Article https://doi.org/10.

1038/s41467-023-44602-3

Alkali metal bilayer intercalation in graphene

Received: 3 August 2023 Yung-Chang Lin 1,2 , Rika Matsumoto3, Qiunan Liu2, Pablo Solís-Fernández 4
,
Ming-Deng Siao5, Po-Wen Chiu 5,6, Hiroki Ago 4,7 & Kazu Suenaga 1,2
Accepted: 21 December 2023

Alkali metal (AM) intercalation between graphene layers holds promise for
Check for updates electronic manipulation and energy storage, yet the underlying mechanism
remains challenging to fully comprehend despite extensive research. In this
study, we employ low-voltage scanning transmission electron microscopy (LV-
STEM) to visualize the atomic structure of intercalated AMs (potassium,
rubidium, and cesium) in bilayer graphene (BLG). Our findings reveal that the
1234567890():,;
1234567890():,;

intercalated AMs adopt bilayer structures with hcp stacking, and specifically a
C6M2C6 composition. These structures closely resemble the bilayer form of fcc
(111) structure observed in AMs under high-pressure conditions. A negative
charge transferred from bilayer AMs to graphene layers of approximately
1~1.5×1014 e−/cm−2 was determined by electron energy loss spectroscopy (EELS),
Raman, and electrical transport. The bilayer AM is stable in BLG and graphite
superficial layers but absent in the graphite interior, primarily dominated by
single-layer AM intercalation. This hints at enhancing AM intercalation capa-
city by thinning the graphite material.

Intercalation, the insertion of foreign atoms or molecules within a host high energy consumption in which the pursuit of increased charge
material, has garnered significant attention for its potential applica- capacity remains crucial.
tions in energy storage, catalysis, and electronics1–3. The weak inter- To gain a deeper understanding of GIC devices and enhance
layer interaction in layered materials creates a natural two-dimensional their properties, it is essential to elucidate the arrangement of AM
(2D) nano-container capable of hosting foreign species for self- atoms between graphene layers. However, the atomic structure of
assembly4. The intercalation of AMs in graphite has been extensively AMs intercalated in graphite remains unclear, primarily due to the
studied for nearly a century since its initial report in 1926 by K. Fre- challenge of directly visualizing the intercalants. Previous studies
denhagen and G. Cadenbach5. Alkali metals (AMs), including lithium have relied on X-ray diffraction (XRD)4 and electron diffraction
(Li), sodium (Na), potassium (K), rubidium (Rb), and cesium (Cs), have techniques11 to infer the intercalation stage and atomic arrange-
attracted particular interest due to their low electronegativity, high ment. Based on XRD measurements, it has been widely accepted
reactivity, catalytic behavior, and ability to store charge in graphite4. that each individual interlayer space within the GIC can accom-
The AM-graphite intercalation compounds (AM-GICs) exhibit tunable modate only a monolayer of AM atoms. However, these methods
optical properties, transitioning from black to gold in color as the have limited spatial resolution, hindering atomic-level character-
intercalation density increases6. The transfer of charge from AMs ization. Recently, a study utilizing transmission electron micro-
enhances the carrier concentration in the graphene layers, resulting in scopy (TEM)12 and density functional theory (DFT) calculation13
GICs with high conductivity and even superconductivity7–10. The indicated the presence of multilayer stacking of Li within the
application of AM-ion batteries is particularly significant in this era of interlayer gap of BLG. These findings suggest that the number of

1
Nanomaterials Research Institute, National Institute of Advanced Industrial Science and Technology (AIST), Tsukuba 305-8565, Japan. 2The Institute of
Scientific and Industrial Research (ISIR-SANKEN), Osaka University, Osaka 567-0047, Japan. 3Department of Engineering, Tokyo Polytechnic University,
5-45-1 Iiyamaminami, Atsugi, Kanagawa 243-0297, Japan. 4Global Innovation Center (GIC), Kyushu University, Fukuoka 816-8580, Japan. 5Department
of Electrical Engineering, National Tsing Hua University, Hsinchu 30013, Taiwan. 6Institute of Atomic and Molecular Sciences, Academia Sinica, Taipei
10617, Taiwan. 7Interdisciplinary Graduate School of Engineering Sciences, Kyushu University, Fukuoka 816-8580, Japan. e-mail: [email protected];
[email protected]

