This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
CEA-LETI, MINATEC, 17 rue des Martyrs, 38054 Grenob le, France [email protected] +33 438 782 069
Adrien DANEL, UCPSS 2006 2
2006 Agenda
� 1) Contamination in microelectronics
� 2) Metrology: focus on metallic contamination
� 1) Challenges� 2) What is named contamination ?� 3) Main sources of contamination� 4) Impact on devices� 5) Management issues� 6) Metrology: general information
� 1) Spectroscopic and quantitative methods• TXRF
• Collection of contamination • ICPMS
• AAS
� 2) Indirect methods• Lifetime
Adrien DANEL, UCPSS 2006 3
2006 1.1 Contamination: challenges
Yield
A big part of yield losses is due to contamination
Manufacturer notoriety
Contaminants affect devices reliability
A "scientific" production monitoring
IC manufacturing uses numerous expert methods to monitor and qualify the production
Adrien DANEL, UCPSS 2006 4
2006 1.2 What is named contamination ?Anything undesirable and potentially dangerous for the production
Fe, Cr, Ni, Mn, MoRobots, chucks, boats • Particles and possible cross contamination
Chuck footprint
Diffusion of Cu from a contaminated quartz boat
Adrien DANEL, UCPSS 2006 8
2006 1.4 Impact on devices
Main effects
• Junction leakage• Dielectric breakdown• Interface segregation• Surface and interface roughness• Carriers lifetime degradation• Defect decoration• Electrical shortcut• Doping modification• Resist poisoning• Interconnects corrosion• Haze• Degradation of molecular bonding• Degradation of stepper optics • Modification of etch and dep parameters
MetalsParticles
Volatil contamination
Post etch panic !
Bug during litho ?
God save our copper …
Adrien DANEL, UCPSS 2006 9
2006
� Metals on gate oxide integrity
Detrimental impact depends on the nature of the contaminant
0
2
4
6
8
10
12
0 5 10 15 20 25 30 35Mea
n B
reak
dow
n F
ield
(M
V/c
m)
Contamination (x1E12 at/cm2)
Ca
Fe
Al
7 nm dry oxides
17 mm2 capacitors
Ca highly detrimental : roughness !
1.4 Impact on devices: examples
Detrimental impact depends on the technology
10-1
100
101
102
103
104
105
1010 1011 1012 1013 1014 1015
Densité de défauts (cm
-2)
Concentration du Fer (at/cm3)
20 nm16 nm13 nm10 nm
Critère :
claquage à
8MV/cm
Epaisseurs d'oxyde :
After B. Henley : thin oxides are more sensitive to contaminants
Iron concentration
Def
ect d
ensi
ty
Gate oxide thickness
Breakdown criteria: 8MV/cm
Adrien DANEL, UCPSS 2006 10
2006
Ionic species• Corrosion of interconnects by acids • Resist poisoning by bases (amines)
• Growth of post etch residues
� Volatil contamination
1.4 Impact on devices: examples
after M. Yamachika, 1999
Adrien DANEL, UCPSS 2006 11
2006 1.5 Management issues
Yield
Management of contamination
Clean ability ?
Monitoring capabilities ?
What do we face ?
Detrimental impact ?
Cross contamination ?
Different fab areas
Adrien DANEL, UCPSS 2006 12
2006
� Efficient management of contamination is mandatory
Fab competitiveness • Fast and safe introduction of new materials• Small volume production = shared equipments • High value production
� Ultra-cleanliness costs a lot
Define "just enough" levels and rules
1.5 Rationale
Measurement of new species ("exotic metals", Hf as a n example)
Detection of "standard" species on/in new layers (C u in HfO 2 gate oxide)
� New knowledge and metrology required
Understanding of detrimental impact
Understanding of dissemination
Adrien DANEL, UCPSS 2006 13
2006 1.5 Recommendations Generalities
Specification of critical particles simply follows the technology: size = ½ dimension of minimal size of the device
Requirement for metals : a threshold independent of the technology
Yield vs Contamination: very complex , case to case study
ITRS 2005 : 1E10 and 5E9 at/cm 2 for critical metals
To know the baseline of the processes and to keep i t
To clean !! Realistic and simple management rules
referring to ITRS, http://public.itrs.net
Requirements for volatile species : lack of knowledge and precaution suggest an overstatement of ultra cleanliness based on the extrapolation of present data
Adrien DANEL, UCPSS 2006 14
2006 1.5 What do we face ?trends in advanced IC manufacturing
• New substrates• High K dielectrics at gate• Salicides, metals, alloys for contact
• Cu + barriers
FEOL
BEOL
Many new metals and precursors introduced into production lines
That's all ?
