Basics of electron beam microanalysis

(Feb 2009)

Bernd Altrichter / Bruker AXS MA Berlin

Talk outline

interaction particles – solid state

elastic and inelastic scattering

origin of bremsstrahlung and characteristic radiation

Auger- and Xray-yield

energy levels

x-ray range (lateral, depth)

energy resolution

quantitative analysis (standard based, standardless)

ZAF correction model
Particle- solid state interaction

Excitation
Electrons Ions X-rays Light
Emission
Electrons SEM, AES XPS UPS
Ions SIMS
X-rays EDS, WDS PIXS XRF
Light CL

SEM - scanning electron microscopy EDS - energy dispersive X-ray spectroscopy

WDS - wavelength dispers. X-ray spectrosc AES - Auger electron spectroscopy

XPS - X-ray photoelectron spectroscopy UPS - UV-light photoelectron spectroscopy

SIMS - secondary ion mass spectrometry XRF - X-ray fluorescence spectroscopy

PIXE - proton induced X-ray emission CL - cathodoluminescence
Advantage of EPMA

Example:

Linear Analysed Assumed Amount

dimension volume density of material

1 µm  10–12 cm3  7 g/cm3  7 x 10-12 g

Detection limit: 0,1%

Mass detection limit:

But: EDX analysis is no trace analysis !

Interaction electron beam - specimen

We get information about :

-Topography
-composition
-crystallography
-electrical potential
-local magnetic field
-and many others

12.02.2009
 EDX spectrum

Bremsstrahlung Charact. X-rays

Elastic scattering - Bremsstrahlung

Bremsstrahlung or continuum: resulting

from decelaration of the beam electrons in
the electro-magnetic field of the atom 
combined with change direction and energy
loss  creation of photon of same energy
Inelastic scattering - Characteristic X-rays

Process of inner-shell ionization

The difference in energy between the shells
is expressed either by the ejection of an
energetic electron with characteristic energy
(Auger process) or by the emission of a
characteristic x-ray photon
Fluorescence- and Auger yield
ω = fraction of ionisation events producing
ω+A=1
characteristic X-rays (rest produce Auger electrons)

Ge

C e.g.:
ω ≈ 0.005 for C K
ω ≈ 0.5 for Ge K

Analytical is this one reason for the low yield of light elements !
 Energy of emitted radiation

If beam energy E > EK then a K-electron may be removed and

replaced by electron from L-shell !

Energy of emitted photon can be calculated:

EPhot = EVac - ECasc

X-ray energy is the difference
between two energy levels !
e.g.: transition L → K for Fe
EVac = EK = 7.11 keV
ECasc = EL = 0.71 keV
EPhot = 6.40 keV

12.02.2009
X-ray energy is equivalent to
wavelength

X-rays are electromagnetic radiation with

energy and a equivalent wavelength:

λ (nm) = 1.24 / E (keV)

Example: Cu Ka
λ = 1.24/8.048 = 0.154 nm

E Kα = 8.048 keV

Compared with visible ligth:

E = 0,0023 keV

12.02.2009
MOSELEY´s law

X- rays are called characteristic because

their specific energies or wavelengths are
characteristic of the particular element
which is excited.

λ = B / (Z – C)2

with B, C = Z- independent constants

Bases of qualitative X-ray analysis ! Moseley´s relation between

wavelength and atomic number
for Ka, La and Ma lines
The K-family of lines (1)

• K-lines : vacancy in K-level is

filled
• K-lines are L-K or M-K transitions
• α- lines are L -K transition
(1 und 2)
• β- lines are M-K transitions
(1 bis 5)
The K–family of lines

• higher atomic number means higher peak energy

• K lines in EDX spectra are either a combination of Ka and Kb
peaks or a seperated pair ( Kb weight then about 1/8..1/10)
• below element S (Ka = 2308 eV) it is not possible to resolve
the two peaks with EDX  a Kb shoulder may be visible on the
high energy side of the Ka
• line energy difference (Kb-Ka) is increasing with atomic number

• for in the SEM (Umax=30kV) we can excite K lines up

to atomic number 42 (Mo)
The K–family of lines

Al Si P S Ca Mn
Kα1,2 1,486 1,740 2,010
2.010 2,309 3,690 5,900
Kβ - K α1,2 89 97 129 166 323 408
The L-family of lines (1)

• L-lines occur: vacancy in L-level

is filled
The L–family of lines

• The L-family has a principal Lα line, which below ~2 keV (Zr

La) is the only visible peak in the spectrum
• Above ~2 keV it becomes increasingly possible to resolve Lα,
Lβ1, Lβ2, Lγ1, Lγ3 and Ll lines

• If L lines fit, there must be a higher energy Kα/Kβ pair,

assuming the beam energy is sufficient to generate the K lines
The L–family of lines

Mo (42) Ce (58)
2292 eV La 1,2 4839 eV
2394 eV Lb 5262 eV
2014 eV Ll 4287 eV
Line intensity relations (1)

Intensity of a x-ray line is determined by the transition

probability of electrons from the outer to inner shell. This
values are fixed for the lines of one series.