Nature Communications | (2024)15:425 1

Article https://doi.org/10.1038/s41467-023-44602-3

graphene layers may play a significant role in determining the unit cell, forming a C6Cs2C6 structure, as depicted in  Fig.
pffiffiffi1bpand
ffiffiffi  Sup-
capacity to store AM atoms. plementary Fig. 2c. The two Cs lattices with identical 3 × 3 R30
The formation and dynamic behavior of the intercalants within structure are represented by different shades of pink for better
the confined interlayer space can be significantly different from their visualization. The chemical composition of the C6Cs2C6 is further
3D counterparts. However, without the ability to visualize their atomic- confirmed by EELS quantification, as presented in Supplementary
level structures, these distinctions would remain elusive. BLG is a Fig. 3. This raises the question of whether the Cs atom in this super-
promising host substrate for visualizing intercalant structure using LV- dense C6Cs2C6 structure arrange itself in single-layer form, or should
(S)TEM14, as it lacks the additional blocking layers found in thicker separate into two layers with two individual atomic planes.
graphite. In this study, we aimed to unravel the intercalation behavior To address this question, we employed both experimental and
and the storage capacity of BLG and graphite for AM intercalation. In theoretical approaches. In the ADF image shown in Fig. 1a, a clear
contrast to the electrochemical intercalation process, vapor phase distinction can be observed between the upper and lower domains in
intercalation represents a relatively “clean” process and can allow a terms of atomic arrangement, as well as in the corresponding FFT
clear visualization of the atomic structure of AM in BLG. Our focus was pattern (Supplementary Fig. 4). The pink-colored spheres overlaid on
primarily on K, Rb, and Cs due to their larger atomic size, which makes the Cs atoms of the two domains indicate a hexagonal arrangement for
them easier to detect and image using STEM. Our observations reveal the upper domain, while the lower domain exhibits a laterally dis-
that the intercalated AMs adopt a bilayer configuration with a hcp placed atomic configuration along the x-axis. Figure 1f illustrates the
structure within the interlayer gap of BLG. No single-layer or three- ADF profile of these two domains, obtained from the pink and light
layers AM configurations were found in BLG. The interlayer spacing green boxes in Fig. 1a. The shortest distance between neighboring Cs
can be expanded to up to 11.25 Å by the presence of two layers of Cs atoms in the honeycomb Cs lattice of the upper domain measures
atoms. This bilayer AM structure is also observed in thin graphite films approximately 0.25 nm, which is consistent with the atomic model of
with a thickness of 5 nm but not in thicker graphite films. The amount C6Cs2C6 structure shown in Fig. 1d. In contrast, the lower domain
of charge transfer from the intercalated AMs to BLG is further mea- exhibits a shorter Cs-Cs distance of 0.14 nm. The nearly 56% reduction
sured by EELS, providing valuable insights into the electronic inter- in the Cs-Cs distance along one direction is highly unusual for a
action between the intercalant and the host material. monolayer system under equilibrium conditions at room temperature,
without any external strain. Our experiments therefore revealed mul-
Results and discussion tiple Cs intercalated domains in BLG displaying distorted structures
Cs intercalation in bilayer graphene with inequivalent shortest Cs-Cs distances (Supplementary Fig. 5). The
The intercalation of AMs (K, Rb, and Cs) into the BLG on TEM grids was migration of Cs domains with the same structural configuration was
achieved through a vapor phase intercalation process. We began our also monitored during in-situ STEM imaging. Our analysis indicates
analysis with the intercalation of Cs, the heaviest AM. The number of that this migration can be attributed to the displacement of two indi-
synthesized graphene layers15,16 was determined by the reflecting color vidual Cs lattices (Supplementary Fig. 6). This indicates that the two Cs
contrast after transferring to a SiO2/Si substrate, along with corre- lattices separate into two layers along the c-axis, as illustrated in the
sponding Raman spectroscopy. The quality of the BLG sample was model shown in Fig. 1e. Our DFT geometry optimization calculations
extensively studied, and growth parameters were fine-tuned to achieve also support this bilayer structure model, as the total energy of K, Rb,
nearly 95% BLG coverage with less than 1% of trilayer or multilayer and Cs intercalated in BLG shows lower energy for the bilayer form
graphene in the grown samples. To evaluate if the sample is full compared to the monolayer (Supplementary Fig. 7). The interlayer
intercalated, a large pyrolytic graphite sheet (PGS) measuring spacing in bilayer graphene is 3.4 Å, which expands to 5.95 Å with the
0.5 cm × 2 cm × 25 µm was placed in a glass ampoule alongside the BLG intercalation of one Cs layer and to 11.25 Å with two Cs layers. The two
sample on a TEM grid (0.3 cm in diameter). The PGS serves as an layers of Cs exhibit a close-packed stacking structure, allowing for a
indicator for monitoring the intercalation stage. When the PGS turns a slight lateral movement and resulting in distorted hexagonal patterns
golden color, it signifies that the intercalation has reached stage 1 for where Cs atoms appear dimerized. This hcp-stacked Cs bilayer in BLG
the graphite, marking the endpoint of the reaction. Given that the BLG resembles the bulk Cs crystal under high pressure conditions, which
sample shares the same environment as the PGS, and its relatively adopts an fcc structure when viewed from the (111) direction (Sup-
smaller size compared to the PGS, BLG should also meet the conditions plementary Fig. 8). The distance between two Cs layers in the bulk Cs
for stage 1 full intercalation (details are presented in the Method sec- crystal along the fcc(111) plane is approximately 3.45 Å. After inter-
tion). Figure 1a presents an annular dark-field (ADF) image of Cs calation into BLG, Cs metal atoms tend to undergo electron charge
intercalated in BLG. On the left side, BLG is observed with isolated Cs transfer to the outer graphene layers, resulting in a partial positive
atoms anchored, and the twist angle between the two graphene layers charge on each Cs atomic layer. Consequently, this leads to an enlar-
is determined to be 2.7° by analyzing the fast Fourier transform (FFT) gement of the distance between the two Cs layers to 5.02 Å due to the
pattern (Fig. 1b). The right half of the ADF image displays a large and repulsive electrostatic forces involved, therefore resulting in further
continuous domain of Cs, showing a honeycomb lattice with hex- expansion of the BLG interlayer distance. At room temperature and
agonal symmetry, as confirmed by the FFT image shown in Fig. 1c. The ambient pressure condition, AMs are known to crystallize in a bcc
number of graphene layers can also be precisely determined based on structure, while they tend to transform into fcc structure under high
the ADF contrast, and the intercalated Cs is confined within the BLG pressure17–20. For Cs, the bcc-to-fcc phase transformation occurs at
region, as shown in Supplementary Fig. 1. The first-order diffraction pressures ranging from 2.37 to 4.3 GPa21–23. Therefore, the shortest Cs-
spots correspond to the Cs[100] plane, indicating a d-spacing of Cs hcp Cs bilayer distance in projection is calculated to be 2.44 Å which
3.70 Å. The second-order diffraction spots, corresponding to the is nearly identical to that of Cs bilayer found in this study. This finding
Cs[110] plane, exhibit a d-spacing of 2.14 Å, which is commensurate confirms that spatial confinement between BLG layers provides a two-
with the graphene C[100] spots. Based on the structural
pffiffiffi pffiffifficonfiguration,
 dimensional (2D) space for Cs atoms to stack and form high-pressure

it can be concluded that the Cs atoms form a 3 × 3 R30 lattice. phase with a bilayer configuration. Previous predictions have also
At first sight it can be consistent to the C6CsC6 structure overlaid on indicated that molecules trapped in the van der Waals gap between
the graphene lattice illustrated in Supplementary Fig. 2b. However, the two graphene layers can experience pressures as high as 1 GPa24.
Cs lattice in single-layer C6CsC6 structure does not exhibit the hon- Interestingly, we also found that the domain size of the intercalated Cs
eycomb lattice observed in the experimental real space image. Pro- bilayer varies with the twist angle of BLG. This will be discussed in
ducing a honeycomb Cs lattice requires having two Cs atoms in the detail in a follow-up paper.

