Fresnel Fringes
Thermal electron source. Few fringes.
http://em-outreach.ucsd.edu/web-course/Sec-I.D/Sec-I.D.html
High brightness electron beam from a multi-walled carbon
nanotube: Small source many fringes:
N. Jonge, et al. Nature 420, 393 (2002)
J. Elec. Microscopy,
22 (1973) 141
Maxima when constructive interference occurs between
straight beam and scattering beam at edge of sample when defocused.
sample
α
D = Defocus
screen
x
I
Fresnel Fringes
(W&C 27.7, Fig. 5.13, 9.20; C. Hall, “Intro. to E.M.”)
d
source
• Fringes at sample edges or boundaries seen only in
out of focus images
• Fresnel zone theory gives the position of fringe
maxima:
x m = 2( L + r )λ ( n − 5 / 8) n = 1, 2, 3 ...
• Would continue indefinitely laterally if electron
source was a point d = 0;
• If d finite: The distance x at which
the fringes disappear will be such
d
that the path difference ∆ is on the
order of a wavelength. Hence:
θ
x
∆ λ
< =
L+r d d
( L + r )λ
x<
d
( L + r )λ
d<
x
r
θ
sample
xm
I
∆
L
screen
So, the smaller the source the larger x and hence the number of fringes n.
x
Phase Contrast
W&C Chap. 27
Electron optics and sample introduce phase differences between transmitted and
scattered beams which then are recombined by objective lens, producing lateral
amplitude fluctuations.
Lattice Fringes (section 27.2)
• Fringes that occur with a period comparable to the atomic lattice < 1 nm.
• Obtained by allowing more than one beam to interfere by using a larger objective
aperture (or no aperture at all)
• Assuming a thin sample where we can ignore dynamical scattering effects:
ψ = ϕ o ( z )e 2πi ( k •r ) + ϕ g ( z )e 2πi ( k
I
DI
•r )
where k D = k I + g + s = k I + g '
assuming ϕ o ( z ) ≈ A a constant
and ϕ g ( z ) = Beiδ where B =
π
π sin πts
and δ = − πts
2
ξ g πs
ψ = e 2πi ( k •r ) [ A + Be 2πi ( g '• r +δ ) ]
I
I = A2 + B 2 + AB[e 2πi ( g '• r +δ ) + e − 2πi ( g '• r +δ ) ]
I = A2 + B 2 + 2 AB cos(2πg '•r + δ )
I = A2 + B 2 − 2 AB sin(2πg ' x − πts )
I = A2 + B 2 − 2 AB sin(2πg ' x − πts )
s = 0 only a sinusoidal function parallel to g and x of spacing 1/g’ = dhkl
s ≠ 0 then a small shift depending on thickness and |s|
No direct relationship to the position of actual atoms in the sample.
Need computer simulation to interpret a lattice image and then must have a
very thin sample (no absorption, no multiple scattering) – Weak phase object.
Good example of problem see Si <110> lattice image: Fig. 27.3 W&C.
SF
InGaAs
InGaAs(14% In)/GaAs(001)
<110> TEM cross-section
(JEOL 4000,
0.18 point to point resolution)
SF has a small shift along the
line intercepting the two
regions.
<111>
<112>
GaAs
Interface
somewhere
here
Lab #1 polycrystalline Aluminum, 200 keV
• Example from a Tecnai image taken by Ryan
and Shawn Penson (2007 students).
• Finer fringes are lattice fringes
• Larger ones are moiré fringes from
overlapping grains
• No fringes visible does not necessarily mean
amorphous since grains probably misaligned
ZnSe nanowire:
zincblende – wurtzite stacking faults.
Fringe spacings: 0.35 ± 0.02 nm
Fringe spacings: 0.25 ± 0.01 nm
(Al d111 = a/√3 = 0.234 nm)
Angles between the two rows 70.5o
also correct for {111} planes.
Moiré Fringes
Two overlapping fringe images, translational, rotational or a combination:
Translational:
∆gtm = g2 – g1
I = A2 + B 2 − 2 AB sin[2π ( g 2 − g1 ) x − πts ]
g1 g2
Rotational:
∆grm = 2gsin(β/2)
g1
β
I = A2 + B 2 − 2 AB sin [ 4π(g sin (β / 2 ))x − πts]
∆gr
g2
GaSb on GaAs
Fine fringes
perpendicular to
g = 004
translational Moiré
spacing 2.1 nm
Fe/GaAs (001)
translational Moiré
10 nm
Moiré Fringe and Dislocation Spacing Prediction GaSb/GaAs
GaAs: ao = 5.65 A
GaSb: ao = 6.10 A
Mismatch = f = ∆ a/a = 0.08 = 8%
If the interfacial dislocations are pure edge type:
a
b =  [110]
2
 a 
a
b =   12 + 12 = 

