DSpace at VNU: Transition Metal (Fe and Cr) Adsorptions on Buckled and Planar Silicene Monolayers: A Density Functional Theory Investigation

DSpace at VNU: Transition Metal (Fe and Cr) Adsorptions on Buckled and Planar Silicene Monolayers: A Density Functional...

The Journal of Physical Chemistry C is published by the American Chemical Society.

Silicene Monolayers: A Density Functional Theory Investigation

Viet Quoc Bui, Tan-Tien Pham, Hoai-Vu Si Nguyen, and Hung Minh Le

J Phys Chem C, Just Accepted Manuscript • DOI: 10.1021/jp407601d • Publication Date (Web): 10 Oct 2013

Downloaded from http://pubs.acs.org on October 11, 2013

Just Accepted

“Just Accepted” manuscripts have been peer-reviewed and accepted for publication They are posted

online prior to technical editing, formatting for publication and author proofing The American Chemical

Society provides “Just Accepted” as a free service to the research community to expedite the

dissemination of scientific material as soon as possible after acceptance “Just Accepted” manuscripts

appear in full in PDF format accompanied by an HTML abstract “Just Accepted” manuscripts have been

fully peer reviewed, but should not be considered the official version of record They are accessible to all

readers and citable by the Digital Object Identifier (DOI®) “Just Accepted” is an optional service offered

to authors Therefore, the “Just Accepted” Web site may not include all articles that will be published

in the journal After a manuscript is technically edited and formatted, it will be removed from the “Just

Accepted” Web site and published as an ASAP article Note that technical editing may introduce minor

changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers

and ethical guidelines that apply to the journal pertain ACS cannot be held responsible for errors

or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Transition Metal (Fe and Cr) Adsorptions on Buckled and Planar Silicene Monolayers: A Density

Functional Theory Investigation

Viet Q Bui, Tan-Tien Pham, Hoai-Vu S Nguyen, Hung M Le *

Faculty of Materials Science, University of Science, Vietnam National University, Ho Chi Minh City, Vietnam

AUTHOR EMAIL ADDRESS hung.m.le@hotmail.com

RECEIVED DATE

TITLE RUNNING HEAD Transition Metal (Fe and Cr) Adsorptions on Buckled and Planar Silicene Monolayers

CORRESPONDING AUTHOR FOOTNOTE

Corresponding author: correspondence should be addressed to Hung M Le at hung.m.le@hotmail.com

ABSTRACT

The adsorption of metals on silicene monolayer may potentially offer advantageous applications in electronic and spintronic devices In this study, by employing first-principles

calculations, we investigate the attachment of two 3d transition metals (Fe and Cr) on buckled

and planar silicene surfaces Besides examining structural stability, we also explore interesting

ferromagnetic as well as half-metallic features of the material All Fe adsorption cases are found

to be more stable (with the lowest binding energy being 3.39 eV) than Cr adsorption cases When

the metal adsorption rate is high, Fe tends to penetrate into both buckled and planar silicene

layers This insertion behavior allows the 3d shells of Fe to enhance bonding interactions with all

3p x , 3p y , and 3p z orbitals of Si, thus produce more stable structures The adsorptions of Cr with

high distribution ratio are found to be more stable than the low-Cr-distribution structures It is

observed that Cr does not penetrate into the silicene layer like Fe Overall, ferromagnetism is

dominant with five nanostructures, while two Cr adsorption cases on planar silicene

preferentially behave as anti-ferromagnets, and one Fe adsorption case is non-magnetic From

our observation, there is an inversed interplay between structural stability and magnetic

moments, i.e FeSix nanostructures (more stable) tend to exhibit lower ferromagnetic moments

The half-metallic characteristic is found in four nanostructures, which can be potentially applied

in spin-electronic devices The gaps derived from spin-down states for those half-metallic

nanostructures vary from 0.28 to 0.57 eV

Keywords: silicene, DFT, metal-silicene interaction, magnetism, half-metallic, metal adsorption

