Вы находитесь на странице: 1из 16

Analytical modeling of subthreshold characteristics by

considering quantum confinement effects in ultrathin


dual-metal quadruplegate (DMQG) MOSFETs
Visweswara Rao Samoju1, Kamalakanta Mahapatra1 and Pramod Kumar Tiwari2 *

1
Department of Electronics & Communication Engineering, National Institute of Technology, Rourkela-
769008, Odisha, India
2
Department of Electrical Engineering, Indian Institute of Technology, Patna-801103, Bihar, India
*
pktiwari@iitp.ac.in, 91-612-3028351

In this work, analytical model of subthreshold characteristics for an ultrathin dual-metal

quadruple gate MOSFET have been developed by considering the quantum confinement effects.

The 3D Poisson’s and 2D Schrödinger’s equations have been solved analytically to obtain the

threshold voltage and subthreshold current of ultrathin dual-metal quadruple gate MOSFET with

minimum cross-section of 3 nm×3 nm. The integrated charge at the threshold condition is

obtained by considering minimum sub-band energy levels with a square potential well

approximation for a two dimensional (width and height) carrier confinement. The subthreshold

characteristics have been analyzed by varying different device parameters like channel cross-

section, gate length ratio, gate metal work functions and gate oxide thickness. Further, the effect

of 2D charge carrier confinement on drain induced barrier lowering (DIBL) has also been

discussed using the proposed model. The proposed models are validated against the numerical

simulation results obtained from 3D ATLAS device simulator.

Key words: ultrathin MOSFETs, Quantum confinement effects, Bohm quantum potential,

threshold voltage, subthreshold current, DIBL.


1. Introduction

The nanoscale multigate (MG) MOSFET structures have been of great interest owing to their

attractive electrostatic properties and potential applications [1-3] that cannot be achieved with

their conventional counterparts [4-5]. Among the innovative MG MOSFET structures [6-11],

quadruple-gate (QG) MOSFETs are found to be attractive building blocks for ultra large scale

integrated (ULSI) circuits [9-13]. In quadruple-gate (QG) MOSFET, the surrounded gate

electrodes produce maximum electric field in the channel region to minimize short-channel

effects [9-13]. Further, the optimization of short-channel effects has been achieved by ultrathin

channel thickness and gate engineering techniques [12-13]. In ultrathin QG MOSFETs, the

quantum mechanical effects (QMEs) start dominating as the width and height are scaled beyond

8 nm [13-14]. Because of the charge carrier confinement in nanoscale devices, the quantized

energy levels lead to reduction in gate capacitance, shift in threshold voltage and degradation of

subthreshold slope etc. [14-15]. Therefore, the quantum confinement effects need to be

incorporated in modeling of ultrathin QG MOSFETs so that the subthreshold characteristics

could be investigated more accurately.

Many attempts have been made to include the quantum mechanical effects in MOSFET devices

[13-16]. Firstly, in 1993 Omura et. al. [14] predicted the shift in threshold voltage due to the

quantization of charge carriers in a thin silicon layer. Later, Colinge et. al. [16] observed the

variations in subthreshold characteristics like reduced subthreshold current and increased

threshold voltage for a small cross-sectional tri-gate SOI MOSFETs. But, the work [16] was

based on only numerical simulation results. Kumar et. al. [9] presented the quantum threshold

voltage model for a triple gate MOSFETs. The authors [9] thoroughly observed the effects of

carrier confinement and effective mass variation of charge carrier on threshold voltage; however
the model [9] was limited to long channel devices. Further, Kumar et.al. [13] developed a

threshold voltage model for quad-gate nanowire MOSFETs by solving the Poisson’s and

Schrödinger equations. The authors [13] successfully explained the effect of charge carrier

quantization in short-channel devices.

In our previous work [12], the gate engineering technique was employed on quadruple gate

MOSFETs and the models of threshold voltage and DIBL were analyzed for short-channel

devices. However, the model [12] was restricted to cross-section of 10 nm×10 nm to avoid the

complexity owing to quantum confinement effects [14-15]. The act of further downscaling of the

cross-section of dual-metal (DM) QG MOSFETs beyond the 10 nm regimes makes the device

more prone to the effects like quantization of charge carriers, tunneling and hot-carrier induced

effects [14-15]. Therefore in this work a quantum model of dual-metal quadruple gate (DMQG)

MOSFETs is proposed.

