You are on page 1of 27

Schrödinger-Poison Solvers Description and

Quantum Corrections in classical simulators:
1. Schrödinger-Poisson solvers
- major concerns for future integrated circuits
- time-dependent Schrödinger wave equation (SWE)
- discretization of the SWE for variable effective mass
- time-independent Schrödinger equation
(A) Airy functions method
(B) Variational approach
(C) Shooting method for 1D problems – SCHRED description
(D) Finding the roots of a characteristic polynomial
(E) The Lanczos algorithm
(F) Numerov algorithm – note on the integration of the SWE
(G) 2D and 3D eigenvalue solvers

- open systems
2. The effective potential approach
- Madelung and Bohm’s reformulation of quantum mechanics
- other quantum potential formulations
- the effective potential approach due to Ferry
- simulation examples that utilize the effective potential approach
(A) Simulation of a 50 nm MOSFET device
(B) Simulation of a 250 nm FIBMOS device
(C) Simulation of a SOI device structure

- conclusions regarding the use of the effective potential approach

Computational Electronics

1. Schrodinger-Poisson Solvers
1.1 Major concerns for future integrated circuits

‹

While current production devices are only at 0.1 µm, predictions are
that they will be at 50 nm by year 2007. Issues that are currently
investigated in these device structures include:
❤ Tunnel currents in gate oxide
❤ Source to drain tunneling
❤ The role of the electron-electron interactions on carrier thermalization
❤ Statistical fluctuations: gate-length, gate-width, oxide thickness and
doping density variation:

‹

❣ Must go 3D instead of 2D
❣ Must model ensemble of devices instead of a single device
❣ Dopant solubility, activation and segregation at the surface
must be addressed

At these sizes, we should begin to see quantum effects, as λD~3-5
nm at 300 K. The questions are:
❤ How large is the electron wave packet?
❤ How do we include space quantization effect into classical device
simulators?
Computational Electronics

1

Example 1: Degradation of Ctot for Different Device Technologies
Gate

1
0.9
0.8

Cox =

Ctot/Cox

Cpoly
ox

t ox

Cdepl

Cinv

T=300 K, NA=1018 cm-3

0.7

classical M-B, metal gates

0.6

classical F-D, metal gates
quantum, metal gates

0.5

quantum, poly-gates N =6x1019 cm-3

0.4

D

quantum, poly-gates N =1020 cm-3
D

0.3

20

2

3

4

9

5

6

7

8

10

Oxide thickness t [nm]

Substrate

C tot =

cm

D

0.2
1

-3

quantum, poly-gates N =2x10

ox

Cox
1+

C ox
Cox
+
Cpoly Cinv + Cdepl

Cox
1+

Cox
C
+ ox
Cpoly Cinv

Conclusion: Quantization effects must be taken into account to properly
describe the operation of future MOS devices.
Computational Electronics

Example 2: Heterojunction devices
• Esaki and Tsu (1969) => ionized donors can be spatially
separated from the free electrons using modulation doping
• Dingle (1978) => grew the first modulation doped
GaAs/AlGaAs superlattice

• The impact of the heterostructure is that the potential is abrupt,
which leads to a more vigorous control of the channel charge
density by the externally applied voltage.
Computational Electronics

2

t ) = exp − Ht ψ(r.8 Conduction band profile Ec 0.0 Fermi level EF -0.3 0. more complicated wavefunctions can be constructed as superpositions of plane waves: ψ(r.2 0.t ) = ∫∫∫ dkψ 0 (k ) exp{i [k ⋅ r − ω(k )t ]} • The evolution of the wavefunction is described by the timedependent Schrödinger equation: ih 2 ∂ψ(r.7 T = 4.4 0.t ) ⇒ ψ(r.2 E0 0.2 Time-dependent Schrödinger equation • One of the goals of the quantum mechanics is to give quantitative description on a macroscopic scale of individual particles which behave both like particles and waves.6 EF-E0 = 12 meV 0.1 µm ud -GaAs substrate SI GaAs Semi .00 0.04 0.4 0. The simplest wavefunction is a plane-wave: ψ = ψ 0 exp[i (k ⋅ r − ωt )] .08 0.2 0.10 y-axis [µm] 0.1 0 EF 20 40 60 80 100 120 140 160 y-axis [nm] Computational Electronics 1.0) ∂t 2m  h  Computational Electronics 3 . p = hk and E = hω • To describe realistic situations.insulating GaAs substrate Nd1=1x1012 cm-2 Si Conduction band edge [eV] Nd2=3x1012 cm-2 Si Conduction band [eV] 0.6 Energy of the ground subband E0 0.5 nm GaAs (cap layer) 200 Å GaAs undoped 40 nm Si -doped Al x Ga As 1-x (barrier layer) 4x50 Å AlGaAs undoped 15 nm Al xGa Nd1 250 Å AlGaAs undoped Nd2 As (spacer layer) 1-x 400 Å AlGaAs undoped 1 µm GaAs undoped 0.02 0.1 0 -0.t ) h 2  i  =− ∇ ψ(r.2 K 0.5 0.06 0. t ) + V (r )ψ(r.