I INTRODUCTION

Advanced two-dimensional (2D) materials have highly attracted attention of the research community for years Silicene,1 one of such interesting materials, is an infinite monolayer of silicon, whose structure is very similar to that of graphene.2 Purely consisting of silicon atoms, silicene can be integrated into electronic components, and is expected to have a great deal of potential applications in electronic transporting devices In the most stable form, each silicon atom in silicene connects to three surrounding others by sp2-sp3-hybridized bonds, which consequently results in a “low-buckled” honeycomb structure.1, 3-5 The electronic structure of silicene has been proved to establish a zero band gap when the bonding (π) and anti-bonding orbitals (π*) are shown to contact at the Dirac point, which consequently results in very high electron mobility.1 Besides the low-buckled structure, first-principles calculations also suggests the existence of planar derivative of silicene, whose structural configuration is even more similar

to the conformation of graphene.6, 7 In this study, we investigate the bonding interactions between two different silicene conformations (both low-buckled (B, more stable) and planar (PL, less stable) forms) and two transition metal atoms (Cr and Fe)

There have been successful efforts in synthesizing silicene on metal and semiconductor

surface for electronic applications Lalmi et al.8 showed an experimental evidence in which silicene had epitaxial development on Ag(111) by condensing a silicon flux on the surface in vacuum condition Nevertheless, those results highly relied on scanning tunneling microscopy (STM) observations, which approximately resulted in a Si-Si distance of 1.9 Å This bond distance was, however, much smaller than the theoretically-predicted value varying in the range

of 2.22 to 2.24 Å.1, 9, 10 By employing tunneling microscopy and angular-resolved photoemission

spectroscopy in conjunction with first-principles simulations, Vogt et al.11 provided an

experimental evidence of epitaxial silicene sheets on Ag(111) The distance between Si-Si was

then determined as 2.22 Å in such a study, which established consistency with the previous

theoretical results.1 In term of metal contact investigations, Feng et al.12 also presented an

experimental investigation showing a procedure for synthesizing silicene on Ag(111)

Since the early initialization of silicene investigations, computational efforts dedicating to study metal-attached silicene have attained remarkable achievements thanks to the rigorous

development of Density Functional Theory (DFT)13, 14 and noticeable efforts in improvement of

computational packages for condensed matter calculations There have been several DFT-based

investigations conducted to study silicene-metal interactions, especially their coordination

chemistry and physical properties In a theoretical work conducted by Sahin and Peeters,15 the

attachments of alkali, alkaline-earth, and 3d transition metal atoms were investigated using DFT

methods, and several possible absorption positions (hexagonal, bridge, valley, and top sites, as

shown in Fig 1) of a metal atom on silicene were suggested It was reported by Dzade et al.16

that many transition metals (such as Ti, Nb, Ta, Cr, Mo, and W) tended to preferably locate on

the H site of silicene when they occupied all honeycomb units on the silicene monolayer (with

empirical formulas of MSi2) They witnessed that the electronic and magnetic properties of

silicene changed significantly due to metal adsorptions Particularly, CrSi2 became a

two-dimensional magnet and exhibited a strong piezomagnetic property with a magnetic moment in

the range of 3.08 and 3.33 µB When inspecting the electronic and magnetic properties as well as

interactions of silicene with H and Br, Zheng and Zhang17 reported that the investigated

structures displayed either ferromagnetic semiconducting or half-metallic behaviors

In this study, we concentrate on two 3d transition metals, Cr and Fe, which are known for their interesting magnetic behaviors Particularly, Fe is known as a ferromagnet, while

Cr exhibits spin-density-wave antiferromagnetism.18 In addition, such transition metals are believed to establish stable coordination bonds with two-dimensional structures, such as graphene.19, 20 Being motivated by the recent experimental results of silicene,8, 11, 12, 21-23 in this study, we attempt to conduct a theoretical investigation of structural stability, electronic structure, and magnetic property of two-dimensional metal-silicene nanostructures (MSix, M =

Fe, Cr) using a DFT-based approach

II METAL-SILICENE ADSORPTION STRUCTURES (MSix)

The distribution rates of Cr/Fe on silicene and the silicene conformation itself (B or PL) have a significant impact on the stability of the investigated structures, which can be evaluated

by estimating strength of coordination bonds (via binding energy) In this investigation, we study different metal absorption ratios on two silicene conformations (B/PL), which include MSi2(B), MSi2(PL), MSi6(B), and MSi6(PL) as clearly shown in Fig 2 Indeed, the H site is most favored when M adsorbs on either buckled or planar silicene,15 which accordingly produces a two-dimensional lattice that has a 2D hexagonal unit cell

We first consider metal adsorptions with high M concentration (MSi2(B) and MSi2(PL))

