. See the text for the details.

-515-

state (small polaron), can be stabilized in the case of
a strong electron-phonon interaction. In the small polaron, the depth of the deformation potential for electron
must be deeper than the kinetic energy. If the Coulomb
repulsive interaction U between two electrons bound in
the deformation potential well is smaller than the energy
gain by the lattice distortion for two electrons at n < 2,
the small bipolaron will be stabilized, as shown in Fig. 4,
where adiabatic potentials for different types of polarons
are illustrated for n < 6 and n > 6. Small bipolarons
have a heavy effective mass and are immobile. They
have a very small contribution to the electrical conductivity. Small bipolarons have a closed electronic shell and
are non-magnetic (spin-singlet). Hence, the hopping of
an electron to neighboring small bipolaron states will be
suppressed. The small Curie constant and the high resistivity at n < 2 in Figs. 2(a) and 3(a), respectively, can be
explained by small bipolarons. However, at n ≈ 3, small
tripolarons become more stable than small tetrapolarons,
because the Coulomb repulsion energy among four electrons is significant in small tetrapolarons. Tripolarons
are paramagnetic and can contribute to the hopping conduction because of the open electronic shell. Adiabatic
potentials of these small multiple-polarons are illustrated
schematically in Fig. 4. The increases in the Curie constant and the hopping conduction at n ≈ 3 can be explained.
With increasing n, small multiple-polarons are generated successively. These small multiple-polarons can become unstable suddenly above a certain critical value of
n, and large polarons, which are mobile, may become stable, indicating that the stability of large polarons show
the insulator-to-metal transition. This type of insulatorto-metal transition has been observed in Nan /Na12 -LSX
at n ≈ 12 [12]. In Kn /K12 -LSX, a similar insulator-tometal transition may occur at n ≈ 6. The smaller critical
value of n in the K-system is due to the weaker electronphonon interaction compared with the Na-system. The
electrons (polarons) in supercages mainly contribute to
the electrical conductivity, because of the large windows

(12MRs) of supercages. Electrons in β cages, however,
may have no contribution to the conductivity, because
of both the well-localized wave functions in β cages and
the high potential barriers by D6MRs between them, as
shown in Fig. 1. An electron in β cage can have magnetic moment and contribute to the remarkable increase
in the Curie constant at higher loading densities in Fig.
2(a). A sudden decrease in the resistivity at n ≈ 6, however, has no correlation to the Curie constant. Hence,
the insulator-to-metal transition is independent of β cage
clusters, but occurs in the clusters in the supercage network.
The localized electronic state in β cage can have a
finite hybridization with supercage electrons through
6MRs. In order to explain the ferrimagnetism observed
at n ≈ 9, an antiferromagnetic interaction through 6MRs
is supposed between non-equivalent magnetic sublattices
of clusters in β cages and supercages. This interaction


-516-

Journal of the Korean Physical Society, Vol. 63, No. 3, August 2013

Kn /K12 -LSX. The Curie constant has a clear peak at
n ≈ 3, and the electrical resistivity suddenly decreases
simultaneously. A sudden decrease in the electrical resistivity is observed at n ≈ 6, and a metallic phase appears
at n > 6. These properties are explained by the polaron
effect. Ferrimagnetic properties are observed at n ≈ 9. A
remarkable increase in the resistivity is observed at very
low temperatures at n ≈ 9. This result is interpreted in
terms of the hypothesis of the Kondo insulator.

ACKNOWLEDGMENTS

Fig. 5. (Color online) Schematic illustration of density
of states at the supercage network and the localized state
at β-cage. One-electron and two-electron states of β-cage
cluster are located at below and above the Fermi energy of
the supercage metallic network, where the Fermi energy is
located at the center of the narrow band.

and the electron correlation in narrow β cage can lead to
the model of the Kondo lattice, as discussed in the next
section.

4. Possibility of a Kondo insulator

As seen in Fig. 3(b), the electrical resistivity at the
metallic phase shows a remarkable increase at very low
temperatures. At least, a very narrow gap may exist at
n ≈ 9, but no gap at n ≈ 6.2. Such a narrow gap at
n ≈ 9 is hardly expected from the ordinary electronic
model. Hence, we propose a model shown in Fig. 5 [5],
where the Fermi energy is located at the center of the
narrow band provided by the clusters in the supercage
network, and the localized state at β cage is located below (above) the Fermi energy for one- (two-) electron
state. The Coulomb repulsion energy U is supposed for
two electrons in the β cage. Differently from the ordinary Kondo scheme, metallic electrons at the supercage
network have the spin polarization, because both of the
supercage and the β-cage networks have magnetic moments in the ferrimagnetic state. If a small gap can be
opened at the Fermi energy, likely a Kondo insulator,
the electrical resistivity increases at very low temperatures. This model is quite speculative, and further study

is needed.

IV. CONCLUSION
Remarkable loading-density dependences are observed
in the Curie constant and the electrical resistivity in

This work was supported by Grant-in-Aid for Scientific Research (24244059 and 19051009) and by G-COE
Program (Core Research and Engineering of Advanced
Materials-Interdisciplinary Education Center for Materials Science).

REFERENCES

[1] T. Nakano, N. H. Nam, T. C. Duan, D. T. Hanh, S.
Araki and Y. Nozue, to be published in J. Kor. Phys.
Soc.
[2] Y. Nozue, T. Kodaira and T. Goto, Phys. Rev. Lett. 68,
3789 (1992).
[3] Y. Nozue, T. Kodaira, S. Ohwashi, T. Goto and O.
Terasaki, Phys. Rev. B 48, 12253 (1993).
[4] T. Nakano and Y. Nozue, J. Comput. Meth. Sci. Eng. 7,
443 (2007).
[5] D. T. Hanh, T. Nakano and Y. Nozue, J. Phys. Chem.
Solids 71, 677 (2010).
[6] V. I. Srdanov, G. D. Stucky, E. Lippmaa and G. Engelhardt, Phys. Rev. Lett. 80, 2449 (1998).
[7] T. Nakano, M. Matsuura, A. Hanazawa, K. Hirota and
Y. Nozue, Phys. Rev. Lett. 109, 167208 (2012).
[8] T. Nakano, R. Suehiro, A. Hanazawa, K. Watanabe, I.
Watanabe, A. Amato, F. L. Pratt and Y. Nozue, J. Phys.
Soc. Jpn. 79, 073707 (2010).
[9] T. Nakano, K. Goto, I. Watanabe, F. L. Pratt, Y. Ikemoto and Y. Nozue, Physica B 374-375, 21 (2006).

[10] D. T. Hanh, T. Nakano and Y. Nozue, J. Phys. Chem.
Solids 71, 677 (2010).
[11] T. Nakano, T. Mizukane and Y. Nozue, J. Phys. Chem.
Solids 71, 650 (2010).
[12] Y. Nozue, Y. Amako, R. Kawano, T. Mizukane and T.
Nakano, J. Phys. Chem. Solids 73, 1538 (2012).
[13] Y. Toyozawa, Prog. Theor. Phys. 26, 29 (1961).
[14] M. Igarashi, T. Nakano, P. T. Thi, Y. Nozue, A. Goto,
K. Hashi, S. Ohki, T. Shimizu, A. Krajnc, P. Jeglic and
D. Arcon, Phys. Rev. B 87, 075138 (2013).
[15] Y. Ikemoto, T. Nakano, M. Kuno, Y. Nozue and T.
Ikeda, J. Mag. Mag. Mat. 226-230, 229 (2001).