The Collisionless Diffusion Region: An Introduction Michael Hesse NASA GSFC SECTP LANL GSFC UMD Overview: Diffusion region basics The (electron) diffusion region for anti-parallel reconnection The (electron) diffusion.

Download Report

Transcript The Collisionless Diffusion Region: An Introduction Michael Hesse NASA GSFC SECTP LANL GSFC UMD Overview: Diffusion region basics The (electron) diffusion region for anti-parallel reconnection The (electron) diffusion.

The Collisionless Diffusion Region:
An Introduction
Michael Hesse
NASA GSFC
SECTP
LANL GSFC UMD
Overview:
Diffusion region basics
The (electron) diffusion region for anti-parallel reconnection
The (electron) diffusion region for guide-field reconnection
An avenue toward fast MHD reconnection without Hall terms
Acknowledgements: J. Birn, M. Kuznetsova, K. Schindler,
M. Hoshino, J. Drake
SECTP
LANL GSFC UMD
Magnetic Reconnection: Dissipation Mechanism
(How does it work?)
DR
DR
time
DR
Conditions:
IMPOSSIBLE (for species s) if
E  vs  B  0
SECTP
LANL GSFC UMD
Electric Field Equations
z
x
Electron eqn. of motion
1
m e  ve

E  ve  B 
  Pe 

 ve  ve 
ne e
e  t

 vi  B 
m  v
1
1

j  B
  Pe  e  e  v e  ve 
nee
ne e
e  t

At reconnection site
1  Pxy Pyz Pyy  me

 
Ey  


ne e  x
z
y  e
important?
SECTP
 v y 


 ve  v y 
 t

small, limited by me?
LANL GSFC UMD
Results for anti-parallel reconnection:
Brief review
SECTP
LANL GSFC UMD
Magnetic field and ion-electron flow velocities
P. Pritchett
M. Hoshino
SECTP
LANL GSFC UMD
Normal Magnetic Flux:
0

F  Bz (x, z  0)dx
X
normal magnetic flux
4.0
3.5
3.0
2.5
2.0
mi/me = 9
mi/me = 25
mi/me = 64
mi/me = 100
1.5
1.0
0.5
0.0
0
5
10
15
20
i t
25
30
35
40
evolution electron-mass independent!
=> Local electron physics adjusts to permit large scale evolution
SECTP
LANL GSFC UMD
mi/me=9, i t = 18
Compare extremes
along dashed lines
- ion quantities
- electron quantities
SECTP
mi/me=100, i t = 16
LANL GSFC UMD
Large (ion) Scale Features
Bz magnetic field
0.60
ion v x
1.0 100
0.40
Bz(9)
Bz(100)
5.0
0.20
vix(9)
vix(100)
10-1
0.00
0.0 100
-0.20
-0.40
-5.0 10-1
-0.60
-0.80
0
5
10
15
shif ted x
20
-1.0 100
25
0
5
10
15
shif ted x
20
-> Ion scale features approx invariant.
SECTP
LANL GSFC UMD
25
Small (electron) Scale Features
electron v
6.0
100
4.0 100
x
vex(9)
vex(100)
vex(9)
2.0 100
0.0 100
-2.0 100
-4.0 100
-6.0 100
0
SECTP
5
10
15
shif ted x
20
25
LANL GSFC UMD
Pressure Tensor
P xye
6.0 10-2
4.0 10-2
2.0 10-2
0.0 100
-2.0 10-2
-4.0 10-2
pxye(9)
pxye(100)
-6.0 10-2
0
p vx
Pxy ~ 
 z x
SECTP
5
10
15
shif ted x
20
25
LANL GSFC UMD
Pxye
Pyze
SECTP
LANL GSFC UMD
Sample Electron Distribution (Pxye)
10.0<x< 11.0
-0.5<z< 0.5 log f
0.076
0.4
-0.739
0.2
-1.555
uy 0.0
-2.370
-0.2
-3.185
-0.4
-4.000
-0.4 -0.2 0.0 0.2 0.4
ux
Thermal inertia (nongyrotropic pressure)-based dissipation
seems key to anti-parallel reconnection
SECTP
LANL GSFC UMD
Can be explained by trapping scale:
“bounce motion” [Horiuchi and Sato, 1996]

