www.nature.

com/npj2dmaterials

ARTICLE OPEN

An atom-to-circuit modeling approach to all-2D

metal–insulator–semiconductor field-effect transistors
Biswapriyo Das1 and Santanu Mahapatra1

Vertical stacking of heterogeneous two-dimensional (2D) materials has received considerable attention for nanoelectronic
applications. In the semiconductor industry, however, the process of integration for any new material is expensive and complex.
Thus, first principles-based models that enable systematic performance evaluation of emerging 2D materials at device and circuit
level are in great demand. Here, we propose an ‘atom-to-circuit’ modeling framework for all-2D MISFET
(metal–insulator–semiconductor field-effect transistor), which has recently been conceived by vertically stacking semiconducting
transition metal dichalcogenide (e.g., MoS2), insulating hexagonal boron nitride and semi-metallic graphene. In a multi-scale
modeling approach, we start with the development of a first principles-based atomistic model to study fundamental electronic
properties and charge transfer at the atomic level. The energy band-structure obtained is then used to develop a physics-based
compact device model to assess transistor characteristics. Finally, the models are implemented in a circuit simulator to facilitate
design and simulation of integrated circuits. Since the proposed modeling framework translates atomic level phenomena (e.g.,
band-gap opening in graphene or introduction of semiconductor doping) to a circuit performance metric (e.g., frequency of a ring
oscillator), it may provide solutions for the application and optimization of new materials.
npj 2D Materials and Applications (2018)2:28 ; doi:10.1038/s41699-018-0073-3

INTRODUCTION demonstrated the operation of a MIS diode, where carrier

Functionality of an electronic device originates from the interfacial tunneling is the principle transport process. In a similar device
properties of its constituent materials. Advancement of nanofab- Vu et al.17 have obtained high photocurrent/dark-current ratio
rication technology has opened up the possibility of realizing >105 and ultrahigh photodetectivity of 2.6 × 1013 Jones. Using first
interfaces at their ‘ultimate-limit’ by vertical stacking1–3 or parallel principles-based analysis, Zan et al.18 have demonstrated the
stitching4 of 2D materials. Since these new materials inherit enhancement of interlayer coupling and linear charge transfer
diverse electronic and opto-electronic properties, novel device between graphene and MoS2 layers upon application of homo-
functionalities could be engineered from such atomically thin geneous electric field, and conversely the weakening of interlayer
interfaces.5 In such vertically stacked van der Waal’s heterostruc- coupling under applied biaxial strain.
tures (vdWH),6 the individual layers are ‘glued’ together by weak It should, however, be noted that introduction of any new
material in the process integration phase of technology develop-
van der Waal’s (vdW) forces of interaction,7 whereas the in-plane
ment in semiconductor industry is an expensive and time-
atoms are strongly bound by covalent or ionic bonds. The hetero-
consuming affair. It is also a difficult task to select appropriate
interfaces thus produced, are atomically sharp and self-passivated,
2D materials from the plethora19 without assessing their
i.e., free of dangling bonds and trapped charges. These subtle performance at circuit level. Thus, a modeling framework, that
attributes have encouraged the application of vdWHs as a enables systematic performance evaluation of new materials at
platform for constructing sophisticated nano-devices such as device and circuit levels before entering into capital-intensive
field-effect transistors (FETs),8–10 tunnel devices,11,12 photo-detec- manufacturing phase, is in great demand. Such models must be
tor,13 light-emitting diode,14,15 solar cell,16 flexible electronics10 first principles based so that the assessment could be conducted
etc. even before the wafer is available. Despite significant efforts on
Among all types of FET devices studied theoretically or synthesis and fabrication, community lacks such modeling frame-
fabricated for experimental and commercial purposes, the MIS work, which can predict integrated circuit performance solely from
(metal–insulator–semiconductor) structure, which substitutes the crystallographic information of the constituent transistor
metal (or highly doped polysilicon) by semi-metallic graphene, materials.
SiO2 (or high-K gate dielectric) by insulating h-BN, and Si by MoS2, In this article, we propose hierarchical bottom-up modeling
could pave the way for realizing thinnest possible FET. Roy et al.9 methodology for all-2D MISFET that bridges between three levels
have demonstrated such experimental device that exhibits n-type of abstraction viz. material, device, and circuit. It is noteworthy
behavior with an ON/OFF current ratio >106 and ~33 cm2/Vs that unlike conventional MOSFETs (Metal Oxide Semiconductor
electron mobility. The work of Lee et al.10 evidences implementa- Field Effect Transistors), in this device, the gate electrostatics is
tion of a flexible and transparent FET claiming field-effect dictated by the effects of dipole–dipole interactions,20–22 prevail-
mobilities up to 45 cm2/Vs. At the same time, Jeong et al.12 have ing at the interfaces of a typical vdWH. Thus, modeling

1
Department of Electronic Systems Engineering, Nano-Scale Device Research Laboratory, Indian Institute of Science (IISc) Bangalore, Bangalore 560012, India
Correspondence: Biswapriyo Das ([email protected])

Received: 10 April 2018 Revised: 5 July 2018 Accepted: 13 July 2018

Published in partnership with FCT NOVA with the support of E-MRS

 An atom-to-circuit modeling approach to all-2Dy
Biswapriyo Das and Santanu Mahapatra
2
methodologies reported earlier for 2D-semiconductor FETs23–25 bandgap might be perturbed due to material defects, externally
are not applicable here. Apart from this, the band-gap opening in applied strain or any uncertainties in process variations, we
graphene due to vdW interactions poses additional complexity to emulate such bandgap variation by changing the graphene–hBN
the device physics. To capture all these intricate atomic level interlayer distance d1 (3.2 ± 0.2 Å) of the graphene-hBN-MoS2
phenomena at circuit level, we adopt several sophisticated system.
techniques, e.g., density functional theory (DFT)-based atomistic Figure 2(b) illustrates the band structures of the vdWH with d1
model, tight-binding Hamiltonian for graphene encompassing the varying from 3.0 to 3.4 Å with an interval of 0.2 Å. The Fermi level (EF)
effect of sublattice symmetry breaking, Fermi-Dirac (FD) distribu- for the band structure is referenced to 0 eV. As depicted in Fig. 2(b),
tion of mobile charge carriers, drift-diffusion (DD) formalism26 with graphene has a bandgap of 0.06 eV at equilibrium separation and
bias-dependent diffusivity, and piecewise charge linearization the bandgap almost doubles at 0.2 Å decrease of d1. Besides, the
(PWCL) technique.27 Finally, we arrive at the closed-form expres- energy dispersion near Dirac point of graphene resembles more of a
sions for drain current and terminal charges for the all-2D MISFET, parabolic nature28 rather than linear one and the parabolicity
which are implemented in professional circuit simulator using increases with the reduction of d1. It also reveals that the otherwise
Verilog-AMS interface in order to conduct static and transient pristine MoS2 becomes p-type in nature due to vdW stacking at
simulation of integrated circuits. We induce bandgap variation in equilibrium spacing (d1 = 3.2 Å), which is in agreement with the
graphene by tuning the interlayer distance between hBN and reported hBN-MoS2 heterostructure.43 We further observe that the
graphene28, and observe its effect at transistor and circuit level. p-type nature becomes more evident with decreasing values of d1,
Since doped MoS2 is now commercially available,29 we further whereas it tends to behave like n-type semiconductor while d1
demonstrate a material-device-circuit co-assessment scheme increases from its equilibrium value. However, the bandgap of MoS2
based on the semiconductor doping. The Verilog-AMS module remains unaltered with a value of 1.71 eV. The absolute numeric
developed in this work connects the material modeling tools30–32 values of ΔGC, ΔGV, ΔMC, ΔMV, which represent the shifts in energy of
with industrial electronic design automation tools33,34 and thus graphene-CBM (conduction band minima), graphene-VBM (valence
promises solution to the DTCO (design-technology co-optimiza- band maxima), MoS2-CBM and MoS2-VBM respectively, as measured
tion) challenges for new materials. from EF and the values of effective masses (viz. m
eG , m
hG , m
eM , m
hM ;
where the subscripts denote e: electron, h: hole, G: graphene, M:
1234567890():,;