Nature Communications | (2024)15:425 2

Article https://doi.org/10.1038/s41467-023-44602-3

pffiffiffi pffiffiffi  
Fig. 1 | Scanning transmission electron microscopy (STEM) image and atomic highlights the unit cell of the Cs lattice with 3 × 3 R30 lattice, where a = b =
structure of Cs bilayer in bilayer graphene (BLG) (C6Cs2C6). a A STEM annular 4.26 Å. e Side and perspective view of the C6Cs2C6. The Cs atoms in different
dark-field (ADF) image of Cs-intercalated BLG displaying the C6Cs2C6 structure. atomic planes are color-coded with two different shades of red. f ADF profile of the
b, c Fast Fourier transform (FFT) image of BLG and Cs domains in (a) (scale hexagonal Cs layer along the pink and light green boxes in (a). The red profile
bar = 5 nm−1). BLG exhibits a twist angle of 2.7° between its two graphene layers. The displays the shortest distance between two Cs atoms as 0.25 nm, consistent with
[100] and [110] spots of C6Cs2C6 domain correspond to interplanar d-spacings of the atomic model shown in (d). The green profile shows the shorter distance
3.70 Å and 2.14 Å in real space, with the Cs [110] spots overlapping with the gra- between the Cs atoms, indicating the lateral displacement of the two Cs
phene [100] spots. d Top-view atomic model of C6Cs2C6. The yellow rhombus atomic plans.
highlights a graphene (1 × 1) unit cell with a = b = 2.46 Å, while the red rhombus

Charge transfer of alkali metal intercalated BLG confined within the region of the BLG domain after the e-beam damage
Similar to Cs, the intercalation of K and Rb in BLG was also performed to the sample (see also Supplementary Fig. 1), suggesting that the AM
and examined using STEM. Compared to Cs intercalation, K and Rb bilayer is confined in the interlayer gap between two graphene layers.
were found to be more sensitive to the electron beam. Lowering the More interestingly, the intercalated AM layer is uniformly distributed
probe current to 5 pA was a key step to capture the ADF image of K and in BLG (Supplementary Fig. 10), and we have not observed single-layer,
Rb in BLG. Our observations revealed that both K and Rb exhibit the three-layer, or four-layer arrangements of these AM layer in BLG. We
same hcp bilayer arrangement as Cs, forming a C6M2C6 structure, as conducted STEM image simulations for 1–4 layers of AM, as illustrated
shown in Figs. 2a, c. Single crystal BLG15 and polycrystalline BLG25 were in Supplementary Fig. 11. Through the quantitative ADF contrast
both employed for AM intercalation, and we observed a C6M2C6 comparison, we can effectively distinguish between 2 layers from 3 and
structure in both cases. Consequently, both edge and defect inter- 4 layers. The results of EELS quantification for K and Rb, confirming the
calation processes should presumably contribute to the overall inter- C6M2C6 structure, are summarized in Supplementary Fig. 12. Based on
calation of AM in BLG. Essentially, we determine whether the AM our findings, we can confidently conclude that the intercalated AM
bilayer resides in the interlayer gap or adheres to the graphene surface consistently adopts a bilayer form. This consistent bilayer form across
based on structural stability under e-beam irradiation and spectro- all three heavy AMs suggests a common structural pattern for inter-
scopy. The AMs adsorbed on the surface are mostly oxidized and calation, with the C6M2C6 structure being the most stable configura-
amorphous; even if they initially form a crystalline structure, the sur- tion in BLG. These bilayer K, Rb, and Cs in BLG are well protected from
face AM layers are simply wiped out by the e-beam. However, AMs oxidation by the graphene layers. The low electronegativity nature of
intercalated in BLG show lattice defects such as multiple vacancies AM tends to donate electron when in contact with other elements,
during prolonged scanning e-beam observation, as demonstrated in contributing to stabilization. In the intercalation of two AM layers,
Supplementary Fig. 9. The intercalated AM bilayers are consistently both layers can undergo charge transfer to the outer graphene layers.

Nature Communications | (2024)15:425 3

Article https://doi.org/10.1038/s41467-023-44602-3

Fig. 2 | Structure and spectroscopy of K, Rb, and Cs bilayers in BLG. a–c ADF the surface of BLG, and Cs2CO3. The oxygen K-edges for these materials are pre-
image of C6K2C6, C6Rb2C6, and C6Cs2C6 in BLG. d The electron energy loss spec- sented in the insets. g–i, EELS valence loss spectra of K, Rb, and Cs intercalated
troscopy (EELS) core loss spectra for the K L2,3-edge of C6K2C6, amorphous oxi- samples. The black lines represent the original valence loss spectra, while the
dized compounds of K adsorbed on the surface of BLG, and K2CO3. e The EELS core colored lines depict the background-subtracted edges. The grey dashed lines serve
loss spectra for the Rb L2,3-edge of C6Rb2C6, amorphous oxidized compounds of as a reference spectrum of AB stacking BLG. The charge carrier plasmon in all three
Rb adsorbed on the surface of BLG, and Rb2CO3. f The EELS core loss spectra for samples appears as a broad peak around 1.5 ~ 3 eV, as highlighted by the corre-
the Cs M4,5-edge of C6Cs2C6, amorphous oxidized compounds of Cs adsorbed on sponding colors.

