dynamical structure factor in warm dense matter and...
TRANSCRIPT
Dynamical Structure Factor in Warm Dense Matter andApplications to X-Ray Thomson Scattering
Carsten Fortmann
Lawrence Livermore National LaboratoryUniversity of California Los Angeles
IPAM, May 24, 2012
This work performed under the auspices of the U.S. Department of Energy
by Lawrence Livermore National Laboratory under Contract DE-AC52-07NA27344.
C. Fortmann (LLNL & UCLA) IPAM 2012 1 / 32
Contributors
S. H. Glenzer, T. Doppner, A. L. Kritcher, O. N. Landen, T. Ma,A. Pak
R. Redmer, K.-U. Plagemann, P. Sperling, G. Ropke A. Wierling
Ch. Niemann
P. Davis, L. Fletcher, R. Falcone
H. J. Lee
SNL M. Desjarlais
C. Fortmann (LLNL & UCLA) IPAM 2012 2 / 32
Outline
Introduction: X-Ray diagnostics of WDM
Many-body theory of light-matter interaction
The Dynamical Structure Factor
Semi-analytic calculations of structure factors in WDM
Applications: X-ray scattering from warm dense plasmas
Ab-initio calculations for the dynamical structure factor
Summary and Conclusions
C. Fortmann (LLNL & UCLA) IPAM 2012 3 / 32
Studying matter under extreme conditions is relevant tounderstand astrophysical and laboratory plasmas
Plasma parameters:
Coupling parameter Γ =Ze2
4πε0 kBT
(4π n
3
)1/3
Degeneracy parameter θ =kBT
EF, EF =
~2
2m(3πn)2/3
C. Fortmann (LLNL & UCLA) IPAM 2012 4 / 32
Probing the bulk properties of solid density plasmasrequires intense x-ray sources
Critical wavelength vs. electron density
10-2
10-1
100
101
102
103
104
1017 1019 1021 1023 1025
100
101
102
103
104
105
0.1 1 10 100
probe wavelength (nm)
electron density (1/cm3)
1ω
2ω3ω
4ω
FLASH
LCLS
probe photon energy (eV
)
λc=2πc/ωpl
λc/100
λc/10
Va K-αZn K-α
XFEL Mn K-α
plasma frequency hωpl (eV)
Cl Ly-α
X-rays penetrate deepinto solid matter
volumetric heatingpossible, mitigate ne/Tgradients
probe bulk properties ofdense matter, discardsurface effects
Need accurate theoryfor the interaction ofcorrelated matter withx-rays
C. Fortmann (LLNL & UCLA) IPAM 2012 5 / 32
The QED Lagrangian governs the particle and fielddynamics
L(x) =∑c
ψc(x)(i~cγµ∂µ −mcc
2)ψc(x)−ε0c
2
4Fµν(x)Fµν(x)
+∑c
Zcejµc(x)Aµ(x)
Free fermions, Dirac equation, particle current jµc = ψc(x)γµψc(x)
Free gauge fields, Maxwell equations, fields Fµν = ∂µAν(x)−∂νAµ(x)
minimal particle-field coupling⇒ emission/absorption, collisions, radiative damping
C. Fortmann (LLNL & UCLA) IPAM 2012 6 / 32
The current-current correlation gives the linear response toexternal fields (EM, pressure, particles)
4-current dynamical structure factor Sµν(k)
Sµν(k) =
∫d4x ′ exp(ikλx
′λ)〈j†µ(x + x ′)jν(x)〉x k = (ω/c , ~k), x = (ct,~x)
Correlation functions are expressed via Green functions
〈j†µjν〉k =1
πImGjj(k)
Green functions are systematically evaluated using MB Feynmandiagrams
Gjj(k) =
C. Fortmann (LLNL & UCLA) IPAM 2012 7 / 32
Dyson-Schwinger Equations
Fermion propagator G = G(0) + G(0)Σ G
Self-energy: Σ =Γ0 Γ
QP shift+damping (lifetime)
Field propagator: W = V + VΠ
W
Polarization: Π =
G
G
Γ(0) Γ Dynamical screening
Vertex: = + K eff. coupling
Dyson, Phys. Rev. 75 (1949), Schwinger, PNAS 37 (1951)
Hedin, Phys. Rev. 139 (1965)
C. Fortmann (LLNL & UCLA) IPAM 2012 8 / 32
The spectral function describes damped quasiparticles in adense medium
A(p, ω) = −2ImG (~p, ω + i0+)
A(~p, ω): probability for particle withmomentum ~p at energy ~ω.