MoS2) as calculated from band structures, are presented in Table 1

RESULTS (m0 being rest mass of electron).
The proposed ‘atom-to-circuit’ modeling framework has been The charge transfer from C0 to CB atom can be quantified by the
synopsized in Fig. 1. We start with the DFT-based atomistic model Mulliken population analysis as listed in Table 2. It reveals the fact
development of vdWH in order to probe the energy band that the amount of charge transfer (Δq) increases with decreasing
structure and interlayer charge transfer. Dielectric constant values values of d1, indicating wider bandgap opening in graphene. To
and carrier mobilities of the materials have been acquired from further investigate the charge redistribution at the hetero-
reported first principle-based calculations.35,36 The material level interfaces, we calculate the electron density difference (EDD),
attributes thus obtained are then used to develop physics-based Δρ, for three different values of d1, averaged along the z direction
compact device model to understand the properties of MIS as shown in Fig. 2(c). The EDD is computed as
capacitor and FET. Limited by the computational budget, we Δρ ¼ ρGþhBNþMoS2  ρG  ρhBN  ρMoS2 , where ρ represents elec-
consider only two layers of hBN for developing the atomistic tron density. In Fig. 2(c), both charge accumulation (positive
model, although experiments12,17 were conducted using several peaks) and depletion (negative peaks) regions are found at the
layers of it. However, our device model parameters could easily be interfaces. Clearly, the charge redistribution is more pronounced
calibrated to any number of hBN-layer based vdWH. Finally, those at the graphene–hBN interface rather than hBN–hBN or
closed-form expressions of current and charges are implemented hBN–MoS2 interfaces and it increases at the graphene–hBN
in professional circuit simulator to enable design and simulation of interface with decreasing d1, while remains almost unaltered at
a digital logic inverter and a 15-stage ring oscillator. the other two interfaces. When d1 < 3.2 Å, the surface charge
repulsion (Pauli repulsion) dictates the charge redistribution in
graphene–hBN interface, whereas for d1 > 3.2 Å, an accumulation
Energy band structure and interlayer charge transfer
region at the middle of the interlayer spacing indicates strong
The geometrically optimized atomic structure of graphene-hBN- orbital hybridization22 between graphene and hBN. Quite
hBN-MoS2 is shown in Fig. 2(a) (see methods for details). The obviously, there is a concoction of Pauli repulsion and orbital
equilibrium spacings between graphene-hBN (d1), MoS2-hBN (d2) hybridization prevalent at d1 = 3.2 Å. This orbital hybridization
and hBN-hBN (d3) were found to be 3.2, 4.9 and 3.1 Å respectively. originates from the orbital contributions of distinct atoms to the
It is well-known28,37–42 that the energy gap at Dirac point of band structure (detailed in Supplymentary Information). The VBM
graphene is strongly influenced by the substrates and the stacking and CBM of graphene at K-point in Brillouin zone are contributed
pattern of the layers. This is due to the fact that there is always an by the pz orbitals (π-bonding and π*-antibonding states respec-
interplay between the strain energy of graphene, when placed on tively) of C atoms, which are prone to be perturbed by the
a substrate38 and the reduction of symmetry from C6v to C3v point interactions with π-electron clouds of hBN, localized around N
group due to breaking of the chemical equivalency of carbon centers. But this should not be the case for MoS2 since its CBM and
lattice sites.21 For graphene on top of hBN (CB configuration), VBM are mostly composed of dz2 and dx2 y2 orbitals of Mo atoms;
there is an effective charge transfer from C0 (C atom on top of BN whose interactions with s and p orbitals of B and N atoms are very
hexagon center) atom to CB (C atom on top of B atom) atom limited. However, a comparatively small accumulation region at
owing to the difference in electronegativities of B and N atoms of MoS2–hBN interface indicates orbital hybridization between s and
hBN.41 It results into an onsite energy difference of C atoms at p orbitals of hBN and nearer S atoms. Interestingly, there is a
different sublattices. Thus, sublattice symmetry is broken and the dipole formation at MoS2–hBN interface which in turn causes the
energy levels of pz orbitals of carbon atoms shift to open a work-function modification,22 signifying band alignment of MoS2
bandgap, exactly equal to the onsite energy difference at the and hBN. This effectuates a charge transfer between MoS2 and
Dirac point of otherwise gapless semi-metallic graphene. The hBN, thereby dictating the p-type character of MoS2. To further
band-gap opening in graphene (analogous to depletion in poly- quantify the charge redistribution, we have calculated the area
silicon gate of Si MOSFETs) is expected to play a very crucial role in under EDD curve at MoS2–hBN (between hBN and nearest S
dictating the MIS capacitor and transistor characteristics. Since this atoms) and graphene–hBN interfaces as detailed in Table 3. The

npj 2D Materials and Applications (2018) 28 Published in partnership with FCT NOVA with the support of E-MRS
 An atom-to-circuit modeling approach to all-2Dy
Biswapriyo Das and Santanu Mahapatra
3

Fig. 1 Synopsis of ‘atom-to-circuit’ multi-scale modeling methodology. First, the band structure parameters are extracted from first principles-
based atomistic calculations and the charge transfer in atomic level is analyzed. The extracted parameters are then used for developing a
compact device model of the vdWH-based MISFET. Then that model is implemented in commercial circuit simulators using its Verilog–AMS
interface