In those structures, M atoms occupy all available honeycomb units of the surface As a result, there are two Si atoms and one M atom in a two-dimensional unit cell In Fig 2(a) and 2(c) respectively representing FeSi2(B) and CrSi2(B), M atoms absorb all honeycomb rings in the low-buckled silicene sheet The nanostructures of FeSi2(PL) and CrSi2(PL) (Fig 2(b) and 2(d), respectively) has one M atom located at the center of every planar honeycomb silicon ring

In the case of MSi6(PL) and MSi6(B), there are six Si atoms and one M atom in a primitive hexagonal unit cell The metal atom in these structures tends to occupy a center

honeycomb unit and leave six adjacent (surrounding) units empty (unoccupied) In Fig 2(e) and

2(g), Fe and Cr atoms are respectively located on a low-buckled silicene sheet, while in Fig 2(f)

and Fig 2(h), Fe and Cr are located on planar silicene (denoted as FeSi6(PL) and CrSi6(PL),

respectively) There are two types of bonding in those structures: coordination bonds between

M-Si (under the hybridization effect of 3d orbitals of M and 3p orbitals of M-Si), and M-Si-M-Si

interactions

In the hexagonal unit cell of each investigated nanostructure (with lattice parameter a

listed in Table 1), the two-dimensional characteristic orientation is established in the x and y

directions The vacuum assumption is constituted in the z direction by employing large lattice

parameter c (30 Bohr or 15.87 Å) in all cases

III COMPUTATIONAL DETAIL

In this study, we employ the Perdew-Burke-Ernzerhof (PBE)24, 25 exchange-correlation functional within generalized gradient approximations and the ultrasoft pseudopotentials26, 27 for

Cr, Fe, and Si to perform first-principles calculations All calculations are executed using the

Quantum ESPRESSO package.28 In addition, we utilize spin polarization implementation to

inspect the electronic and magnetic properties

The nanostructures are optimized by relaxing atomic positions and unit-cell vectors simultaneously using the Broyden-Fletcher-Goldfarb-Shanno (BFGS)29 algorithm with the

energy and gradient convergence criteria being 10-5 eV and 10-4 eV/Bohr, respectively The

k-point mesh for all hexagonal lattices (with lattice parameter a given in Table I) is selected as (12

× 12 × 1) in all calculations to ensure consistency in total energy calculations, and the kinetic

energy cut-off of 45 Rydberg (612 eV) is chosen for plane-wave expansions

For each optimized structure, we employ the following formula to analyze the binding energy of M atoms attached on silicene:

where Esilicene , E M, and Estructure are the total energies of silicene, M layer, and the investigated silicene adsorption nanostructure given by DFT calculations, respectively

M-IV RESULTS AND DISCUSSION

1 FeSi2(B) and FeSi2(PL) nanostructures

In the FeSi2 (as well as CrSi2) structures, the metal atoms occupy all available honeycomb units on buckled/planar silicene Particularly, in FeSi2(B), Fe atoms have a tendency

to penetrate into the silicene layer, and interact with both upper and lower Si atoms (as shown in Fig 2(a)) Fe, therefore, forms bonding interactions with the surrounding Si atoms and heavily alters the buckled silicene structure by stretching the Si-Si interaction The Si-Si and Fe-Si distances in FeSi2(B) are 2.60 and 2.27 Å, respectively Recall that in an isolated buckled silicene monolayer, the Si-Si bond is only 2.29 Å, which is much shorter than the Si-Si bond in FeSi2(B) The Fe-Si bond is, however, almost equal to the Si-Si bond in isolated buckled silicene according to our DFT calculations In addition, the buckled gap between the upper and lower Si layers is much distorted (estimated as 1.45 Å), while the original buckled gap in silicene is only 0.45 Å Hence, we believe that such Fe penetration with a high distribution ratio would cause a significant change in structural configuration to the buckled silicene structure Equation (1) is then employed to derive the binding energy, and FeSi2(B) is found to be highly stable with a binding energy of 3.67 eV/cell