i. one can adopt the SWE with varying effective mass.e. Hψ = − h2  1  ∇⋅ ∇ψ  + Vψ 2 m *  For uniform mesh size. There are two ways one can make the Hamiltonian of the system to be Hermitian: (a) Bring the effective mass inside the differential operator. the discretized version of Hψ is: − h 2  ψ i +1 − ψ i ψ i − ψ i −1  + Vi ψ i − 2 * mi*−1/ 2  2∆  mi +1/ 2 Computational Electronics 4 . one solves the tri-diagonal system of equations ih∆t ih∆t k 2ψ j = − χ − χ   2 j +1 2 j −1 K K 4 4 mh mh 2  − 1ψ = χ − ψ ⇒  i∆t  ih∆t 1+ i H∆t  V  χ j +1 + 1 + +   2h 2 2h j  2mh  [ ] and then calculates the value of the wavefunction at time-step (k+1) as: k +1 k k ψ = χ −ψ Computational Electronics 1.3 Discretization of the SWE for variable effective mass • Under the assumption of slowly varying material composition.• A stable and norm-preserving discretization scheme for the timedependent SWE is the Crank-Nicholson semi-implicit scheme: i 1 − H∆t K +1 K ψ −ψ i K +1 K K +1 K 2h =− Hψ + Hψ ⇒ ψ = ψ i 2h ∆t 1 + H∆t 2h • In the actual implementation.

one has a boundary value problem where the energy of the wanted solution is specified and one has to specify the spatial variation of ψ(r) • For 1D confinement (triangular or rectangular quantum well problems) one can use either: . then: ψ(r. it gives the same discretized form of the equations as the fist method i +1/ 2 i +1/ 2 1 ∂ 2ψ i +1/ 2 ∂ψ ∂  1  1 ∂ψ + = dz ∫ ∫   2 i −1/ 2 m * ∂z i −1/ 2 ∂z ∂z  m *  m * ∂z i −1/ 2 i +1/ 2 i +1/ 2 ∂ψ ∂  1  i +1/ 2 ∂ψ ∂  1  1 ∂ψ −  ∫ + ∫  = i −1/ 2 ∂z ∂z  m *  i −1/ 2 ∂z ∂z  m *  m * ∂z i −1/ 2 Computational Electronics 1. the SWE is an eigenvalue problem (b) For an open system.analytical approaches .4 Time-independent Schrödinger equation • If one is considering stationary states.(b) Another Hermitian operator proposed for variable mass has the form: h2  1 2  1  ∇ ψ + ∇2  ψ  + Vψ  4 m *  m *  h 2  1 ∂ 2ψ ∂ψ ∂  1  1 ∂ 2  1   =−  +  + ψ    + Vψ 2  m * ∂z 2 ∂z ∂z  m *  2 ∂z 2  m *   Hψ = − If one applies the box-integration method to the first two terms of this Hamiltonian.t ) = ψ(r ) exp( −iωt ) = ψ(r ) exp( −iEt / h) • The time-independent SWE then reduces to: − h2 2 ∇ ψ(r ) + V (r )ψ(r ) = Eψ(r ) 2m * (a) For a system confined by a potential well.numerical approaches Computational Electronics 5 .