Fig. 2 a Atomic model for graphene-hBN-hBN-MoS2 vdWH with the equilibrium interlayer spacings indicated. Color codes for the atoms are:-
cyan: molybdenum, yellow: sulfur, pink: boron, blue: nitrogen, gray: carbon. b Energy band structures of the vdWH with d1 varying from 3.0 to
3.4 Å with a stepping of 0.2 Å. Red and blue-dashed bands correspond to graphene and MoS2 respectively. It depicts that with decreasing d1,
band gap of graphene increases and MoS2 becomes more p-type with an unaltered band gap of 1.71 eV. c Electron density difference (EDD)
plot of the vdWH showing greater charge redistribution in graphene–hBN interface than MoS2–hBN interface. This charge redistribution at
graphene–hBN interface increases with decreasing d1. d Energy band structures of the vdWH comprising of n-type and p-type MoS2
monolayers with a doping concentration of 1 × 1019 cm−3 (both n and p), indicating unaltered band gaps of both graphene and MoS2. It
shows that although graphene is kept pristine, effectively it becomes n or p-type in accordance to the type of doping due to interlayer charge
transfer

higher the value, more pronounced is the charge redistribution threshold voltage of MIS capacitor is expected to increase with the
with more chemical interactions occurring at the interface. As bandgap broadening in graphene and the upward shifting of
depicted, the comparatively smaller area under EDD at MoS2–hBN MoS2 band structure. Furthermore, since the charge distribution at
interface remains almost unaltered with variation of d1, while for the interlayer spacing can be perceived as a way of energy storing,
graphene–hBN interface it increases with decreasing d1. The the increment of EDD area at the graphene–hBN interface with

Published in partnership with FCT NOVA with the support of E-MRS npj 2D Materials and Applications (2018) 28
 An atom-to-circuit modeling approach to all-2Dy
Biswapriyo Das and Santanu Mahapatra
4
Table 1. Band-structure parameters

d1 (Å) ΔGC (eV) ΔGV (eV) ΔMC (eV) ΔMV ðeVÞ m
eG =m0 m
hG =m0 m
eM =m0 m
hM =m0

3.0 0.059 0.059 1.058 0.657 0.017 0.017 0.56 0.632

3.2 0.032 0.032 0.923 0.791 0.009 0.009 0.56 0.632
3.4 0.018 0.018 0.835 0.879 0.005 0.005 0.56 0.632
n-3.2 0.008 0.058 0.828 0.885 0.009 0.009 0.56 0.632
p-3.2 0.057 0.008 1.018 0.696 0.009 0.009 0.56 0.632

quantum capacitance of MoS2 (CqM). Considering the semicon-

Table 2. Mulliken population ductor body to be grounded, if we apply a gate voltage VG in
graphene, equal amount of charge will be stored in all three
d1 (Å) CB C0 Δq
capacitors, thereby causing a potential drop of ΨG across CqG, ΨI
3.0 3.99752 3.98684 0.01068 across CI and ΨS across CqM. For convenience, the relevant band
3.2 3.99608 3.99392 0.00216
structure parameters used in our model are depicted in Fig. 3(c).
The calibrated tight-binding Hamiltonian (see methods section)
3.4 3.99796 3.99648 0.00148 and corresponding energy dispersion relations, used for device
model development, are found to be in good agreement with DFT
calculation as shown in Fig. 3(d), where m
eG ¼ ΔGC =vF2 (vF being
Table 3. Area under EDD (Å−2) the reduced Fermi velocity of graphene over hBN) and
m
hG ¼ ΔGV =vF2 . For a given bias condition, the variation of the
d1 (Å) hBN-MoS2 Interface (×10−5) G-hBN Interface (×10−5) inversion charge density in MoS2 layer (Q0I ) along the MISFET
channel as a function of surface potential ΨS is demonstrated in
3.0 0.71 1.77 Fig. 3(e). Similar to conventional Si-MOSFETs, Q0I holds a linear
3.2 0.72 1.19 relationship with ΨS as long as the band gap opening in graphene
3.4 0.72 1.18 is small. However, significant non-linearity creeps in as the
bandgap broadens, which necessitates PWCL technique based
drain current and terminal charge modeling as discussed in
decreasing d1 signifies possible enhancement of the quantum methods section.
capacitance of graphene as will be elucidated in the next section. Figure 4(a) and (b) respectively depicts the plots of CqG and CqM
The electronic band structure of the aforesaid vdWH comprising as a function of VGS (swept from 0 V to 1.5 V) with VDS set to 0 V. It
of n-type and p-type MoS2 monolayer is depicted in Fig. 2(d). The is found that both CqG and CqM operate within the quantum
downward or upward shifting of the bands with respect to the capacitance limit (i.e., q2g2D, where q is fundamental electronic
Fermi level is clearly visible for n-type and p-type doping charge and g2D is 2D density of states). The value of CqG increases
respectively; although the band gaps and effective masses of as we decrease the interlayer separation due to the increased
graphene and MoS2 remain unaltered. The values of effective charge redistribution at graphene–hBN interface as mentioned
masses and absolute positions of conduction and valence band before. However, unlike CqG, CqM doesn’t even tend to saturate at
edges of graphene and MoS2 (n or p-type) in the energy scale are large gate bias of 1.5 V. The total gate capacitance of the device as
provided in Table 1. Figure 2(d) reveals that even though the seen

from the gate terminal,
1 i.e. Cgg, can be calculated as Cgg ¼
graphene layer is kept pristine and only MoS2 layer undergoes 1=CqG þ 1=CI þ 1=CqM using the value of CI, computed to be
electrostatic doping, due to interlayer charge transfer and thereby ~26.5 fF considering the relative permittivity of bilayer hBN, i.e. ε
charge redistribution, graphene also becomes effectively n-type or to be 1.9.35 The variation of Cgg as a function of VGS has been
p-type according to the doping in MoS2. Since this is mediated by plotted in Fig. 4(c) keeping VDS = 0 V. It reveals that saturated Cgg
interlayer charge transfer, graphene should eventually tend to increases with increasing bandgap of graphene and the crossing
remain pristine if we increase the number of hBN layers in between individual graphs is a direct consequence of threshold
between graphene and MoS2. voltage increment of the device. In the overall capacitance-voltage
characteristics, Cgg is mostly dominated by CqG.
Figure 4(d) and (e) respectively provides the transfer and drain
MIS capacitor and transistor characteristics characteristics of the MISFET, where IDC denotes the dc drain
It is worth noting that in previous studies18 the electric field within current flowing in the channel. The transfer characteristics of Fig. 4
the material system was considered to be homogeneous. The (d) clearly points out the threshold voltage increment of the
charge distribution inside a material system under spatially transistor as bandgap opening in graphene becomes more
varying electric field is beyond the scope of pure DFT calcula- pronounced with decreasing d1. However, the subthreshold slope
tions.44 The proposed model is thus useful to understand the C-V remains almost unaltered in all three cases. The subthreshold
characteristics of the vdWH system. The transistor schematic of swings were calculated to be ~60.48 mV/decade for all cases. This
the concerned vdWH-based MISFET is illustrated in Fig. 3(a), where is because in subthreshold regime CqM is found to be much
‘S’ and ‘D’ denotes source and drain (deemed to be ‘ideal’) smaller than CqG and CI, which makes the subthreshold slope
respectively, VGS and VDS refer to gate-to-source and drain-to- factor nearly unity. In the drain characteristics, delineated in Fig. 4
source voltage respectively. The non-ideal effects arising from the (e), IDC attains a maximum value of ~60 μA/μm for d1 = 3.2 Å. Since
charges, unintentionally present in semiconductor substrate have the threshold voltage is maximum for d1 = 3.0 Å, it bears the
been neglected in this work. The coined device symbol is lowest pinch-off voltage (VP) among all three cases and that’s why
portrayed in inset of Fig. 3(a). The electrical equivalent circuit of the corresponding drain current saturates at earliest having
this MIS capacitor is shown in Fig. 3(b) that consists of three smallest saturation drain current value (ID,sat). From similar
capacitors in series connection viz. quantum capacitance of analogy, the transistor with d1 = 3.4 Å has highest Vp and
graphene (CqG), insulator capacitance of bilayer hBN (CI) and therefore it should have maximum ID,Sat. However, this is not the

npj 2D Materials and Applications (2018) 28 Published in partnership with FCT NOVA with the support of E-MRS
 An atom-to-circuit modeling approach to all-2Dy
Biswapriyo Das and Santanu Mahapatra
5

