M Reza Rezapour1. 1. Department of Atomic, Molecular and Nuclear Physics, Faculty of Science, Campus de Fuente Nueva, University of Granada, 18071 Granada, Spain.
Abstract
The development of quantum information and quantum computing technology requires special materials to design and manufacture nanosized spintronic devices. Possessing remarkable structural, electronic, and magnetic characteristics, graphitic carbon nitride (g-C3N4) can be a promising candidate as a building block of futuristic nanoelectronics and spintronic systems. Here, using first-principles calculations, we perform a comprehensive study on the structural stability as well as electronic and magnetic properties of triazine-based g-C3N4 nanoribbons (gt-CNRs). Our calculations show that gt-CNRs with different edge conformation exhibit distinct electronic and magnetic characteristics, which can be tuned by the edge H-passivation rate. By investigating gt-CNRs with various possible edge configurations and H-termination rates, we show that while the ferromagnetic (FM) ordering of gt-CNRs stays preserved for all of the studied configurations, half metallicity can only be achieved in nanoribbons with specific edge structure under full H-passivation rate. For spintronic application purposes, we also study spin-transport properties of half-metal gt-CNRs. By determining the suitable gt-CNR configuration, we show the possibility of developing a perfect gt-CNR-based spin filter with a spin filter efficiency (SFE) of 100%. Considering the above-mentioned notable electronic and magnetic characteristics as well as its high thermal stability, we show that gt-CNR would be a remarkable material to fabricate multifunctional spintronic devices.
The development of quantum information and quantum computing technology requires special materials to design and manufacture nanosized spintronic devices. Possessing remarkable structural, electronic, and magnetic characteristics, graphitic carbon nitride (g-C3N4) can be a promising candidate as a building block of futuristic nanoelectronics and spintronic systems. Here, using first-principles calculations, we perform a comprehensive study on the structural stability as well as electronic and magnetic properties of triazine-based g-C3N4 nanoribbons (gt-CNRs). Our calculations show that gt-CNRs with different edge conformation exhibit distinct electronic and magnetic characteristics, which can be tuned by the edge H-passivation rate. By investigating gt-CNRs with various possible edge configurations and H-termination rates, we show that while the ferromagnetic (FM) ordering of gt-CNRs stays preserved for all of the studied configurations, half metallicity can only be achieved in nanoribbons with specific edge structure under full H-passivation rate. For spintronic application purposes, we also study spin-transport properties of half-metal gt-CNRs. By determining the suitable gt-CNR configuration, we show the possibility of developing a perfect gt-CNR-based spin filter with a spin filter efficiency (SFE) of 100%. Considering the above-mentioned notable electronic and magnetic characteristics as well as its high thermal stability, we show that gt-CNR would be a remarkable material to fabricate multifunctional spintronic devices.
Spintronic
devices aim to simultaneously
utilize the charge and
spin of electrons to deliver, store, and process information.[1,2] Since the performance efficiency of any spintronic device depends
on the spin polarization ratio of the currents provided, finding materials
capable of producing 100% spin-polarized current at the Fermi level
is a necessity. A half metal, i.e., a material in which only one spin
direction possesses metallic characteristics, fully meets this demand.[3,4] In recent years, tremendous efforts have been devoted to exploring
low-dimensional half metals, including organic and inorganic two-dimensional
(2D) sheets,[5−8] as well as one-dimensional (1D) nanowires.[9−12] In particular, it has been shown
that graphene and graphene analogues,[13−15] which have attracted
a great deal of attention in terms of 2D physics and chemistry,[16−22] may exhibit intriguing spin states on their edges.[23−26] Graphene nanoribbons (GNRs) having localized electronic edge states[27−30] also exhibit magnetic features with the finite-size effect.[31] Owing to the 1D ballistic transport characteristic
of GNR, various nanoelectronics and spintronics applications of GNR-based
devices have been investigated.[32−39] Several techniques have been introduced to induce half metallicity
in the electronic structure of graphene or GNR such as the application
of electric field,[40] edge modification,[41] B/N dopants,[42] and
introduction of particular atomic-scale defects.[43]Recently, g-C3N4, a 2D semiconductor
benefitting