 2meTe

z   2
2
 e (Bx / z ) 
1/ 4
[Biskamp and Schindler, 1971]
=> Estimate of reconnection electric field
E  2me Te v x'
SECTP
[Hesse et al., 1999]
[Kuznetsova et al., 2000]
LANL GSFC UMD
realistic electron mass
Ricci et al.
SECTP
3D – no LHD, kink, …
Zeiler et al.
LANL GSFC UMD
But, some questions remain…
Sausage mode,
Buechner et al.
SECTP
Kink, LHD,
Ozaki et al.
Ion sound mode…
LANL GSFC UMD
…and other limitations, such as
-Finite (small) system size
-Finite (small) ion/electron mass ratio
-Finite (small) speed of light
-Periodicity
…there is work to be done!
SECTP
LANL GSFC UMD
What changes in the presence of guide field?
if guide field strong enough
electrons are magnetized
no bounce orbits
no nongyrotropic pressures(?)
bulk inertia dominant(?)
Method: Theory and PIC simulations
SECTP
LANL GSFC UMD
Simulation Setup
- 1-D “Harris” Equilibrium,
Lx= 2Lz= 25.6 c/wpi
- Flux function: A = -ln cosh(z/a)
- normal magnetic field perturbation (X type, 2.5% of lobe field)
- 0, 40, 80% guide field
- Sheet Full-Width a= c/wpi
- Ti/Te = 5
- mi/me=256
- 100x106 particles
- 800x800 grid
Results averaged over 60 plasma periods
SECTP
LANL GSFC UMD
SECTP
LANL GSFC UMD
Change of symmetry
By
SECTP P. Pritchett
LANL GSFC UMD
Parallel electric field it=16
…also analytic theory by Drake et al.
SECTP
LANL GSFC UMD
Electric Field Equations
z
x
Electron eqn. of motion
E  ve  B 
 vi  B 
1
m  v

  Pe  e  e  ve  ve 
ne e
e  t

m  v
1
1

j  B
  Pe  e  e  v e  ve 
nee
ne e
e  t

At reconnection site

1  Pxy Pyz Pyy  me  v y 

  
E y  Ell  


 ve  v y 
ne e  x
z
y  e  t

important?
small, limited by me?
SECTP
LANL GSFC UMD
Magnitude of Bulk Acceleration Contribution
Time derivative of (negative) electron velocity in y direction:
SECTP
LANL GSFC UMD
Pxye
Pyze
SECTP
LANL GSFC UMD
-(vezBx-vexBz)
1  Pxye Pyze 




ene  x
z 
-me(ve.grad vey)/e
SECTP
LANL GSFC UMD
Electron Distribution Functions
F(vx,vy)
vy
F(vx,vz)
vz
vx
SECTP
F(vy,vz)
vz
vx
vy
LANL GSFC UMD
..pressure tensor nearly(?) gyrotropic
Pe  Peg  Peng
p||  p  
Peg  p 1 
BB
2
B
But:

p||  p p||  p 
  Peg | y  By B  

B  By  0
2
2
B
B
if Bx, Bz=0
-> nongyrotropy important. How to estimate?
SECTP
LANL GSFC UMD
Scaling the pressure tensor evolution equation



T
Pe


 T e
   ve Pe   Pe  ve  Pe  ve  
Pe  B  Pe  B    Q
t
me
Q
v
P

P
e
P
L
L

Assume

   e1 , L / v
Pii  Pij
ignore heat flux…
SECTP
LANL GSFC UMD
Pressure tensor approximations
Pxxe vey
Bz
Pyze 
 Pyye  Pxxe 
e x
By
Pzze vey
Bx
Pxye  
 Pyye  Pzze 
e z
By
Hesse, Kuznetsova, Hoshino, 2001
SECTP
LANL GSFC UMD
Electron Pressure Tensors
from simulation
approximation
Pxye
Pyze
Pxye
Pyze
critical difference at reconnection site!
SECTP
LANL GSFC UMD
P
yze
0.004
at x=13.15, t=16
0.003
0.002
0.001
0
-0.001
-0.002
-0.003
-0.4
-0.2
0
z
SECTP
0.2
0.4
coll. skin depth
LANL GSFC UMD
Pxxe vey
Bz
1  Qxxye Qxyze 


Pyze 
 Pyye  Pxxe  

 e x
By  e  x
z 
P vey
B
1 Qxyze
 xxe
 Pyye  Pxxe  z 
 e x
By  e z
Qxyze
Qxxye
Pyza approximation
SECTP
LANL GSFC UMD
Heat Flux Tensor Time Evolution