Fig. 3 a Schematic representation of the vdWH-based MISFET. The coined device symbol is shown in the inset. b Electrical equivalent circuit
of the MIS capacitor comprising of graphene quantum capacitance CqG, insulator capacitance CI of bilayer hBN and quantum capacitance of
MoS2 CqM. c Showcase of relevant band structure parameters used in the model development. Here EF is the Fermi level; ECG, EVG, ECM, EVM,
respectively denotes conduction band minima and valence band maxima of graphene and MoS2. d Comparison between analytical dispersion
relations obtained from tight-binding Hamiltonian and DFT data. e Plot of non-linear relation between inversion charge density in MoS2 (Q0I )
and its surface potential ΨS. Non-linearity of the relation increases with decreasing d1

case due to the trade-off between Vp reduction and Cgg simulator34 or SPICE (simulation program for integrated circuit
increment. The higher capacitance value for d1 = 3.2 Å drives the emphasis) using its Verilog–AMS interface in order to conduct
drain current of the transistor with d1 = 3.2 Å to be greater than static and transient simulation of integrated circuits. As shown in
that with d1 = 3.4 Å. On the other hand, upon introduction of n- Fig. 5, SPICE engine assigns terminal voltages (viz. VG, VD and VS) to
type doping in MoS2, the threshold voltage of the device the Verilog-AMS module, which computes terminal currents (viz.
decreases and accordingly IDC attains a much higher value iG, iD and iS) according to Eqn. (F1)–(F5), and then returns them to
(~105 μA/μm) at saturation (see Supplementary Information). the individual terminals to be further processed by the SPICE
Conversely, the threshold voltage of the transistor with p-type engine. A direct implementation of these equations in SPICE
MoS2 as channel material increases and IDC saturates with a value without any numerical pitfall is very tedious and time-consuming
as small as ~17 μA/μm (see Supplementary Information). We also process. For simulation of basic logic circuits, however, a look-up
plot the nature of two important transcapacitances: Cgd (calcu- table approach will suffice and be followed in this work. First, we
lated at VGS = 1.5 V) and Cdg (calculated at VDS = 1.5 V) in Fig. 4(f), have designed a resistive-load inverter (see Fig. 6(a)) and
which follow the similar trends as that of Cgg. For all these simulated its voltage transfer characteristics (VTC) with the input
characteristics, the mobility value is taken to be 400 cm2/Vs as voltage varying from 0 V to 1.5 V, as plotted in Fig. 6(b). It reflects
estimated by the DFT + Bolzmann formalism.36 the effect of bandgap broadening of graphene by indicating the
positive shift of threshold voltage of the inverter, designed with
Digital logic performance the MISFET having smaller d1. Also plotted in Fig. 6(b) are the VTC
The equations for proposed compact device model are shown in of the logic inverters designed with the MISFETs having n and p-
Fig. 5 (where VCB is channel potential/imref; W and L—both taken type MoS2 monolayers as semiconducting channel materials. As
to be 1 μm, specifies the channel width and length respectively; μ depicted in Fig. 6(b), the inverter corresponding to n-type MoS2
and D, respectively denotes low-field mobility and field- channel experiences a smaller threshold voltage, whereas the
dependent diffusivity; QG, QD, QS and iG, iD, iS refers to terminal inverter with p-type MoS2 has a greater threshold voltage. Clearly,
charges and currents at gate, drain and source respectively; and this effect of threshold voltage modification is quite similar to the
finally Q0In and ψSn symbolizes the values of Q0I and ΨS at nth effect of band gap modulation of graphene. Subsequently, we
breakpoint). The surface potential equation ζ = 0 (Eqn. (F1) in Fig. have designed 15-stage ring oscillator circuits with these resistive-
5) is a function of gate bias, imref and material parameters, which load inverters to study the transient response. The output voltage
are obtained from the DFT calculations. We solve this equation waveforms of three ring oscillators (each designed with the
numerically for different bias conditions to calculate the surface MISFETs with different values of d1) are shown in Fig. 6(c). It
potentials and inversion charge densities, and thereby the dc exhibits moderate changes in oscillation frequencies as men-
drain current (IDC) and terminal charges (QG, QD, and QS) become tioned in the figure itself. To be precise, variation of d1 by ± 0.2 Å
explicit polynomials of them. Efficient algorithms45,46 could be results in increment of the oscillating frequency by a factor of
developed in this regard to solve such equations inside the circuit 1.085 and conversely, a decrement of the same by a factor of 0.85.
simulator. We implement this model in professional circuit The ring oscillator featuring d1 = 3.0 Å possesses the lowest

Published in partnership with FCT NOVA with the support of E-MRS npj 2D Materials and Applications (2018) 28
 An atom-to-circuit modeling approach to all-2Dy
Biswapriyo Das and Santanu Mahapatra
6

Fig. 4 Plot of a CqG, b CqM, and c Cgg with respect to gate-to-source voltage VGS (keeping drain-to-source voltage VDS = 0 V) for different values
of d1. In a and b, both CqG and CqM are confined within the quantum capacitance limit. In c, Cgg is mostly dominated by CqG. d and e,
respectively depicts the transfer and drain characteristics of the vdWH-based MISFET. f Plot of the transcapacitances Cgd (with respect to VDS,
keeping VGS = 1.5 V) and Cdg (with respect to VGS, keeping VDS = 1.5 V). For c and f, all the terminal capacitances are calculated as CXY = ∂QX/
∂VY, where X and Y are the terminals of the device