• Non volatile memories • Above IC
New electrical, magnetic, optical, mechanical … or even biologic properties push the fast introduction of new materials.
Adrien DANEL, UCPSS 2006 15
2006 1.5 What do we face ?
B, Na, Mg, Al , P, S, Cl, K, Ca, Ti, V, Cr, Mn, Fe, Co, Ni, Cu, Zn, Ga, Ge, As , Br , Sr, Y, Zr, Nb, Mo, Ru, Rh, Pd, Ag, Cd, In, Sn, Sb, Te, Ba, La, Ce, Pr, Nd, Gd, Dy, Er, Yb, Hf , Ta, W, Re, Ir, Pt, Au, Hg, Tl, Pb
� Volatile species:• Condensable organics originated from plastic boxes and clean room
materials (phthalates as an example) and litho solvents • SO2, HF, HCl, HBr, NH3 from chemicals
SSD (Solid State Detector, SiLi diode)SDD (Silicon Drift Detector)
X-ray fluorescence is specific for each element
Non invasive, non contact method
Number of photons is proportional to element concen trationQuantitative analysis
The selected monochromatic X-ray source defines the measurement range: ZW Lβ @ 9.67keV = P to Zn on Si, Kα linesW Mα @ 1.77keV = F, Na, Mg, Al on SiW continuum @ 24keV = → Ru K lines, → U L lines
Contamination identification
Adrien DANEL, UCPSS 2006 25
2006 2.1 TXRF: principle
Spectrum background
time
I
I
C3LLD bkd
i,net
i=
Low Limit of Detection
2
1
2t
1t
t
t
LLD
LLD =
Practical Low limit of Quantification
LOQ ≈ 3LLDDefined at σ=40% of [mean]
Incident beam W-Lβ peakSubstrate Si-Kα peakUseful spectrum area
No excitation for energy < absorption edge
Global fluorescence efficiency decreases(absorption factor X fluorescence desexcitation factor X detection factor)
Adrien DANEL, UCPSS 2006 26
2006 2.1 TXRF: performances
after D. Hellin
Adrien DANEL, UCPSS 2006 27
2006
• mapping-TXRF: local information, 90% surface
• TXRF with multiple sources: Na U
• VPD-TXRF: ultra low Low Limits of Detection
What can TXRF –based method offer ?
2.1 TXRF: equipments
Issue of constant quantification whatever the contaminant shape!
Adrien DANEL, UCPSS 2006 28
2006 2.1 Direct–TXRF: plus and minus
Hf-LααααHf-Ll
Hf-Lαααα Esc
Hf-Mαααα
Hf-Mζζζζ
Hf-Mγγγγ
HfO2 high-k gate film
Hf-LββββW-Lββββ
Mg Al Fe Co Ni Cu Zn
Hf-LααααHf-Ll
Hf-Lαααα Esc
Hf-Mαααα
Hf-Mζζζζ
Hf-Mγγγγ
HfO2 high-k gate film
Hf-LββββW-Lββββ
Mg Al Fe Co Ni Cu Zn
+ Multiple sources cover all elements of interest (except B, F, Li)
Small surface coverage, Throughput
≈ 0.6 W per h4.8%10.8%17 points
≈ 1.0 W per h2.5%5.7%9 points
≈ 1.6 W per h1.4%3.2%5 points
300sec. /point300mm wafer200mm wafer
one beam≈ 2cm 2 per point
Challenge: overlaps
Al (Kα 1.49 keV) Br (Lα)
Mg (Kα 1.25 keV) As (Lα)
K (Kα 3.31 keV) In (Lα)
Na (Kα 1.04 keV) Zn (Lα)
• Peak separation (detector and data processing)• Specific sources
+ Low Limits of Detection
W-Mα 1.77keV source:• LLD – Al = 1.3E11 at/cm2
W-Lβ 9.67keV source:• LLD – Ni = 6.5E8 at/cm2
W-HE 24keV source:• LLD – Mo = 5.0E9 at/cm2
time
I
I
C3LLD
Bgd
i
i=
@ 1000s
Data for a Rigaku FAB300 system
Adrien DANEL, UCPSS 2006 29
20062.1 Surface Profiling–TXRF: plus and minus
+ Low Limit of Detection
+ Local and average information with 90% surface coverage
SIMS Direct method dedicated to bulk analysis LLD of about E16 at/cm 3 (Cu)
Applied to near surface (1µm), LLD is equivalent to E12 at/cm2
XPS Direct method dedicated to surface analysis LLD of about 0.1% to 1% Information on chemical bonds
TOF SIMS Direct method dedicated to surface Very good LLD (< E9 at/cm 2) Local analysis capabilities (on pattern, a few µm 2)Information on elements, molecules and fragments
Industrial aspects for IC manufacturing under validation
