Quantum Transport Simulacon: A few case studies where it is ...

Post on 27-Dec-2016

218 views 2 download

Transcript of Quantum Transport Simulacon: A few case studies where it is ...

Quantum  Transport  Simula0on:  A  few  case  studies  where  it  is  necessary  

Sayeef  Salahuddin  Laboratory  for  Emerging  and  Exploratory  Devices  (LEED)  

EECS,  UC  Berkeley  

sayeef@eecs.berkeley.edu  

Low  power  Architectures;Lecture  #1:IntroducKon,  Dr.  Avi  Mendelson  Source:  “New  Microarchitecture  Challenges  in  the  Coming  GeneraKons  o  fCMOS  Process  Technologies”  –Fred  Pollack,  Intel  

Corp.  Micro32  conference  key  note  -­‐1999.  

The  celebrated  Moore’s  Law  

The  celebrated  Moore’s  Law  

Hydrodynamic  model  

Dri<-­‐Diffusion  

Boltzmann  Transport    Equa0on  

Quantum  Transport  

A  few  numbers  to  remember  

M:  number  of  current  carrying  modes  l:  length  of  the  conductor  λ:  mean  free  path    

λ:  10-­‐20  nm  in  Si            20-­‐30  nm  in  III-­‐V  materials  

We  are  approaching  an  era  where  quantum  transport  simulaKon  is  necessary!  

A  Few  Examples  

Direct  Source  to  Drain  tunneling  

source  

drain  

source  

drain  

Off-­‐State  leakage  through  band-­‐to-­‐band  tunneling  

Conven0onal  Device  geometry  

Novel  Devices  

Band-­‐to-­‐band  tunneling  transistor  

Non-­‐Semiconductor  quantum  devices  

(presentaKon  by  Dr.  Luisier  this  morning)  

Spin  Transfer  Torque  Devices  

Bottom Electrode!

CoFe (2.5)!

Ru (0.85)!

Insulator!

Top Electrode!

CoFeB (3)!

CoFeB (3)!

MgO (0.85)!

Pinned  layer  

So<  layer  

Oxide   0  

0  R  

Current  Pinned  layer   So<  layer  

Kubota  et.  al.,  JJAP,  44,  40,  1237,2005  

Slonczewski,  JMMM,  96  Berger,  PRB,96  KaKne,  PRL  2000  

Why  STT  Devices?  

 Can  we  explain  the    (i) Amplitude  of  the  switching  current  (ii) Resistance  

With  the  same  set  of  device  parameters?  

Key  challenges  for  device  simulaKon  

Kubota  et.  al.,  JJAP,  44,  40,  1237,2005  

Electron Transport: NEGF

Equations Voltage Current

LLG

Spin Dynamics

Torque Magnet- ization

NEGF (Non Equilibrium Green’s Function)

S.  Salahuddin  and  S.  Daga:    IEDM  (2006)  

Self  Consistent  solu0on  of  the  transport  and  magne0za0on  dynamics  

The  difference  of  spin  currents  is  absorbed  by  the  so<  magnet  

Fixed  magnet   oxide   So<  ferromagnet  

Torque  from  conserva0on  of  angular  momentum  

Ef  

• Band  spliing  

• Fermi  level  

• Barrier  Height  

• EffecKve  mass  in  barrier  

• Kup  

• Kdown  

Band  structure  dependent  parameters  

Fixed  magnet   oxide   So<  ferromagnet  

Effec0ve  mass  treatment  of  the  transport  

W.  H.  Butler  et  al,  PRB,  2001  

 S0les  Group,  PRL,  2008,  Macdonald  Group,  PRL,  2008  

Fixed  magnet   So<  ferromagnet  

Oxide  

Ef  

E  

ky  

kx=0  kz=0  

Kx>0  kz=0  

Kx>0  kz>0  

Incorpora0ng  transverse  modes/k  sampling  

x  

y  

z  

Cross  secKon  is  typically  larger  than  50  nm  X  50  nm  

Effect  of  the  transverse  modes  

Pure  1D  100  modes  

Pure  1D   10  modes  

Fixed  magnet   So<  ferromagnet  

Oxide  

x  

y  

z  

Non  Equilibrium  distribu0on  for  the  right  magnet  

+  

Torque  felt  by  the  delectrons  Torque  felt  by  the  selectrons    

Torque  

Physics  of  Spin  Torque  

=  

Known  from  band  spliing  

Calculated  from    NEGF  CalculaKons  

MagneKzaKon  of  The  magnet  

Can  be  calculated  from  the  other  three  

Bottom Electrode!CoFe (2.5)!Ru (0.85)!

Insulator!

Top Electrode!

CoFeB (3)!

CoFeB (3)!MgO (0.85)!

Fuchs  et.  al.  ,  PRL  96,  186603,  

2006  

Theory  Experiment  

-0.4   -0.2   0   0.2   0.4  22  

24  

26  

28  

30  

32  

Current (mA)   T

MR

(%)  

Experimental  Benchmark  

Experimental  Benchmark  

0   0.2   0.4   0.6   0.8  0  

20  

40  

60  

80  

100  

120  

Voltage (V)  

TMR

(%)  

Cur

rent

(  µ   A

)  

Experiment  

T.  Kawahara  et.  al.,  ISSCC,  2007  

Theory:  Salahuddin  group  and    Daea  Group  

Ef  =  2.25  V  ∆  =  2.15  eV  m*  =  0.2  m0  m*FM=    m0  Ub  =  1.4  V  

PRL    99,    226602  (2007)    

Theory:  UCB  and  Purdue  http://arxiv.org/abs/

0910.2489  

Experimental  Benchmark  

A  typical  hysteresis  from  self  consistent  NEGF-­‐LLG  simulaKon    

Typical  contour  of  voltage  induced  switching  

             Nat.  Phys.  4,    67-­‐71  (2008)  

Theory:  UCB  and  Purdue  http://arxiv.org/abs/

0910.2489  

Experimental  Benchmark  

Perpendicular Component  

Perpendicular Component  

Ef = 2.25 V ∆ = 2.15 V m*= 0.32 m0 mFM

*= 0.73 m0 Ub = 0.9 V

PRB  79,  224416  (2009)  

Theory:  UCB  and  Purdue  http://arxiv.org/abs/

0910.2489  

Experimental  Benchmark  

Nature  Physics,  4,  37,  2008  Theory:  UCB  and  Purdue  

http://arxiv.org/abs/0910.2489  

Experimental  Benchmark  

Nature  Physics,  4,  37,  2008  

Ef = 2.25 V ∆ = 2.15 V m*= 0.18 m0 mFM

*= 0.73 m0 Ub = 0.77 V

Theory:  UCB  and  Purdue  http://arxiv.org/abs/

0910.2489  

Experimental  Benchmark  

Ef  

• Band  spliing  

• Fermi  level  

• Barrier  Height  

• EffecKve  mass  in  barrier  

• Kup  

• Kdown  

http://arxiv.org/abs/0910.2489  

Integrated  system  

Quantum  Device  Simulator  

Circuit  Simulator  

V  

H    

I  

R    

Puing  it  with  circuit  

Device  and  magneKzaKon  dynamics  

Circuit  Variability  

Li,  Augus0ne,  SS,  Roy,  DAC,  2008,  pp.  278-­‐283.  

Op0mal    opera0ng  point  from  device  

circuit  co-­‐simula0on  

First  Device  Circuit  Simula0on  

Conclusion  

 Quantum  Transport  SimulaKon  is  going  to  be  necessary  for  many  devices  in  the  nano  scale  regime!  

Acknowledgement:  

NSF/NRI