• Summary of approaches used to solve the SWE Triangular well (analytical approaches) 1D Numerical Approaches 2D and 3D Airy functions method Variational Approach 1. Shooting method 2. Implicitly restarted Arnoldi method (ARPACK) Computational Electronics (A) Airy functions method • Suppose. Housholder method 2. n = ∑ Ni ψ i2 (z) ε A i • The analytical solution of the Poisson equation is of the form: e 2N AW ε z   z  e2 2 z 1 − ∑ Ni z − ∫ (z − z’ )ψ i (z’ )dz’ +  2W  ε i  0  2 e Ns z av 2εEd W = . Lanczos algorithm 1. Ed = (Ec − EF )bulk + (E F − E0 ) + E0 − ε e 2N A V ( z) = Computational Electronics 6 . Finding the roots of a characteristic polynomial 3. we want to solve self-consistently the 1D SchrödingerPoisson problem in a MOS structure: h 2 ∂ 2ψ(z) + V (z )ψ(z ) = Eψ(z) − 2m * ∂z 2 2 ∂ V ∂z 2 =− e (N + n ) .

In this case. the Airy functions approach fails.• When the inversion charge density does not make significant contribution to the band-bending. at low temperatures one can use the variational approach due to Fang and Howard. Fs = 2 ε 2m * ∂z • The solutions are:  2m * eEs  E   z − i   ψ i (z) = Ai  2 eFs     h 1/ 3  h2   Ei ≈   2m *     3 πeFs  2   3   i +   4  2/3 Computational Electronics (B) Variational Approach • When the electrons significantly affect the band bending near the interface. in which the ground state wavefunction is assumed to be of the form: 3 − bz / 2 ψ 0 (z) = b2 ze • The energy of the lowest subband is: < Eo >=< T > + < Vd > + < Vs > + < VI > h 2b 2 8m * 2 2 3e N AW 6e N A − 2 εsc b ε sc b 2 33e Ns 16εsc b εsc − εox e 2b εsc + εox 32πε sc Computational Electronics 7 . The 1D SWE reduces to: − eN AW h 2 ∂ 2ψ(z ) + eFs zψ( z) = Eψ(z).

it follows that: 12m * e 2N *  11 b= .• When one minimizes the total energy per electron. where N * = Ndepl + 32 Ns 2  ε sc h  N A = 1015 cm −3 Ns = 1012 cm −2 T = 0K Computational Electronics (C) Shooting Method for 1D problems • To describe the shooting method. lets rewrite the timeindependent SWE in the form: 2m * d 2ψ 2 [E − V (z)] + k ( z)ψ( z) = 0 . one gets that: ( 2 2 ) ψ i +1 − 2 − k i ∆ ψ i + ψ i −1 = 0 • The schematics of the potential used in the discussion is: Computational Electronics 8 . k (z) = 2 h2 dz • When discretized on a uniform mesh.

196m0 ∆∆4-band: 4-band: 1/2 mm⊥=m =mt=0.196m =0. one has to solve self-consistently the 1D Schrödinger equation self-consistently with the 1D Poisson equation. m||=mt=0. • For the case of Si.916m0. in the solution of the 1D Schrödinger equation one must worry about the mass anisotropy Ε14 Ε13 Ε12 Ε11 ∆4-band Ε22 Ε21 EF ∆2-band VG>0 z-axis [100] (depth) ∆∆2-band: 2-band: mm⊥=m l=0.196m0 ⊥=m l=0. one matches the solutions ψ< and ψ> (4) The eigenvalue is then signaled by equality of the derivatives at zm dψ < dz = zm dψ > dz ⇒ f = zm [ 1 < > ψ ( zm − ∆ ) − ψ ( z m − ∆ ) ψ ] Computational Electronics (C1) Description of SCHRED • For actual device simulations. mm||==(m (ml mmt) )1/2 ⊥ t 0 || l t Computational Electronics 9 . .• The solution strategy is the following one: (1) Integrate the SWE towards larger z from zmin 2 2 ψ i +1 = (2 − k i ∆ )ψ i − ψ i −1 (2) Integrate the SWE towards smaller z starting from zmax 2 2 ψ i −1 = (2 − k i ∆ )ψ i − ψ i +1 (3) At the matching point zm. m||=mt=0.196m0 .916m0.