However, if trilayers or more layers of AM are intercalated, no charge deposited AMs or their oxides (Supplementary Fig. 13a), and the
transfer will occur in the interior layers of AM, result in less energetic crystalline intercalated AMs remained and were clearly visualized in
favorability compared to the bilayer case. The theoretical calculation13 Supplementary Fig. 13c. Based on these observations, we conclude that
also suggests that the trilayer form of K, Rb, and Cs is less energetically surface deposition does not adversely affect the intercalated structure.
favorable than the bilayer structure when intercalated in BLG. Addi- We used EELS to investigate the oxidation behavior of AM bilayers in
tionally, an increasing number of AM layers also results in less favor- BLG, comparing them with the oxidized AMs on the surface of BLG, as
ability in formation energy in BLG. well as reference oxides (M2CO3). The results are displayed in
Surface absorption is an inherent aspect of the gas phase inter- Figs. 2d–f. The L3 and L2 edges of K in BLG (C6K2C6) show peaks at
calation sample. However, even when the BLG is coated with amor- 295.8 and 298.5 eV, respectively. The oxidized clusters exhibit an
phous or oxidized AMs, the intercalated structure maintains the amorphous structure with covalent bonds to hydrocarbons, leading to
consistent bilayer form of the C6M2C6 structure. In Supplementary a shift in the K-L3,2 edges to 296.3 and 298.9 eV. In the case of K2CO3,
Fig. 13b, an ADF image of Rb-intercalated BLG with surface deposition the K-L3,2 peaks shift to higher energies at 297.1 and 298.5 eV. For Rb,
is presented, revealing the crystalline structure of Rb in BLG. However, the Rb-L3,2 peaks are less sensitive to the oxidation state compared to
due to the presence of a surface amorphous layer, the imaging quality K. The Rb-L3 peak for C6Rb2C6 is observed at 1812.4 eV, exhibiting a
is somewhat diminished (blurred FFT spots of Rb in BLG). To address blue shift of 1 eV for oxidized Rb absorbed on BLG surface, and another
this, we subjected the sample to an e-beam shower for 5 min over a 1 eV shift in the case of Rb2CO3. As for Cs, the Cs-M5,4 edges of C6Cs2C6
large area (~ 100 µm2). This treatment effectively removed surface- are observed at 735.6 and 749.5 eV. The Cs-M5,4 peaks for oxidized Cs

Nature Communications | (2024)15:425 4

Article https://doi.org/10.1038/s41467-023-44602-3

on BLG exhibit a shift to 736.1 and 750.1 eV and to 744.6 and 759.5 eV in To further study the charge transfer between the bilayer graphene
the case of Cs2CO3. These results indicate that BLG can effectively and intercalated AMs, we performed Raman spectroscopy to the AM
protect intercalated AMs from forming covalent bonds with oxygen, in intercalated BLG. The Raman spectra of K-, Rb-, and Cs-intercalated
contrast with AMs on the BLG surface. In the bilayer AM structure, each BLG are acquired within a sealed glass tube immediately following the
AM atom forms metallic bonds with its neighboring atoms, which intercalation process, ensuring that the samples remain unexposed to
involves the sharing and delocalization of valence electrons among the air. This setup is depicted in Fig. 3a. The exclusion of exposure to air is
metal atoms in the bilayer crystal lattice. The AMs form positively crucial for preserving the integrity of the intercalated samples during
charged metal cations and are held together by the delocalized elec- the Raman measurements. Notably, the Raman G peak of BLG
trons. The presence of the negatively charged electron cloud and its experiences a shift from 1580 cm−1 to 1560 cm−1 following K, Rb, and Cs
mobility facilitates efficient charge transfer to the outer graphene intercalation as shown in Fig. 3c. The observed shift indicates a
layers. To quantitatively measure the charge carrier transport from the remarkably high level of doping, reaching nearly 0.01 ~ 0.03 e−/C atom
bilayer AMs to BLG, we utilized an advanced EELS system with a which is equivalent to 1013 ~ 1014 e−/cm2 for graphene layers30–32. This
monochromatic electron source offering an energy resolution of value is consistent with the estimation derived from the charge carrier
25 meV. Figures 2g–i present the valence loss spectra of the inter- plasmon in our EELS data. The extent of n-doping is further corrobo-
calated bilayer AMs in BLG. The black curve represents the original rated by our transport measurements. This high-level doping will also
spectra, while the colored curves are the background-subtracted result in the enhance of electrical conductivity. In terms of transport
edges of the C6M2C6. As a reference, a spectrum of AB stacking BLG properties, BLG graphene with prepatterned electrodes was fabricated
without intercalation is included as a grey dashed line in each figure, on a SiO2/Si substrate (Fig. 3b). The Id-Vg characteristics of BLG, as
showing the interband π plasmon peak at 5.06 eV and a smooth compared to K and Rb intercalated BLG, are illustrated in Fig. 3d. The
absorption background below 4 eV. In the intercalated sample, the heavily n-doped intercalated BLG exhibits a charge neutrality point
interband π plasmon peak is significantly downshifted to 4.76 eV that shifts toward the conduction band, extending beyond the gate
without any interband transition peak26, indicating complete decou- voltage of −200V. The sheet resistance of BLG undergoes significant
pling of the BLG layers. Additionally, a broad peak in the range of changes upon K and Rb intercalation. Specifically, the sheet resistance
2 ~ 2.5 eV is observed for all of the K, Rb, and Cs samples, which can be transitions from 1303.8 Ω/□ to 306.1 Ω/□ and 367.5 Ω/□ after K and
attributed to the intraband plasmon or so-called charge carrier Rb intercalation, respectively. These findings underscore the sub-
plasmon27,28. This observation is consistent with previous studies on stantial impact of intercalation on the electronic characteristics of
Stage 1 of K intercalated graphite29. According to the Drude model, the BLG, as evidenced by both Raman spectroscopy and transport
plasma angular ffi frequency (ωp ) is given by the equation
pffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi measurements.
ωp = ne2 =ε0 m* , where the n is the carrier density, e is the electron
charge, ε0 is the permittivity of free space, and m* is the effective mass Alkali metal intercalation in graphite
of electrons. Using this formula, the amount of charge transfer can be Since we have established that bilayer AM structures can form in BLG,
calculated to be approximately 1 ~ 1.5 × 1014 e−/cm2. an intriguing question arises: can we achieve higher AM storage