from nitrogen doping and structural defect techniques, has attracted
considerable scientific interest due to its appealing electronic and
magnetic structures, excellent chemical and thermal stability, and
environmentally friendly features.[44−46] The uniform distribution
of vacancies in the structure of g-C3N4 also
makes it a highly capable material to trap foreign atoms without forming
clusters. This feature makes g-C3N4 a good candidate
for designing nanosize spintronic devices by doping or adsorbing foreign
atoms, such as B, C, Al, or transition metals (TM).[47−50] It also has been shown that g-C3N4 possesses potential applications in electrocatalysis,
photosynthesis, solar energy conversion, bioimaging application, and
photocatalytic performance.[51−54] Various allotropes of g-C3N4 have been synthesized,[55−60] among which heptazine g-C3N4 (gh-C3N4) and triazine g-C3N4 (gt-C3N4) are generally viewed to be the most energetically
stable and have been broadly investigated.[54,56,59,61−63] Although it has been verified that gh-C3N4 is the most stable modification under ambient conditions,[64] gt-C3N4 also has been
successfully synthesized starting from dicyanamide.[65] Despite its porous framework with well-ordered vacancies,
g-C3N4 is a nonmagnetic (NM) material.[47,66] However, Du et al. showed that the hole injection via replacing
a nitrogen atom with a carbon atom can cause the transition of NM
to ferromagnetic (FM) phase in the magnetic structure of g-C3N4.[67] Several methods have
been proposed to induce ferromagnetism in the electronic structure
of g-C3N4 such as introducing defects,[68] doping external ions,[69] fluorine dangling bonds,[70] and boron
bonds.[71] In particular, for spintronics
applications, inducing half metallicity is essential. It should also
be noted that although plenty of studies have been conducted to investigate
the various characteristics of 2D g-C3N4, despite
its experimental realization,[72−74] there is still no systematic
study on characteristics and possible electronic and spintronic applications
of 1D g-C3N4.Motivated by all of the
above-mentioned facts, in the present work,
we perform a comprehensive study on the structural, electronic, and
magnetic properties of gt-CNRs, the induction of half metallicity
in their magnetic structure, and their spin-transport characteristics.
To this end, first, we study the possible structural configurations
of gt-CNRs in terms of their edge conformations and energetic stability.
Next, the possibility of inducing half metallicity in the electronic
structures of gt-CNRs is investigated in terms of their edge structure
as well as the H-passivation rate. Our calculations indicate that
the electronic and magnetic characteristics of gt-CNRs, hence the
induction of half metallicity in its electronic structure, depend
on both edge hydrogenation and the edge structure of nanoribbons.
In this regard, we show that gt-CNRs with a certain edge configuration
can undergo a magnetic phase transition, from NM to half metal, by
tuning the H-passivation rate. Finally, we study the spin-transport
properties of gt-CNRs and show that a half-metal gt-CNR can be employed
to develop a perfect spin filter device.
Computational Methods
Our first-principles calculations
are performed based on the density
functional theory (DFT). The Vienna ab initio simulation package (VASP)[75] is employed for geometry relaxations, cohesive
energy calculations, and investigation of the electronic and magnetic
structures of the systems. The exchange–correlation effects
are treated within the form of the generalized gradient approximation
(GGA) of Perdew, Burke, and Ernzerhof (PBE).[76] The electron–ion interactions are described by the plane-augmented
wave (PAW) method and the Kohn and Sham orbitals are expanded in a
plane wave basis set.[77] A 500 Ry cutoff
energy for the grid-mesh and a k-point mesh of 1
× 1 × 64 are employed along the x-, y-, and z-directions. All of the structures
are fully relaxed until energies and forces are converged to 10–5 eV and 0.01 eV/Å, respectively. DFT combined
with nonequilibrium Green’s function (NEGF)[78] as implemented in the TranSIESTA code[79] is employed to investigate the spin-transport characteristics
of the proposed spin filter device. The spin-dependent transmission
is given bywhere Tr is the trace, Γ = [∑ – ∑] with ∑ as the self-energy
of the left/right electrode, and =
[ – – ∑ – ∑]−1 is the Green’s
function with the scattering region Hamiltonian H. Within the Tr[], all quantities implicitly depend on the energy
(E), Vb, and the spin
σ. The current is calculated using the Landauer–Büttiker
formalismwhere (, μ) is the Fermi–Dirac function with the associated chemical
potential μ= ± /2,
which is a shifted value relative to the Fermi level of a neutral
system EF.