Subsequently, spin polarizations are inspected to predict the magnetic property of FeSi2(B) By observing the total density of state (DOS) and the corresponding partial density of

state (PDOS) of Fe 3d and Si 3p subshells (as illustrated in Fig 3(a)), we are able to address a

non-magnetic behavior (polarization is not found in the DOS) In addition, this nanostructure is

believed to be metallic because of electronic state distribution around the Fermi level (positioned

at 0 in the plot) We also conceive that the 3d shells of Fe and 3p shells of Si highly overlap,

which consequently results in a strong bonding interaction Especially, it can be seen that the

electron distribution in 3 2

z

d gives a strong peak near the Fermi level, which indicates a high

electron accepting behavior of Fe The 3p z subshell of Si, unsurprisingly, overlaps much with 3d

orbitals of Fe As mentioned earlier, the penetration of Fe into the silicene layer also allows the

metal 3d orbitals to have more interactions with 3p x and 3p y of Si, and consequently results in

high binding stability Furthermore, such geometric configuration in general allows spin-up and

spin-down states to form exactly similar interactions and therefore align identically (no spin

polarization) This is a unique behavior when Fe is attached on the surface of buckled silicene,

and we do not observe such similarities in other cases

In the FeSi2(PL) structure, Fe is located at the centers of all honeycomb rings in the infinite planar (PL) silicene sheet The occupancy of such metal atoms results in an interwoven

network with an infinite planar structure as illustrated in Fig 2(b) At equilibrium, we have

found in the relaxed FeSi2(PL) structure that Fe atoms again penetrate into the surface of planar

silicene; hence, the resulted structure is perfectly planar and can be considered as the most

compressed structure in this investigation We observe that the Si-Si and Fe-Si bond distances

are identical (2.33 Å) The Si-Si bond in this case is slightly longer than the Si-Si bonds in an

isolated planar silicene sheet (given by our DFT calculations as 2.25 Å) Consequently, the

silicene network is slightly loosened under the effect of Fe penetration by 3.4% For comparison purposes, we summarize bonding distances and binding energies of FeSi2(B) and FeSi2(PL) (as well as the other investigated nanostructures) in Table 1 To evaluate structural stability, we subsequently calculate the binding energy of FeSi2(PL) using equation (1) Indeed, its binding energy is 3.76 eV, which is the highest among eight investigated nanostructures

In spin-polarized DOS analysis, it can be seen that there is a difference in distributions of the spin-up and spin-down states Unlike the previous structure (FeSi2(B)), we observe ferromagnetism and electron conductivity in FeSi2(PL) In the bonding aspect, when bonding

orbitals (3d subshells of Fe and 3p subshells of Si) are analyzed, the overlapping behavior is

similar to the previous case study of FeSi2(B) More specifically, the 3d orbitals of Fe strongly hybridize with not only Si 3p z but 3p x and 3p y orbitals as well However, we do not observe equal distributions in spin-up and spin-down states, which adequately produces a small ferromagnetic moment There are two types of magnetic terms reported in this study, i.e the total (MT) and absolute magnetizations (MA) which are derived in the following equations:

The total magnetization of FeSi2(PL) indicates ferromagnetism with a magnitude of 1.20

µB/cell, while the absolute magnetization is 1.48 µB/cell For convenience, all total and absolute magnetizations of the investigated nanostructure are summarized and reported in Table 2 A calculation of a 2 × 1 supercell is performed to validate ferromagnetism of this structure In fact,

the ferromagnetic configuration is proved to be more energetically stable, which provide a valid

evidence to conclude ferromagnetism

All 3d orbitals tend to align ferromagnetically, especially Fe 3 2

d − and 3dxy are observed to contribute small magnetic moments of 0.08

µB In the other hand, all Si 3p orbitals give negative magnetic terms (anti-ferromagnetic);

especially, the largest anti-ferromagnetic contribution comes from Si 3p z (-0.09 µB) The orbital

contributions to magnetism can be consulted in Table 3

We perform additional calculations for double adsorption cases in order to validate the ferromagnetic properties of FeSi2(B) and FeSi2(PL) In such calculations, Fe atoms are

considered to adsorb on both sides of a silicene monolayer In the case of buckled silicene, we

observe that the condensed Fe adsorption on both sides does not result in a stable structure In the

later case, we are able to obtain an equilibrium structure where Fe atoms adsorb on both sides of

planar silicene (with the empirical formula of Fe2Si2(PL)) This structure is found to exhibit a

total ferromagnetic moment of 4.50 µB/cell, which is almost three times the ferromagnetic

moment given by FeSi2(PL) (1.20 µB/cell)

2 CrSi2(B) and CrSi2(PL) nanostructures

The interacting configurations of CrSi2 structures (in both buckled and planar forms) are very different from that of FeSi2 In fact, it can be seen from Fig 2(c) and 2(d) that in both cases,

the Cr atoms do not penetrate into the silicene monolayer like the previous cases Therefore, we

consider CrSi2 as real adsorption cases This observation is contradicting to a previous study, in

which Cr atoms were shown to penetrate into the silicene layer.16 Unfortunately, no experimental evidences are available to validate the contradicting theoretical results