Q  m s  d 3 u (u  v)( u  v)( u  v) f s
lots of work

Qijk
t

l

l
es

ms
SECTP

(ijkl  Pkl vi v j  Pil v j v k  Pjl vi v k  Qijk vl )
xl



Qlij
v k   Qljk
vi   Qlik
vj
xl
xl
xl
l
l

r s
[Qijs Br  Qijr B s ] rsk 


 [Qiks Br  Qikr B s ] rsj   0
 [Q B  Q B ] 
jks r
jkr s
rsi 

LANL GSFC UMD
Approximations for Qxyze
x,y,x component:

(2 Pxl v x v y  Pyl v x v x  Qxyxvl )
xl

l

 0; neglect 
t
  Qlxy
l




v x   Qlyx
v x   Qlxx
vy
xl
xl
xl
l
l
e
2Qxyy Bz  2Qxyz By  Qxxx Bz  Qxxz Bx   0
me
Assume near gyrotropy, By>>Bx, Bz
Qxyz  
1
y






2
2 
P
v
v

0
.
5
P
v

P
v
v

0
.
5
P
v
xy x
xz x y
yz x 
 x xx x y
z

Leading order, Pii>>Pjk
Qxyz
SECTP
Pxxvy vx
1 

( Pxxvxv y )  
 y x
 y x
LANL GSFC UMD
Approximations for Qxyze
From simulation:
Qxyze and approximation, x=13.15
0.008
Q
0.006
xyze
Q
xyze
approximation
0.004
0.002
0
Approximation:
-0.002
-0.004
-0.006
-0.008
-0.4
-0.2
0
0.2
0.4
z
Ok in center, difference
due to 4-tensor?
SECTP
LANL GSFC UMD
Scaling of diffusion region
me v y
1 B0vz me
c2 1
c2 1
| Einertial |~
vz
~ 2
 B0vz
| Econvection |
2
2
2
e
x L1 0 e ne
w pe L1
w pe 2 L12
Pxxe vey
Bz
1   Pxxv y vx 
Pyze 
 Pyye  Pxxe  
e x
By e z   y x 
| E pressure
1 1  2  Pxxv y vx  1 1  2  Pxxv y vz 
|~
~
nee e z 2   y x  nee e z 2   y z 
2
1 P
1
r
~| Einertial | 2 xx2
| Einertial | L 2
L2  y ne me
L2
=> 2 Scale lengths:
SECTP
Collisionless skin depth
Electron Larmor radius in guide field
LANL GSFC UMD
Physical Mechanism:
Larmor orbit interacts with “anti-parallel” B components
SECTP
LANL GSFC UMD
3D Modeling
M. Scholer et al.: Formation of
“2D” channel
J. Drake et al.: Buneman modes,
electron holes, anomalous resistivity
SECTP
LANL GSFC UMD
P. Pritchett: inertia important
SECTP
LANL GSFC UMD
…and other limitations, such as
-Finite (small) system size
-Finite (small) ion/electron mass ratio
-Finite (small) speed of light
-Periodicity
…there is work to be done!
SECTP
LANL GSFC UMD
Results from GEM reconnection challenge:
•Hall effect (dispersive waves) speeds up reconnection rate
•Reconnection rate otherwise independent on model
•MHD models with simple resistivity show only slow reconnection rates
Question:
Are Hall effects the only way to include fast reconnection in MHD
models?
SECTP
LANL GSFC UMD
Approach:
•Hall effect result of ion-electron scale separation
•Eliminate scale separation by
- Choosing equal ion and electron mass
- Choosing equal ion and electron temperatures
•Simple and cheap…, includes ion and “electron” kinetic physics
•“Small” GEM runs with and without guide field
•“Large” runs, with and without guide field
SECTP
LANL GSFC UMD
GEM-size run, no By
SECTP
LANL GSFC UMD
GEM-size run, no By
me=1
SECTP
me=1/256
LANL GSFC UMD
GEM-size run, By=0.8
SECTP
LANL GSFC UMD
GEM-size run, By=0.8
me=1
SECTP
me=1/256
LANL GSFC UMD
large run, By=0.
SECTP
LANL GSFC UMD
large run, By=0.8
SECTP
LANL GSFC UMD
large run, By=0.
large run, By=0.8
Reconnection rates similar to GEM problem
SECTP
LANL GSFC UMD
By, both large runs, t=40
initial By=0.
initial By=0.8
no quadrupole or quadrupolar modulation!
SECTP
LANL GSFC UMD
large run, By=0., t=40
SECTP
Pxye
vix
Pyze
jiy
LANL GSFC UMD
large run, By=0.8, t=40
SECTP
Pxye
vix
Pyze
jiy
LANL GSFC UMD
Electric Field Equations
z
x
Electron eqn. of motion
E  ve  B 
 vi  B 
1
m  v

