Influence of radiation damage on single-particle

Influence of radiation damage
on single-particle XFEL diffraction imaging
Stefan Hau-Riege
X-ray Science and Technology Group
Physical and Life Sciences, LLNL
[email protected]
International Workshop on Science with and Instrumentation for
Ultra-fast Coherent Diffraction Imaging of Single Particles, Clusters and Bio-Molecules
at the European XFEL
This work was performed under the auspices of the U.S. Department of Energy
by Lawrence Livermore National Laboratory under Contract No. DE-AC52-07NA27344.
X-ray interaction with materials
1. Absorption
• Bound-bound, bound-free, free-free
• Bound-free (photoionization) dominates for harder x-rays
• Free-free (inverse Bremsstrahlung) may be important for
softer x-rays
2. Scattering
• Elastic (Rayleigh scattering)
• Inelastic (Compton scattering)
3. Emission
• Inverse process
• Fluorescence occurs for hot plasmas on a longer timescale
• Auger electron emission tends to dominate for low-Z atoms
Total interaction cross sections
Nitrogen (Z=7)
106
104
Cross 102
section
(barn) 1
photoabsorption
K
elastic scattering
10-2
inelastic scattering
10-4
102
103
104
105
X-ray energy (eV)
106
Veigele, Atomic Data 5, 51 (1973)
• For large x-ray energies, inelastic scattering dominates over
elastic scattering
• Photoabsorption dominates in the XFEL energy regime
• Low-Z materials absorb fewer photons than high-Z materials
Effect of ionization on x-ray material interaction
• Since absorption occurs predominantly through photoionization, the
ionization state can easily be accounted for by correcting the cross sections
• Scattering strongly depends on the subshell ionization state since the atomic
form factor is the Fourier transform of the electron density:
2
Example: Carbon
2s
F
f 1
2p
0
0
neutral C
C w/ core hole
q = 4"
sin #
$
1s
!
1
q (a-1
)
0
2
photoionization
core relaxation
Details of ionization states influences the diffraction pattern
Hau-Riege, PRA 76, 042511 (2007)
Processes in short-pulse x-ray-material interaction
• Processes in materials during the pulse:
• Photoionization
• Auger relaxation
• Electron impact ionization
• Electron equilibration
• Three-body recombination
• Electron-ion coupling
• (Recombination)
• Fast electrons escape the sample and charge it up
• Electrons may be trapped
(first secondaries, then Auger e-, then photo e-)
• Sample expands due to Coulomb and
hydrodynamic forces
High-field effects
• Multi-photon ionization and excitation (e.g. above-threshold ionization, highharmonic generation etc) are well known phenomena in the optical and
(recently) EUV regime.
• In the x-ray regime, inner-shell phenomena dominate.
• Currently, little is known about multi-photon inner shell processes.
• With a 100 fs pulse and 100 nm focus, we are not in the strong-field regime
yet (Keldysh parameter γ ~ 10, assuming a flat-top pulse profile), but tighter
focus or shorter pulses could lead to γ < 1.
• In charged nanoparticles, the strong quasi-static electric field can lead to
tunneling ionization
• We are currently implementing high-field effects into our models.
=> We will need to investigate under which conditions this is needed,
and adjust our models accordingly.
Continuum model for XFEL-induced dynamics
in single particles and clusters
• Computationally very efficient to survey a large parameter space
• Our two-fluid hydrodynamic model for large systems assumes that
• Sample is initially a homogeneous continuum
• Sample has spherical symmetry
• Separate free electrons and ions fluids interact by the Coulomb force
• Rate equations describe the ionization of each atomic species
• Time-dependent electron trapping
• We do not assume local charge neutrality
“real” molecule
Continuum model
Hau-Riege et al.,
PRE 69, 051906 (2004),
PRE 71, 061919 (2005),
PRL 98, 198302 (2007),
PRE 77, 041902 (2008)
Example: Dynamics of Anthrax Lethal Factor
R=50A, τ=40fs, fluence=6x1012 photons/100nm diameter
Number
of Atoms
2x104
Carbon
(2,4)
(K,L)
(2,3)
4
1x10
(1,0)
(2,2)
(2,1)
(1,2)
0
0
(2,0)
(0,0)
(1,1)
Time (fs)
40
• Collisional ionization is dominant initially
• Heavier atoms get ionized faster
• Captured photoelectrons accelerate ionization
Expansion dynamics
2.0
neutralized
hot core
1.5
R/R0
charged
layer
1.0
0.5
0.0
0
10
20
Time (fs)
• Higher-charged outer layers explode
faster than inner layers
• Inner part of molecule is more
strongly ionized
30
40
Atomistic model for XFEL-induced dynamics
in single particles and clusters
• Continuum model does not describe single-atom effects (e.g. escape of H)
• We are developing a massively-parallel MD code to study XFEL-material
interactions (based on our hot-dense radiative plasmas code (Glosli,
Hau-Riege, et al., PRE 78, 025401 (2008))
• Includes relevant physics that is known today and allows studying XFELrelevant sizes (> 108 atoms)
Example: Protein with 93,102 atoms
Size-effects for XFEL-induced dynamics in biomolecules
Evolution of biomolecules during the pulse strongly depends on the electron
distribution, and with that on the particle size.