e_nsub1=4.tox[nm]=1.Ea[meV]=45. ionize=no.Ed[meV]=45 Ed[meV]=45 Vmin[V]=0. Vmin[V]=0. toleranc=5. exchange=no. Nb[cm-3]=1.edu/labs/punch/) Computational Electronics (2) Sample input file used in SCHRED simulations: device device params params body body gate gate ionize ionize voltage voltage charge charge ++ calc calc save save ++ converge converge bulk=yes bulk=yes temp[K]=300.dat toleranc=5. file_ch=chrg.file_cv=cv.5. charges=no.e-6.1 Vstep[V]=0.dat file_wf=wfun. Nb[cm-3]=1.dat. Vmax[V]=2. file_ch=chrg.dat charges=no.9 kox=3.ecn.0e19 ionize=no. tox[nm]=1. e_nsub1=4.e19 metal=false. wavefunc=no. max_iter=2000 max_iter=2000 Computational Electronics 10 .edu/labs/punch/) (http://www.kox=3.5.ecn.purdue. Fermi=yes.Vmax[V]=2.Ng[cm-3]=-6. wavefunc=no. metal=false. Ea[meV]=45.(1) Existing Features of SCHRED: Classical and quantum-mechanical charge description Fermi-Dirac and Maxwell-Boltzmann statistics Single-valley and multiple-valley conduction bands Exchange and correlation corrections to the ground state energy of the system Metal and poly-silicon gates Partial ionization of the impurity atoms å  ê ‚ ƒ Features Recently Being Implemented in SHRED: Á Hole quantization Non-parabolicity of the bands Á Current Currenthome homeof ofthe thesolver: solver: Purdue PurdueSemiconductor SemiconductorSimulation SimulationHub Hub (http://www.e_nsub2=2 e_nsub2=2 CV_curve=yes.dat file_cv=cv.file_wf=wfun. quantum=yes.Vstep[V]=0.e19 uniform=true.1 quantum=yes.5.purdue.dat. temp[K]=300.9 uniform=true.0e19 Ng[cm-3]=-6. Fermi=yes.exchange=no.e-6. CV_curve=yes.5.

Technol.06 ➯ Contracted wavefunctions first subband Thick (thin) lines correspond to the case when the exchange-correlation corrections are included (omitted) in the simulations.18 Exchange-Correlation Correction: ➯ Lower subband energies ➯ Increase in the subband separation ➯ Increase in the carrier concentration at 0. Phys. J.14 which the Fermi level crosses into the second subband second subband 0. B 14.. T=0 K) Veff Energy [meV] 0.5x10 12 2x10 Ns [cm -2 ] 12 2. Ns=4x1012 cm-2. 0. data [Kneschaurek et al.02 0 20 40 60 Accounts for the reduction of the Ground State Enery due to the inclusion of the Pauli Exclusion Principle 80 100 Distance from the interface [Å] Computational Electronics (4) Sample of simulation results obtained with SCHRED (Cont’d): Kneschaurek et al. B 13.5x10 12 3x10 Transmission-Line Arrangement Computational Electronics 11 .2 K. 1610 (1976) Energy E 10 [meV] 60 T = 4.8x1015 cm-3.] Veff(z)=V (z)+V (z)+V (z) 40 H im SiO2 Veff(z)=V H(z) 30 Al-Gate LED exc far-ir Veff(z)=V H(z)+V im (z) Si-Sample radiation 20 10 Vg 0 0 11 5x10 12 1x10 12 1.1 Normalized wavefunction 0. Ndepl=10 50 11 -2 cm Infrared Optical Absorption Experiment: Exp. Rev. Vac. 1841 (1995) (Na=2.(3) Sample of simulation results obtained with SCHRED: E = E HF + E Total Ground State Energy of the System = E HF + E HF +E kin exchange corr corr Accounts for the error made with the Hartree-Fock Approximation Hartree-Fock Approximation for the Ground State Energy Vasileska et al.. Sci.

n −1Pn − 2 (E ) Computational Electronics 12 . P2 (E ) = ( A2.2 − E )P1(E ) − A12.i ψ i + Ai .n − E )Pn −1(E ) − An2.i −1ψ i −1 + Ai .(4) Sample of simulation results obtained with SCHRED (Cont’d): Computational Electronics (D) Finding the roots of a characteristic polynomial • The finite-difference approximation of the 1D SWE is given by: d 2ψ + V ( z)ψ(z ) = Eψ(z) dz 2 1 ψ + Vi − 22 ψ i + 12 ψ i +1 = Eψ i 2 i −1   ∆ ∆  ∆ ⇓ Ai .2 .i +1ψ i +1 = Eψ i • The eigenvalues are the zeroes of the Nth degree characteristic polynomial of A N PA (E ) = det(A − EI) = Π (E i − E ) i =1 ⇓ P1(E ) = A1. Pn (E ) = ( An.1 − E .

