33
William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April 19, 2004 The Onset of Magnetic Reconnection

William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

Embed Size (px)

Citation preview

Page 1: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

William Daughton

Plasma Physics Group, X-1

Los Alamos National Laboratory

Presented at:

Second Workshop on Thin Current Sheets

University of Maryland

April 19, 2004

The Onset of Magnetic Reconnection

Page 2: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

Motivation for this work Current sheet geometry is often employed to study the

basic physics of collisionless magnetic reconnection

Kinetic Simulations are typically 2D with large initial perturbation:

a. Does not allow instabilities in direction of currentb. Avoids the question of onset completely

www-spof.gsfc.nasa.govwww-spof.gsfc.nasa.gov

rB

rJ Courtesy of Hantao Ji (PPPL)

Page 3: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

Basic Approach

cost ∝mi

me

⎝ ⎜

⎠ ⎟

1+n / 2

n → dimensionsFor a given problem with fixed box size

Explicit PIC must resolve all relevant scales

cΔt < Δx ωpeΔt <1 ΩceΔt <1 Δx ≈ λ D

3D Simulations - Must choose very artificial parameters

2D Simulations - More realistic parameters are possible

mi

me

,ωpe

Ωce

, etc

Bx

z

Jy

x

y

z − x plane → Tearing →γ

Ωci

~ 0.05

z − y plane → LHDI →γ

Ωci

~ 5

Page 4: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

Harris Current Sheet

fs =n(z)

π 3 / 2V||sV⊥s2

exp −vx

2

V||s2

−vy −Us( )

2+ vz

2

V⊥s2

⎢ ⎢

⎥ ⎥

MainDistribution

fbs =nb

π 3 / 2v th3

exp −vx

2 + vy2 + vz

2

Vbs2

⎣ ⎢ ⎢

⎦ ⎥ ⎥

BackgroundDistribution

n(z) = no sech2 z

L

⎝ ⎜

⎠ ⎟

V||s =2T||s

m s

V⊥s =2T⊥s

m s

Us =2cT⊥s

qsBoL

Anisotropy

T⊥s

T ||s

Thickness

ρi

L=

U i

V⊥s

Bx (z) = Bxo tanh(z /L)

Jy (z) =cBo

4π Lsech2 z

L

⎝ ⎜

⎠ ⎟

x€

z

Page 5: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April
Page 6: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

2D Simulations of Tearing

Consider 3 simulations - Only change the box length

1. Single island saturation

2. Two island saturation

3. Four island saturation

ρi

L=1

mi

me

=100Ti

Te

=1ωpe

Ωce

= 5Equilibrium Parameters

γΩci

≈ 0.11 kxL ≈ 0.5

Reduced by 30% for

mi

me

=1836

Box Size → 4πL × 4πL 640 × 640 grid 50 ×106 particles

Box Size → 8πL × 4πL 1280 × 640 grid 100 ×106 particles

Box Size →16πL × 4πL 2560 × 640 grid 200 ×106 particles

Page 7: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

Single Island Tearing Saturation

γΩci

kxL€

T⊥e

T||e

=1

z

L

x /L

T⊥e /T||e

T⊥e

T||e

= 0.95

Linear Growth Rate Mode Amplitude

tΩci

PIC Simulation

Ay

QuickTime™ and aYUV420 codec decompressor

are needed to see this picture.

Page 8: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

Two Island Coalescence

z

L

x /L

T⊥e /T||e

Mode Amplitudes

Ay

kxL

tΩci

γΩci

Linear Growth Rate

M=1

M=2

T⊥e

T||e

=1

0.95

0.9

QuickTime™ and aYUV420 codec decompressor

are needed to see this picture.

Page 9: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

QuickTime™ and aYUV420 codec decompressor

are needed to see this picture.

Four Island Coalescence

z

L €

x /L€

z

L

Onset Stage• Central region of box

• Linear tearing islands

• Coalescence

• Very slow process

Fast Reconnection

• Show entire box

• Large scale reconnection

• Saturation limited by box

tΩci = 0 →190

tΩci =190 → 244

x /L

QuickTime™ and aYUV420 codec decompressor

are needed to see this picture.

Page 10: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

Reconnection Onset from Tearing

How might this change in 3D?• LHDI is much faster than tearing

• 2D simulations in oblique plane

• Can the LHDI modify onset physics ?

• Single island tearing saturates at small amplitude

• Onset requires coalescence of many islands

• Finite Bz is stabilizing influence

1−Te⊥

Te||

⎝ ⎜

⎠ ⎟>

ρ e

LLaval & Pellat 1968Biskamp, Sagdeev, Schindler, 1970

Scholer et al, PoP 2003

Horiuchi

Shinohara & Fujimoto

Pellat, 1991Pritchett, 1994Quest et al, 1996

Sitnov et al, 1998 -> can go unstable?