size and electron dynamics at 8keV
Content removed
(pending publication).
Hau-Riege et al.,
submitted
X-ray material interaction at lower x-ray energies (FLASH)
0.15
• Absorption is complicated by inverse
Bremsstrahlung absorption
• We calculated the opacity using a
screened average-ion model (LTE)
32nm
0.10
!
0.05
Hau-Riege et al., Phys. Rev. E 76, 046403 (2007)
Ryszard Sobierajski et al.
0.2
!
0.1
Bergh, Timneanu, Hau-Riege et al., Phys. Rev. E 77, 026404 (2008),
Coupled with hydrodynamic codes, these
models have been incredibly successful in
guiding and analyzing experiments
performed at FLASH.
C
Si
latex
0.00
0
0.3
• No major change in the index of
refraction during the pulse, in
agreement with FLASH measurements
• At larger fluences, the system becomes
non-LTE.
SiC
0.0
0
Si3N4
25
50
75 100
Temperature (eV)
Si3N4
C
SiC
latex
Si
20 40 60 80 100
Temperature (eV)
We were the first to demonstrate ultrafast coherent diffraction
imaging at FLASH (13.5 nm wavelength)
Single pulse FEL
Focussed FEL
25 fs, 3x1013 W/cm2
Resolution limited
to ~50 nm due to
wavelength
rad-hydro calculations
Ab initio reconstruction
w/ H. Chapman
1 micron
We used a new method to measure FEL-induced explosion
with fs time resolution (‘one-color pump-probe’)
300 fs time delay
time delay = Δz/c
d
object
reference
‘One-color pump-probe’ experiments
match hydrodynamics simulations
1.5 I/I0
0
polystyrene, λ = 32 nm,
F = 3.2 J/cm2, tL = 25 fs, LEOS
FLASH
beam
140 nm
Diffraction Intensity (a.u.)
5
t = 26 fs
130 fs
830 fs
1.5 ps
7.8 ps
4
3.2 ps
3
2
0.5 ps
1
2.5
5.0
7.5
q
(µm-1)
10.0
• The structure factor narrows,
showing the particle exploding
• The lower resolution shape of the
explosion is different than expected
• This is the first high-resolution
observation of particle explosions
We interferometrically measure the change in optical density
of the particle at short delays
d1
d2
Object, phase
change of ϕ
reference
l
Rings occur when ϕ1 + d1= ϕ2 + d2+Nλ
Fitting the ring
radii gives l and
Δϕ
The calculated optical density shows the same trends as the
experiments, but some disagreement remains
Phase shift (degrees)
30
Δn
0.02
1.6×1014
1.2×1014
20
0.01
10
0.8×1014
0.4×1014
0
0.00
0
200
400
600
Delay (fs)
800
W/cm2
1000
Disagreement may be due to uncertainties in the real part of the index (δ).
Discovery at FLASH: Multilayer mirror works ONCE
During 25 fs pulse (1014 W cm-2)
32 nm wavelength
After the pulse
Si/C multilayer
40µ
mm
Reflectivity unchanged
B
A
B
A
200nm
50nm
Hau-Riege et al., Phys. Rev. Lett. 98, 145502 (2007)
Hau-Riege et al., Rev. Sci. Instr. 78, 013104 (2007).