A! K- series: Z < 12 α1 : α2 : β1 = 100 : 50 : 15

13 > Z < 50 α1 : α2 : β1 = 100 : 53 : 18

L- series: 15 > Z < 90 α1 : β1 : γ1 = 100 : 52 : 10

12.02.2009
Line intensity relations (2)

Barium L series

Lα1 (100)

Lβ1 (49)

Lβ3,4
Lα2 (8)
Ll Lγ1
Lγ2/3

α1 : β1 : γ1 = 100 : 52 : 10
Line intensity relations (3)

Barium
BaTiO3
Lα1 (100)

Ti-Kα1

Lβ1 (31)

Ti-Kβ1
Lβ2
Ll Lα2 Lγ1 (5)
LγLγ2/32/3

Spectrum Barium L-series, 15 kV Spectrum BaTiO3, 15 kV

α1 : β1 : γ1 = 100 : 52 : 10 Overlapped Ba L-series and Ti K-series
Probability of producing a hole:
Ionisation cross-section for electrons

• Ionisation cross section: probability of excitation

• maximum ionisation cross section: 2,5 x Ebind
X-ray range

Important:
Different excitation ranges for different
interaction signals:
characteristic X-ray radiation and
Bremsstrahlung, secondary (SE) and back-
scattered (BSE) electrons
EDX resolution (depth and lateral)

1. X-ray range
A
Rd = 0.033(E01.7 − Ec1.7 )
Zρ Z = atomic number
A = atom weight
2. Spatial resolution
1 E = primary electron energy
Rs = 0.048(E − E
1.5
o
1.5
c )ρ EC = critical energy for
excitation
ρ = density
Excitation area depend on:
A! - primary energy
- density of material

The only way to improve spatial EDX resolution – go to low kV!

EDX nanoanalysis

What means low voltage for EDX:

Excitation voltage 3…5 kV

Lost information of K-lines for elements with AN > 17 (Cl)

Strong overlap of K-, L-, M- and N-Lines

1. Detector energy resolution !

2. Correct and complete database !
Low kV- line overlap

M-lines
L-lines
X-ray range

Monte Carlo electron-trajectory simulations of interaction

volume as function of atomic number

With higher density penetration depth is decreasing !

X-ray range

0.5 um 1.5 um 3 um

Monte Carlo electron-trajectory simulations of interaction volume

in iron as function of primary beam energy

With higher primary energy penetration depth is increasing!

Example: 5 or 15 kV A

Alu- Tab 15 kV
5 kV
Peak intensity ratio

5 kV 3 kV
12.02.2009
 Energy resolution (2)

∆ES = 2,35 E ⋅ε⋅ F

With:
E… x-ray energy
F… Fano factor
ε… energy for creating an
energy hole pair

Energy resolution measurement:

full width at half maximum peak
intensity (FWHM): Mn Kα- line

Best available resolution:

Si(Li)- detector:
127 eV
Ge- detetor: 109 eV
Silicon drift detector: 123 eV
Energy resolution

energy resolution Si(Li)

140

120
resolution (eV)

100

80

60

40

20
0 5 10 15 20 25 30
atomic number

- Energy resolution depend on energy !

12.02.2009
Theoretical energy resolution of
Kα lines for Si-based detectors

FWHM Mn-Kα=125 eV Mn-Kα=133 eV

[eV]
160

150 Cu
Ni
140
Co
Fe
130
Mn

120 Cr
V
Ti
110
Sc - 8eV at Mn-Kα
100 Ca
K
90
Ar
Cl
80
S
70 P
Si
60 Al - 17eV at C-Kα
Mg
50 Na
C-Kα=65 eV Ne
40
F
O
30
N
C
20
C-Kα=48 eV B
10 Be

0
0 50 FWHM, elektronic [eV] 100
 Energy resolution of Kα lines
at 60 kcps output
FWHM Mn-Kα=123 eV Mn-Kα=150 eV
[eV]
160

150 Cu
Ni
140 Co
Fe
130

120
Mn Mn
Cr -27 eV at
V
Ti
Mn-Kα
110
Sc
100 Ca
K
90
Ar
Cl
80
S
70 P

C-Kα=94 eV Si

-48 eV at
60 Al
Mg
50
Na

40
Ne
F
C-Kα
30 C O
N
20

C-Kα=46 eV
C
B
(50% !)
10
Be

0
0 50 FWHM, elektronic [eV] 100
 Comparison 125 eV / 133 eV
(calculated spectra)

B C N O F

Meaning of good energy resolution for analysis:

- better possibilities of separation overlapping lines
- better peak/background ratio
- higher detection limit
Comparison 129eV – 123 eV

other SDD XFlash®

129eV Mn-Ka 123eV Mn-Ka
68eV Mn-La 56eV Mn-La
340 s 200 s

Clear separation
Mn-Ll 0.556keV
Mn-La 0.633keV

Faster results

15 kV, 1000 cps

Quantitative analysis (1)
Standard based analysis

Base:

Intensity of a characteristic x- ray line is direct proportional to
the concentration of the corresponding element (1.
approximation, Castaing 1951)