frequency of oscillation among all. The waveforms of the and extensive scope of applicability should encourage our
oscillators featuring MISFETs with n and p-type MoS2 channel modeling framework to stand high.
have also been plotted in Fig. 6(c) and similar trends of increment
and decrement of oscillating frequencies are observed. For n-type
doping, the frequency increases to 1.125f0 and it decreases to METHODS
0.85f0 for p-type doping, with f0 = 5.75 MHz being the frequency Atomistic model development
of the ring oscillator featuring undoped MoS2 channel and d1 = In order to carry out the first principles-based calculations, the DFT code as
3.2 Å. Good convergence of the simulations in all cases advocates implemented in atomistix tool kit (ATK)30 is employed in conjunction with
generalized gradient approximation (GGA) exchange correlation and the
the practicality and applicability of the proposed model for large-
Perdew-Burke-Ernzerhof (PBE) functional.47 Apart from that, we have used
scale circuit simulation in order to assess the device performance the OpenMX (Open Source package for Material eXplorer) norm-
at circuit level at the early stage of technology development. conserving pseudopotentials48,49 as implemented in ATK database with
pseudo-atomic orbitals (PAO) and ‘medium’ basis sets for the constituent
atoms. To be precise, the LCAO (linear combination of atomic orbitals)
DISCUSSION basis sets for ‘C’, ‘B’, ‘N’ and ‘S’ atoms are adopted to be s2p2d1, and s3p2d1
In this article, we propose a first principle-based ‘atom-to-circuit’ for ‘Mo’ atom. The k-points in the Monkhorst-Pack grid50 were set to 9 ×
modeling methodology, which can assess the impact and predict 9 × 1 along with the density mesh cutoff of 90 Hartree. The maximum
iteration steps were set to 200 using Pulay mixer algorithm and the
the performance of a material or material-system at device and Poisson solver we followed was the fast Fourier transform (FFT). To account
circuit level even in the absence of any experimental data. We for the vdW interactions in the heterostructures, we have incorporated
demonstrate two applications of the proposed model: (i) effect of Grimme DFT-D2 semi-empirical correction51 as defined in ATK database in
band-gap fluctuation of graphene on the circuit performance and combination with counterpoise correction52 that deals with the basis set
(ii) material-device-circuit co-assessment scheme considering superposition error (BSSE) of LCAO basis sets. Furthermore, for all the
semiconductor doping as a design parameter. A schematic view structures, we have provided sufficient vacuum in the perpendicular, i.e.,
normal-to-the-plane direction to avoid spurious interactions between
as depicted in Fig. 6(d) captures the basic philosophy of the
periodic images. The geometry optimization of the unit cells of graphene,
proposed modeling scheme, which bridges between first hBN, and semiconducting 2H-MoS2 were performed using the LBFGS
principle-based material modeling tools and industry standard (Limited-memory Broyden-Fletcher-Goldfarb-Shanno) algorithm53 with
circuit simulators. The model equations could further be simplified maximum stress tolerance value of 0.001 eV/Å3 and force tolerance of
to be applicable for any 2D-material-channel MOSFET with 0.01 eV/Å. Keeping in mind the commensurability condition, the interfaces
conventional gate stack. Furthermore, since we use industry- between these constituent materials were formed by 4 × 4 MoS2 supercell
standard DD formalism for SPICE model development, the (lattice parameter = 12.77 Å), 5 × 5 hBN supercell (lattice parameter =
12.63 Å) and 5 × 5 graphene supercell (lattice parameter = 12.36 Å), leaving
proposed model is ‘core’ in nature and standard techniques (i.e., 0.75% and 1.88% mean absolute strain on hBN and graphene respectively,
pre-correction) for including several small geometry effects (such which are reasonably small. In order to model n and p-type doped MoS2
as drain induced barrier lowering, velocity saturation etc.) could monolayers, we have exercised the method of electrostatic doping (using
easily be conjoined with it. Henceforth, this stupendous flexibility n and p-type atomic compensation charges54,55) available in ATK30 instead

npj 2D Materials and Applications (2018) 28 Published in partnership with FCT NOVA with the support of E-MRS
 An atom-to-circuit modeling approach to all-2Dy
Biswapriyo Das and Santanu Mahapatra
7

Fig. 5 Equations (F1)–(F5) are mathematical expressions for the proposed compact device model. The closed form expressions of dc drain
current IDC and terminal charges viz. QG, QD, QS are obtained employing piecewise charge linearization technique. Inset: SPICE engine assigns
terminal voltages (viz. VG, VD, and VS) to the Verilog-AMS module, which computes terminal currents (viz. iG, iD, and iS) according to Eqn. (F1)–
(F5), and then returns them to the individual terminals to be processed by the SPICE engine

of explicitly substituting the Mo atoms by elemental dopant atoms (e.g., Thereafter, we obtain the expressions for density of states (DOS) of
Nb for p-type56 and Au, Re for n-type57). This is because in order to achieve sG gvG ΔGC
conduction and valence bands of graphene and MoS2 as: gGn ¼ g2π hv Þ2
ð
a practical n or p-type doping concentration in MoS2 (typically 1 × 1017 − gsM gvM m
eM
F
sG gvG ΔGV
1 × 1019 cm−3)58 by substitutional doping, a very large supercell is needed, (for graphene CB), gGp ¼ g2π ðhv Þ2
(for graphene VB), gMn ¼ 2πh2
(for
F
which demands very large computation cluster. Such electrostatic doping g g m

MoS2 CB) and gMp ¼ sM2π vM hM
h2
(for MoS2 VB). Here, gsG, gsM are spin
scheme is very effective and advantageous since it does not depend on
exact detail of dopant atoms.54 Here, the doping concentration is set to degeneracies (both taken as 2) and gvG, gvM are valley degeneracies (both
1 × 1019 cm−3 for both n and p-type doping. taken as 2) of graphene and MoS2 respectively.

Development of energy dispersion relations MIS capacitor modeling

We start with the 2-band Hamiltonian of graphene that incorporates the Employing Fermi-Dirac statistics, we first calculate the intrinsic carrier
effect of sublattice symmetry breaking.42,59,60 In nearest-neighbor π– concentrations in graphene (MoS2) as:
electron tight-binding parlance, where sublattice symmetry breaking can   
ΔGCðMCÞ
be parameterized by a mass term, the Hamiltonian describing electronic nGðMÞ ¼ gGnðMnÞ KT ln 1 þ exp  (4a)
KT
properties of graphene near the Fermi level can be approximated as59:

!   
Δ hvF kx  iky ΔGVðMVÞ
H¼
 (1) pGðMÞ ¼ gGpðMpÞ KT ln 1 þ exp  (4b)
vF kx þ iky
h Δ KT
where, k is wave vector relative to Dirac (K or K′) point, ħ is modified where, nG (nM) and pG (pM) stands for electron and hole concentrations in
Planck’s constant and Δ is the onsite energy difference between two graphene and MoS2 respectively, K is the Boltzman constant and T is
sublattices A and B, which in turn equals to m
vF2 (m
m* = m
eG or m
hG ). temperature (taken as 300 K). Now, if we symbolize net electron and hole
ΦA concentrations in graphene and MoS2 (upon application of VG) by nnetG,
This Hamiltonian operates on spinors Ψ ¼ (where ΦA and ΦB are
ΦB pnetG, nnetM and pnetM respectively, then the total charge densities in these
the amplitudes of the wavefunctions ofq two two layers can be written as Q0G ¼ qðpnetG  nnetG Þ for graphene and Q0I ¼
ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi to produce the
sublattices),
qðpnetM  nnetM Þ for MoS2, where
energy eigenvalues of the form: E ¼ ± Δ2 þ ð hvF k Þ2 . For graphene on   
top of hBN, Δ ≠ 0 and the energy dispersions near the Dirac points can be ΔGCðMCÞ qΨGðSÞ
nnetGðnetMÞ ¼ gGnðMnÞ KT ln 1 þ exp  ∓ (5a)
approximated to be parabolic having the form28: KT KT
" #
vF k Þ2
ðh   
E ðk Þ  ± Δ þ (2) ΔGVðMVÞ qΨGðSÞ
2Δ pnetGðnetMÞ ¼ gGpðMpÞ KT ln 1 þ exp  ± (5b)
KT KT
Here, ‘±’ sign corresponds to the conduction band (CB) with Δ = ΔGC and
valence band (VB) with Δ = ΔGV respectively, defining the non-zero Here, we consider the charge distribution inside MoS2 to be an ideal 2D
bandgap introduced at Dirac point as: EgG = ΔGC + ΔGV. Although for sheet having no spatial variation, which removes the burden of solving
±0.2 Å variation of d1, the bandgap opening is symmetric (i.e., ΔGC = ΔGV) Poisson’s equation beforehand. On the other side, ΨG, ΨI and ΨS sum up to
about EF, in general ΔGC may not be equal to ΔGV for further reduction of satisfy the potential balance equation, which is:
d1. On the other hand, the band structure of MoS2 can also be VG ¼ ΨG þ ΨI þ ΨS (6)
approximated to be simple parabolic in nature, written as23,25:
! where ΨI can be calculated as:
h2 k 2
E ðk Þ ¼ ± ΔMCðMVÞ þ (3)
2m
eMðhMÞ ΨI ¼ Q0G =CI ¼ Q0I =CI (7)