Free particle:AFP(~p, ω) = 2π δ(~ω − p2/2m)
Dense medium:A(~p, ω) = −2Γ(~p,ω)
[~ω−Ep−∆(~p,ω)]2+[Γ(~p,ω)]2
Σ(~p, ω) = ∆(~p, ω) + iΓ(~p, ω)
C. Fortmann (LLNL & UCLA) IPAM 2012 9 / 32
Example: GW0 spectral function
Spectral function for diluteplasma (ne = 1022/cc,T = 1keV)
]-1
B
momentum p [a
0 1 2 3 4 5 6 7 8 910 [Ry]eµ+ω
frequency 0 20 40 60 80 100
) [1
/Ry]
ωA
(p,
00.20.40.60.8
11.21.4
Narrow width, ideal dispersion
No quasiparticle satellites
Spectral function for denseplasma (ne = 1026/cc,T = 1keV)
]-1
B
momentum p [a
0 1 2 3 4 5 6 7 8 910 [Ry]eµ+ω
frequency -40 -20 0 20 40 60 80 100 120140
) [1
/Ry]
ωA
(p,
00.05
0.10.150.2
0.250.3
0.350.4
Strongly damped & shifted
Plasmaron satellitesFortmann, J.Phys.A (2008) & PRE (2009)
C. Fortmann (LLNL & UCLA) IPAM 2012 10 / 32
The spectral function yields important observables
Equation of state
na(µa,T ) =
∞∫−∞
dω
2π
∫d3p
(2π)3fa(ω)Aa(~p, ω)
Vorberger et al. PRE 69, 046407 (2004)
Band structure
Ep = ~2p2/2m + ReΣ(~p,Ep)
Optical properties
Πe(~k, ωµ)=∑zν ,~p
∞∫−∞
dω
2π
dω′
2πΓ0
Ae(~p + ~k)
ω − zν − ωµAe(~p, ω′)ω′ − zν
Γ(ωµ + zν , zν ;~p + ~k , ~p)
Fortmann et al., Contrib. Plasma Phys. 94 (2007)
C. Fortmann (LLNL & UCLA) IPAM 2012 11 / 32
Applications to X-ray processes in WDM
Bremsstrahlung
σbrems(~k, ω) ∝ ImΠee(~k , ω)
Important energy loss/heatingprocess in dense plasmas
Spectrum typically ∝ e−~ω/kBT
⇒ Temperature diagnostic forhot electrons
Direct measure of collisionalityin the plasma
Electron distribution functioncan be calculated bydeconvolution of spectrumfrom elementary cross-section
Inelastic x-ray scattering
σIXS(~k , ω) ∝ See00(~k, ω)
Important ne,T ,Zf , ρ/ρ0
diagnostic
Sensitive to many-bodycorrelations (collisons, localfield effects)
Spectrum gives informationabout free-free, bound-free,and bound-bound correlations
C. Fortmann (LLNL & UCLA) IPAM 2012 12 / 32
X-ray scattering is a versatile dense matter diagnostic tool
A diagnostic for complex systems
scattering from free andbound electrons
sensitive to static and dynamiccorrelations
e-e and e-i interaction
strongly non-ideal system
Key advantages of x-rayscattering
undercritical for solids andcompressed matter
deep penetration
probe bulk properties
measures both dynamic andstatic plasma properties
direct ne, Te measurement viafirst principle relations (plasmafrequency, detailled balance)
high spatial and temporalresolution thanks to modernx-ray sources (backlighters,fel’s)
C. Fortmann (LLNL & UCLA) IPAM 2012 13 / 32
Scattering spectrum reflects the correlation of densityfluctuations
Scattering
on free
electrons
S
*
Optical Laser
D
Plasma 0
v
(v/c)sin(/2)]
Non-collective Thomson Scattering (* < D)
0
Inte
ns
ity
Wavelength
Te
Boltzmann distribution