ISample With: Z = atomic number correction

≈ k i ≈ ci ⋅ Z ⋅ A ⋅ F A = absorption correction
IStan dard F = fluorescence correction

Measurement: spectra from standards for all elements and spectrum of

unknown sample (under defined and reproducible conditions: TOA,
calibration, resolution, same excitation …)
Example: steel sample (Fe, Ni, Cr, Si)
High requirements on a microprobe standard:
Not only a known composition - the same composition everywhere !
Quantitative analysis (2)
Standardfree analysis

dΩ br dΩ ch
N ibr = i 0 t εi ( ZAF )i N ich = i 0 t ωi ε i q i ci ( ZAF )i
4π 4π
dΩ spatial angle of the detector
4π
i 0t number of incident electrons
ωi fluorescence yield
εi detector efficiency for X-ray radiation of the characteristic
energy of the observed element I
qi relative emission rate of selected line
ci concentration of the element i in % mass share
br Bremsstrahlung
ch characteristic radiation

ch
ch

= N ibr = ci ωi q i
[ ZAF]
( P B) i
br
i
N i [ ZAF] i

Measurement: only the spectrum of unknown sample

ZAF correction

Measured are k-ratios ( ratio of net intensities between sample

and standard) or the P/B- ratios for the spectrum and then
following correction:
•Z: atomic number
• differences in deceleration of the primary electrons
•A: absorption

absorption of primary emitted characteristic x- ray
•F: (secondary) fluorescence
• generation of secondary x- ray fluorescence by
characteristic radiation
Z ... atomic number correction

Reason:

A not negligible part of primary electrons will be backscattered
(depending on atomic number Z !).

Stopping power of material for the incident electrons.

Effect:

Composition will be calculated to small/ to high.

Z is smaller with higher excitation voltage

0.8 < Z < 1.2
A … absorption correction

Reason:

1st order x-rays are partly absorbed on their way to
sample surface.

Effect:

Intensity of 1st order radiation will be diminished and
calculated percentage will be too small.

A is increasing with higher acceleration voltage

A is increasing with lower take off angle
A … absorption correction (2)

e- X-ray detector e-- X-ray detector

d
d 2
1
d
2 d1 ,d2 … path length

Loss of intensity of x- ray in specimen depends on the distance

length and the mass attenuation coefficient of the specimen
material for the produced x-rays.

 CONCLUSION: take off angle of x- ray detector should be as
big as possible!
F … fluorescence correction

Reason:
• 1st order x-rays can create additional characteristic 2nd order
x-rays if E1st > E2nd

Effect:

2nd order radiation will be amplified and calculated percentage
will be too high.

This effect is especially high for elements with atomic number

differences between 1 ...3
e.g.: Cr excitation by Fe radiation!
F … fluorescence correction (2)

• F is decreasing with lower atomic number of the excited

element
• F is decreasing with increasing atomic number difference
between primary produced x- ray and excited element
Comparision quantification methods
Standard based

Determination of absolute element
concentrations normalised to
standard
+
Influence of matrix corrections
similar if standard similar to
sample

Small statistical error for net
counts of intensive lines

At least two measurements

Only accurate for homegeneous
samples

Unstable electron current, electron
- energy, detector cause errors

No spectrum evaluation in steps,
concentration calculation is “black
box”
standardless

No standard needed

Non normalised concentrations

Electron energy can be verified

Spectrum evaluation can be
checked step by step

No errors due to TOA, detector,
surface roughness

Larger statistical error, especially
for low background

Accurate determination of
background necessary
Galvanically produced NiP layers

Position of the 12 measurement points on the surface

Points on smooth
areas

Points on rough
areas

Typical surface structure, 2000x mag., 15 KV

Galvanically produced NiP layers

Spectra of points 7-12 (rough surface)

 Galvanically produced NiP layers

Φ(ρZ) analysis with standards P/B-ZAF analysis with Ref sample

Smooth areas Rough areas Smooth areas Rough areas

Point P [%] Point P [%] Point P [%] Point P [%]

1 4,72 7 4,99 1 4,73 7 4,68
2 4,71 8 5,28 2 4,71 8 4,69
3 4,74 9 4,68 3 4,67 9 4,69
4 4,62 10 5,36 4 4,69 10 4,7
5 4,71 11 5,36 5 4,72 11 4,79
6 4,75 12 3,94 6 4,7 12 4,65

mean 4,71 mean 4,94 mean 4,7 mean 4,7

Target value: 4,72 % P Permitted tolerance: ± 0,25 %

Detection limits of standardless P/B ZAF
quantification
www.bruker-axs-microanalysis.com
 Line intensity relations (2)

Das Lambert-Beersche Gesetz ist eine Vereinigung des Bouguer-Lambertschen

Gesetzes über die Schwächung der Strahlungsintensität mit der Weglänge beim
Durchgang durch eine absorbierende Substanz mit dem Beerschen Gesetz über den
Zusammenhang der Intensitätsschwächung mit der Konzentration der
absorbierenden Substanz.