Post on 05-Jan-2016
description
MiniSIMS
Secondary Ion Mass Spectrometer
Dr Clive Jones
Millbrook Instruments Limited
Blackburn Technology Centre, England
www.millbrook-instruments.com
Depth Profiling 101
Contents
• Depth profiling overview
• Sputter rate
• Calibration
• Depth resolution
• Detection limit
• Noise
• Reproducibility
Depth Profiling Overview
• Continuously sputter sample to make a crater
• Peak switch spectrometer through a selected list of up to 10 masses
• Acquire data for the selected number of scans at each mass
• Record raw data as counts per second versus etch time
• Convert raw data to concentration versus depth
Raw data Processed data
cps
etch timeco
ncen
trat
ion
depth
Dopant isotope
Matrix isotopematrix
dopant
Depth Profiling OverviewDepth Profiling Overview
AsSi/100
As Implant 60 keV 1e16 at/cm2
1E+18
1E+19
1E+20
1E+21
1E+22
0 500 1000 1500 2000
Sputtered Depth (Angstroms))
Con
cent
ratio
n (a
t/cc)
Depth Profiling OverviewDepth Profiling OverviewMiniSims Depth Profile
Contents
• Depth profiling overview
• Sputter rate
• Calibration
• Depth resolution
• Detection limit
• Noise
• Reproducibility
6 nm/minSi matrix
1.5 nm/minSi matrix
100 microns
200 microns
D
Sputter rate = K / D2
K is a matrix dependant constant2
D is the lateral crater dimension
2sputter rate also proportional to beam current but this fixed on MiniSIMS
Sputter rateSputter rate is a function of crater size1
1For a given primary beam incident angle
Sputter RateSputter rate is a function of angle of incidence
Sputter rate can be increased by using an angled stub
Ga 6keV in Si - Results of Detailed SRIM Calculations
0
2
4
6
8
10
12
14
16
18
0 10 20 30 40 50 60 70 80 90
Angle from Normal (Degrees)
Spu
tte
r Y
ield
Si i
n a
tom
s/io
n
Contents
• Depth profiling overview
• Sputter rate
• Calibration
• Depth resolution
• Detection limit
• Noise
• Reproducibility
Concentration = (I/M) . RSF
Calibration
I = Impurity secondary ion counts per second
M = Matrix secondary ion counts per second
RSF = Relative sensitivity factor of impurity for that matrix
Depth sputtered in dt = S.dt
I(t) = impurity counts at time t
dt
time
coun
ts
Concentration = (I/M) . RSF
M = matrix countsDose in one slice = (I/M). RSF. S.dt
Sputter rate = S
Total dose = (I/M). RSF.S .dt
I
S.dt .
RSF .S .dt .
Therefore, RSF =
M
I
M . dose
=
RSF1 calculation from profile of implant of known dose
1 relative sensitivity factor
Calibration
I = impurity counts at peak
time
coun
ts M = matrix countsConcentration = (I/M) . RSF
Where P is the known concentration of the implant at its peak
Then RSF = M . P
I
How to calculate an RSF from a profile of known concentration
Also please note that the sputter rate is not required for this calculation
Calibration
Si+
O+
Cou
nts
Time
Surface ion yield transient region
Presence of oxygen at the surface enhances the positive ion yield
Calibration
Calibration
The surface ion yield transient can distort an impurity profile
Raw Data
This may be corrected to some extent during data reduction
Data normalized to matrix profile
matrix
impurity
The normalized profile is closer to the true distribution
Contents
• Depth profiling overview
• Sputter rate
• Calibration
• Depth resolution
• Detection limit
• Noise
• Reproducibility
coun
ts
depth depthco
unts
exponential
gaussian
buried layer
How ion beam mixing affects depth resolution1
1 exaggerated for illustration
Depth Resolution
This effect can be reduced by using by using an angled sample stub
ideal actual
Depth Resolution
Depth resolution improves with increasing angle of incidence
Detailed SRIM calculations for 6kV Ga ion bombardment of Si
0
10
20
30
40
50
60
70
80
90
100
0 10 20 30 40 50 60 70 80
Degrees from normal
Ga R
ang
e in
Angst
rom
s
Using angled stub changes angle of incidence of primary beam
Depth Resolution
Orientation of stub important for good secondary ion yields
Using an angled stub leads to better depth resolution and a faster sputter rate
T U R B O
Quadrupole
Sample
Sloping stub
Best analysis area
Direct away from quad / turbo
Depth ResolutionSputter rate is a function of angle of incidence
Be aware of sputter rate
Ga 6keV in Si - Results of Detailed SRIM Calculations
0
2
4
6
8
10
12
14
16
18
0 10 20 30 40 50 60 70 80 90
Angle from Normal (Degrees)
Spu
tter
Yie
ld S
i in
atom
s/io
n
conc
entr
atio
n
depth
conc
entr
atio
n
depth
Depth Resolution
Illustration of consequence of sputtering too quickly
Remedy – slow down sputter rate or reduce number of masses per cycle
conc
entr
atio
n
poor gating
Depth resolution
Illustration of the need for gating
Without gating, some ions from the crater wall will be counted
Gate
Beam size
depth depth
conc
entr
atio
n
good gating
100 microns
200 microns