X-ray source Scattering
on free and
weakly
bound
electrons
Solid
density
Plasma
p = mv
S
p = h/c
p = h’/c
2(h/mc2)sin2(/2) (v/c)sin/2]
X-ray ‘Compton’ scattering
Inte
ns
ity
Wavelength
Rayleigh
peak
Compton
peak
Te
*
0
Energy
Fermi or
Boltzmann
distribution e
C
m
kE
2
22
C. Fortmann (LLNL & UCLA) IPAM 2012 14 / 32
XRTS resolves collective and single particle dynamics
C. Fortmann (LLNL & UCLA) IPAM 2012 15 / 32
Dynamical Structure Factor S(k , ω)
Scattered power Ps per frequency interval dω per solid angle dΩ
Ps(~R, ω)dΩdω =Pi r
20dΩ
2πA
∣∣∣∣~k f × (~k f × ~E 0i )
∣∣∣∣2 NS(~k, ω)dω
S(~k , ω) =1
2πN
∫dt eiωt〈ρe(~k , t)ρe(−~k , 0)〉
S : density-density correlation function ⇔ coherent superposition ofdensity fluctuationsS determined via
scattering experimentstheoretical calculationscomputer simulations (MD, QMD, WPMD, ...)
r0 = e2/4πε0mec2: classical electron radius
N: number of scattering centers; A: area of illuminated plasma∣∣∣∣~k f × (~k f × ~E0i )
∣∣∣∣2 : laser polarizationC. Fortmann (LLNL & UCLA) IPAM 2012 16 / 32
Chihara formalism for the dynamical structure factor in dense plasmas
S(k, ω) = |fI (k) + q(k)|2 Sii(k, ω) + Zf S0ee(k, ω) + ZcScore(k, ω)
f (k) q(k) 2S (k, ) Z S0 (k, ) Z S (k, ')S (k, ')d( , )Probe
Energy
Inte
ns
ity
E0
Ion feature[tightly bound/screening e-]
Ecompton
Electron feature[free or delocalized e-]
Bound-free[weakly bound]
Ebind
Schematic Scattering spectrum
f (k): ionic form factor, elasticscattering from bound states
q(k): “screening cloud”: freeelectrons following ionsadiabatically
Sii (k , ω): ion-ion structure factor
S0ee(k , ω): Electron OCP
structure factor (e.g. plasmons)
Score(k , ω) ∝ 〈~p|ei~k·~r |nlm〉:bound-free scattering
Chihara, J. Phys.: Cond. Mat. 12 (2000)
C. Fortmann (LLNL & UCLA) IPAM 2012 17 / 32
Extended Born-Mermin theory and local field effects
Electronic subsystem
e − e correlations important atfinite k
treat e correlations via LFC
→ χOCPee = χ
(0)e
1−V (1−Gee)χ(0)e
Ionic subsystem
static structure factor Sii (k)
e.g. HNC, DFT-MDWunsch et al. PRE (2009)
Schwarz et al., HEDP (2010)
Coupling of e and i subsystems
e − i collision frequency
νBorn(ω) ∝ Zf
ω
∫ ∞0
dq q6V 2S (q)Sii(q)
[χ(0)(q, ω)− χ(0)(q, 0)
]Mermin ansatz with correlated electrons and e − i collision frequency
χxM(k, ω) =−izν(ω)
χOCP(k , z)χOCP(k, 0)
χOCP(k , z)− iων(ω) χ
OCP(k , 0), z = ω + iν(ω)
Fortmann et al, PRE 81 (2010)
C. Fortmann (LLNL & UCLA) IPAM 2012 18 / 32
Mermin response function is derived from Linear Response Theory
relevant statistical operator for multi-component plasmaZubarev et al. Statistical Mechanics of Nonequilibrium Processes (1996)
ρrel = exp− Φ(t)− βH + β
∑c
µcNc + β∑c
δµc(k , ω)nckeiωt + c .c .