In CrSi2(B), Cr atoms adsorb on the H site of a buckled silicene honeycomb ring The

Si-Si distance is calculated as 2.38 Å, while two Cr-Si-Si bonds are 2.52 and 3.11 Å The condensed adsorption of Cr on a buckled silicene surface significantly pushes the lower Si atoms away; consequently, the buckled gap between the upper and lower Si layers increases to 0.95 Å Compared to that in the FeSi2(B) case, this buckled gap is smaller but it should be noticed in this case that Cr does not penetrate into silicene The stability of this structure is, however, relatively low (1.77 eV) compared to FeSi2 nanostructures owning to its lower binding energy

Ferromagnetism is found to be more energetically favored than anti-ferromagnetism in the case of CrSi2(B) when we perform total energy calculations for a (2 × 1) supercell The total ferromagnetic moment is estimated as 4.00 µB/cell from spin-polarized calculations In the total DOS of CrSi2(B) and PDOS of Cr 3d and Si 3p orbitals (illustrated in Fig 4(a)), there is a large

distinction in distribution of spin-up and spin-down states of Cr 3d and Si 3p

Besides, we also observe a very importantcharacteristic when the schematic spin-up state indicates conducting, while the spin-down state ends up as a semi-conducting material with a band gap estimated as 0.28 eV In terminology, the nanostructure of this type is often referred to

as “half-metallic” material, which may offer great potential applications in spin-electronic devices The spin polarization effect can be employed to evaluate the conducting behavior, which is empirically defined as following:30

)()(

)()(

F F

F F

E E

E E

ρ ρ

In this equation, ρ↑(E F) and ρ↓(E F) represent the spin-up and spin-down densities of states at the Fermi level, respectively For a half-metallic material (CrSi2(B) in this case), we

expect that the calculated value of P to approach unity Otherwise, a conducting material would

have a lower-than-1 P value As expected, in the FeSi2(B) and FeSi2(PL) cases, the P values are

less than 1 For convenience, such spin polarization values for all investigated nanostructures are

provided in Table 3

The PDOS of Cr 3 2

z

d is highly polarized as we clearly see a dominant peak of the

spin-up states in the valence band (prior to the Fermi level positioned at 0 in the plot) 3d zx and 3d zy

also contribute strong ferromagnetic moments while 3 2 2

y x

d − and 3d xy are observed to contribute

much smaller ferromagnetic terms In the contrary, the 3p z (as well as 3p x and 3p y) subshells of

Si provide insignificant anti-ferromagnetic moments Overall, this nanostructure is reported to be

ferromagnetic with estimated total and absolute magnetizations being 4.00 and 4.60 µB/cell,

respectively

Similarly to CrSi2(B), the relaxed structure of CrSi2(PL) (Fig 2(d)) demonstrates a metal adsorption case The Si-Si bond (2.31 Å) is more slightly compressed than that in CrSi2(B),

while the Cr-Si distance is estimated as 2.56 Å The distance between Cr and planar silicene

layer is observed as 1.11 Å When comparing the interlayer distance (between metal and silicene

layers), CrSi2(PL) is most compressed in the z direction (excluding the penetration cases, i.e

FeSi2(B) and FeSi2(PL)) Compared to CrSi2(B), the nanostructure of CrSi2(PL) is more stable

with a binding energy of 1.94 eV

When decorating highly concentrated Cr on the buckled/planar silicene surface, we conceive less stable nanostructures compared to FeSi2 Deliberately, we believe that this fact is

well associated with the bonding behavior While we suggest the involvement of Si 3p x and 3p y

in forming coordination bonds in FeSi2, a similar behavior is less likely to be observed in CrSi2

as we conceive less overlapping in the PDOS distribution (Fig 4) It should be noticed that the ability to accept more electrons (electronegativity) of Fe is higher than that of Cr.31

From DOS plots, CrSi2(PL) is also found to behave as a metallic material, which exhibits electron conductivity at the Fermi level Unlike CrSi2(B), total energy calculations of a (2 × 1) supercell indicate that CrSi2(PL) favors anti-ferromagnetism with a magnitude of 4.15 µB/cell (absolute magnetization) According to the PDOS distribution in Fig 4(b), the electronic states

of two Cr atoms are opposing to one another All 3d subshells of each Cr atom are polarized

indistinctively (shown in Table 3), which all contribute to anti-ferromagnetic terms

