SERBIAN JOURNAL OF ELECTRICAL ENGINEERING Vol. 8, No.

2, May 2011, 147-154
UDK: 621.382.323.027.26

Threshold Voltage Modeling in (100), (110) and (111) Oriented Nanoscale MOSFET Substrates
Amit Chaudhry1, Jatindra Nath Roy2
Abstract: An analytical model for the inversion layer quantization for nanoscale – Metal Oxide Semiconductor Field Effect Transistor (MOSFET) with different crystallographic substrate orientations, such as (100), (110) and (111) has been developed. The threshold voltage analysis has been studied using the quantum inversion charge model under three substrate orientations. The results indicate a significant impact of crystal orientation on the threshold voltage and the inversion charge density. The results have also been compared with the numerically reported results and show good agreement. Keywords: Orientation, model, Inversion quantization, MOS device, Threshold voltage.

1

Introduction

MOSFET modeling is facing difficulties to achieve accurate description of extremely scaled down devices. The reason is that many complicated new phenomena are arising which are not easy to describe. One such phenomenon arising out of down scaling the MOSFET is the failure of classical physics at nanometer scale. As Complementary Metal Oxide Semiconductor (CMOS) technology scales down aggressively, it approaches a point, where classical physics is not sufficient to explain the behavior of a MOSFET. At this classical physics limit, quantum mechanics has to be taken into account to accurately assess the overall performance of a MOSFET. One of the major effects in the nanoscale MOSFETs is the inversion layer quantization. Accurate modeling of inversion layer quantization in the (100) substrate has received considerable attention [1]. However, very less research has taken place to study inversion layer quantization in other crystal oriented substrates. Whatever models developed so far for other orientations, are mainly numerical. The paper is organized as follows: The paper starts with an overview of the basic MOSFET models. Secondly, modeling of inversion layer quantization effect has been done by considering various crystal
1

University Institute of Engineering and Technology, Panjab University, Chandigarh, India; E-mail: amit_ chaudhry01@yahoo.com. 2 Solar Semiconductor Private Limited, Hyderabad, India.

147

Roy orientations.A. Van Dort Model [11]. 1. Thirdly. Solving the Poisson equation in the inverted MOSFET substrate channel. 2 Mosfet Models The charge based models include the basic Simulation Program with Integrated Circuit Emphasis (SPICE) Level 1. The inversion layer results in the lifting of the energy sub bands in different energies because the electron quantization masses are different in three orientations as shown in Fig. Also. The other models which include the energy quantization effects though empirical in nature are Hansch model [11]. MOS Model 11.N. Level 2. 3 Inversion Layer Quantization As the MOSFET dimensions approach deep sub-micron and nanometer regions. the electrons are confined in this potential well. the electron energies are quantized and hence the electrons occupy only the discrete energy levels. we get the total charge density “ Qs ” [13]: 148 . yet a MOSFET model largely analytical including inversion layer quantization in all the crystal orientations of the nanoscale MOSFET substrate is still to be developed. Due to aggressive scaling of the MOSFETs. Due to confinement. Hiroshima Starc Insulated gate Field effect transistor model (HiSIM) model etc. the conductance based models like the Enz. Inversion charge model [12] etc. 1. Berkeley Short Channel Insulated gate Field effect transistor model (BSIM) models and the other advanced models such as BSIM 4 and 5 [2-5].Vittoz (EKV) model [10]. the classical movement of the charge carriers is greatly affected by the non-classical behavior of electrons in the MOSFET. Secondly. It can be concluded that attempts are being made to include inversion layer quantization in nanoscale MOSFET models in (100) crystal orientations. the substrate doping is increased tremendously to negate the short channel effects at the deep sub-micrometer or nanometer scales. This results in very high electric fields in the silicon/silicon oxide interface and hence the potential at the interface becomes steep. During the inversion condition. [6-9] and thirdly. This results in a potential well between the oxide field and the silicon potentials. Level 3. Krummenacher. Chaudhry and J. the potential based models include the Surface Potential (SP) model. the gate oxides are also scaled to nanometer regions. This results in the electrons residing in some discrete energy levels which are above the classical energy level by some fixed value of energy as shown in Fig. threshold voltage has been formulated by considering various crystal orientations and the inversion quantization effect in the substrate.