Lagrange multipliers β, µ, δµc(k, ω) from self-consistency relations
Mermin response functionMermin, PRB (1970) Ropke et al., Phys. Lett. A (1999)
χMc (k , ω) =
iz
ν
χ(0)c (k, z)χ
(0)c (k , 0)
χ(0)c (k , z)− iω
ν χ(0)c (k , 0)
, z = ω + iν(ω) .
χM conserves local density
ν(ω): dynamical collision frequency, describes e-i collisions
C. Fortmann (LLNL & UCLA) IPAM 2012 19 / 32
Calculation of the collision frequency
Force-force correlation function
ν(ω) =βΩ0
ε0ω2pl
⟨Jz0, J
z0
⟩ω+iη
∝ G4
Approximations
Born ap-proximation
e−
V scee
Dynamicalscreening
T-matrix
C. Fortmann (LLNL & UCLA) IPAM 2012 20 / 32
Collision frequency in Born approximation
Redmer et al., IEEE Trans. Plasma Sci (2005)
Fortmann et al., Laser Particle Beams (2009)
νBorn(ω) ∝ Zf
ω
∫ ∞0
dq q6V 2S (q)Sii(q)
[εRPAe (q, ω)− εRPA
e (q, 0)]
Be, Zf = 2, T = 13 eV
0.1 1 10frequency [ω
pl,e]
0
0.05
0.1
0.15
0.2
colli
sion
fre
quen
cy ν
(ω) [
Ry]
-200 -100 0frequency shift [eV]
0
0.01
0.02
stru
ctur
e fa
ctor
See0 (k
,ω)
[1/e
V]
Born-MerminRPA
ne=3x1023
cm-3
, θ=0.73
ne=3x1024
cm-3
, θ=0.15
ne=3x1022
cm-3
, θ=3.4
Collisions most importantnear θ = kBT
EF' 1
At high & low degeneracyLandau damping dominates.
Inclusion of dynamicalscreening and strongcollisions is straightforward.Thiele et al., J. Phys. A (2006)
C. Fortmann (LLNL & UCLA) IPAM 2012 21 / 32
Treatment of e-e correlations via local field corrections (LFC)
χxM(~k , ω) =−izν(ω)
χOCP(~k , z)χOCP(k , 0)
χOCP(~k , z)−(
iων(ω)
)χOCP(~k , 0)
, z = ω + iν(ω)
χOCP(~k , ω) =χ0e(~k , ω)
1− Vee(k)(1− Gee(~k , ω))χ0e(~k , ω)
The LFC is calculated using Pade interpolationbetween low and high ω limits Wierling et al. CPP 2011
0 1 2 3 4k [units of k
F]
0
0.5
1
1.5
2
2.5
3
Gee
(k)
HubbardBornSTLSPathak, VashishtaIchimaru, UtsumiFarid et al.Moroni et al.
rs=2
static LFC
0 2 4 6 8 10frequency ω [units of E
F/h_
]
-0.2
0
0.2
0.4
0.6
0.8
Re G(k,ω), k = 1 kF
Im G(k,ω), k = 1 kF
Re G(k,ω), k = 2 kF
Im G(k,ω), k = 2 kF
rs=2
dynamic LFC
C. Fortmann (LLNL & UCLA) IPAM 2012 22 / 32
1) XRTS in isochorically heated Be
Glenzer et al., PRL 98, 065002 (2007)
µ
!"#$%&!!
&'&
&!(!
(&
!)&*+,
-)./&0)+
1! !