Overall, we observe that CrSi2(B) and CrSi2(PL) are close in bonding stability, but have different magnetic behaviors (ferromagnetic versus anti-ferromagnetic) Even though planar silicene is less stable than low-buckled silicene, the high-concentrated Cr attachment on the planar structure helps to stabilize planar silicene and results in a higher binding energy

We also consider introducing Cr atoms to both sides of silicene to produce Cr2Si2(B) and

Cr2Si2(PL) In both cases, the resulted structures favor anti-ferromagnetism As illustrated in Fig

5, the PDOS distributions of two Cr atoms in these two structures are opposing to each other and thereby produce anti-ferromagnetism The absolute magnetizations of Cr2Si2(B) and Cr2Si2(PL) are reported as 7.00 and 7.39 µB/cell, respectively

3 FeSi6(B) and FeSi6(PL) nanostructures

In FeSi6(B) (Fig 2(e)), we no longer observe a penetration behavior in the optimized structure like in the FeSi2 case In fact, Fe actually adsorbs on the buckled silicene surface and

coordinatively bonds to six Si atoms As mentioned earlier, in FeSi6(B) (and other MSi6

nanostructures), the metal atom tends to occupy one center honeycomb unit while leaving six

surrounding units empty (unoccupied) In this buckled structure, there are two different Fe-Si

bonds with the distances reported to be 2.38 and 2.64 Å The original Si-Si bond in buckled

silicene is 2.28 Å; however, under the effect of coordination bonds with Fe, the Si-Si bond

distance is 2.29 Å, which indicate a very slight stretching Unlike the previous FeSi2(PL) case,

the buckled gap between the upper and lower Si layers is only 0.66 Å In FeSi2(PL), due to the

lack of electron donation from Si (because of high distribution rate of Fe), the metal atom (Fe)

has a tendency to approach closer and penetrate into the silicene layer to get more electron

density from Si 3p orbitals As a result, both upper and lower Si atoms are stretched due to the

insertion of Fe The metal distribution rate in MSi6 is, however, much lower than in the case of

MSi2, and we believe that the Si 3p z orbitals sufficiently provide electron donation toward the 3d

shell in metal atoms Therefore, in MSi6, the metal atom is less likely to form bonding

hybridization with 3p x and 3p y orbitals Using equation (1), we again evaluate the stability based

on binding energy analysis With a binding energy of 3.49 eV, the nanostructure of interest is

believed to be highly stable

The FeSi6(PL) nanostructure (Fig 2(f)) has been optimized using a similar calculation method, and the Fe atoms do not penetrate into planar silicene as in the case of FeSi2(PL)

Instead, Fe forms a separated layer, and the interlayer distance between Fe and silicene is 1.10 Å

The Fe-Si distance is found to be 2.49 Å, while the nearest-neighbor Si-Si distance is 2.24 Å

According to equation (1), the FeSi6(PL) nanostructure is proved to be stable with a reported

binding energy of 3.39 eV Compared to the previous structure (FeSi6(PL)), we obtain a slightly

lower binding energy Considerably, this energy is relatively high when it is compared to the binding energies of Cr adsorption cases

Interestingly, the overlapping of Fe 3d and Si 3p in FeSi6(B) and FeSi6(PL) are very similar as illustrated in Fig 6 In both cases, we conceive a good overlap between 3d zx (3d zy) of

Fe and 3p z of Si By examining PDOS distributions at the Fermi level, FeSi6(B) and FeSi6(PL) are both found to be half-metallic because there are gaps in the spin-up states (0.51 and 0.49 eV, respectively) The spin polarization effects are then calculated using equation (4), and we obtain

P to be almost unity as expected for half-metallic materials

The total magnetizations of FeSi6(B) and FeSi6(PL) are calculated as 2.10 and 2.05

µB/cell, respectively There are, however, some particular distinctions in the 3d orbital

contributions to total magnetization More specifically, the 3 2

z

d contribution (0.78 µB) is dominant in FeSi6(PL), while in FeSi6(B), the polarizations of 3d zx and 3d zy give the highest

contributions The 3 2 2

y x

d − and 3d xy orbitals in both cases contribute minor ferromagnetic

moments to the total Again, our theoretical evidences show that buckled/planar silicene plays the role as an anti-ferromagnet (with a negative magnetic moment causing opposing effects to the ferromagnetic moment caused by Fe)