(2) Fig. Therefore. e0 is permittivity of free space. (110) and (111) Oriented Nanoscale MOSFET. 1 – Inversion layer quantization in the substrate for different crystal orientations. The main problem with (3) is that the surface-potential has to be evaluated explicitly in all the regions of electron inversion and then only. (1) q is electron charge.. ϕs is surface potential. ϕ f is fermi potential. Qs = −(2qN a esi e0 )1/2 ⎡ ϕs + Vt e ⎣ − 2 ϕ f Vt (e ϕs Vt −1 ⎤ ⎦ ) 1/2 . ⎜ ⎟⎥ q kT ⎪⎢ ⎪ ⎝ ⎠⎦ ⎣ ⎩ ⎭ (3) γ = (2qN a esi eo )1/2 / Cox is body effect parameter and Cox is the unit area oxide capacitance (Fcm–2). the inversion charge density Qinv is given by (1) and (2): Qinv 1/ 2 ⎧⎡ ⎫ ⎛ q ( ϕ s − 2ϕ f ) ⎞ ⎤ kT ⎪ 1/2 ⎪ exp ⎜ = −γCox ⎨ ⎢ ϕs + ⎟ ⎥ − (ϕ s ) ⎬ . The wave function solution of the Schrödinger’s equation is given by using variation approach [1]: ψ( x) = b3 2 x ⎛ −bx ⎞ exp ⎜ ⎟. and Vt = kT q is thermal potential. the (3) can be solved. the depletion charge Qb is approximated as Qb = −(2esi eo qN a ϕs )1/ 2 . Similarly. esi is silicon relative permittivity.Threshold Voltage Modeling in (100). 2 ⎝ 2 ⎠ (4) where 149 . N a is substrate concentration.. An explicit solution has been evaluated in the [14].

16 γ ) − 40 f + 1⎬ . we can calculate explicitly inversion charge density and depletion charge density. the corresponding shift in the sub band electron energy [1] is 3h 2b 2 Eo = . (7) f = ϕ f + 0. (8) Using the surface potential model (8) in (2) and (3). Roy ⎡ 48π2 m∗ q 2 ⎤ b=⎢ ( 0. The inversion layer charge density further reduce in these orientations. The results show that the inversion layer quantization leads to reduced inversion charge densities in the substrate. After solving the Schrödinger’s equation using variation approach. 2 match quite closely with the reference 1 (BSIM 5 results) [15] and reference 2 results [16]. (5) 8m ∗ and shift in the surface potential “ δϕ ” = Eo q . −1/ 2 2 ⎧ ⎫ −2 a = 0.5 ⎡(ϕswi − 2ϕ f ) 2 + 0. ⎥ ⎦ 2 and ϕswi is the weak inversion surface potential.25γ 2 ) ⎢ ⎣ y = ϕswi − f .A. ϕsqm = ϕs + δϕ = 2ϕ f + δϕ . ⎢ ⎥ ⎣ ⎦ ⎩ ⎭ ϕswi = ⎡(Vgs − V fb + 0. Using (6-8). 150 . The results in Fig. 2 ⎣ esi e0 h ⎦ 13 and m* is effective quantization mass. the quantum depletion charge density (2) and inversion charge density (3) can be evaluated for quantum mechanical case.33Qinv + Qdep )⎥ .5γ ⎤ . and The quantum surface potential is hence given by adding the classical surface potential with the shift in the surface potential due to inversion layer quantization from (5).0016 ⎤ ⎣ ⎦ 1/ 2 .N. (6) The (6) is then included in the explicit surface potential expression given by [13]: ϕs = f + a . 1/ 2 − 0. The shift in the surface potential due to inversion layer quantization in the substrate can hence be calculated from (6). Chaudhry and J.5ϕswi − 0. The situation further worsens as the substrate orientation changes from (100) to (110) and (111). x = Vgs − V fb − f .025 ln ⎨ ⎡ x − y (1 + 100 y 2 ) ⎤ ( 0.