In this work, the analytical models of the subthreshold characteristics like threshold voltage and

subthreshold current for ultrathin DMQG MOSFETs have been proposed. The Poisson-

Schrodinger equations have been solved analytically to obtain the discrete energy levels in

channel region, which is required for the calculation of inversion charge density to formulate the

subthreshold characteristic models. For simplification, the parabolic potential well at virtual

cathode position has been approximated to square potential well. The proposed quantum models

are validated against the numerical simulation results obtained from ATLAS 3D device simulator

[17].
2. Modeling

The three dimensional simulated structure of ultrathin dual-metal quadruple gate (DMQG)

MOSFET, obtained by using 3D ATLAS device simulator has been shown in Fig.1. The lightly

doped Si channel of width W = 5 nm, height H = 5 nm and length L = 30 nm is surrounded by

SiO2 of thickness t ox = 1 nm. Covering this dielectric layer, two different metals are surrounded

side by side, with work functions ∅𝑚1 and ∅𝑚2 , (where ∅𝑚1 > ∅𝑚2 ). In this way, the metal with

higher work function (∅𝑚1 ) works as control gate whereas the metal with lower work function

(∅𝑚2 ) functions as screen gate. By keeping ∅𝑚1 near source-end, the channel region gets

virtually divided into two regions. The region I under the control gate of length L1 governs the

electron flow and maintains the electrostatic control in the device, whereas the region II under

the screen gate of length L2 aims to minimize the variations in the electrical characteristics

occurred due to the drain voltage.

2.1 Channel potential

The proposed model is limited to subthreshold regime of ultrathin DMQG MOSFETs. Thus, in

weak inversion regime the inversion charge concentration is ignored such that the 3D Poisson

equation can be approximated to Laplace equation [10, 12] and the Poisson-Schrodinger

equations could be decoupled and solved analytically. Thus, the 3D Laplace equation is given as

𝜕2 𝜙𝑘 (𝑥,𝑦,𝑧) 𝜕2 𝜙𝑘 (𝑥,𝑦,𝑧) 𝜕2 𝜙𝑘 (𝑥,𝑦,𝑧)


+ + =0 (1)
𝜕𝑥 2 𝜕𝑦 2 𝜕𝑧 2

where, 𝜙𝑘 (𝑥, 𝑦, 𝑧) is the channel potential, which is assumed to follow the parabolic distribution

and is given as

2𝑥 2𝑦 2 2
𝜙𝑘 (𝑥, 𝑦, 𝑧) = 𝜙̂𝑘 (0,0, 𝑧) (1 − ( 𝑎 ) ) (1 − ( 𝑏 ) ) + 𝑉𝑔𝑠 − 𝑉𝐹𝐵𝑘 (2)
where 𝑘 = 1, 2 denotes the potential distribution in the channel region I and channel region II

respectively and 𝜙̂𝑘 (0,0, 𝑧) is the center channel potential distribution. Following our previous

work [12], the channel potential distribution is obtained by solving the Eq. (1) with the help of

Eq. (2) and appropriate boundary conditions as mentioned in [10-12]. The expression of channel

potential distribution is given as

𝑧⁄ ) 𝑧 2𝑥 2 2𝑦 2
𝜙1 (𝑥, 𝑦, 𝑧) = ( 𝑘1 𝑒 ( 𝜆 − 2𝐵1 sinh (𝜆)) (1 − ( 𝑎 ) ) (1 − ( 𝑏 ) ) + 𝑉𝑔𝑠 − 𝑉𝐹𝐵1 (0 ≤ 𝑧 ≤ 𝐿1 )

(3)

𝑧⁄ ) 𝑧 𝑧−𝐿1 2𝑥 2
𝜙2 (𝑥, 𝑦, 𝑧) = ( 𝑘1 𝑒 ( 𝜆 − 2𝐵1 sinh (𝜆) + (𝑉𝐹𝐵2 − 𝑉𝐹𝐵1 ) (1 + cosh ( ))) (1 − ( 𝑎 ) ) (1 −
𝜆

2𝑦 2
( 𝑏 ) ) + 𝑉𝑔𝑠 − 𝑉𝐹𝐵2 (𝐿1 ≤ 𝑧 ≤ 𝐿1 + 𝐿2 ) (4)

𝐿
1 𝑘1 𝑒 ((𝐿1 +𝐿2 )⁄𝜆) −𝑘2 −𝑉𝑑𝑠 −(𝑉𝐹𝐵1 −𝑉𝐹𝐵2 ) cosh( 2 )
𝜆
where, 𝐵1 = 2 ( 𝐿 +𝐿 ) (5)
sinh( 1 2 )
𝜆

𝑘1 = 𝑉𝑏𝑖 − 𝑉𝑔𝑠 + 𝑉𝐹𝐵1 (6)

𝑘2 = 𝑉𝑏𝑖 − 𝑉𝑔𝑠 + 𝑉𝐹𝐵2 (7)