in which one starts with an initial guess for the eigenvector ψn(1) associated with a given eigenvalue En and refines the guess by evaluating: ( 2) (1) ψ n = [A − (E n + δ)I]−1ψ n Computational Electronics (E) The Lanczos algorithm • This algorithm is very suitable when one is interested in many of the lowest eigenvalues.i − ∑ Ai . i  i  j ≠i j ≠i   (3) Once the eigenvalues are determined. j . in which A is explicitly tri-diagonal matrix (1) Choose an arbitrary first vector in the basis ψ 1 such that: T ψ1 ψ1 = 1 (2) One then forms a second vector in the basis as: ψ 2 = C 2 ( Aψ1 − A11ψ1) [ ] −1/ 2 2 A11 = ψT1 Aψ1 . P2(E). the eigenvectors are found by using the inverse vector iteration procedure.• Several features help considerably in finding the roots of PA: (1) The number of times the sequence 1. C2 = (Aψ1 )T (Aψ1 ) − A11 (3) Subsequent vectors in the basis are then constructed recursively: ψ n +1 = Cn +1( Aψ n − Ann ψ n − Ann −1ψ n −1) [ 2 2 Cn +1 = (Aψ n )T (Aψ n ) − Ann Ann −1 ] −1/ 2 Computational Electronics 13 . j . a guidance can be the Gerschgorin’s bounds:     En ≥ min  Ai . En ≤ max  Ai . • The strategy is to construct a set of orthonormal basis vectors {ψ n}.PN(E) changes sign equals the number of eigenvalues less than E (2) To make a systematic search for all of the eigenvalues. P1(E).i + ∑ Ai .….

Computational Electronics (F) Numerov Algorithm – note on the integration of the SWE • There is particularly simple and efficient method for integrating second-order differential equations called Numerov or Cowling’s method. Suppose that one is interested in the 10 lowest eigenvalues of a matrix of dimension 1000.• The real utility of the Lanczos method is when many. of the eigenvalues of a large matrix are required. -> diagonalizing the tridiagonal matrix -> using the normalized sum of the lowest eigenvectors as the first vector in the next basis converges to the required eigenvalues and eigenvectors. The procedure is then the following: (1) Generate some number of states in a basis larger than the number of eigenvalues being sought (2) A linear combination of these eigenvectors is then used in constructing a new basis of dimension 25 (3) The continued procedure of: -> generating a limited basis. but bot all. based on the idea sketched below: ψ i −1 − 2ψ i + ψ i +1 ∆2 = ψ i’’ + ψ i’’’’ 12 ∆2 k 2ψ i −1 − 2 k 2ψ i + k 2ψ i +1 ∂ 2  ∂ 2ψ  ∂ 2 = − k 2ψ = − ∂x 2  ∂x 2  ∂x 2 ∆2 2 2    5∆2 2    1 + ∆ k 2 ψ 1 − ψ + 1 + ∆ k 2 ψ 2 k =0 −  12 i +1  i +1  12 i  i  12 i −1  i −1    ( ) ( ) ( ) ( ) • Solving the linear system of equations for either ψi-1 or ψi+1 then provides a recursion relation for integrating either forward or backward Computational Electronics 14 .

rice. one needs to solve self-consistently the Poisson equation: e ∇ 2ϕ = − (ND − N A − n ) ε and the 1D Schrödinger equation: h 2 ∂ 2ψ + Ec ( x )ψ( x ) = Eψ( x ) 2m ∂x 2 ⇓ 2 2 ∂ ψ 2m ∂ ψ + 2 [E − Ec ( x )]ψ( x ) = 0 ⇒ 2 + k x2 ( x )ψ( x ) = 0 2 ∂x ∂x h − Computational Electronics 15 . 91. Appl.E. based on Implicitly Restarted Arnoldi Method. Laux. pp. and F. Stern.(G) 2D and 3D Eigenvalue Solvers • When solving 2D eigenvalue problems. 1986). one can transform the matrix to tri-diagonal form by using either the Householder method.caam.edu/software/ARPACK Computational Electronics 1. Lett.caam.rice.5 Open Systems To calculate the conduction-band edge of the RTD structure. • For 3D problems. 49. Lanczos iteration or Rayleigh quotient iteration method (S. Phys. the best eigenvalue solvers of large and sparce matrices. can be found in the publicly available software package ARPACK: -> the codes are available by anonymous ftp from ftp.edu -> or by connecting directly to the URL http://www.