Threshold Voltage Modeling in (100). electrons from the substrate. <110> model. In the case of inversion layer quantization in the substrate. This results in increased threshold voltage. 4 Threshold Voltage The voltage required to turn on the MOSFET is called the threshold voltage. This is called threshold voltage shift. <111> model). Effective quantization Substrate crystal transverse mass (m*) and orientation mo= electron free mass <100> 0. this shift in the threshold voltage is added to the classical threshold voltage to obtain the net threshold voltage. As the gate voltage is increased above the threshold voltage. (110) and (111) Oriented Nanoscale MOSFET. the energy band-gap widens due to the electrons staying at higher energy levels..315mo 0. source and drain start accumulating at the surface of the substrate thus creating a sheet of charge called as inversion layer which is enough for the conduction from the source to the drain. So less electron contribution is available at the same gate voltage. 2 – Simulated Results of quantized inversion charge density in three crystal orientations (lines on the graph from top: Model <100>. An analytical model of this effect has been developed in 151 . in order to invert the channel. Table 1 Variation of effective electron quantization mass with crystal orientation in silicon substrate [1]. more band bending is required and voltage greater than classical voltage has to be applied at the gate terminal.258mo Fig.916mo <110> <111> 0.. So. So.

(10) Vgs = V fb + ϕs + (2ε0 ε si qN b ϕs )1/2 / Cox .5(ε0 ε si qN b )1/ 2 / (ϕ f )1/2 Cox . (14) The shift in the surface potential due to inversion layer quantization is given by (6) which is multiplied with (14) to get the shift in the threshold voltage. (9) total substrate charge = = Qs = −Qb (before inversion). substrate doping variation from 1×1015 cm–3 to 1×1018cm–3. Chaudhry and J. (11) Hence at ϕs = 2ϕ f . before ϕs = 2ϕ f . VT = V fb + 2ϕ f + 2(ε0 ε si qN b ϕ f )1/2 / Cox . Differentiating (10) with respect to surface potential ( ϕs ). Therefore (9) becomes. Using (2) in (10). Therefore. VT is the classical threshold voltage. we get. (12) d Vgs d ϕs d Vgs d ϕs = 1 + 0. (13) Putting the condition ϕs = 2ϕ f .5 nm.N. dϕ dϕ As shown in Fig. The parameters taken are oxide thickness 2.e. Therefore.5(2ε0 ε si qN b )1/ 2 / (ϕs )1/2 Cox . At higher substrate concentrations. Fgs = VT . the shift in the threshold voltage is d Vgs dV (15) δϕs = T δϕs . 3. the shift in the threshold voltage increases and is of the order of 50 mV and more. The gate to source voltage is given by: Vgs = V fb + ϕs + Qs / Cox . threshold voltage shift in the crystal orientations (110) and (111) is larger as compared to (100) crystal orientations. Vgs = V fb + ϕs − Qb / Cox . Here the inversion charge density is negligible and hence can be ignored. The results have been compared for the (100) crystal orientation with the numerical results as predicted in reference 3 [17]. Various crystal orientations have been taken to study the impact of inversion layer quantization. The threshold voltage is defined for the depletion potential at the onset of inversion i. This shows the extent of impact of inversion layer quantization on the (110) and (111) crystal orientations making them less useful for the nanoscale MOSFETs. 152 .A. we get = 1 + 0. we get. Roy this section.

12. Fig. 5.Threshold Voltage Modeling in (100).. The threshold voltage with inversion layer quantization analytically derived shows that crystal orientations also have a significant effect on the threshold voltage and inversion charge density of MOSFET at nanoscale levels. The authors would like to thank to Panjab University.5nm. The case of three crystal orientations viz. Physics Review B. Vol. India for providing excellent research environment to complete this work. 3 – Threshold voltage shift with substrate doping in the presence of inversion layer quantization in the substrate at oxide thickness of 2. Eng-Siew Kang of University Teknologi Malaysia for useful discussions on the topic. India for allowing to carry out the work. Chandigarh. 4891 – 4899. Stern: Self-consistent Results for n-Type Si Inversion Layers. (110) and (111) has been taken. (110) and (111) Oriented Nanoscale MOSFET. Chandigarh. 153 . (100). 6 Acknowledgements The authors thank the Director. June 1972 pp. The authors wish to thank all individuals who have contributed directly or indirectly in completing this research work. We are grateful to Ms. 5 Conclusion Quantization effects on carrier distribution in substrate are studied based on the variation approach. No. 7 [1] References F. Panjab University. UIET..