2
 2
Spacing of dislocations to relieve all mismatch strain along <110>:
b
a
=
f
2f
5.6
=
= 50Α
0.08 2
D=
Black lines are dislocations with average spacing 5 nm.
Moiré fringe spacing along the <100> direction is 1.8 nm or ∆ gtm = 1/1.8 = 0.55 nm-1
Predicted Moiré spacing ∆ gtm = gGaAs – gGaSb = (4/0.565)-(4/0.61) = 0.52 nm-1
High Resolution Imaging
1. Our sample consists of a certain density of material ρ(r). The diffracted wave amplitude
is mathematically the Fourier transform of ρ(r):
1
F ( ρ ( r )) =
2π
+∞
∫ ρ ( r )e
− 2πi∆k • r
d 3r
−∞
ρ(r)
This is just the structure factor again.
If ρ(r) has short periodicities (eg. atomic planes)
then F contains large reciprocal lattice
vectors: ∆k
eg. r = 0.2 nm, ∆k = 1/r = 2sinθ/λ = 5 nm-1
2. The electron microscope objective lens adds
a phase distortion function to the sample
scattering factor called the:
contrast transfer function H(∆k )
worse for higher ∆k (larger scattering angles)
Adds a phase factor to F:
F ( ρ ( r ))e
− iH ( ∆k )
Objective lens
Diffraction pattern
Image:
F
−1
[F ( ρ (r))e
− iH ( ∆k )
]
Contrast Transfer Function H(∆k ) (W&C pg. 463)
Each point on the sample will be imaged as a blurred disk with a radius δ(θ) as a function of
spherical aberrations (Cs ) and defocus of the objective lens:
if ∆f = 0 there still is the objective lens spherical aberation
δ (θ ) = ∆fθ + Csθ 3
We average this over all scattering angles to get the average blurred disk:
θ
∆fθ 2 Csθ 4
D (θ ) = ∫ δ (θ )dθ =
+
( nm )
2
0
4
Phase angle χ:
2π
2π  ∆fθ 2 Csθ 4 


χ=
D(θ ) =
+
λ
λ  2
4 
1 2 sin θ 2θ
=
≈
≡ u (nm -1 )
d
λ
λ
π∆fλ πCs λ3
(radians)
χ= 2 +
4
2d
1
χ = π∆fλu 2 + πCs λ3u 4
2
H ( ∆k ) = 2 A(u ) sin χ (u )
d
Surfaces of
constant
phase
Sample and lens
Phase
varies
laterally
Where A(u) is an amplitude function related to the objective aperture.
Both are sample independent, depends on your microscope objective lens Cs and your
choice of ∆ f.
Plots of 2sin χ as a function of Cs and ∆f .
2
0
-1
-2
2
4
6
Cs = 1.2 mm
∆f = -58 nm
1
0
-1
-2
0
2
Transfer Function (2 sin
1 mm
CsC=s =1.0
mm
= -58
∆f f =
-58nmnm
1
Transfer Function (2 sin
Transfer Function (2 sin
2
8
0
2
4
6
0
-1
-2
8
-1
-1
0
2
4
6
8
-1
Reciprocal Lattice Vector (u (nm ))
Reciprocal Lattice Vector (u (nm ))
mm
s = 2.0
Cs =C2.0
mm
f = -58 nm
∆f = -58 nm
1
Reciprocal Lattice Vector (u (nm ))
2
1
1
0
Cs = 1.2 mm
∆f = 0
-1
-2
2
4
6
8
-1
Reciprocal Lattice Vector (u (nm ))
Cs = 1.2 mm
∆f = -63 nm
0
-1
-2
0
2
Transfer Function (2 sin
2
Transfer Function (2 sin
Transfer Function (2 sin
Tecnai 20 at 200 keV
0
2
4
6
8
-1
Reciprocal Lattice Vector (u (nm ))
1
Cs = 1.2 mm
∆f = -100 nm
0
-1
-2
0
2
4
6
8
-1
Reciprocal Lattice Vector (u (nm ))
The ideal shape of sin χ as a function of u is one that is flat over the largest reciprocal frequencies
Optimal condition for the best resolution?
1
2
χ = π∆fλu 2 + πCs λ3u 4
dχ
= 2π∆fλu + 2πCs λ3u 3 = 0
du
0 = ∆f + Cs λ2u 2
−
2π
1
= π∆fλu 2 + πCs λ3u 4
3
2
Phase angle (combining focus and aberration constant)
To find the zero slope relationship between u and ∆f, λ, and Cs
(1)
(2) Best curve is when χ is near -120° (-2π/3)
1/ 2
∆f Sch
4

=  Cs λ 
3

uSch = 1.5Cs
rSch =
−1 / 4 − 3 / 4
λ
1
1/ 4
= 0.66Cs λ3 / 4
uSch
Combining (1) and (2) gives best focus value called:
Scherzer defocus
Cross-over u at ∆fSch when χ = 0
Scherzer resolution
Tecnai 20: Cs = 1.2 mm; λ = 0.0025 nm: ∆fSch = - 63 nm; uSch = 4.0 nm-1; rSch = 0.25 nm
For larger u (smaller real space distances) phase angle rapidly oscillates and will not translate
sample information reliably.
Minimum contrast is what we have called focus. (28.9)
This occurs when:
sin χ ~ 0.3 or
∆fMC = - 0.44 (Csλ)1/2 = -25 µm (Tecnai at 200 keV)
On the TEM the in situ Fourier analysis of the image that you can use to adjust
astigmatism is essentially this contrast transfer function.
Bright rings are when 2sin χ = 2 and χ (u) = nπ/2 with n odd
and dark rings when 2sin χ = 0 and χ (u) = nπ/2 with n even
We can measure the microscope Cs using these fringe spacings. (Fig. 30.6)
Transfer Function (2 sin
2
1
0
-1
-2
0
2
4
6
Reciprocal Lattice Vector (u (nm-1))
8