1.• For a 1D domain limited to x∈[0.L].…. Computational Electronics 16 . ψ −1 = I0 exp( − jk0 ∆ ) + R0 exp( jk0 ∆ ) ψ exp( jk0 ∆ ) − ψ −1 I0 = 0 i 2 sin( jk0 ∆ ) • The procedure for inflow from the right boundary is identical:   2m∆2 ψ i +1 = 2 + 2 (Vi − E )ψ i − ψ i −1 h   and is started assuming ψ 0 = 1 and ψ −1 = exp(− jk0 ∆ ). N-1: [ ] ψ i −1 = 2 − ∆2ki2 ψ i − ψ i +1 (3) For i=0. one has that: ψ 0 = I0 + R0 . the SWE discretized on a uniform grid. the procedure is the following one: (1) one sets A(L)=1 at the output boundary to get that: ψ N = exp( jkN L ) ψ N +1 = exp[ jkN (L + ∆ )] = ψ N exp( jkN ∆ ) Computational Electronics (2) The next step is to calculate the wavefunction for i=0. with mesh size ∆ is: ψ i −1 − 2ψ i + ψ i +1 2 + ki ψ i = 0 ∆2 • The traveling wave at a specified energy is assumed to be of a plane-wave type of the form: ψ ( x ) = A( x ) exp[± jxk ( x )] • For a traveling wave that enters the open system at x=0.

The effective potential approach 2. r. one can proceed with the calculation of the current through the structure: J(x) = − ∂ψ ( k j ) ∂ψ * ( k j )   eh − ψ (k j ) ∑ n( k j )ψ * ( k j )  ∆k 2πm k ∂x ∂x   j • The electron density is then given by: n( x ) = 21π ∑ n( k j )ψ * (k j )ψ (k j )∆k kj Computational Electronics 2. t )2 . t )   1 + ∇ ⋅  ρ ∇S  = 0.• To find the total wavefunction from the renormalized results of the two recursions steps (from the left and from the right). = ( 2) − ∂t 2m ↓ (1) Q(ρ. v = ∇S/m ∂t   m ∂S(r. r. t ) + Q(ρ. t ) = R(r. one needs to specify the INJECTION conditions: +∞ dk n( x ) = ∫ −∞ x n(k ) where x 2π n(kx ) =   E − ECL − Ex   ln1+ exp F   kBT πh2    m • Once the wavefunction is determined. Hamilton Jacobi eq. t ) = − m h2 1 2 h 2 −1 / 2 2 1 / 2 ∇ R=− ρ ∇ ρ 2m R 2m dv = −∇(V + Q ) = fc + fq dt Computational Electronics 17 . R = n = ψ − ∂ψ h2 2 ∇ ψ + Vψ = ih 2m ∂t Š The resultant real and imaginary parts give: ∂ρ(r. ρ(r. t ). t ) 1 2 (∇S ) + V (r.1 Madelung and Bohm’s Reformulation of QM Š The hydrodynamic formulation is initiated by substituting the wavefunction into the time-dependent SWE: ψ = ReiS / h .

2. Ω) = ln − a + Va 2 ( x’) β βΩ / 2 2 Computational Electronics (c) The minimization with respect to Ω and the minimization with respect to a2 then give: a2 = 1  βΩ  βΩ   coth   − 1 2  2 βΩ   2   and Ω 2 ( x0 ) = 2 ∂ ∂a 2 Va 2 (x0 ) Š Gardner and Ringhofer derived a smooth quantum potential for hydrodynamic modeling that is valid to all orders of h2.2 Other Quantum Potential Formulations Š An alternate form of the quantum potential has been proposed by Iafrate. and is based on moments of the WignerBoltzmann equation: VQ = − h 2 ∂ 2 (ln n ) 8m ∂x 2 Wigner potential or densitygradient correction Š Ferry and Zhou derived a form for a smooth quantum potential based on the effective classical partition function of Feynman and Kleinert. Grubin and Ferry. a 2 . The Feynman and Kleinert idea is as follows: (a) Calculate the smeared version of the potential V(x) as follows: Va 2 ( x) = ∫ dx’ (2πa ) 2 1/ 2  1  exp − 2 ( x − x’)2 V ( x’)  2a  (b) Introduce a second parameter Ω and form the auxiliary potential: 1 sinh(β Ω / 2) Ω 2 2 ~ W ( x. x) = ∫ Computational Electronics 18 . that involves smoothing integration of the classical potential over space and temperature:   2mβ   ∫ d 3 x ’ × 2 0 β β  π(β − β’)(β + β’)h   2mβ 2 exp− ( x ’ x ) − V (x’) 2  π(β − β’)(β + β’)h  β dβ’ β’ 2 V (β.