At this point, it is seen that all FeSix nanostructures are highly stable and exhibit small or intermediate ferromagnetic moments Especially, in the FeSi2(PL) case, we even observe a non-magnetic case Because of having higher electronegativity, Fe tends to approach closer to Si (considering Fe-Si bonds and metal-silicene interlayer distances) in order to receive more electron donation not only from 3p z, but also from 3p x and 3p y orbitals of Si

FeSi6 nanostructures exhibit higher magnetic moments than FeSi2 even though the concentration of Fe in FeSi6 is lower Hence, there is an inversed interplay between bonding

stability (bond strength) and magnetism In FeSi6, we observe smaller 3d-3p orbital overlap

(because Si 3p x and 3p y do not seem to involve in the hybridization) but higher spin polarization

(causing higher ferromagnetism)

When Fe atoms are distributed on both sides of silicene (Fe2Si6(B) and Fe2Si6(PL)), we have found that Fe atoms can stably bind to both sides of a silicene monolayer The equilibrium

Fe2Si6(B) and Fe2Si6(PL) lattices exhibit total ferromagnetic moments of 4.26 and 4.27 µB/cell,

respectively, which strongly imply ferromagnetism of Fe in the nanostructures

4 CrSi6(B) and CrSi6(PL) nanostructures

The CrSi6(B) and CrSi6(PL) nanostructures are considered less stable since their binding energies are relatively lower than the previous FeSix cases Both nanostructures are reported to

have somewhat similar binding energies (above 2.6 eV) Compared to CrSi2, both CrSi6

nanostructures are more stable Observationally, two Cr-Si bonds in CrSi6(B) are 2.50 and 2.74

Å, and the Cr-Si bond in CrSi6(PL) is 2.55 and 2.82. The Si-Si bond in CrSi6(B) is 2.32 Å, longer

than the Si-Si bond in CrSi6(PL) (2.27-2.29 Å) We also find in CrSi6(B) a short buckled gap

between Si atoms (0.62 Å) As shown in Table I, the buckled gap in FeSi6(B) (0.66 Å) is higher

than in CrSi6(B) We suggest there is a clear correlation between binding strength and buckled

gap distance Comparing CrSi6(B) to FeSi6(B), the compound with Fe adsorption is much more

stable (with 29% difference between two binding energies) Similarly, when comparing CrSi2(B)

to FeSi2(B), we observe that the Fe compound has higher binding stability while its buckled gap

is shorter We are convinced at this point that the strong binding behavior of metal-Si would then weaken the Si-Si bond and thereby extend the silicene buckled gap

Interestingly, the adsorption of Cr on planar silicene with low concentration pushes two

Si atoms out of the plane (Fig 2(h)), thus produces another buckled structure (for consistency in naming, this structure is still referred to as CrSi6(PL)) Two different Cr-Si distances are reported

as 2.55 and 2.82 Å Because of the distortion of the planar structure, there is a silicene buckled gap in this case, which is 0.21 Å

By performing DFT calculations for (2 × 1) supercells and comparing total energies,

CrSi6(B) is found to favor ferromagnetic The ferromagnetic spin polarizations of 3 2

z

d in

CrSi6(B) is 0.88 µB, which is equal to 3d zx and 3d zy polarizations, while 3 2 2

y x

d − and 3d xy

magnetic contributions, as shown in Table 3, are somewhat indistinctive (0.67 µB) The PDOS distributions of Cr 3d and Si 3p subshells in CrSi6(B) are illustrated in Fig 7(a) Buckled silicene has insignificant effects on the total magnetic moments by contributing small negative amounts Overall, the total magnetization of CrSi6(B) is found to be 4.00 µB/cell Again, it is interestingly observed that the CrSi6(B) nanostructure has a half-metallic characteristic when we conceive discontinuity in the spin-up electron density (with a gap of 0.57 eV estimated from the PDOS plot) CrSi6(PL), in the other hand, tends to be an anti-ferromagnet (with 0.03 eV lower in total energy than the ferromagnetic configuration) As shown in Fig 7(b), the electronic states of two

Cr atoms are found to be opposing like the earlier case (CrSi2(PL)) The 3d subshell

contributions of each Cr atom to ferromagnetism are found to be large, and the ferromagnetic contribution from six Si atoms (especially from the 3p z orbitals) in the unit cell is