23& !
!-)+3.4
5(. !
(&!)&*+,
-)+./&
1! !
6237!
-)8.4
&'9 &!!"%:!$;2&(!!
!1&!!&! !
<)=9! 7) !&9 &! '1&
(.&!!!& 6237(! !*
&! 1 ('&&
! (. !(&!&!
C. Fortmann (LLNL & UCLA) IPAM 2012 23 / 32
Collisions are crucial to fit the scattering spectrum
Plasmonscattering
Best fitne = 3 x 1023 cm-3
Te = 12 eV
with collisions
Inte
nsit
y
0
3Ion feature
[elastic scattering]
Detailed
balance
Energy (keV)
3.02.9
1
2
)a
Glenzer et al., PRL 98 (2007)
BMA: Te = 12 eV (consistent w/ independent measurement)RPA-fit: Te ' 30 eV.First experimental evidence for the importance of collisions in X-rayplasma scattering.
C. Fortmann (LLNL & UCLA) IPAM 2012 24 / 32
Two component fully ionized plasma model
Fluctuation-dissipation theorem
See(~k , ω) = − ~πne
Imχee(~k, ω)
Dyson equations for pair correlation function and screened potential
χcc ′(~k, ω) = χ0c(~k , ω)δcc ′ +
∑d=e,i
χ0c(~k , ω)V sc
cd (~k , ω)χdc ′(~k, ω)
V sccc ′(
~k, ω) = Vcc ′(~k) +
∑d=e,i
Vcd(~k)χ0d(~k , ω)V sc
dc ′(~k, ω)
Vcc ′(~k) =
qcqc ′
ε0k2(fully ionized, replace by eff. potential in partially ionized plasma)
χ0c(~k, ω) = =
1
Ω0
∑~p
f c~p+~k/2
− f c~p−~k/2
E c~p+~k/2
− E c~p−~k/2
− ~ω
C. Fortmann (LLNL & UCLA) IPAM 2012 25 / 32
RPA: Self-consistent mean field approximation
Terminate Dyson series after 1st iteration
χRPAcc ′ (~k , ω) = χ0
c(~k , ω)δcc ′ + χ0c(~k, ω)V sc
cc ′(~k , ω)χ0
c ′(~k, ω)
Solution to linearized Dyson equations
χRPAee (~k, ω) =
χ0e − χ0
eViiχ0i
1− Veeχ0e − Viiχ
0i
χRPAei (~k , ω) =
2χ0eVeiχ
0i
1− Veeχ0e − Viiχ
0i
χRPAie (~k, ω) = χRPA
ei (~k , ω) χRPAii (~k , ω) =
χ0i − χ0
i Veeχ0e
1− Veeχ0e − Viiχ
0i
C. Fortmann (LLNL & UCLA) IPAM 2012 26 / 32
Calculation of the DSF
Apply fluctuation dissipation theorem
Imχee = Imχ0ee + V 2
ei
[(Reχ0
ee
)2 −(Imχ0
ee
)2]Imχii+
+ 2Reχ0ee Imχ0
ee Reχii
χ0ee =
χ0e
1− Veeχ0e
Compare to “Chihara formula” (fully ionized case)J. Chihara, J. Phys. Cond. Mat. (2000)
See(k, ω) = Zf S0ee(k, ω) + |q(k)|2 Sii (k, ω)
Identified high frequency component Zf S0ee and screening cloud q(k)
in TCP-RPA model
TCP-RPA contains additional dynamic terms
C. Fortmann (LLNL & UCLA) IPAM 2012 27 / 32
TCP-RPA model largely agrees with Chihara formula
DSF calculations for fully ionized H, ne = 1022 cm−3,T = 1 eV
0.0001
0.001
0.01
0.1
1
10
100
0.001 0.01 0.1 1
See
0
TCP-RPAChihara
S(k
,ω)
ω (ωpl)
Dynamical structure factor for H,Zf = 1, α = 3
0
0.05
0.1
0.15
0.2
0.25
0.3
0 0.5 1 1.5 2
TCP-RPAChihara
low
fre
qu
en
cy s
ca
tte
r W
(k)
k (1/aB)
Integrated low frequency signal
Chihara DSF evaluated in RPA, q(k) =√Zf Sei (k)/Sii (k)
Additional dynamical features not important for integrated signal
C. Fortmann (LLNL & UCLA) IPAM 2012 28 / 32
Application 2: Angle and energy resolved XRTS on shockcompressed boron
5 and 8 keV photons are used to probeshock-compressed B at various k
X‐rayprobebeam(250J,10ps)
Borontarget
Goldshield(CHcoated)
Drivebeam(527nm,400J,10ns)
Probefoil(V,Cu)
K‐α
tovanHamosspectrometer
Experiments performed at theTitan Laser, LLNL
HELIOS simulations of thedrive-beam target interaction
Neumayer, Fortmann et al., PRL (2010)
C. Fortmann (LLNL & UCLA) IPAM 2012 29 / 32
We completely characterize the plasma via XRTS
Plasmon position gives electron densitysolid: w/ collisions, dashed: w/o coll.