for CB (VB) (where ‘+’: CB and ‘–’: VB), featuring a bandgap of EgM = ΔMC + Here, we neglect any leakage current through the hBN layers. Now using
ΔMV. Eqns (5)–(7), the charge balance equation can be solved numerically for ΨS,

Published in partnership with FCT NOVA with the support of E-MRS npj 2D Materials and Applications (2018) 28
 An atom-to-circuit modeling approach to all-2Dy
Biswapriyo Das and Santanu Mahapatra
8

Fig. 6 a Circuit diagram for a resistive load inverter, designed using the graphene-hBN-hBN-MoS2 vdWH based MISFET. RL is the load resistor
and CL is external capacitive load at output. b DC transfer characteristics of the logic inverter, indicating a positive shift of the threshold
voltage of the inverter circuit with decreasing d1. It also depicts the DC characteristics of the inverters designed with MISFETs featuring n and
p-doped MoS2 monolayers. For n-type doping the threshold voltage decreases, whereas the opposite happens for p-type doping. c Output
frequency waveform of a 15-stage ring oscillator, designed with these inverters. It shows the change in output frequency as a result of varying
d1 from its equilibrium value and also as a result of doping in MoS2. Here f0=5.75 MHz. The legend convention is same as that of b. d A
schematic view of the basic philosophy behind the proposed model which reveals that, the model (i.e. AMS module) works as a bridge
between first principles-based material modeling tools and industry standard circuit simulators

which can be written as: MISFET modeling

To formulate the drain current and terminal charges, we adopt the
Q0G þ Q0I ¼0 (8) semiclassical DD formalism,26,61 which has been practiced in industrial
that ultimately takes the form of Eqn. (F1) of Fig. 5 with VCB set to zero. top–down compact modeling methodology over the years. We can take
Once we get the numeric values of ΨS, the other two potentials ΨG and the benefits arising from well-established techniques available to add
ΨI can be easily computed, thereby facilitating the computation of different small geometry effects to the core model as future work. It also
quantum capacitances viz. CqG and CqM using the following expressions: allows us to develop the model without using any unphysical model
8 9 parameters or interpolating function. Since the proposed model involves
< ΔGC 1 þ exp
ΔGC þqΨG  1 = only material parameters (CBMs, VBMs, effective masses, dielectric
2 gsG gvG constant, and mobility), which could be calculated by first principles-
CqG ¼ q ´

KT
 (9a)
2πðhvF Þ2 : þΔ 1 þ exp ΔGV qΨG 1 ;
GV based methods, it can predict the device and circuit characteristics just
KT
from the crystallographic information of the constituent materials.
8
Δ qΨ  1 9 In DD formalism, the dc drain current equation reads61:
gsM gvM < m
eM 1 þ exp MCKT S = " Z Z Q0 #
CqM ¼q 2 ´:
2

 (9b) W ΨSL

1 ;
IL
2πh Δ þqΨ
þm
hM 1 þ exp MVKT S IDC ¼ μ Q0I dΨS þ D dQ0I (10)
L ΨS0 Q0I0
For a similar device with more than two hBN layers, the model equations
will remain all the same, however, the parameters viz. ΔGC, ΔGV, ΔMC, ΔMV, where, ΨS0, ΨSL and Q0I0 , Q0IL are the values of ΨS and Q0I at source (x = 0)
vF, m
eM , m
hM and ε need to be recalculated using DFT. Similarly, for the and drain (x = L) end of the transistor respectively, with x being considered
heterostructures with n or p-type MoS2 monolayers, the whole set of as the direction of transport. Since we explicitly use FD statistics to
equations will essentially remain same, provided the aforesaid parameters describe carrier occupation probability, instead of using typical Einstein
62
are obtained from DFT calculations of that particular heterostructure. This relation, we treat the diffusivity

coefficient D to be bias dependent. We
0 0
is because the effect of doping has been effectively captured here through express D as D ¼ μQI dΨS =dQI VDS ¼0 to ensure zero drain current at VDS
63

the positions of the band extrema in the energy scale and corresponding = 0. Due to complex nature of ζ, an analytical solution of Eq. (9) is not
effective masses and that basically rules out the necessity of introducing a possible. Hence, we apply charge linearization technique27 to obtain a
doping term explicitly in the charge neutrality equation. It is noteworthy closed form expression of the drain current. In case of conventional Si-
that the band structure obtained from DFT calculation is qualitatively MOSFET, for a given bias condition, Q0I changes quasi-linearly with ΨS along
similar to the previously reported work,18 which uses different simulation the channel and hence Q0I is approximated as a linear function of ΨS in Eqn.
toolkit and deals with single hBN layer. In this aspect, the proposed model (10), which results in a quadratic relationship between IDC and ΨS. Such
equations are independent of the exchange-correlation functionals or relationship is very useful to obtain closed form expression of terminal
pseudopotentials; only the relevant model parameters need to be charges under Ward-Dutton (WD)64 charge partitioning scheme. However,
calculated accordingly. as shown in Fig. 3(e), due to the band-gap opening in graphene, Q0I does