Wan. pp.W. M. No. Vol. M. 10. M. Xi.H. J. Vol. X . Puerto Rico. C.C.1. Vittoz: An Analytical MOS Transistor Model Valid in All Regions of Operation and Dedicated to Low-voltage and Low-current Applications. No. Hu: BSIM5 MOSFET Model. Suetake. pp. WA. Cai: SP: An Advanced Surface-potentialbased Compact MOSFET Model. S. 2000. H. Roy [2] [3] D. 2010. No. N. 1997. Vol. pp. G. 920 – 923. pp. Analog Integrated Circuits and Signal Processing. J. H. May 1994. International Conference on Simulation of Semiconductor Processes and Devices. University of California. pp. pp.3. Solid-state Electronics. USA. No.L. Vol. Lin. Chaudhry and J. R. SpringerVerlag. Chen. NJ. F.Gu. D. Principles and Practice. No. 2004. Berkeley. B. Chan. pp. T.N. K. 185 – 193. No. Analog Integrated Circuit Signal Processing. Chan. He. C. 75. 9. User’s Manual. No. F. Vol. Seattle. Dunga. A. China. 3. Gildenblat. B. C. CA. March 2000. 723 – 732. 433 – 444. March 2007. He. Foty: Perspectives on Analytical Modeling of Small Geometry MOSFETs in SPICE for Low Voltage/Low Power CMOS Circuit Design. Klaassen: An Explicit Surface-potential-based MOSFET Model for Circuit Simulation. Cheng. Niknejad. T. Vol.1. K. Chen. 409 – 418. Suematsu. 229 – 252. No. International Conference on Simulation of Semiconductor Processes and Devices. J. 109 – 112. [4] [5] [6] [7] [8] [9] [10] [11] [12] [13] [14] [15] [16] [17] [18] 154 . J. 3. 10. Wang. Prentice Hall. April 2007. Hu: ICM – An Analytical Inversion Charge Model for Accurate Modeling of Thin Gate Oxide MOSFETs.A. N. Vol. 83 – 114. N.V. A. Cambridge. E. International Conference on Modeling and Simulation of Microsystems. M.H. 1819 – 1825. K. R. 2004. HiSIM 1. A.A. Z. Wan. Oct. pp. S. Miura-Mattausch. 1997. M. 2003. 1394 – 1406. Sept. Mattausch. Odanaka. Miura-Mattausch. S. Vol. USA. 47. 5186 – 5190.H. 39. Nakatama: HiSIM: Self-consistent Surface-potential MOS-model Valid Down to Sub-100 nm Technologies. San Juan. International Conference on Solid-state and Integrated Circuits Technology. Dutton. Gamiz. 261 – 264.B. H.based Advanced MOSFET Model for Circuit Simulation with Easy Parameter Extraction. Nov. Nakayama: HiSIM: A Drift Diffusion. H. 54. Kumashiro. Kiehl: Circuit/device Modeling at the Quantum Level. Sept. Yamaguchi. Yu. Yamaguchi. pp.1 User’s manual. J.P. Vol. 44. USA. USA. Majkusiak: Influence of Carrier Energy Quantization on Threshold Voltage of Metal Oxide Semiconductor Transistor. 51. Semiconductor Technology Academic Research Centre. 3. NJ. Mattausch.N. Janik. IEEE Transactions on Electron Devices. Ueno. IEEE Transactions on Electron Devices. pp. Foty: MOSFET Modeling with SPICE. July 1995. M. Langevelde. No. R. pp.0 MOSFET Model. Y. 2. X. IEEE Journal of Solid-state Circuits. 1. 2. Beijing. Arora: MOSFET Models for VLSI Circuit Simulation – Theory and Practice. Chaudhry. 678 – 681. Nikneja: BSIM 5: An Advanced Charge based MOSFET Model for Nanoscale VLSI Circuit Simulation.M. pp. Serbian Journal of Electrical Engineering. Journal of Applied Physics. USA. 7. T. Kumashiro. M. MA. BSIM4. Dunga. 2000. Rodriguez. April 2002. Roy: A Comparative Study of Hole and Electron Inversion Layer Quantization in MOS Structures.J. T. 4.C. Solid-state Electronics. N. 1999. Enz. Dec.M. pp. X. X. Imai. Xi.H.P. Oct. Sept. 21. Roldan: Modeling of Inversion Layer Centroid and Polysilicon Depletion Effects on Ultrathin-gate-oxide MOSFET Behavior: The influence of Crystalline Orientation. 1993. F. Krummenacher. Department of Electrical Engineering and Computer Sciences. H. 8. Nagakura. J. Vol. Vol.

Sign up to vote on this title
UsefulNot useful