0
0.2
0.4
0.6
0.8
1
1.2
4x1023 cm-3
2x1023 cm-3
6x1023 cm-3
sign
al [a
.u]
θ = 31o
α=1.7
ΔEpl
k=1.3 Å-1 compressed
uncompressed
source
(a)
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
4850 4875 4900 4925 4950 4975
BMA+LFCRPA
sign
al [a
.u]
photon energy E [eV]
67o
62o
47o
α=0.83
α=0.90
α=1.2
(b)
k=2.0 Å-1
k=2.8 Å-1
k=2.6 Å-1
Non-collective spectrum sensitive toionization
g/cc g/cc
g/cc
k=4.3/A a=0.56
Best fit parameters:ne = 4× 1023 cm−3, Zfree = 2.3,⇒ ρ/ρ0 = 1.3. .
C. Fortmann (LLNL & UCLA) IPAM 2012 30 / 32
Plasmon dispersion and width reveal influence ofmany-particle correlations
Plasmon dispersion data supportstatic LFC model
10
20
30
40
50
60
70
80
90
1 2 3 4
RPA
BMA+LFC
BMA
plas
mon
shi
ft Δ
E plas
mon
[eV
]
wavenumber k [1/Å]
ωpl
2k2/2me
Collisional plasmon width mainly dueto e-i collisions
0
10
20
30
40
1 2 3 4
RPA
BMA+LFC
BMA
wid
th σ
S [eV
]
wavenumber k [1/Å]
Still unresolved discrepancies between experiment and theory.
C. Fortmann (LLNL & UCLA) IPAM 2012 31 / 32
Plasmon dispersion and width reveal influence ofmany-particle correlations
Plasmon dispersion data supportstatic LFC model
10
20
30
40
50
60
70
80
90
1 2 3 4
RPA
BMA+LFC
BMA
plas
mon
shi
ft Δ
E plas
mon
[eV
]
wavenumber k [1/Å]
ωpl
2k2/2me
Collisional plasmon width mainly dueto e-i collisions
0
10
20
30
40
1 2 3 4
RPA
BMA+LFC
BMA
wid
th σ
S [eV
]
wavenumber k [1/Å]
Still unresolved discrepancies between experiment and theory.
C. Fortmann (LLNL & UCLA) IPAM 2012 31 / 32
Summary
Many-body field theory is applied to calculate radiation effects instrongly coupled, non-ideal plasmas.
Dyson-Schwinger Equations are solved self-consistently yielding thequasi-particle shift and damping
We developed a reliable Thomson scattering code as a basis forXRTS plasma diagnostics
Collisional plasmon width and dispersion are succesfully described viathe externded Born-Mermin approximation.
First attempts to apply ab-initio methods for DSF calculations arepromising
C. Fortmann (LLNL & UCLA) IPAM 2012 32 / 32