npj 2D Materials and Applications (2018) 28 Published in partnership with FCT NOVA with the support of E-MRS
 An atom-to-circuit modeling approach to all-2Dy
Biswapriyo Das and Santanu Mahapatra
9
not always maintain a linear relationship with ΨS in case of vdWH MISFET. 14. Withers, F. et al. Light-emitting diodes by band-structure engineering in van der
Thus, in this work we adopt PWCL technique27 (see Supplementary Waals heterostructures. Nat. Mater. 14, 301–306 (2015).
Information). Using the expression of drain current, the terminal charges 15. Ross, J. S. et al. Electrically tunable excitonic light-emitting diodes based on
could be computed under WD charge partitioning64 scheme as: monolayer WSe2 p-n junctions. Nat. Nanotechnol. 9, 268–272 (2014).
RL RL 16. Furchi, M. M., Pospischil, A., Libisch, F., Burgdörfer, J. & Mueller, T. Photovoltaic
QG ¼ W Q0Ix ð x Þdx, QD ¼ W xL Q0Ix ð x Þdx and QS = −(QG + QD), where effect in an electrically tunable van der Waals heterojunction. Nano Lett. 14,
0 0
4785–4791 (2014).
Q0Ix is the value of Q0I at any position x in the channel. Similar to state-of-
17. Vu, Q. A. et al. Tuning carrier tunneling in van der Waals heterostructures for
the-art surface potential-based Si-MOSFET models,65 Eqns. (F2)–(F5) are
ultrahigh detectivity. Nano Lett. 17, 453–459 (2017).
‘single-piece-continuous’ equations, valid for all regimes of transistor
18. Zan, W., Geng, W., Liu, H. & Yao, X. Electric-field and strain-tunable electronic
operation.
properties of MoS2/h-BN/graphene vertical heterostructures. Phys. Chem. Chem.
Phys. 18, 3159–3164 (2016).
Data availability 19. Mounet, N. et al. Two-dimensional materials from high-throughput computa-
The authors declare that the main data supporting the findings of this tional exfoliation of experimentally known compounds. Nat. Nanotechnol. 13,
study are available within the paper and its Supplementary Information 246–252 (2018).
file. MATLAB and Verilog-AMS codes are freely available at osf.io/me236. 20. Tung, R. T. Formation of an electric dipole at metal-semiconductor interfaces.
Other relevant data are available from the corresponding author upon Phys. Rev. B 64, 205310 (2001).
request. 21. Yakovkin, I. N. Dirac cones in graphene, interlayer interaction in layered materials,
and the band gap in MoS2. Crystals 6, 143 (2016).
22. Gong, C., Colombo, L., Wallace, R. M. & Cho, K. The unusual mechanism of partial
ACKNOWLEDGEMENTS Fermi level pinning at metal-MoS2 interfaces. Nano Lett. 14, 1714–1720 (2014).
23. Cao, W., Kang, J., Liu, W. & Banerjee, K. A compact current-voltage model for 2D
This work was supported by Science and Engineering Research Board (SERB),
semiconductor based field-effect transistors considering interface traps, mobility
Department of Science and Technology (DST), Government of India, under Grant SB/
degradation, and inefficient doping effect. IEEE Trans. Electron Devices 61,
S3/EECE/0209/2015. We would like to thank G.N. Kadloor, DESE-IISc, for his help in
4282–4290 (2014).
preparing Fig. 6(d).
24. Marin, E. G., Bader, S. J. & Jena, D. A new holistic model of 2-D semiconductor
FETs. IEEE Trans. Electron Devices 65, 1239–1245 (2018).
25. Yadav, C., Agarwal, A. & Chauhan, Y. S. Compact modeling of transition metal
AUTHOR CONTRIBUTIONS dichalcogenide based thin body transistors and circuit validation. IEEE Trans.
B.D. performed the DFT calculations, developed the compact device model, Electron Devices 64, 1322–1329 (2017).
conducted SPICE simulations, and analyzed final results. S.M. conceived the problem 26. Zebrev, G. I. & Useinov, R. G. Simple model of current-voltage characteristics of a
statement and overall supervised the work. All authors contributed in the writing. metal–insulator–semiconductor transistor. Fiz. Tekhn. Polupr. (Sov. Phys. Semi-
cond.) 24, 777–781 (1990).
27. Jandhyala, S., Abraham, A., Anghel, C. & Mahapatra, S. Piecewise linearization
ADDITIONAL INFORMATION technique for compact charge modeling of independent DG MOSFET. IEEE Trans.
Supplementary information accompanies the paper on the npj 2D Materials and Electron Devices 59, 1974–1979 (2012).
Applications website (https://doi.org/10.1038/s41699-018-0073-3). 28. Fan, Y., Zhao, M., Wang, Z., Zhang, X. & Zhang, H. Tunable electronic structures of
graphene/boron nitride heterobilayers. Appl. Phys. Lett. 98, 083103 (2011).
Competing interests: The authors declare no competing interests. 29. 2D Semiconductors. URL: http://www.2dsemiconductors.com/n-type-mos2/.
30. QuantumWise Atomistix ToolKit (ATK) with Virtual NanoLab, Synopsys Denmark,
URL: http://quantumwise.com/. Last accessed: 18 Mar 2018.
Publisher's note: Springer Nature remains neutral with regard to jurisdictional claims
31. Vienna Ab initio Simulation Package (VASP), Computational Materials Physics
in published maps and institutional affiliations.
Sensengasse, Austria, URL: https://www.vasp.at/.
32. CAmbridge Serial Total Energy Package (CASTEP), Dassault Systèmes BIOVIA, URL:
http://www.castep.org/.
REFERENCES 33. Eldo, Mentor Graphics. URL: https://www.mentor.com/.
1. Shih, C-J. et al. Tuning on-off current ratio and field-effect mobility in a MoS2- 34. SmartSpice, Analog Circuit Simulator–Users’ Manual, version 4.10.6.R (Silvaco Inc.,
graphene heterostructure via Schottky barrier modulation. ACS Nano 8, Santa Clara, CA, 2014). URL: https://www.silvaco.com/.
5790–5798 (2014). 35. Kumar, P., Chauhan, Y. S., Agarwal, A. & Bhowmick, S. Thickness and stacking
2. Deng, Y. et al. Black Phosphorus-monolayer MoS2 van der Waals heterojunction dependent polarizability and dielectric constant of graphene-hexagonal boron
p–n diode. ACS Nano 8, 8292–8299 (2014). nitride composite stacks. J. Phys. Chem. C. 120, 17620–17626 (2016).
3. Lee, G. -H. et al. Highly stable, dual-gated MoS2 transistors encapsulated by 36. Gunst, T., Markussen, T., Stokbro, K. & Brandbyge, M. First-principles method for
hexagonal boron nitride with gate-controllable contact, resistance, and threshold electron-phonon coupling and electron mobility: applications to two-
voltage. ACS Nano 9, 7019–7026 (2015). dimensional materials. Phys. Rev. B 93, 035414 (2016).
4. Ling, X. et al. Parallel stitching of 2D materials. Adv. Mater. 28, 2322–2329 (2016). 37. Wang, E. et al. Gaps induced by inversion symmetry breaking and second-
5. Pant, A. et al. Fundamentals of lateral and vertical heterojunctions of atomically generation Dirac cones in graphene/hexagonal boron nitride. Nat. Phys. 12,
thin materials. Nanoscale 8, 3870–3887 (2016). 1111–1115 (2016).
6. Geim, A. K. & Grigorieva, I. V. Van der Waals heterostructures. Nature 499, 38. Jung, J., DaSilva, A. M., MacDonald, A. H. & Adam, S. Origin of band gaps in
419–425 (2013). graphene on hexagonal boron nitride. Nat. Commun. 6, 6308 (2015).
7. Ma, Z. et al. Tunable band structures of heterostructured bilayers with 39. Bokdam, M., Amlaki, T., Brocks, G. & Kelly, P. J. Band gaps in incommensurable
transition–metal dichalcogenide and MXene monolayer. J. Phys. Chem. C. 118, graphene on hexagonal boron nitride. Phys. Rev. B 89, 201404(R) (2014).
5593–5599 (2014). 40. Skomski, R., Dowben, P. A., Driver, M. S. & Kelber, J. A. Sublattice-induced symmetry
8. Roy, T. et al. Dual–gated MoS2/WSe2 van der Waals tunnel diodes and transistors. breaking and band-gap formation in graphene. Mater. Horiz. 1, 563–571 (2014).
ACS Nano 9, 2071–2079 (2015). 41. Kan, E. et al. Why the band gap of graphene is tunable on hexagonal boron
9. Roy, T. et al. Field-effect transistors built from all two-dimensional material nitride. J. Phys. Chem. C. 116, 3142–3146 (2012).
components. ACS Nano 8, 6259–6264 (2014). 42. Sławińska, J., Zasada, I. & Klusek, Z. Energy gap tuning in graphene on hexagonal
10. Lee, G. -H. et al. Flexible and transparent MoS2 field–effect transistors on hex- boron nitride bilayer system. Phys. Rev. B 81, 155433 (2010).
agonal boron nitride–graphene heterostructures. ACS Nano 7, 7931–7936 (2013). 43. Komsa, H. -P. & Krasheninnikov, A. V. Electronic structures and optical properties
11. Cheng, R. et al. Multifunctional tunneling devices based on graphene/h-BN/ of realistic transition metal dichalcogenide heterostructures from first principles.
MoSe2 van der Waals heterostructures. Appl. Phys. Lett. 110, 173507 (2017). Phys. Rev. B 88, 085318 (2013).
12. Jeong, H. et al. Metal-insulator-semiconductor diode consisting of two- 44. Kresse, G., Marsman, M., & Furthmüller, J. VASP–The Guide, sec. 6.67.3 Finite Electric
dimensional nanomaterials. Nano Lett. 16, 1858–1862 (2016). Fields (2016). URL: http://cms.mpi.univie.ac.at/vasp/vasp.pdf.
13. Zhang, K. et al. Interlayer transition and infrared photodetection in atomically 45. Jandhyala, S. & Mahapatra, S. An efficient robust algorithm for the surface-
thin type-II MoTe2/MoS2 van der Waals heterostructures. ACS Nano 10, potential calculation of independent DG MOSFET. IEEE Trans. Electron Devices 58,
3852–3858 (2016). 1663–1671 (2011).