Results and Discussion
Figure a illustrates
the optimized geometry of a primitive unit cell of a gt-C3N4 monolayer. The calculated lattice parameter a = 4.78 Å is in good agreement with previous studies.[80,81] The band structure of the introduced unit cell is represented in Figure b. It is followed
from the calculated band structure that gt-C3N4 is a direct band gap NM semiconductor with a band gap of Eg = 1.57 eV. The obtained Eg is also in agreement with previously reported values.[82,83] It should be noted that the theoretical Eg, obtained by DFT calculations, heavily depends on the constructed
model as well as the employed functional. It has been shown that the
calculated Eg follows the order of HSE06
> GGA-PW91 > GGA-PBE based on the employed functionals.[83]
Figure 1
(a) Geometric structures of the gt-C3N4 monolayer.
The lattice vector is indicated by a and the unit
cell is presented by dashed lines. C and N atoms are depicted in gray
and blue. (b) Band structure of the gt-C3N4 monolayer.
The Fermi level is shifted to zero and represented by a dashed line.
(a) Geometric structures of the gt-C3N4 monolayer.
The lattice vector is indicated by a and the unit
cell is presented by dashed lines. C and N atoms are depicted in gray
and blue. (b) Band structure of the gt-C3N4 monolayer.
The Fermi level is shifted to zero and represented by a dashed line.To investigate the structural, electronic, and
magnetic properties
of gt-CNRs, we employ nanoribbons with different possible carbon-terminated
edge structures and H-passivation rates. The optimized exemplary unit
cells of the studied configurations are illustrated in the upper panel
of Figure . It can
be deduced from Figure that, unlike structures B1, B2, C1, and C2, the opposite edges of
the A1 and A2 systems possess a different topology. First, we examine
the energetic stability of the introduced structures by calculating
the cohesive energy () per atom for each configuration using the following equationHere, is the total energy of gt-CNR’s
unit
cell; , , and are energy per carbon,
nitrogen, and hydrogen atoms, respectively; and is the number of atoms in the unit cell of gt-CNR. Table summarizes the calculated values. It is
inferred from Table that structures A1 and A2 exhibit slightly lower energetic stability
than structures B1, B2, C1, and C2. It is also deduced from Table that although nanoribbons
with low H-passivation rate exhibit better energetic stability, the
difference between cohesive energies of similar configurations is
negligible.
Figure 2
Upper panel illustrates the geometric structures of the exemplary
studied gt-CNRs. The lower panel represents the corresponding band
structures. The Fermi level is shifted to zero and represented by
the black dashed line. C, gray; N, blue; and H, white.
Table 1
Calculated Values of the Studied gt-CNRs
structure
A1
A2
B1
B2
C1
C2
Ecoh (eV)
0.46
0.45
0.47
0.47
0.49
0.49
Upper panel illustrates the geometric structures of the exemplary
studied gt-CNRs. The lower panel represents the corresponding band
structures. The Fermi level is shifted to zero and represented by
the black dashed line. C, gray; N, blue; and H, white.Since gt-CNRs may host spin-polarized electronic states,
it is
worthy to investigate the electronic and magnetic structures of the
introduced gt-CNRs. To this end, we calculate the spin-polarized band
diagrams of the introduced nanoribbons. The lower panel of Figure represents the corresponding
band structures of the studied gt-CNRs. It can be seen from the plotted
band structures that at a low H-passivation rate, structure A1 exhibits
a nonmagnetic electronic structure with an indirect band gap of 1.55
eV, while other systems possessing a metallic band diagram with spin-polarized
states in the vicinity of the Fermi level. At a high H-passivation
rate, although all systems show spin-polarized band structures, only
structure A2 exhibits a half-metallic feature while other nanoribbons
show metallic characteristics. This feature is maintained regardless
of the width of nanoribbons as discussed in the Supporting Information (SI) and illustrated in Figure S1. For further investigation of the electronic
and magnetic properties of gt-CNR, we calculate and plot spin densities
throughout the suggested structures in their final magnetic ordering
as depicted in Figure . Since structures C1 and C2 exhibit band diagrams similar to those
of structures B1 and B2, we only represent spin density distributions
for structures A1, A2, B1, and B2. As can be deduced from Figure , the spin density
appears to be distributed almost periodically along the nanoribbons’
width. This is different from the spin density distribution patterns
observed in pristine or defected GNRs.[39,84] Our calculations
also show that regardless of different initial spin orientations,
i.e., FM or antiferromagnetic (AFM) ordering between opposite edges
of studied gt-CNRs, the converged magnetic ordering after self-consistent
iterations is always FM. This is an interesting finding since it would
be advantageous from the practical viewpoint to design and provide
a gt-CNR-based spintronic system with a solid magnetic configuration.