Tearing is stablein magnetotail

Page 11: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

Lower-hybrid Drift Instability (LHDI)• Driven by density gradient

• Fastest growing modes

• Real frequency

• Growth rate

• Stabilized by finite beta

• Primarily electrostatic and localized on edge

kyρ e ~ 1

β =8π n(z)(Te + Ti)

Bx2(z)

ω ≤Ωlh =ωpi

1+ ωpe2 /Ωce

2( )

1/ 2 ≈ ΩciΩce( )1/ 2

γ≤Ωlh

ˆ φ (z) = ˜ φ (z) exp −iωt + iky y[ ]€

z /L

Example Eigenfunction

GoodAgreement

Carter, Ji, Trintchouck, Yamada, Kulsrud, 2002Davidson, Gladd, Wu & Huba, 1977

Huba, Drake and Gladd, 1980Theory Experiment

˜ φ (z)

Bale, Mozer, Phan 2002 Observation

U i < Vthi ⇒ kinetic (dissipative)

U i > Vthi ⇒ fluid - like (reactive)

Page 12: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

Established Viewpoint on LHDI

z

y€

˜ φ (z)• Localized on edge of layer

• Small anomalous resistivity

• Wrong region to modify tearing

• Not relevant to reconnection

New results challenge this conclusion

1. Direct penetration of longer wavelength linear modes

ky ρ iρ e ~ 1

ρi

L>1

ρi

L≤12. Nonlinear development of

short wavelength modes

kyρ e ~ 1

Page 13: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

Penetration of LHDI

mi

me

=1836ρ i

L= 2

Δx ≈1.4λ D Δt Ωce ≈ 0.08 Box Size =12L ×12L

1280 ×1280 cells 500 ×106 particles

z

L

z

L

z

L

yL

tΩci=3

tΩci=11

tΩci=8

Bx (y,z) − Bxo tanh(z /L)

yL

kyL = 2.62 ⇒ ky ρ iρ e ≈ 0.8

tΩci=13

tΩci=13

Bx€

Jy€

z

L

z

L

Page 14: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April
Page 15: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

2D Simulation of Lower-Hybrid

ρi

L=1

mi

me

= 512Ti

Te

= 5ωpe

Ωce

= 4nb

no

= 0.02Tb

Te

=1Equilibrium Parameters

Box Size →12L ×12L 2560 × 2560 grid 1.6 ×109 particles

ΔtΩce = 0.08 Δz = Δy ≈ λ D 256 processors

Simulation Parameters:

Thicker Sheet Colder Electrons

More relevant to magnetospheric plasmas

Background

Page 16: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

QuickTime™ and aYUV420 codec decompressor

are needed to see this picture.

Electrostatic Fluctuations

ωr /Ω lh = 0.54 γ /Ωci =1.93

z /L

˜ φ (z)

ωr /Ω lh = 0.57 γ /Ωci = 2.26

˜ φ (z)

z /L

Two fastest Growing modes

kyρ e ≈ 0.75

z

L

y /L

Lower-Hybrid Drift Mode

Lower-Hybrid Drift Mode

Fluctuations are confined to the edge of the sheet

˜ φ (z)

Ey (y,z)

Page 17: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

QuickTime™ and aYUV420 codec decompressor

are needed to see this picture.

Evolution of Current Density

z

L

y /L

Jy

Jo

Jy (z) = Jo sech2(z /L)Initial

Y-averaged

z

L

Jy = −eneVey + eniViy

Contours of

Jy (z,y)

Jy =1

Ly

Jy∫ (z,y) dy

Page 18: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

z

L

y /L

ni(z) = no sech2(z /L)Initial

Y-averaged

z

L

Contours of

ni(z,y)

ni =1

Ly

ni∫ (z,y) dy

Evolution of Ion Density

ni

no

QuickTime™ and aYUV420 codec decompressor

are needed to see this picture.

Page 19: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

z

L

y /L

Initial

Y-averaged

z

L

Contours of

Viy (z,y)

Viy =1

Ly

Viy∫ (z, y) dy

Viy

Vthi

Evolution of Ion Velocity

Viy (z) =U i

1+ nb cosh2(z /L)

QuickTime™ and aYUV420 codec decompressor

are needed to see this picture.

Page 20: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

QuickTime™ and aYUV420 codec decompressor

are needed to see this picture.

z

L

y /L

Initial

Y-averaged

z

L

Contours of

Vey (z,y)

Vey =1

Ly

Vey∫ (z,y) dy

Vey

Vthe

Evolution of Electron Velocity

Vey (z) =Ue

1+ nb cosh2(z /L)

Page 21: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

QuickTime™ and aYUV420 codec decompressor

are needed to see this picture.

z

L

y /L

Y-averaged

z

L

Contours of

Evolution of Electron Anisotropy

T⊥e

T ||e

T⊥e

T ||e

T⊥e

T ||e

Page 22: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

Resonant Scattering of Crossing Ions

z

y

Bx (z) = Bo tanhz

L

⎝ ⎜

⎠ ⎟

δi ≈ 2ρ iL

Scale forCrossing Orbit

vy

vz

U i

ωky

≈U i

2

Noncrossing

Crossing

Crossing Example of scattering

Lower-hybrid fluctuations

˜ φ (z)

zlh

Page 23: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

QuickTime™ and aYUV420 codec decompressor

are needed to see this picture.