′ ′ ′
𝑎 = 𝑊 + 2𝑡𝑜𝑥 and 𝑏 = 𝐻 + 2𝑡𝑜𝑥 are the extended body width and height respectively, 𝑡𝑜𝑥 =
𝑡𝑜𝑥 𝜀𝑠𝑖
is electro-statically extended silicon layer thickness [10], where 𝜀𝑠𝑖 and 𝜀𝑜𝑥 are the relative
𝜀𝑜𝑥

𝑁𝑎 𝑁𝑑
permittivity of silicon and silicon oxide. 𝑉𝑏𝑖 = 𝑉𝑇 𝑙𝑛 ( ) be the built-in potential, Vfbk =
𝑛𝑖 2

Eg N
𝜙mk − (χsi + 2q + VT ln ( na )) is the flat band voltage of the regions 1 and 2 with k = 1 and 2,
i

𝐾𝑇
respectively; χsi is the electron affinity and 𝐸𝑔 is energy band gap of the silicon; and 𝑉𝑇 = is
𝑞

the thermal voltage.


Threshold voltage model

The threshold voltage of ultrathin devices is defined as the value of gate voltage at which the

inversion charge density (Qinv ) reaches to a critical charge (Qth ) which is obtained from

numerical simulation [10, 12-13]. The classical approach of finding the inversion charge in

DMQG MOSFET is reported in [12, 13] and is given as


H W 𝜙𝑣𝑐
− −
2 2
Qinv,C = ∫ −H −W ∫ qni e VT dx. dy (8)
2 2

2𝑥 2 2𝑦 2
where, 𝜙𝑣𝑐 (𝑥, 𝑦) = 2√B1 (k1 − B1 ) (1 − ( 𝑎 ) ) (1 − ( 𝑏 ) ) + 𝑉𝑔𝑠 − 𝑉𝐹𝐵1 (9)

The virtual cathode potential 𝜙𝑣𝑐 (𝑥, 𝑦) is a locus of minimum potential values distributed along

x- and y- directions at virtual cathode position Zmin . The virtual cathode position is obtained by

𝜕𝜙1 (0,0,𝑧)
solving the equation = 0, and given as
𝜕𝑉𝑔𝑠

λ k
Zmin = − 2 (ln (B1 − 1)) (10)
1

𝑎𝑏
where, λ is the characteristic length given by [12 ], λ =
√8(𝑎2 +𝑏2 )

Solving Eq. (8) at 𝑉𝑔𝑠 = Vth,C and Qinv,C = Qth gives the classical threshold voltage expression

as given in [12].

In ultrathin DMQG MOSFETs, the electron energy gets quantized due to the 2D structural

confinement in both x- and y- directions. The quantized sub-band energy levels could be

obtained by solving the two dimensional Schrodinger equation given as

ℏ2 𝜕 2 𝜓 ℏ2 𝜕 2 𝜓
+ 2𝑚 + (𝐸 − 𝐸𝑐,𝑚𝑖𝑛 )𝜓 = 0 (11)
2𝑚𝑥 𝜕𝑥 2 𝑦 𝜕𝑦 2
where, 𝜓 represents wave function of electron, 𝐸 is confined energy levels, 𝐸𝑐 is minimum

energy level in the conduction band, ℏ is the reduced plank constant and 𝑚𝑥 , 𝑚𝑦 are the electron

effective mass in x- and y- direction respectively.

From the numerical simulation of ultrathin DMQG MOSFET, it is evident that the potential

distribution at virtual cathode is forming a parabolic potential well in x- and y- directions.

Solving Eq. (11) with a parabolic potential well will be a complex mathematical practice. Hence,

to reduce the complexity in solving the Eq.(11), the parabolic well has been approximated with

square potential well by considering the bottom of the potential well as the minima of conduction

band energy (𝐸𝑐,𝑚𝑖𝑛 ).

𝐸𝑔 3𝑎 3𝑏
𝐸𝑐,𝑚𝑖𝑛 = − 𝑞𝜙𝑣𝑐 (𝑥 = , 𝑦 = 14) (12)
2 14

Solving Eq.(11) with variable separable technique gives the solutions as [13],

4 𝑛𝑥 𝜋(𝑥−(𝑎⁄2)) 𝑛𝑦 𝜋(𝑦−(𝑏⁄2))
𝜓𝑛𝑥 𝑛𝑦 (𝑥, 𝑦) = √𝑎.𝑏 sin ( ) sin ( ) (13)
𝑎 𝑏

𝐸𝑛𝑥 𝑛𝑦 = 𝐸𝑐,𝑚𝑖𝑛 + 𝐸𝑁 (𝑛𝑥 , 𝑛𝑦 ) (14)

where, 𝑛𝑥 = 1,2 … and 𝑛𝑦 = 1,2 …. are the principal quantum number and 𝑁 = 1,2,3 are three