typically the ∂x dependence of the density upon the 2 2α 2 ∂ ln( n / n 0 ) potential is as a factor exp(-βV) Veff ( x) = V ( x ) − + ...... Microstruc. 59 (2000).. the effective role of the potential can be rewritten in terms of the non-local density as (Ferry et al... Ferry. V = ∫ drV (r )∑ ni (r) i  r − r’2  δ(r ’− ri ) ~ ∫ drV (r )∑ ∫ dr ’exp −  α 2  i   r − r ’2   ~ ∑ ∫ drδ(r − ri ) ∫ dr ’V (r ’) exp −  α 2  i  ~ ∑ ∫ drδ (r − ri )Veff (r) Effective potential approximation Quantization energy i “Set back” of charge -quantum capacitance effects Smoothed. 27. Superlatt. + V ( x ) + ξ 2 2 ∂x ∂x 2π α −∞   2 ∂V Veff ( x) = V ( x) + α 2 2 + . K. In semiconductors. e −ξ / 2α dξ .2. in press. 2 β n ∂x Q= 2 h2 ∇ ψ 2m* ψ Within a factor of 2... Computational Electronics • The connection of Ferry’s Approach to the Bohm Potential is given below: Veff ( x ) = ∞ 1 2π α ∫ V ( x + ξ )e −ξ 2 / 2α 2 dξ −∞ ∞   2 2 ∂V ξ 2 ∂ 2V + .3 The Effective Potential Approach due to Ferry In principle.1): Built-in potential for triangular potential approximation. β ∂x 2 ≅ 1 ∫ = V (x) − 2α 2 ∂ 2 n + . but is more commonly known as the Bohm potential Computational Electronics 19 . VLSI Design. effective potential Classical density 1 D. the second term is now recognized as the density gradient term.

SCHRED 1e17 .5 0 1.SCHRED 1e17 .5 nm.5 1 0.5 (b) 1e17 .0 3.Effective Potential 3 2.0 5.classical 1e17 .0 2.0 3.0 Gate voltage V 4.0 2.0 g 5.classical 1e17 . Computational Electronics 20 .classical 1e18 .0 [V] The Gaussian fitting parameter a0 = 0.5 10 13 1 10 13 5 10 12 0 10 (a) 1e17 . Computational Electronics (2) Validation of the approach on the example of a MOS capacitor with tox = 6 nm – results for the average carrier displacement: Average distance [nm] 4 3.5 2 1.Effective Potential 1e18 .Effective Potential 1e18 .Effective Potential s Inversion charge density N [cm -2] (1) Validation of the approach on the example of a MOS capacitor with tox = 6 nm – sheet density results: 0 1.0 Gate voltage V [V] g The Gaussian fitting parameter a0 = 0.SCHRED 1e18 .5 nm.classical 1e18 .SCHRED 1e18 .2 10 13 1.0 4.

8 µm Junction depth = 36 nm Substrate doping: NA=1018 cm -3 • Doping of the source-drain regions: ND = 1019 cm -3 Computational Electronics (1) Conduction Band Profile Energy [eV] Distance [nm] VG = VD = 1 V source drain Distance [nm] Without Effective Potential Veff With the Effective Potential Veff Computational Electronics 21 .4 Simulation examples that utilize the effective potential approach (A) Simulation of a 50 nm MOSFET Device Device specification: gate source N+ SiO2 drain N+ channel p-type substrate • • • • • Oxide thickness = 2 nm Channel length = 50 nm Channel width = 0.2.

without V D Average distance [nm] -2 Sheet density [cm ] Electron sheet density reduction due to quantum-mechanical band-gap widening effect 10 50 100 Distance [nm] eff D eff V =1 V.4 V. without V 0 40 12 eff V =0. with V 15 150 Additional 2-3 nm charge set-back from the interface due to quantummechanical space-quantization effect Computational Electronics 22 .(2) Electric field profile Distance [nm] Distance [nm] Ex – field [V/cm] source drain source Ey – field [V/cm] drain Distance [nm] Distance [nm] Negative electric field responsible for charge setback from the interface With Veff Without Veff Computational Electronics (3) Carrier density and average distance Applied Appliedbias: bias: VVGG==11VV 20 10 13 V =0.4 V. with V D 10 eff 5 50 60 70 80 90 100 110 Distance [nm] Without the effective potential With the effective potential 0 D V =1 V.