C
C
D
D
E
E
F
F
Example for exploding single-pulse optics: Spherical lenses
intensity of photon field
32 nm, 25 fs
high
z
155 nm Ø
polystyrene
low
SEM
(Michael Bogan)
0 123
100nm
w/ M.Bogan
Intensity (I0)
r
20 nm Si3N4
1.5
Intensity inside membrane
1
2
3
4
1.0
0.5
0.0
0
100
200
R (nm)
300
Diffraction image single biological molecules with ultrashort
pulses before the absorbed energy has time to alter the structure
Particle injection
XFEL output:
8 keV,
< 70 fs,
2x1012 photons
in 100 nm spot
Neutze et al,
Nature 406, 752 (2000)
CCD collecting
diffraction pattern
Effect of the limited temporal coherence of the LCLS beam
Coherence time ~ maximum time delay of two beams scattered
into the highest resolution part of the diffraction pattern
Scattering factor for a single SASE spike
(Gaussian-shaped):
Diffraction pattern is the incoherent sum
of multiple SASE spikes.
At LCLS at 8 keV, the particles have to be smaller than 500 nm
to achieve atomic resolution
Hau-Riege, Optics Express, Vol. 16 Issue 4, pp.2840-2844 (2008)
Effect of XFEL radiation on diffraction pattern
instantaneous
diffraction pattern
cumulative
diffraction pattern
105
atoms
104
atoms
103
atoms
Effect of radiation damage is surprisingly minor
Contribution of the later part of the pulse to the diffraction
pattern is smaller since
1.) Atomic ionization increases with time, and ionized
atoms have less scattering power.
2.) Atomic motion increases with time, and the degraded
diffraction pattern changes quickly in time, so that its
contribution is averaged out
• For a given fluence, short pulses are preferable
Ionization damage averages out…
Ionization damage is surprisingly minor and can
be corrected for
1.) Need statistical information about the ionization process:
(i) For the case of mono-atomic ionization-damaged particles, a
perfect correction (“repair”) of pulse- and shot-averaged diffraction
pattern is possible!
(ii) For the case of a more generic particle, partial “repair” of the
diffraction pattern is possible
Works for
• stationary atoms
• randomly-ionized atoms (on average spatially homogeneous)
2.) Can be applied blindly even when atoms move and
ionization is not spatially homogeneous
The good news is that we do not have to know the details of the ionization process.
Hau-Riege et al., PRL 98, 198302 (2007)
Using a tamper to reduce expansion dynamics
Encapsulating the molecule with a sacrificial layer reduces damage
with tamper
without tamper
charged
layer
tamper
e.g. H2O
neutralized
hot core
60
60
40
40
R (A)
20
R (A)
20
0
0
5
10
15
time (fs)
20
Tamper (H2O)
0
0
Hau-Riege et al., PRL 98, 198302 (2007)
molecule
5
10
15
time (fs)
20
Combining biomolecular tamper with repair methods of
diffraction patterns: Anthrax lethal factor
Repair
25
-1.6%
Tamper & repair
-2.4%
20
distortion
R
of
factor 15
diffraction
(%)
pattern 10
averaged
Averaged and repaired
-11.1%
-18.0%
5
0
C0.31H0.52N0.08O0.09S0.01,
D=80A, 40A tamper, 10fs pulse,
3x1012 photons/100nm
1
10
1
10
number of diffraction patterns
Diffraction pattern of damaged molecule can be substantially “repaired”!
Pulse lengths > 50 fs may be acceptable!
FLASH experiment to demonstrate tamper concept
Content removed
(pending publication).
Using one-color pump-probe at FLASH to demonstrate a Si tamper
to reduce the effect of damage in x-ray free electron laser imaging
Content removed
(pending publication).
Early imaging experiments at LCLS to verify
XFEL-material interaction models
Experiments at longer wavelength and low pulse energies (3x1011 photons):
Measure the change in the Mie pattern of latex spheres (the position of first
minimum) as a function of x-ray focus and energy
reduced trapping
beam size (µm)
relative shift
in the position
of the Mie
minimum
reduced σphoto
Hau-Riege et al., Phys. Rev. E 77, 041902 (2008)
Conclusions
Demonstrating atomic-resolution diffraction imaging
of biomolecules and nanostructures will require
• Classification under non-ideal conditions
(not only due to noise but also damage)
• Alternatively molecular alignment or determination
of molecular orientation from explosion fragments
• Image reconstruction under non-ideal conditions
• Modification of particles upon injection