Fig. 3 | Electronic characteristics of alkali metal (AM) intercalated BLG. and electrical transport measurements. c Raman spectra of C6K2C6, C6Rb2C6,
a Photograph of sample prepared by vapor-phase intercalation. b Optical image of C6Cs2C6, and BLG. d Id-Vg characteristics of C6K2C6, C6Rb2C6, and BLG measured at
the BLG with prepatterned Au electrodes on a SiO2/Si substrate for intercalation room temperature.

Nature Communications | (2024)15:425 5

Article https://doi.org/10.1038/s41467-023-44602-3

Fig. 4 | Characterization of K and Cs intercalation in graphite. a ADF image of K top view of the AM intercalation in HOPG, highlighting the C6M2C6 domains in pink.
intercalated highly oriented pyrolytic graphite (HOPG) revealing the honeycomb e Schematic illustration of the side view of AM intercalation in HOPG, emphasizing
C6K2C6 structure in three domains with different numbers of graphene layers. The the wrapping of the AM bilayer domain by the surface graphite layers and the
imaged region has an approximate thickness of 5 nm (Supplementary Fig. 15). The single-layer AM intercalated in the deeper layer of graphite. The cross-sectional
inset displays the corresponding FFT image (scale bar = 5 nm−1). b Cross-sectional specimen, fabricated by focused ion beam (FIB), has a specimen thickness in
ADF image of Cs intercalated HOPG. Bilayer Cs is found near the surface layers, y-direction of approximately 50 nm. The darker ADF contrast of the Cs intercalation
while single layer Cs is found in the deeper layers of graphite. c Corresponding EELS domain is attributed to its incomplete intercalation within the graphite layers.
Cs intensity mapping from the same area as in (b). d Schematic illustration of the

capacity by intercalating bilayer AMs into graphite? To investigate this, intercalated HOPG, as shown in Fig. 4b. Additionally, Fig. 4c presents
we prepared exfoliated graphite from highly oriented pyrolytic gra- an EELS mapping overlaid onto the ADF cross-sectional image, high-
phite (HOPG) and performed AM intercalation experiments. Interest- lighting the Cs intensity distribution. The Cs signal was found to be
ingly, we observed that in the thicker graphite films, the majority of the widely distributed within the graphite layers, with a distinct dark
intercalated K and Rb atoms exhibited fast mobility and could not be contrast layer displaying a significant Cs signal. Furthermore, a two-
clearly resolved in STEM observation (see Supplementary Fig. 14 and layer Cs intercalated structure was observed within the dark contrast
Movie 1 and 2). However, we found that stable K hcp bilayers could be region expanding the interlayer spacing between the graphite layers to
observed in thinner graphite films. Figure 4a displays an ADF image of 1.09 nm, consistent with the model obtained from our DFT calcula-
K-intercalated few-layer HOPG, revealing a honeycomb atomic tions (Fig. 1e and Supplementary Fig. 7). Additionally, single-layer Cs
arrangement similar to that observed in BLG. The thickness of the few- intercalated structures were also visualized in the deeper layers of
layer HOPG sample where C6K2C6 structure was observed was HOPG, with the graphite interlayer spacing being expanded to 0.6 nm.
approximately 5 nm (Supplementary Fig. 15). We hypothesize that The darker contrast of Cs in the cross-sectional image can be attrib-
these relatively unstable structures correspond to the intercalation of uted to incomplete intercalation within the graphite layers. In our
single-layer AM situated in the interior of the thick graphite layers. In bilayer graphene intercalation study, it is worth noting that the AM
contrast, the surface layers of graphite demonstrate the ability to bilayer intercalants do not fully occupy the interlayer gap in BLG
accommodate bilayer AM intercalation, which tends to be more stable (Supplementary Fig. 10). This could be due to some deintercalation
under e-beam irradiation. Consequently, we conclude that the hon- during specimen preparation or STEM observation. The void observed
eycomb structure depicted in Fig. 4a represents an overlapped con- in the upper part of Fig. 4b is likely a consequence of surface oxidation
figuration, comprising AM bilayers in the top few layers of graphite and in the Cs-intercalated HOPG during the sample preparation for cross-
single K layers within different graphene sheet beneath. The absence sectional imaging (Supplementary Fig. 16).
of visualizing bilayer AMs in thick graphite may be attributed to the Schematic illustrations of the intercalation structure are provided
fact that these bilayer AMs near the surface layers are out of focus in in Figs. 4d–e. The pink-colored regions represent domains with the
our STEM imaging when viewed from the top. The presence of a well-ordered C6M2C6 structures, while the remaining regions may also
substrate could potentially limit the expansion of graphene layers on contain intercalated AMs but with greater mobility under the electron
the side attached to the substrate. However, for both BLG and the top beam, potentially indicating single-layer AM intercalation areas situ-
surface of HOPG, the expansion of interlayer spacing remains effective ated in the middle of the thick graphite layers. From the side view, the
and is not significantly influenced by the substrate allowing AM bilayer small AM bilayer domains are surrounded by the graphite layers, with
intercalation. the cross-sectional specimen prepared using focused ion beam (FIB)
To further investigate the bilayer structure of intercalated AM having a thickness of approximately 50 nm. As a result, even for heavy
from a side view, we performed cross-sectional STEM imaging on Cs AMs such as Cs, the small Cs intercalation domains still exhibit darker