2 Š The shift in the threshold voltage 0. 0.without Veff 0.0 .6 VG=1.with Veff 0 0.7 Gate voltage V [V] G 1.2 0.6 0.(4) Transfer and output characteristics Š The observed threshold voltage shift D Drain current I [mA/µm] 0.5x10 7 1x10 6 5x10 0 0 50 100 150 1.without Veff 0.05 0 0 0.2 0.1 leads to a reduction in the on-state current. 0.8 VG=1.4 0.4 1.with V G 0.6×1018 cm- Substrate doping 1016 cm-3 Implant length 70 nm 3 P Computational Electronics 23 .2 0 Distance [nm] eff VG=1.5 0.15 0.3 0.2 .0 .5x10 V =1V G VD = 1 V 7 1.2 [V] (B) Simulation of a 250 nm FIBMOS device Source and drain length 50 nm Source and drain junction depth 36 nm Gate length 250 nm Device width  Bulk depth 400 nm Oxide thickness 5 nm Source and drain doping 10×19 cm-3 Substrate doping 1016 cm-3 Implant doping 1.8 D 1 1.2 1 0.2 .25 0.35 is mainly due to the degradation of the device transconductance and is not due to mobility degradation in the channel.6 Drain voltage V Computational Electronics 0.3 0.1 0.6 7 without Veff with Veff 2x107 Drain current I D [mA/µm] Velocity [cm/s] 2.4 V =1.4 0.4 without effective potential with the effective potential 0.

(1) Concentration of electrons in the channel Average electron displacement from the interface and the total channel width increase when quantum effects are included! Computational Electronics (2) Sheet electron density When Quantum effects are included: Š Š Š sheet density is reduced inversion more difficult to achieve threshold voltage is increased Computational Electronics 24 .

Computational Electronics (4) Device transfer and output characteristics With quantum effects: Š Drive current is lowered Š Threshold voltage is higher Š Transconductance is degraded Computational Electronics 25 .(3) Energy and velocity along the channel Š Average energy increases and velocity decreases when quantum effects are included. Š Prominent velocity overshoot evidence of non-stationary transport.

y .(C) Simulation of a SOI Device Structure Experimental data1: Gate oxide (34nm thick) Threshold voltage [V] Gate Thin SOI layer (7nm) Channel width [nm] n+ A pn+ B A B ˆ Simulate this structure using full 3D Poisson solver coupled with 2D Schrödinger solver. y . 396 (2000). Ishikuro. Computational Electronics (1) Proof of concept for 2D quantization Approach 1: Solve the 2D Schrödinger equation   2 2 ∂ h2 ∂ 2 − h − + V ( x. z) . y. y. ˆ Do the same calculation with effective potential Buried Oxide Substrate (400nm) Channel (~10nm wide) 1 Majima. 21. z) = E νj ( x )ψ νj ( x. Lett. z)ψ νj ( x. N νj ( x ) = ν =1 j  E ν (x) − E 1 j F 2m xν k BT ⋅ F−1 / 2  πh k BT       Approach 2: Use the effective potential approach to obtain the line electron density. z)  2m ν ∂y 2 2m ν ∂y 2  y z   where the electron density is given by: 3 2 n( x. Computational Electronics 26 . Hiramoto. y. IEEE El. z) = 2 ∑ ∑ N νj ( x ) ψ νj ( x. Dev.

Š It is still “quasi-local”.Effective Potential 7 nm .5 Gate Voltage V G 2 2.Effective Potential 10 nm .2x10 7 15 nm . so it does not allow proper treatment phase interference effects Computational Electronics 27 .(2) Simulation results for a quantum wire Line density [1/cm] 1.5 1 1.64 nm. Computational Electronics 2.2D Schrodinger 8x106 4x106 0 0 0.6x107 1.Effective potential 15 nm . however: Š Model valid within the random phase approximation.5 [V] Excellent agreement is observed between the two approaches when using the theoretical value for the Gaussian smoothing parameter of 0.2D Schrodinger 10 nm .2D Schrodinger 7 nm .5 Conclusions regarding the use of the effective potential approach Š The effective potential approach in many cases allows for a quite accurate approximation of the quantization effects in real semiconductor devices Š The numerical cost of including the effective potential is low – “more bang for the buck” Some challenges remain.