Published in partnership with FCT NOVA with the support of E-MRS npj 2D Materials and Applications (2018) 28
 An atom-to-circuit modeling approach to all-2Dy
Biswapriyo Das and Santanu Mahapatra
10
46. Chen, T. L. & Gildenblat, G. Analytical approximation for the MOSFET surface 60. Hunt, B. et al. Massive Dirac fermions and Hofstadter butterfly in a van der Waals
potential. Solid State Electron 45, 335–339 (2001). heterostructure. Science 340, 1427–1430 (2013).
47. Perdew, J. P., Burke, K. & Ernzerhof, M. Generalized gradient approximation made 61. Tsividis, Y. & McAndrew, C. Operation and Modeling of The MOS Transistor, 3rd edn
simple. Phys. Rev. Lett. 77, 3865 (1996). (Oxford University Press, New York, 2011).
48. Ozaki, T. Variationally optimized atomic orbitals for large-scale electronic struc- 62. Ghatak, K. P., Bhattacharya, S. & De, D. Einstein Relation in Compound Semi-
tures. Phys. Rev. B 67, 155108 (2003). conductors and Their Nanostructures. (Springer, Berlin, Germany, 2009). Springer
49. Ozaki, T. & Kino, H. Numerical atomic basis orbitals from H to Kr. Phys. Rev. B 69, Series in Material Science.
195113 (2004). 63. Chakraborty, A. S. & Mahapatra, S. Compact model for low effective mass channel
50. Monkhorst, H. J. & Pack, J. D. Special points for Brillouin-zone integrations. Phys. common double-gate MOSFET. IEEE Trans. Electron Devices 65, 888–894 (2018).
Rev. B 13, 5188 (1976). 64. Oh, S.-Y., Ward, D. & Dutton, R. Transient analysis of MOS transistors. IEEE Trans.
51. Grimme, S. Semiempirical GGA-type density functional constructed with a long- Electron Devices 27, 1571–1578 (1980).
range dispersion correction. J. Comput. Chem. 27, 1787–1799 (2006). 65. Gildenblat, G. et al. PSP: An advanced surface-potential-based MOSFET Model for
52. Boys, S. F. & Bernardi, F. The calculation of small molecular interactions by the circuit simulation. IEEE Trans. Electron Devices 53, 1979–1993 (2006).
differences of separate total energies. Some procedures with reduced errors. Mol.
Phys. 19, 553–566 (1970).
53. Liu, D. C. & Nocedal, J. On the limited memory BFGS method for large scale Open Access This article is licensed under a Creative Commons
optimization. Math. Program. 45, 503–528 (1989). Attribution 4.0 International License, which permits use, sharing,
54. Stradi, D., Martinez, U., Blom, A., Brandbyge, M. & Stokbro, K. General atomistic adaptation, distribution and reproduction in any medium or format, as long as you give
approach for modeling metal semiconductor interfaces using density functional appropriate credit to the original author(s) and the source, provide a link to the Creative
theory and nonequilibrium Green’s function. Phys. Rev. B. Condens. Matter 93, Commons license, and indicate if changes were made. The images or other third party
155302 (2016). material in this article are included in the article’s Creative Commons license, unless
55. Saha, D. & Mahapatra, S. Asymmetric junctions in metallic–semiconducting–metallic indicated otherwise in a credit line to the material. If material is not included in the
heterophase MoS2. IEEE Trans. Electron Devices 64, 2457–2460 (2017). article’s Creative Commons license and your intended use is not permitted by statutory
56. Suh, J. et al. Doping against the native propensity of MoS2: Degenerate hole regulation or exceeds the permitted use, you will need to obtain permission directly
doping by cation substitution. Nano Lett. 14, 6976–6982 (2014). from the copyright holder. To view a copy of this license, visit http://creativecommons.
57. Park, H. -Y. et al. Wide-range controllable n-doping of molybdenum disulfide org/licenses/by/4.0/.
(MoS2) through thermal and optical activation. ACS Nano 9, 2368–2376 (2015).
58. Pham, V. P. & Yeom, G. Y. Recent advances in doping of molybdenum disulfide:
Industrial applications and future prospects. Adv. Mater. 28, 9024–9059 (2016). © The Author(s) 2018
59. Oostinga, J. B., Heersche, H. B., Liu, X., Morpurgo, A. F. & Vandersypen, L. M. K.
Gate-induced insulating state in bilayer graphene devices. Nat. Mater. 7, 151–157
(2008).

npj 2D Materials and Applications (2018) 28 Published in partnership with FCT NOVA with the support of E-MRS