  Pe  e  e  ve  ve 
ne e
e  t

m  v
1
1

j  B
  Pe  e  e  v e  ve 
nee
ne e
e  t

Approximate representation in MHD:


  1
E  vi  B 
  Pi
ni e
SECTP
LANL GSFC UMD
Additional slides
SECTP
LANL GSFC UMD
By
jye
jyi
A tour of the reconnection region…
Pxye
SECTP
Pyze
LANL GSFC UMD
Mass Dependence of Electron Diffusion Region:
Simulation Setup
- 1-D “Harris” Equilibrium,
Lx= 2Lz= 25.6 c/wpi
- Flux function: A = -ln cosh(z/a)
- normal magnetic field perturbation (X type, 5% of lobe field)
- Sheet Full-Width a= c/wpi
- Te/Ti = 0.2
- me/mi=1/9-1/100
- wpe/wce=5
- 50x106 particles
- 800x400 grid
SECTP
LANL GSFC UMD
mi=me, By=1
rate slightly reduced due to higher plasma mass
SECTP
LANL GSFC UMD
Additional Material
SECTP
LANL GSFC UMD
Magnitude of Pressure Tensor Contribution
Pyze
ne
SECTP
1
1 Pyz
Ey  
  Pe | y  
ene
ene z
1 0.5E  2
~
 0.18
0.35 8E  2
LANL GSFC UMD
Particle Picture:
Straight Acceleration and Thermalization
Question: Are electrons transiently accelerated while crossing
the diffusion region, or is some of the energy thermalized?
Relevance:
me
straight acceleration ->  v e  vey
e
1
thermalization ->

  Pe
ene
Approach: Integrate 104 electron orbits in vicinity of reconnection region
SECTP
LANL GSFC UMD
kinetic energy change as function of delta y
delta y-component of kinetic energy vs. delta y
2
2
y = -2.5605e-05 - 0.17785x R= 0.98882
y = -0.027939 - 0.16877x R= 0.9873
1.5
1.5
1
1
0.5
delta Eyk
0.5
delta Ek
0
0
-0.5
-0.5
-12
-10
-8
-6
-4
-2
delta y
Ekin  1.78y
Ekin, y  1.67y
SECTP
0
2
-12
y  
-10
Ekin
-8
-6
-4
-2
0
2
delta y
eE y
Approximately 6% of energy
is thermalized
LANL GSFC UMD
orbit( 6293): x-z plane
0.06
0.04
0.02
0
-0.02
-0.04
-0.06
-0.08
13.15
13.2
13.25
13.3
x
13.35
13.4
13.45
orbit( 6293): z-x acceleration phase
0.06
0.04
0.02
z
0
-0.02
-0.04
-0.06
-0.08
13.15
13.2
13.25
13.3
13.35
13.4
13.45
x
SECTP
LANL GSFC UMD
Contours of Poloidal Magnetic Field
Scale length related to electron Larmor radius
SECTP
LANL GSFC UMD
Vmax= 0.65
Vmax= 2.8
SECTP
LANL GSFC UMD
Scaling the pressure tensor evolution equation




T


 T e
  ve Pe   Pe  ve  Pe  ve  
Pe  B  Pe  B  0
me

 
 ,
y
x z

  ve Pxy   Pxx x v y  Pxz z v y  Pxy x vx  Pxy x vx  Pyz z vx
  x Pxz   z Pyy  Pxx    y Pyz  0
xy component
near reconnection site:
B y  B x , B z
Pii  Pij
SECTP
LANL GSFC UMD
magnetic flux normal to current sheet
3.5
3
recflux/0.0
recflux/0.4
recflux/0.8
2.5
2
1.5
1
0.5
0
0
5
10
15
20
25
30
time
Reconnection faster for smaller guide fields
SECTP
LANL GSFC UMD