different series of quantized energy levels for silicon MOSFET with <100> orientation [8] and is

given as

ℏ2 𝜋 2 1 𝑛 2 1 𝑛𝑦 2
𝐸1 (𝑛𝑥 , 𝑛𝑦 ) = (𝑚 ( 𝑎𝑥 ) + 𝑚 ( 𝑏 ) ) (15)
2 𝑙 𝑡

ℏ2 𝜋 2 1 𝑛 2 1 𝑛𝑦 2
𝐸2 (𝑛𝑥 , 𝑛𝑦 ) = (𝑚 ( 𝑎𝑥 ) + 𝑚 ( 𝑏 ) ) (16)
2 𝑡 𝑙

ℏ2 𝜋 2 1 𝑛 2 1 𝑛𝑦 2
𝐸3 (𝑛𝑥 , 𝑛𝑦 ) = (𝑚 ( 𝑎𝑥 ) + 𝑚 ( 𝑏 ) ) (17)
2 𝑡 𝑡

where, 𝑚𝑙 = 0.916×𝑚0 and 𝑚𝑡 = 0.19×𝑚0 are the longitudinal and transversal effective

masses of electron, respectively and 𝑚0 is rest mass of electron.


The total quantum inversion charge (𝑄𝑖𝑛𝑣,𝑄 ) at virtual cathode formed due to the sub-band

energy levels is given by [13]



𝑄𝑖𝑛𝑣,𝑄 = 𝑞 ∫𝐸 𝑁1𝐷 (𝐸, 𝐸𝑛𝑥 𝑛𝑦 )𝑓(𝐸)𝑑(𝐸) (18)
𝑛𝑥 𝑛𝑦

where, 𝑁1𝐷 (𝐸, 𝐸𝑛𝑥 𝑛𝑦 ) is 1D density of states (DOS) and 𝑓(𝐸) is the Fermi-Dirac distribution

function:
1
𝑚𝐷 −
2
𝑁1𝐷 (𝐸, 𝐸𝑛𝑥 𝑛𝑦 ) = √ . (𝐸 − 𝐸𝑛𝑥 𝑛𝑦 ) (19)
2𝜋ℏ2

1
𝑓(𝐸) = 1+𝑒𝑥𝑝((𝐸− 𝐸 (20)
𝐹 )⁄𝐾𝑇 )

where, 𝑚𝐷 is the effective mass of electron perpendicular to the direction of quantization (𝑚𝐷 =

𝑚𝑡 for 𝑁=1,2 and 𝑚𝐷 = 𝑚𝑙 for 𝑁=3).

The quantum inversion charge density in Eq.(18) is dependent on gate voltage via the position of

fermi level with respect to minimum value of quantized energy level and with respect to different

energy levels. However, in weak inversion region the fermi level is found to be much below the

conduction band energy level. Therefore, using Boltzmann statistics the quantum inversion

charge at virtual cathode is given as [8, 13]

𝑚𝐷 𝑘𝑇 𝐸𝐹 −𝐸𝑛𝑥 𝑛𝑦
𝑄𝑖𝑛𝑣,𝑄 = 𝑞 ∑𝑁 ∑𝑛𝑥 ∑𝑛𝑦 √( ) 𝑒𝑥𝑝 ( ) (21)
2ℏ2 𝑘𝑇

Using Eqs.(12), (14)-(17), the quantum inversion charge density given in Eq.(21) could be

rewritten as

2𝑚𝑡 𝑘𝑇 𝐸𝑐,𝑚𝑖𝑛 +𝐸1 (𝑛𝑥 ,𝑛𝑦 )


√( ) 𝑒𝑥𝑝 (− )+
ℏ2 𝑘𝑇

𝐸 𝐸𝑐,𝑚𝑖𝑛 +𝐸2 (𝑛𝑥 ,𝑛𝑦 )


𝑄𝑖𝑛𝑣,𝑄 = 𝑞𝑒𝑥𝑝 (𝑘𝑇𝐹 ) ∑𝑛𝑥 ∑𝑛𝑦 √(2𝑚𝑡2𝑘𝑇) 𝑒𝑥𝑝 (− )+ (22)
ℏ 𝑘𝑇

2𝑚𝑙 𝑘𝑇 𝐸𝑐,𝑚𝑖𝑛 +𝐸3 (𝑛𝑥 ,𝑛𝑦 )


√( ) 𝑒𝑥𝑝 (− )
[ ( ℏ2 𝑘𝑇 )]
Each sub-band energy level has its own minimum energy levels. However, from the simulation,

it is found that the lowest energy level 𝐸𝑛𝑥 𝑛𝑦 with 𝑛𝑥 = 1 and 𝑛𝑦 = 1 has most of the electrons.