Nature Communications | (2024)15:425 6

Article https://doi.org/10.1038/s41467-023-44602-3

ADF contrast compared to the surrounding graphite layers. Theore- Cs intercalation

tical calculations have estimated that the charge capacity of BLG can The intercalation of Cs into the graphite specimen was carried
exceed the theoretical value by inserting an additional layer of AM out according to the following method. At first CsC8 was prepared
atoms13. Our discovery proves that the AM bilayer can intercalate in from small pieces of Grafoil sheet (GrafTech Co., GTA grade).
few-layer graphite and suggests a possible way to improve the battery Metallic Cs (Sigma Aldrich 99.95%) and small pieces of Grafoil
capacity. (typically 15 mm × 5 mm × 0.4 mm) were put in a glass ampoule
In conclusion, our study has demonstrated the formation of under vacuum where the molar ration of C/Cs was adjusted to be
stable bilayer AM structures through intercalation in BLG. While 8. The glass ampoule was then set in an oven held at 230 °C for a
the presence of bilayer AM structures can also be observed in week. The formation of CsC8 was confirmed by the coloration to
multilayer graphite, it is typically limited to small domain sizes or characteristic brownish golden35.
more easily observed in few-layer graphene (FLG). Interestingly,
despite utilizing high-resolution imaging techniques, we did not STEM and EELS
obtain a clear atomic image of the single-layer form of AM STEM images were acquired by using JEOL-tripleC#3, an ultrahigh
intercalation in our experiment. This suggests that the single- vacuum JEOL-ARM200F-based microscope equipped with a JEOL
layer AM structure may be highly mobile and less stable under delta corrector and a cold field-emission gun operating at 60 kV.
electron beam irradiation. The interlayer spacing in thick graphite The probe current was approximately 15 pA for intercalated Cs
is less flexible compared to BLG and FLG, leading the AM bilayer and 5 pA for K and Rb, and the convergence semi-angle and the
to predominantly intercalate into the surface layers of graphite. inner acquisition semi-angle were 37 and 76 mrad, respectively. A
This implies that intercalating AM atoms into graphite layers with typical ADF image has a resolution of 1024 × 1024 pixels and was
lower crystallinity or reduced thickness could potentially enhance captured using a 38.5 μs pixel time. EELS core-loss spectra were
the AM storage capacity. Our findings not only contribute insights obtained using a Gatan Rio CMOS camera optimized for low-
into the intercalation of AMs in BLG but also provide valuable voltage operation. EEL spectra were acquired by line scanning
information about the potential behavior of AM ions during with an exposure time of 0.1 s/pixel. EELS low loss spectra were
electrolyte intercalation. performed by using JEOL-tripleC#2, a low-voltage thermal 60 kV
microscope equipped with a double Wien-filter monochromator
Methods and delta correctors for TEM and STEM. A convergence semi-
Material growth and alkali metal intercalation angle and the inner acquisition semi-angle were 43 mrad and 125
Isolated single crystal BLG domains were grown on Cu foil by ambient mrad. The probe current was 8 pA with an energy resolution of
pressure chemical vapor deposition (CVD)15. The Cu foil was heated in 45 meV after inserting a 0.5 μm slit. All STEM images and EELS
a tube furnace up to 1050 °C for 40 min with a constant flows of Ar were collected at room temperature.
(300 sccm) Ar and H2 (10 sccm). The sample was annealed for 90 min,
then methane (80 ppm) mixed with the flows of Ar (300 sccm) and H2 Raman characterization
(15 sccm) was fed into the reaction chamber for 7 min to form BLG. Raman spectra were acquired by using a JASCO NRS-5500 spectro-
After the growth, the Cu foil was moved to the cooling zone under the meter, which was equipped with a 5× objective lens and a DU420_OE
flow of Ar and H2. CCD detector with 1024 channels cooled to −70 °C. The excitation
Another uniform BLG sheet was synthesized by ambient pressure laser had a wavelength of 532 nm and a power of 5.9 mW. The total
CVD using a Cu–Ni alloy thin film deposited on c-plane sapphire25. The exposure time for the measurements was 100 s.
Ni and Cu thin film was deposited by radio frequency (RF) sputtering.
The substrate was set at ~500 °C and 80 °C for Ni and Cu deposition. Electrical transport measurement
The Cu–Ni film was alloyed by heating at 950 °C in a Ar/H2 (Ar: For the electrical characterization of the intercalated BLG, the as-
195 sccm, H2: 5 sccm) flow for 1 h at ambient pressure. Then CH4 grown BLG was initially transferred onto a SiO2/Si substrate. Subse-
(200 ppm) feedstock as supplied for 10 min to 1 h at temperature quently, the BLG was patterned into squares with 1 mm sides for sheet
ramped to 1085 °C. After growth, the sample was immediately moved resistance measurements, and into stripes of 20 μm width to serve as
to the cooling zone to room temperature. channels for the field-effect transistors (FETs). Electrodes were then
The BLG sheet was transferred onto a TEM grid by etching fabricated through the deposition of a 50 nm layer of Au. Following
the Cu foil by diluted HCl or etching a Cu–Ni film with an aqueous this, the samples underwent the intercalation process. To prevent
ammonium persulfate etching solution, followed by a well- exposure to ambient conditions, a custom-made sample handler was
developed clean transfer process33,34. After transference to the employed to transport the intercalated samples into the chamber of a
TEM grid, the BLG was vacuum sealed in a reaction glass ampoule device probe station fabricated by Thermal Block Co. The electrical
for the intercalation of K, Rb, and Cs. The vacuum reaction tube measurements were finally conducted at room temperature and at low
was opened in an Ar atmosphere glove box and the AM@BLG pressure (~10−2 Pa) using a Keysight B1500A semiconductor device
specimen was transferred to the TEM using a JEOL vacuum parameter analyzer
transfer holder without exposure to air.
Density functional theory calculations
K and Rb intercalation The geometries of K, Rb, and Cs intercalation in BLG were optimized
The intercalated specimens were synthesized by the vapor method. using density functional theory (DFT) implemented in the CASTEP
The alkali metals K (Sigma Aldrich 99.95%) and Rb (Sigma Aldrich module of the Materials Studio ver. 7.0 software (Accelrys Co.).
99.95%) were used without further purification. The alkali metals were Structures were generally obtained using ultra-fine quality gradient
put into a glass ampoule with the BLG and PGS graphite sheet (EYGS, approximation (GGA) – PBE functionals36. The calculation parameters
Panasonic Co.), where the molar ratio of K/C and Rb/C was adjusted to for ultra-fine quality are: the convergence threshold for the maximum
be 8 under an Ar atmosphere. The glass ampoules were sealed after change in energy, maximum force, maximum stress and maximum
evacuating and heated to 200 °C for a few days. The reaction end- displacement, which were set to 5.0 × 10−7 eV/atom, 0.01 eV/Å,
points were confirmed by the coloration to characteristic golden col- 0.02 GPa, and 5.0 × 10−4 Å, respectively. An SCF tolerance smaller than
ors of the PGS specimens35. 5.0 × 10−7 eV/atom was regarded as converged. We also used a 390.0 eV