It is also worth mentioning that based on our calculations, the favored
FM configuration of gt-CNRs is preserved for various widths of the
nanoribbon.
Figure 3
Spin density distribution maps of A1, A2, B1, and B2 structures.
C, gray; N, cyan; spin up, red; and spin down, blue.
Spin density distribution maps of A1, A2, B1, and B2 structures.
C, gray; N, cyan; spin up, red; and spin down, blue.To provide a more detailed insight into the origin
of the observed
half metallic feature of gt-CNRs, we calculate and compare the orbital
resolved electronic structures of systems A1 and A2. As illustrated
in Figure a, the conduction
band (CB) of structure A1 is mainly formed by px orbitals
of C and N atoms close to the carbon-terminated edge (atoms C1, C2, C3, and N1), while
py and pz orbitals of N atoms in
the vicinity of the nitrogen-terminated edge (atoms N2 and
N3) construct the valence band (VB). For structure A2,
it is px and pz orbitals of C and
N atoms on both edges (atoms C4, C5, N4, and N7 as depicted in Figure b) that engage in the construction of the
CB. The VB of structure A2 is also formed by px and
py orbitals of N atoms in the middle of nanoribbon
(atoms N5 and N6). A comparison of the orbital
resolved band structures of two systems reveals that the hydrogenation
of unsaturated edge N atoms in structure A2 changes the position of
atoms and the type of orbitals involved in the formation of the CB
and VB. The advent of the half-metallic characteristic in the electronic
structure of system A2 can be explained as follows: unsaturated edge
nitrogen atoms in system A1 that possess localized lone pair electrons
form σ bonding by the hydrogenation in system A2 and therefore
inject additional electrons into its electronic structure. These additional
electrons are transferred to the edge C and N atoms, occupying the
energy states near the Fermi level of system A2 and resulting in the
emergence of half metallicity.
Figure 4
Orbital resolved band structures of systems
(a) A1 and (b) A2.
The Fermi level is shifted to zero energy and indicated by the black
dashed line. C, gray; N, blue; and H, white.
Orbital resolved band structures of systems
(a) A1 and (b) A2.
The Fermi level is shifted to zero energy and indicated by the black
dashed line. C, gray; N, blue; and H, white.Given the intrinsic half metallicity of gt-CNR,
we investigate
its spin-transport properties under a finite bias voltage (Vb) to develop a feasible spin filter device. Figure a represents a two-probe
system employed to investigate spin-resolved transport features of
half-metal gt-CNRs. The calculated current–voltage (I–V) profiles are illustrated in Figure b. It is worth noting
that the Vb range can be chosen with respect
to the band gap discrepancies of spin components to obtain the desired
proportion of spin filtering. Here, we choose the Vb range of 0–1 V to achieve the desired spin filtering
effect. The selected bias window is also a proper choice from the
application viewpoint since it is not large enough to modify the geometrical
structure of the device. It is followed from Figure b that in the range of applied Vb, the spin-up component shows a nonzero current, while
the transport channel for carriers of down spin is blocked. To provide
more insight into the charge transport characteristics of the proposed
spin filter system, the transmission profiles of the introduced two-probe
system are presented in SI, Figure S2 for Vb = 0 and 0.5 V. Figure b also shows that while the spin-down current
remains negligible with increasing Vb,
the spin-up current increases and reaches its maximum value of 11.1
μA at Vb = 0.8 V. However, further
increase in Vb suppresses the spin-up
current and drops to almost 10 μA at Vb = 1.0 V. This feature may be due to the decrease in the slope
of the spin-up band dispersion as it can be seen in Figure b. This indicates that in addition
to the spin filter feature, the introduced structure also shows spin
negative differential resistance (NDR).[85] To evaluate the spin filtering capability of the proposed spin filter
device, we calculate the SFE of the proposed system. The SFE at a
given bias voltage is defined aswhere ↑ and ↓ are the spin-up and spin-down currents,
respectively. At Vb = 0 V, the SFE can
be obtained by replacing
the current values with the corresponding transmission coefficients. Figure c represents the
calculated SFE.