z

L

y /L€

z

L

Contours of

φ(z, y)

Electrostatic Potential

−e φTe

Net gain + + + + + + + + +

Net gain + + + + + + + +

Net loss - - - - - - - - -

Page 24: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

Electron Acceleration

mene

dVe

dt= −∇ • Pe − ene E+

Ve ×B

c

⎝ ⎜

⎠ ⎟

Neglect

Vey =c

eBxne

∂Pe

∂x−

c

Bx

∂φ

∂z

Use EquilibriumProfiles

z

L

y /L€

z

L

Vey /Vthe

QuickTime™ and aYUV420 codec decompressor

are needed to see this picture.

Page 25: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

Inductive Heating of ElectronsEvolution of current profile modifies magnetic field

Jy

Bx

For electrons, magnetic field changes slowly

Changes on the ion time scale

pdq∫

μ =mv⊥

2

2B

z

y

δe ≈ 2ρ eL

p = mv⊥r

dq = dθ

T⊥e (t)

T⊥e0

≈Bx (t)

Bx 0

How to construct adiabatic invariant for these orbits?

Magnetic Moment

Inductive Heating

Adiabatic Invariant

x

Λ(x)

Page 26: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

Anisotropic Electron Heating

z

L

y /L

T⊥eT⊥e 0

T⊥e

T⊥e 0Contours of Y-averaged

T⊥e (t)

T⊥e 0

Y-averaged

Bx (t)

Bx 0

QuickTime™ and aYUV420 codec decompressor

are needed to see this picture.

Page 27: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

Physical Mass

mi

me

=1836

5120 × 5120 grid

6 ×109 particles

Plasma parameters are same butnumerical requirements increase

tΩci = 7

Results show same basic physics

Details are described in preprint

How big of a mass ratio is needed?

Page 28: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

What about lower mass ratio?

z

L

y /L

Jy

z

L

mi

me

= 512

mi

me

=100

QuickTime™ and aYUV420 codec decompressor

are needed to see this picture.

QuickTime™ and aYUV420 codec decompressor

are needed to see this picture.

Page 29: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

1. Critical thickness for process to occur

2. Potential structure accelerates electrons

3. Enhances tearing mode

New Model for Fast Onset of Reconnection

zlh ≈ (1− 2)L

δi ≈ 2ρ iL

ρi

L≈ 0.5

Lower-hybrid drift instability

Lower-hybrid drift instability

1. Current density2. Anisotropy

kxL€

γΩci

= 0.035

γΩci

T⊥e

T||e

=1

T⊥e

T||e

=1.1

γΩci

= 2.2

4. Rapid onset of reconnection

Critical Scale

Tearing Growth RateForslund, 1968

J. Chen and Palmadesso, 1984

Page 30: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

QuickTime™ and aYUV420 codec decompressor

are needed to see this picture.

Test this idea at reduced mass ratio

γΩci

kxL€

T⊥e

T||e

=1

Tearing Growth Rate

T⊥e

T||e

=1.5

z

L

z

L

x /L

Factor of 17 increase in growth rate

Fastest mode shifts to shorter wavelength

Growth of small islands --> Coalescence

Rapid onset of large scale reconnection

Initialize previous 2-Mode case with

T⊥e

T||e

=1.5

mi

me

=100

Page 31: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

Electron Anisotropy Instabilities?Theory of Space Plasma Microinstabilities, S.P Gary

T⊥e

T||e

<1 Ωci < ω << Ωce

kc

ωpi

>1 k × Bo = 0€

T⊥e

T||e

>1 Ωci < ω < Ωce

kc

ωpe

≤1 k × Bo = 0

1. Whistler Anisotropy Instability

2. Electron Firehose Instability

1. Edge region is low beta2. Center has complicated orbits3. Does not appear in simulations?

Should these occur in neutral sheet?

Page 32: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

Neutralization of Electrostatic Potential

γ>>Vthe

D

γDVthe

Ωci

⎝ ⎜

⎠ ⎟Vthi

Vthe

⎝ ⎜

⎠ ⎟ΩciL

Vthi

⎝ ⎜

⎠ ⎟D

L

⎝ ⎜

⎠ ⎟>>1

1

20

1

5

D >> 4L

D

Growth of LHDI

Time scale for electrons to flow in and neutralize

Page 33: William Daughton Plasma Physics Group, X-1 Los Alamos National Laboratory Presented at: Second Workshop on Thin Current Sheets University of Maryland April

Future Work

Working with collaborators to simulate in 3D

However, many things left to examine in 2D:

1. Does predicted critical thickness hold?

2. Role of guide field and/or normal component

3. Influence of background (lobe) plasma

4. More realistic boundary conditions

Possible relevance to recent Cluster observations Runov et al, Cluster observation of a bifurcated current sheet, GRL, 2003