Nature Communications | (2024)15:425 7

Article https://doi.org/10.1038/s41467-023-44602-3

energy cutoff and a 1 × 1 × 1 K-point set for the band structure and DOS 23. Carlesi, S., Franchini, A., Bortolani, V. & Martinelli, S. Cesium under
calculation. pressure: First-principles calculation of the bcc-to-fcc phase tran-
sition. Phys. Rev. B 59, 11716–11719 (1999).
24. Vasu, K. S. et al. Van der Waals pressure and its effect on trapped
Data availability interlayer molecules. Nat. Commun. 7, 12168 (2016).
All data supporting the findings of this article and its Supplementary 25. Solis-Fernandez, P. et al. Isothermal growth and stacking evolution
Information will be made available upon request to the authors. in highly uniform bernal-stacked bilayer graphene. ACS Nano 14,
6834–6844 (2020).
References 26. Lin, Y. C. et al. Coupling and decoupling of bilayer graphene
1. Stark, M. S., Kuntz, K. L., Martens, S. J. & Warren, S. C. Intercalation monitored by electron energy loss spectroscopy. Nano Lett. 21,
of Layered Materials from Bulk to 2D. Adv. Mater. 31 https://doi.org/ 10386–10391 (2021).
10.1002/adma.201808213 (2019). 27. Kramberger, C. et al. Channeling of charge carrier plasmons in
2. Zhou, J. et al. Layered intercalation materials. Adv. Mater. 33, carbon nanotubes. Phys. Rev. B. 85 https://doi.org/10.1103/
e2004557 (2021). PhysRevB.85.085424 (2012).
3. Xu, J. et al. Recent progress in graphite intercalation compounds for 28. Ritsko, J. J., Mele, E. J. & Gates, I. P. Excitations of back-folded
rechargeable metal (Li, Na, K, Al)-ion batteries. Adv. Sci. 4, graphite bands in KC_{8}. Phys. Rev. B 24, 6114–6120 (1981).
1700146 (2017). 29. Ritsko, J. J. Valence- and core-electronic excitations in potassium-
4. Dresselhaus, M. S. & Dresselhaus, G. Intercalation compounds of intercalated graphite. Phys. Rev. B 25, 6452–6459 (1982).
graphite. Adv. Phys. 30, 139–326 (1981). 30. Howard, C. A., Dean, M. P. M. & Withers, F. Phonons in potassium-
5. Fredenhagen, K. & Cadenbach, G. Die Bindung von Kalium durch doped graphene: The effects of electron-phonon interactions,
Kohlenstoff. Z. Anorg. Allg. Chem. 158, 249–263 (1926). dimensionality, and adatom ordering. Phys. Rev. B. 84 https://doi.
6. Hennig, G. R. Optical transmission of graphite compounds. J. org/10.1103/PhysRevB.84.241404 (2011).
Chem. Phys. 43, 1201–1206 (1965). 31. Bianchi, M. et al. Electron-phonon coupling in potassium-doped
7. Hannay, N. B. et al. Superconductivity in graphitic compounds. graphene: Angle-resolved photoemission spectroscopy. Phys. Rev.
Phys. Rev. Lett. 14, 225–226 (1965). B. 81 https://doi.org/10.1103/PhysRevB.81.041403 (2010).
8. Emery, N. et al. Superconductivity of bulk CaC6. Phys. Rev. Lett. 95, 32. Bostwick, A., Ohta, T., Seyller, T., Horn, K. & Rotenberg, E. Quasi-
087003 (2005). particle dynamics in graphene. Nat. Phys. 3, 36–40 (2006).
9. Belash, I. T., Bronnikov, A. D., Zharikov, O. V. & Palnichenko, A. V. 33. Lin, Y.-C. et al. Graphene annealing: how clean can it be? Nano Lett.
Superconductivity of Graphite-Intercalation Compound with 12, 414–419 (2012).
Lithium C2li. Solid State Commun. 69, 921–923 (1989). 34. Lin, Y.-C. et al. Clean transfer of graphene for isolation and sus-
10. Koike, Y., Suematsu, H., Higuchi, K. & Tanuma, S. Superconductivity pension. Acs Nano 5, 2362–2368 (2011).
in Graphite-Potassium Intercalation Compound C8k. Solid State 35. Matsumoto, R., Arakawa, M., Yoshida, H. & Akuzawa, N. Alkali-metal-
Commun. 27, 623–627 (1978). graphite intercalation compounds prepared from flexible graphite
11. Evans, E. L. & Thomas, J. M. Ultramicrostructural Characteristics of sheets exhibiting high air stability and electrical conductivity.
Some Intercalates of Graphite: An Electron-Microscopic Study. Synth. Met 162, 2149–2154 (2012).
J. Solid State Chem. 14, 99–111 (1975). 36. Predew, J. P., Burke, K. & Kieron, E. Generalized gradient approx-
12. Kuhne, M. et al. Reversible superdense ordering of lithium between imation made simple. Phys. Rev. Lett. 77, 3865–3868 (1996).
two graphene sheets. Nature 564, 234–239 (2018).
13. Chepkasov, I. V., Ghorbani-Asl, M., Popov, Z. I., Smet, J. H. & Kra- Acknowledgements
sheninnikov, A. V. Alkali metals inside bi-layer graphene and MoS2: This work was financially supported by the JSPS Grant-in-Aid for Sci-