Figure 5
(a) Schematic illustrations of the proposed two-probe
transport
system. The electrodes and the scattering region are indicated by
red and green dashed lines, respectively. C, gray; N, blue; and H,
white. (b) Spin-resolved current–voltage and (c) spin filter
efficiency profiles of the transport systems.
(a) Schematic illustrations of the proposed two-probe
transport
system. The electrodes and the scattering region are indicated by
red and green dashed lines, respectively. C, gray; N, blue; and H,
white. (b) Spin-resolved current–voltage and (c) spin filter
efficiency profiles of the transport systems.It shows that the device provides a fully spin-filtered
current
for Vb values up to 0.4 V. By increasing Vb, SFE decreases and eventually drops to almost
97.4% at Vb = 1 V, which is still a significant
efficiency. It is worth noting that the SFE of the device is different
at higher Vb values where the bias window
may exceed the band gaps of spin-up and spin-down states and consequently
the current might be nonzero for both spin components. This feature
provides an opportunity to use the proposed system as a nanosized
on/off spintronic switch in which the spin filter characteristic of
the device can be adjusted by the applied Vb.
Conclusions
In summary, we have performed a systematic
study on the structure,
electronic, and magnetic characteristics of triazine-based graphitic
carbon nitride nanoribbons using the first-principles calculations.
To this end, we first examine the energetic stability of gt-CNRs with
possible edge structures and H-passivation rates. Our calculations
show that edge configuration and hydrogenation rate affect, albeit
not significantly, the stability of gt-CNRs. Next, by calculating
spin-polarized band diagrams, we show that electronic and magnetic
characteristics of the studied nanoribbons depend on their edge structures
and H-passivation rate in which only gt-CNRs with a certain edge configuration
and fully hydrogenated edges exhibit the half-metallic feature. Plotting
calculated spin density distribution over various gt-CNRs reveals
that FM ordering is a robust feature of gt-CNRs. To investigate the
origin of the observed half metallicity, we calculate and compare
the orbital resolved band structures of NM and half-metal gt-CNRs.
It is shown that the hydrogenation of unsaturated edge N atoms injects
additional electrons into the electronic structure of gt-CNR and hence
provides an energy level under the CB of gt-CNR in the spin-up channel.
To investigate the possible spintronic applications of gt-CNRs, we
study the spin-transport properties of a two-probe system composed
of a half-metal gt-CNR. The performed calculations show that the proposed
spin-transport system possesses an SFE of 100% for a practically feasible Vb range. It is also shown that aside from its
spin filtering feature, the introduced two-probe system exhibits NDR
as well. Having all the mentioned practical and robust characteristics,
the gt-CNR-based spin filter system is a promising candidate to achieve
a feasible multifunctional atomically thin spintronic device.
Authors: Xinchen Wang; Kazuhiko Maeda; Arne Thomas; Kazuhiro Takanabe; Gang Xin; Johan M Carlsson; Kazunari Domen; Markus Antonietti Journal: Nat Mater Date: 2008-11-09 Impact factor: 43.841
Authors: Michael Ashton; Dorde Gluhovic; Susan B Sinnott; Jing Guo; Derek A Stewart; Richard G Hennig Journal: Nano Lett Date: 2017-08-04 Impact factor: 11.189
Authors: Stanislav M Avdoshenko; Daijiro Nozaki; Claudia Gomes da Rocha; Jhon W González; Myeong H Lee; Rafael Gutierrez; Gianaurelio Cuniberti Journal: Nano Lett Date: 2013-04-17 Impact factor: 11.189