Indicates size of 10 micron beam
Indicates size of 25% gate
Gating ensures the monitored ions come only from the crater bottom
Depth resolution
Poor crater edge rejection Good crater edge rejection
Depth resolution
Indicates size of 10 micron beam
100 microns50% gate
200 microns50% gate
Choosing the appropriate gate size
100 microns
Good
Indicates size of 10 micron beam
Depth resolution
Smaller craters may need smaller gate size to preserve depth resolution
100 microns
Poor
10% gate 50% gate
For deep profiles (microns) the crater bottom may become rounded1
gate
gate
100 microns 200 microns
1Exaggerated for illustration
Larger raster size gives better depth resolution because the curvature is less
Depth resolution
Contents
• Depth profiling overview
• Sputter rate
• Calibration
• Depth resolution
• Detection limit
• Noise
• Reproducibility
Detection Limit
• Count rate limitations
• Background from “residual vacuum” species
• Background from sample doping
• High surface concentration (surface sputtering)
• Interferences from primary beam, matrix and residual vacuum species
Count rate limit possible solutions
• Increase integration (scan) time
• Sputter faster
• Use oblique ion bombardment
Longer scan time
Shorter scan time
Effect of scan duration on data quality
Illustrating reduced noise with longer scan time
Count rate limit - possible solutionsCount rate limit - possible solutions
1
10
100
1000
10000
100000
0 1000 2000 3000 4000 5000 6000
Etch Time (s)
Inte
nsi
ty
Effect of scan time on data quality
Longer scan time
Shorter scan time
Illustrating improved dynamic range with longer scan time
Count rate limit - possible solutionsCount rate limit - possible solutions
1
10
100
1000
10000
0 1000 2000 3000 4000 5000 6000
Etch Time (s)
Inte
nsi
ty (
cps)
500px and 200px single scan profiles of same sample
500px is 13.93s/scan;
200px is 2.19s/scan;
ratio = 6.36
Longer scan times less noise + better dynamic range
But note that longer scan times also result in less data density
Effect of scan time on data quality
Count rate limit - possible solutionsCount rate limit - possible solutions
1
10
100
1000
0 200 400 600 800
Etch Time (s)
Inte
ns
ity
(cp
s)
High surface or near surface concentration
• If there is a peak in impurity e.g. from an ion implant or surface contamination, then the gate needs to be small enough to reject sputtered crater sidewall ions.
• Use smaller gate
• For high concentration at surface do a two stage profile. Profile through top surface with large raster, continue with smaller raster
conc
entr
atio
n
depth
With gating
Without gating
Dyn
amic
Ra
nge
Typical ion implant profile illustrating the need for gating
Without gating, some ions from the crater wall will be counted
Gate
Beam size
High surface or near surface concentrationHigh surface or near surface concentration
Residual vacuum possible solutions
• Bake sample and stub
• Pump down overnight
• Background subtract
• Choose a different isotope
Interferences
Interferences
• Use sloping stub to reduce level of Ga in sample
• Monitor different isotopes, dimers or doubly charged species
Primary beam and matrix interference – possible solutions
Sample doping issue
• Check whether there is real doping or residual vacuum problem
• Run the analysis with faster and slower scan times
• The unknown / matrix ratio will remain unchanged if the unknown is sample doping
Contents
• Depth profiling overview
• Sputter rate
• Calibration
• Depth resolution
• Detection limit
• Noise
• Reproducibility
Statistical counting noise is proportional to n1/2
Counts
(n)
Noise
(root n)
Relative noise
10 3.2 32%
100 10 10%
1000 32 3.2%
10000 100 1.0%
Noise
Size
of
gate
Number of
scans
Actual analysis
time
1 count in units of
cps
10% s 2.19s
10
4.57
s
25% s 2.19s
4
1.83
s
50% s 2.19s
2
0.91
s
CPS Noise is inversely related to scans per cycle
Noise
Contents
• Depth profiling overview
• Sputter rate
• Calibration
• Depth resolution
• Detection limit
• Noise
• Reproducibility
Guidelines for obtaining the best reproducibility
• Use same sample stub each time with same orientation
• Use exactly the same analysis position coordinates each time
• Use the integrated peak counts rather than the peak height for detailed comparison of spectra
Guidelines for obtaining the best reproducibility
• Use exactly the same vacuum conditions each time (either pump down for a given length of time before analysis, or wait until the pressure reading reaches a certain value)
• Use the same raster and gate conditions each time
• Make sure that the peaks used are at count rates in the linear range of the channeltron. A good rule of thumb would be <100,000 cps.
Guidelines for obtaining the best reproducibility
Make sure you select the exact mass