Hence, to obtain the quantum charge density, only one energy level has been considered [8, 13].

Thus, the quantum charge density obtained in Eq. (22) could be rewritten with 𝑛𝑥 = 1 and 𝑛𝑦 =

1 as

𝐸𝐹 −𝐸𝑐,𝑚𝑖𝑛
𝑄𝑖𝑛𝑣,𝑄 = 𝑞𝑒𝑥𝑝 ( ).Υ (23)
𝑘𝑇

where,

(1,1) 𝐸2 (1,1) 𝐸3 (1,1)


Υ = √(2𝑚𝑡2𝑘𝑇) (𝑒𝑥𝑝 (− 𝐸1 ) + 𝑒𝑥𝑝 (− )) + √(
2𝑚𝑙 𝑘𝑇
) 𝑒𝑥𝑝 (− ) (24)
ℏ 𝑘𝑇 𝑘𝑇 ℏ2 𝑘𝑇

Following the definition of threshold voltage for ultrathin MOSFETs [12, 13], the quantum

charge density (𝑄𝑖𝑛𝑣,𝑄 ) at the virtual cathode is equated to the critical charge (𝑄𝑇𝐻 ). However,

there is no standard value to find the critical threshold charge. So, using 3D atlas simulator the

critical threshold charge is calibrated for each devices and an average charge is found to 𝑄𝑇𝐻 =

5×1022 𝑞𝑊𝐻 cm-1 [10, 13].

The threshold voltage of ultrathin DMQG MOSFETs is obtained by equating the right hand side

of Eq. (23) to 𝑄𝑇𝐻 at 𝑉𝑔𝑠 = 𝑉𝑇𝐻,𝑄 , which gives the following equation

𝐸𝑔 𝑄
1.33√𝐵1′ (𝑘1′ − 𝐵1′ ) + 𝑉𝑇𝐻 − 𝑉𝐹𝐵1 = 𝑇𝐻
− 𝐸𝐹 + 𝑉𝑇 𝑙𝑛 ( 𝑞.Υ ) (25)
2

where 𝐵1′ = 𝐵1 and 𝑘1′ = 𝑘1 at 𝑉𝑔𝑠 = 𝑉𝑇𝐻,𝑄

Further simplification of Eq. (25) gives the analytical equation of threshold voltage for an

ultrathin DMQG MOSFETs:

−𝑐2 +√𝑐22 −4𝑐1 𝑐3


𝑉𝑇𝐻,𝑄 = (26)
2𝑐1

𝐿1 +𝐿2 2
Where 𝑐1 = 1 − 1.77 (𝑠𝑒𝑐ℎ ) (27)
𝜆
𝐿1 +𝐿2 𝐸𝑔
𝐿1 +𝐿2 𝑄
𝑐2 = (1.77(𝑉𝑏𝑖 + 𝑉𝐹𝐵1 ) + 𝐶0 )𝑒 ( ) 𝑇𝐻
2𝜆 . 𝑠𝑒𝑐ℎ ( ) − 2 ( 2 − 𝐸𝐹 + 𝑉𝑇 𝑙𝑛 ( 𝑞.Υ )) − 2𝑉𝐹𝐵1 − 𝐶0
2𝜆

(28)

𝐸𝑔 𝑄 2
𝑇𝐻
𝑐3 = ( 2 − 𝐸𝐹 + 𝑉𝑇 𝑙𝑛 ( 𝑞.Υ ) + 𝑉𝐹𝐵1 ) − 1.77(𝐶0 (𝑉𝑏𝑖 + 𝑉𝐹𝐵1 ) − 𝐶0 2 ) (29)

𝐿 +𝐿 𝐿 +𝐿
( 1 2) ( 1 2) 𝐿 𝐿 2
𝑉𝑏𝑖 (𝑒 𝜆 −1)+𝑉𝐹𝐵1 (𝑒 𝜆 −𝑐𝑜𝑠ℎ 2 )+2𝑉𝐹𝐵2 (𝑠𝑖𝑛ℎ 2 ) −𝑉𝑑𝑠
𝜆 𝜆
𝐶0 = 𝐿 +𝐿 (30)
2𝑠𝑖𝑛ℎ 1 2
𝜆

Using Eq. (26), the drain induced barrier lowering (DIBL) of ultrathin DMQG MOSFET could

be expressed as [12]

𝑉𝑇𝐻,𝑄 | − 𝑉𝑇𝐻,𝑄 |
𝑉𝑑𝑠,ℎ𝑖𝑔ℎ 𝑉𝑑𝑠,𝑙𝑜𝑤
𝐷𝐼𝐵𝐿 = − (31)
𝑉𝑑𝑠,ℎ𝑖𝑔ℎ −𝑉𝑑𝑠,𝑙𝑜𝑤

2.2 Quantum subthreshold current

Using the quantum inversion charge obtained in Eq. (23), the subthreshold current for ultrathin

DMQG MOSFET could be modeled using the Pao-Sah’s double integral formula [18-21]. The

modified Pao-Sah’s double integral equation for ultrathin DMQG MOSFETs is given as [20].