2.1 Others
Adrien DANEL, UCPSS 2006 38
2006
Principle All indirect methods detect contaminants via their impact on some electrical properties of the semiconductor, minority carrier lifetime in Si as an example
Measurements on bare wafers: characterization of a few processesWet clean, Thermal treatments, Epitaxy or witness wafers
2.2 Detection of a metallic contamination using indirect method
• At room temperature, Fe contamination into p-type Si bulk forms stable FeB pairs
Assuming that Fe contaminant dominates and with FeB being an efficient recombination center
under medium or high optical flux excitation (µ-PCD ): ττττbulk ≈ ττττFeB
Under low optical flux excitation (SPV) FeB is a poor recombination center but Feinterstitial is a very efficient one
Feibgdbulk
111
τττ+=• After dissociation of FeB pairs
(by heating or optical):
Assuming measurement on
p-type Si (1/τ or Ldiff) is τvolume
with Fe contribution only:
[ ]Fev11
Fei
i
nth
Febulk
σττ
=≅ [ ]bulk
12E82.1Fe
τ≅
with vth = 107 cm/s and σnFei = 5.5E-14 cm2
(µs)(at/cm3)
Adrien DANEL, UCPSS 2006 45
2006 3 Conclusion: summary of methods for metallic contamination
* Automatic equipment for IC manufacturing
Invasive Mapping Throughput LLD Price* Capability
TXRF No
VPD – TXRF
≈ YesResolution 1cm
Edge exclusion 1cm
5min. per point10sec. per point
≈ 1010
at/cm2
> 1M€ GoodNa → U
≠ surfacesbare and smooth
Yes No
Global, entire surface
45min. <109 at/cm2 > 1.5M€ Good
VPD/LPD/LPE– ICPMS/AAS
Yes No 30min. per wafer
<109 at/cm2 < 0.5M€ Very goodAlmost any
element and any substrate
Surface and bulk
+ wet bench+ people
Lifetime(applied to Si)
Yes
Very good resolution
A few min. per wafer
<1011
Fe at/cm3
0.1 to 1M€ Poor
Complex understanding
of results
≈ Yes
IntradiffusionSurface
passivation
Global, entire surface
Na → U≠ surfaces
bare and smooth
Adrien DANEL, UCPSS 2006 46
2006 3 Conclusion
Yield
Management of contamination
Fab competitiveness • Fast and safe introduction of new materials• Small volume production = shared equipments • High value production
Define "just enough" levels and rules
• Cost issues of ultra cleanliness
Understanding of detrimental impact
• Electrical and physical short loops• Yield and crisis expertise
Knowledge on contamination control
• Cleaning – Metrology – Dissemination
Adrien DANEL, UCPSS 2006 47
2006 3 ReferencesMetrology:• "Contamination-Free Manufacturing for Semiconductors and other Precision Products"
Editor R.P. Donavan, M. Dekker Press, NY, 2001.• "Handbook of Silicon Semiconductor Metrology" Editor A. Diebold, M. Dekker Press,
NY, 2001.• "trace-analytical methods for monitoring contaminations in semiconductor-grade Si
manufacturing", L. Fabry et al., J. Anal. Chem. 349, p. 260-271, 1994.• “Organic contamination: Impact, Characterization, Sources and Cleaning during IC
Manufacturing”, M. Claes and S. De Gendt, Proc. of the Electrochem. Soc., PV 2001-29, pp. 320-335 (2001).
TXRF:• "Handbook of X-Ray Spectrometry", Editor R.E. van Grieken, M. Dekker press, NY, 1992.• “Whole surface analysis of semiconductor wafers by accumulating short-time mapping data of
total-reflection X-ray fluorescence spectrometry”, Y. Mori et al., Anal. Chem., 74, pp 1104-1110, 2002.
• "Trends in total reflection X-ray fluorescence spectrometry for metallic contamination control in semiconductor nanotechnology", D. Hellin, to be published in Spectra Chem. Acta B, 2006.
ICPMS:• "Guide to ICPMS", R. Thomas, www.spectroscopyonline.com