Insights from first-principles calculations. Nano Energy 75, entific Research on Innovative Areas “Science of 2.5 Dimensional
104927 (2020). Materials: Paradigm Shift of Materials Science Toward Future Social
14. Lin, Y. C. et al. Polymorphic Phases of Metal Chlorides in the Con- Innovation”. Y.C.L. acknowledge to JSPS-KAKENHI (18K14119,
fined 2D Space of Bilayer Graphene. Adv. Mater. 33 https://doi.org/ 22H05478). K.S. acknowledge to JSPS-KAKENHI (16H06333, 21H05235,
10.1002/adma.202105898 (2021). 22F22358), the JST-CREST program (JPMJCR20B1, JMJCR20B5,
15. Lu, C.-C. et al. Twisting Bilayer Graphene Superlattices. Acs Nano 7, JPMJCR1993), ERC “MORE-TEM” (951215), and the JSPS A3 Foresight
2587–2594 (2013). Program. H.A. acknowledge to JSPS-KAKENHI (18H03864, 21H05232,
16. Takesaki, Y. et al. Highly uniform bilayer graphene on epitaxial Cu- 21H05233, 23K18878). P.W.C. acknowledge to the Ministry of Science
Ni(111) Alloy. Chem. Mater. 28, 4583–4592 (2016). and Technology (MOST) Taiwan Grants MOST 109-2124-M-007-002-
17. Olinger, B. & Shaner, J. W. Lithium, compression and high-pressure MY3, MOST 109-2112-M-007-027-MY3, MOST 106-2628-M-007-003-
structure. Science 219, 1071–1072 (1983). MY3, and MOST 109-2124-M-006-001 as well as Academia Sinica (AS)
18. McMahon, M. I. & Nelmes, R. J. High-pressure structures and phase Grant AS-TP−106-A07. We acknowledge Prof. Noboru Akuzawa for
transformations in elemental metals. Chem. Soc. Rev. 35, conducting the Cs and ethylene intercalation experiments.
943–963 (2006).
19. Degtyareva, V. F. Potassium under pressure: Electronic origin of Author contributions
complex structures. Solid State Sci. 36, 62–72 (2014). Y.C.L., R.M., H.A., and K.S. designed the project. P.S.F., M.D.S., H.A., and
20. Takemura, K. & Syassen, K. High-pressure phase-transitions in P.W.C. synthesized BLG. R.M. performed the AM intercalation experi-
potassium and phase-relations among heavy alkali-metals. Phys. ment. Y.C.L. and Q.L. performed the STEM experiments. P.S.F. and H.A.
Rev. B 28, 1193–1196 (1983). fabricated the devices and performed the electrical transport mea-
21. Hall, H. T., Barnett, J. D. & Merrill, L. High pressure polymorphism in surements. Y.C.L. analyzed the data. Y.C.L. and K.S. co-wrote the paper.
cesium. Science 146, 1297 (1964). &. All authors commented on the manuscript.
22. Takemura, K., Minomura, S. & Shimomura, O. X-Ray-diffraction
study of electronic-transitions in cesium under high-pressure. Phys. Competing interests
Rev. Lett. 49, 1772–1775 (1982). The authors declare no competing interests.

Nature Communications | (2024)15:425 8

Article https://doi.org/10.1038/s41467-023-44602-3

Additional information Open Access This article is licensed under a Creative Commons
Supplementary information The online version contains Attribution 4.0 International License, which permits use, sharing,
supplementary material available at adaptation, distribution and reproduction in any medium or format, as
https://doi.org/10.1038/s41467-023-44602-3. long as you give appropriate credit to the original author(s) and the
source, provide a link to the Creative Commons license, and indicate if
Correspondence and requests for materials should be addressed to changes were made. The images or other third party material in this
Yung-Chang Lin or Kazu Suenaga. article are included in the article’s Creative Commons license, unless
indicated otherwise in a credit line to the material. If material is not
Peer review information Nature Communications thanks the anon- included in the article’s Creative Commons license and your intended
ymous reviewers for their contribution to the peer review of this work. A use is not permitted by statutory regulation or exceeds the permitted
peer review file is available. use, you will need to obtain permission directly from the copyright
holder. To view a copy of this license, visit http://creativecommons.org/
Reprints and permissions information is available at licenses/by/4.0/.
http://www.nature.com/reprints
© The Author(s) 2024
Publisher’s note Springer Nature remains neutral with regard to jur-
isdictional claims in published maps and institutional affiliations.

Nature Communications | (2024)15:425 9