𝑉
𝜇𝑉𝑇 (1−𝑒𝑥𝑝(− 𝑑𝑠 ))
𝑉𝑇
𝐼𝐷 = 𝐿1 −1 𝐿1 +𝐿2 −1 (32)
∫0 𝑄𝑖𝑛𝑣,𝑄1 𝑑𝑧+∫𝐿 𝑄𝑖𝑛𝑣,𝑄2 𝑑𝑧
1

where, 𝜇 is mobility of electrons; 𝑄𝑖𝑛𝑣,𝑄𝑘 is the quantum inversion charge in the channel region I

and channel region II for 𝑘 = 1 and 𝑘 = 2 respectively.

The quantum inversion charge given in Eq. (23) is obtained by integrating the density of sates

and Fermi-Dirac function in the channel region I at virtual cathode. Having higher metal work

function to control gate than screen gate, it has been assumed that the inversion charge in

channel region I would provide major contribution in total subthreshold current of ultrathin

DMQG MOSFETs. Hence, Eq. (32) could be rewritten neglecting 𝑄𝑖𝑛𝑣,𝑄2 and replacing 𝑄𝑖𝑛𝑣,𝑄1

with 𝑄𝑖𝑛𝑣,𝑄 given in Eq. (23), and the subthreshold current is given as
𝑉𝑑𝑠
𝐼𝐷 = 𝜇𝑉𝑇 (1 − 𝑒𝑥𝑝 (− )) 𝑄𝑖𝑛𝑣,𝑄 𝐿1 (33)
𝑉𝑇

In fact, the present proposed model of Eq. (26) is an improved and extended version of threshold

voltage models reported in references [12-13]. Further, the proposed model can also be used to

find subthreshold characteristic of ultra-scaled quadruple gate MOSFET by considering the

length of the screen gate to be zero.

3. Results and discussion

In this section, the threshold voltage and subthreshold current for ultrathin DMQG MOSFETs

have been analyzed by varying device parameters like gate length ratio, difference of gate metal

work function, varying channel thickness (𝑊 = 𝐻). The developed analytical results are

validated by comparing with numerical simulation results obtained from 3D ATLAS device

simulator from SILVACO [17]. The Bohm quantum potential (BQP) model has been

incorporated in the numerical simulation, which could be calibrated to match the Poisson-

Schrodinger solver to address the quantum confinement effects in ultrathin DMQG MOSFETs

[22-23]. The BQP model ensures good calibration for silicon or non-silicon materials, planar or

non-planar devices with proper fitting parameters. The BQP model provides better convergence

properties when compared to other quantum models. Further, the Fermi-Dirac statistics has also

been included in the model [22-23]. The BQP model is initiated using the command line BQP.N

in the models statement.

Figure 2 shows the variation of threshold voltage with channel length for different gate length

ratios. For higher gate length ratio, (𝐿1 : 𝐿2 = 2: 1), the threshold voltage falls to nearly 20 mV as

the channel length is reduced from 90 nm to 20 nm. Whereas, in case of low gate length ratio

(𝐿1 : 𝐿2 = 1: 2), the threshold voltage falls by 50 mV for the same channel length variation. This

high fall of threshold voltage in case of 𝐿1 : 𝐿2 = 1: 2 is due to the decreased electrostatic control
in the channel region. Hence from Fig.3, it may be deduced that considering larger part of

control gate could minimize the threshold voltage roll-off. Further, in Fig.3, the quantum

confinement effect on the threshold voltage has also been shown by comparing the quantum

threshold voltage model with the classical threshold voltage model [12], for a gate length

ratio 𝐿1 : 𝐿2 = 2: 1. It is observed that due to the quantum carrier confinement effects the

quantum threshold voltage has an incremental shift with respect to classical threshold voltage

[12] of DMQG MOSFET having 5 nm×5 nm crosssection. Figure 3 shows the effect of screen

gate metal work function over the quantum threshold voltage. When the gate metal work

function difference [∅𝑚1 (= 4.8 ev) − ∅𝑚2 (= 4.6⁄4.3⁄4.1 ev)] increases by considering

constant control gate work function, the minimum value of the channel potential is observed to

be increasing [12], which results in reduced threshold voltage. Clearly, increasing the metal work

function difference from 0.2 eV to 0.7 eV, the quantum threshold voltage is reduced from 0.379

V to 0.358 V respectively for 20 nm channel length with gate length ratio of 1:1.Therefore, it is

noted that the devices with low value of screen gate metal work function, in other words for large

metal work function difference, will be having high threshold voltage roll-off. It is also important

to note from both Fig. 2 and Fig. 3 that the dual-metal-gate structure has a strong impact over the

threshold voltage of the ultrathin DMQG MOSFETs.

The variation of the quantum, 𝑉𝑇𝐻,𝑄 and classical,𝑉𝑇𝐻,𝐶 threshold voltage differences for

different channel cross-sections of DMQG MOSFETs is discussed in Fig.4. The increase in

threshold voltage due to the quantum confinement effects is shown along the y-axis and the

channel length variation along the x-axis of Fig. 4. Because of the addition of short-channel

effects to the quantum confinement effects, the difference of threshold voltages (𝑉𝑇𝐻,𝑄 − 𝑉𝑇𝐻,𝐶 )

increases for short-channel devices when compared to long channel devices. Further, it is
observed that with decreasing cross-sectional area the difference of quantum and classical

threshold voltage increases. This difference in threshold voltage is due to the effect of two

dimensional confinements of charge carriers in the channel regions. As the channel cross-section

is decreased from 10 nm×10 nm to 3 nm×3 nm, the difference of quantum and classical

threshold voltage is increased from 5 mV to 62 mV for channel length of 50 nm. In Fig. 5, the

variation of subthreshold current is plotted against the gate to source voltage for different gate

length ratios. The effect of gate length ratio could be observed by considering 𝐿1 : 𝐿2 = 1: 1 as

reference. In case of low gate length ratio (𝐿1 : 𝐿2 = 1: 2), the electric field control over the

charge carrier weakens as the control gate length is decreased and thus the leakage current

increases as compare to 𝐿1 : 𝐿2 = 1: 1. Whereas in case of 𝐿1 : 𝐿2 = 2: 1, the larger part of control

gate will strengthen the electric field over the charge carriers and helps in reducing the leakage

currents. Figure 6 shows the subthreshold current variations against the gate to source voltage for

different channel thicknesses (W = H = 10⁄8⁄5/3 nm). Varying the channel thickness from 10

nm to 3 nm, the subthreshold current decreases from 10−11A to 10−14 A respectively. This is

because of strong quantization of charge carriers and increased electrostatic control over the

charge carriers in the ultrathin channel region. Figure 7displays the effect of drain voltage on

threshold voltage. This variation in threshold voltage due to the drain voltage is estimated by

drain induced barrier lowering (DIBL) which is defined in Eq. (31). The DIBL has significant

value as the channel length is scaled beyond 35 nm for a DMQG MOSFET. For reducing the

channel length from 35 nm to 20 nm, the DIBL value increases from 2 mV/V to 15.2 mV/V

respectively as shown in Fig. 7. This effect of drain voltage on threshold voltage could be

minimized by selecting larger screen gate length. Further, it is noticed from Fig.7 that the
DMQG devices show better resistance to the drain voltage when compared to quadruple gate

(QG) MOSFETs.

Conclusion:

The present work deals with the subthreshold characteristics of ultrathin DMQG MOSFETs

which have been derived by an analytical model considering quantum confinement effects. The

proposed models are obtained from 3D Poisson’s and 2D Schrodinger’s equations and are

capable of predicting the subthreshold characteristic for ultrashort and ultrathin body devices

competently. For high performance of the proposed device, it would be preferred to consider

ultrathin channel cross-section, the minimum metal gate work function difference and large

control gate length. Further, the DIBL of ultrathin DMQG MOSFETs has also been compared

with QG MOSFETs and it is inferred that the DMQG MOSFETs possess high resistance towards

drain voltage fluctuations. The presented model results are in perfect agreement with the 3D

numerical simulation results obtained from 3D ATLAS device simulator.

References:

1. J.Appenzeller, J.Knoch, M.T.Bj ̈ork, H.Riel, H.Schmid, W.Riess, Toward nanowire


electronics, IEEE Trans. Electron Dev. 55 (2008) 2827–2845.
2. B.Yu, M.Meyyappan, Nanotechnology: Role in emerging nanoelectronics, Solid-State
Electronics 50 (2006) 536–544.
3. International technology roadmap for semiconductors (ITRS)
http://www.semiconductors.org/clientuploads/Research_Technology/ITRS/2015/0_20
15%20ITRS%202.0%20Executive%20Report%20(1).pdf
4. F. Assad, Z. Ren, D. Vasileska, S. Datta, M. Lundstrom, On the performance limits for Si
MOSFETs: A theoretical study, IEEE Trans. Electron. Devices 47 (2000) 232–240.
5. Y.B.Kim, Challenges for nanoscale MOSFETs and emerging nano electronics, Trans. Electr.
Electron. Mater., 11 (2010) 93-105.
6. J.P.Colinge, FinFETs and Other Multi-Gate Transistors, Springer 2008.
7. T.A.Fjeldly, U.Monga, S.K.Vishvakarma, Compact unified modeling of multigate MOSFETs
based on isomorphic modeling functions, in: Proceeding of 8th international Caribbean
conference 2012 devices, circuits and systems (ICCDCS), 2012, pp. 1-4.
8. J.L.Autran, K.Nehari, D.Munteanu, Compact modeling of the threshold voltage in silicon
nanowire MOSFET including 2D-quantum confinement effects, Molecular Simulation, 31
(2005) 839–843.
9. P.R. Kumar, S. Mahapatra, Analytical modeling of quantum threshold voltage for triple gate
MOSFETs, Solid State Electronics 54 (2010) 1586-1591.
10. D.Sharma, S.K.Vishvakarma, Precise analytical model for short-channel quadruple-gate
gate-all-around MOSFET, IEEE Trans. Nanotechnol. 12 (2013) 378-385.
11. T.K.Chiang, A Novel quasi-3-D threshold voltage model for fully depleted quadruple-gate
(FDQG) MOSFETs: with equivalent number of gates (ENG) included, IEEE Trans.
Nanotechnol.12 (2013) 1022-1025.
12. V.R.Samoju, P.K.Tiwari, Threshold voltage modeling for dual-metal quadruple gate
(DMQG) MOSFETs, Int. J. Numer. Model. 29 (2015) 695-706.
13. P.R.Kumar, S.Mahapatra, Quantum threshold voltage modeling of short channel quad gate
silicon nanowire transistor, IEEE Trans. Nanotechnol. 10 (2011) 121-128.
14. Y. Omura, S. Horiguchi, M. Tabe, K. Kishi, Quantum Mechanical effects on the threshold
voltage of ultrathin-SOI, IEEE Electron Dev. Lett. 14 (1993) 569–571.
15. M. Balaguer, J.B.R.Aranda, F.Gamiz, In-Depth Study of Quantum Effects in SOI
DGMOSFETs for Different Crystallographic Orientations, IEEE Electron Dev. 58 (2011)
4438-4441.
16. J.P. colinge, J.C. Alderman, W. Xiong, C.R. Cleavelin, Quantum Mechanical effects in tri-
gate SOI MOSFETs, IEEE Trans. Electron Dev. 53 (2006) 1131–1135.
17. ATLAS User’s Manual: Device Simulation Software, Santa Clara, CA: Silvaco International,
2008.
18. Y.Taur, T.H.Ning, Fundamentals of Modern VLSI Devices, Cambridge Univ. Press, New
York, 1998.
19. P.K.Tiwari, S.Dubey, K.Singh, S.Jit, Analytical modeling of subthreshold current and
subthreshold swing of short-channel triple-material double-gate (TM-DG) MOSFETs,
Superlattices Microstruct. 51 (2012) 715–724.
20. A.Kumar, S.Bhushan, P.K.Tiwari, Analytical modeling of subthreshold characteristics of
ultra-thin double gate-all-around (DGAA) MOSFETs incorporating quantum confinement
effects, Superlattices Microstruct., https://doi.org/10.1016/j.spmi.2017.05.038.
21. V.R.Samoju, S.Dubey, P.K.Tiwari, Quasi-3D subthreshold current and subthreshold swing
models of dual-metal quadruple-gate (DMQG) MOSFETs. Journal of Comput. Electron. 14
(2015) 582-592.
22. G.lannaccone, G.curatola, G.Flori, Effective Bohm Quantum Potential for device simulators
based on drift and energy transport, simulation of semiconductor process and devices 2004
(2004) 275-278.
23. V.S. Patnaik, A.Gheedia, M.J. Kumar, The simulation standard 2008, 18 (2008) 6-12.

Captions to Figures

Fig.1 3D simulated structure of ultrathin DMQG MOSFET in x-y-z space.

Fig.2 Threshold voltage variation against the channel length with gate length ratio (𝐿1 : 𝐿2 ) as a
varying parameter.

Fig.3 Quantum threshold voltage variation against the channel length with different screen gate
work functions (𝜙𝑚2 ).

Fig.4 Variation of difference of quantum and classical threshold voltages along the channel
length for different channel cross-sections.

Fig.5 Subthreshold current variation with gate to source voltage for different gate length ratios (
L1 : L2 ).

Fig.6 Subthreshold current variation with gate to source voltage for different gate channel cross-
sections

Fig.7 DIBL variation against the channel length for different gate length ratios (𝐿1 : 𝐿2 ).

Вам также может понравиться