text book: silicon vlsi technology fundamentals, …• bunny suits, gloves, air showers, covered...
TRANSCRIPT
Manufacturing, Cleaning, Gettering - Chapter 4
Text Book:Silicon VLSI Technology
Fundamentals, Practice and Modelingg
Authors: J. D. Plummer, M. D. Deal, and P. B. Griffina d G
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ
Manufacturing, Cleaning, Gettering - Chapter 4
Chapter 4• A three tiered approach to control unwanted
impurities in silicon– Clean Rooms
• Minimize particles in the environment– Wafer Cleaningg
• Minimize particles on the wafers– Gettering
• Minimize the effect of particles in the waferMinimize the effect of particles in the wafer
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ2
Manufacturing, Cleaning, Gettering - Chapter 4
FRONT END PROCESSES - CLEANING, LITHOGRAPHY, OXIDATIONION IMPLANTATION, DIFFUSION, DEPOSITION AND ETCHING
• Over the next several weeks, we’ll study front end processes individually.• Cleaning belongs to front end processes and is an important part of fabrication
Reference ITRS Roadmap for Front End Processes
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ3
• Reference - ITRS Roadmap for Front End Processes (http://www.itrs.net/Links/2007ITRS/2007_Chapters/2007_FEP.pdf ).
Manufacturing, Cleaning, Gettering - Chapter 4
Defect and Impurities•• Importance of unwanted Importance of unwanted
impurities increases with impurities increases with shrinking geometries of shrinking geometries of devicesdevicesdevices. devices.
•• 75% of the yield loss is 75% of the yield loss is due to defects caused by due to defects caused by particles (1/2 of the min particles (1/2 of the min feature size)feature size)feature size)feature size)
•• Wafer localized light Wafer localized light scatterersscatterers (LLS) is a (LLS) is a starting wafer teststarting wafer test
C l i i dC l i i d•• Crystal originated Crystal originated (45(45--150nm) particles 150nm) particles (COP) ~1,000Å=void (COP) ~1,000Å=void with with SiOSiOxx --> affect gate > affect gate oxide integrity (GOI)oxide integrity (GOI)g y ( )g y ( )
•• --> anneal in H> anneal in H22 --> oxide > oxide decomposes and surface decomposes and surface reconstructs! & oxide reconstructs! & oxide precipitates from deep precipitates from deep
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ
p p pp p pdepth in Si.depth in Si.
Yield > 90% at the end requires > 99% @ each step
Manufacturing, Cleaning, Gettering - Chapter 4
Wafer Cleaning• Surface films and doped regions must not be
significantly attacked.• Photoresist strip and particle removal typical• Room air and process equipment delivered particles
Significant elements that cause severe problems in silicon– Significant elements that cause severe problems in silicon include: organics, metals (Fe, Au, Cu, etc.) and alkali ions (Na, K, etc.)
Particles
PhotoresistAu
Cu
Fe
ParticlesNa
SiO2 or other thin films
AuFe
Interconnect Metal
N, P
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ5
Silicon Wafer
Manufacturing, Cleaning, Gettering - Chapter 4
Example Sensitivity• Example #1: MOS transistor gate threshold shift
2ε qN (2φ ) qQMOS VTH is given by
VTH = VFB + 2φ f +
2εSqN A (2φ f )CO
+qQ M
CO(1)
– QM is the mobile charge density (number of charges per cm2) of Na+ and K + in the gate oxide
If 10 h 0 1 l V hif b d bIf tox = 10 nm, then a 0.1 volt Vth shift can be caused by QM = 6.5 x 1011 cm-2 (< 0.1% monolayer or 10 ppm in the oxide).
– Prevented MOS technologies initially from being important commercially!
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ6
Manufacturing, Cleaning, Gettering - Chapter 4
Example Sensitivity• Example #2: MOS DRAM
– Storage of charge based on minimal charge leakage in time.• Refresh required to maintain charges after a technology defined period of time. g gy
– Dominated by Shockley, Reed Hall (SRH) recombination (intermediate impurity related energy levels).
• σ is the trap cross sectional area, (10-15cm-2) vth is the minority carrier thermal l it (107 / ) d N i th d it f tvelocity (107 cm/sec), and Nt is the density of traps.
• Deep-level traps (Cu, Fe, Au etc.) Pile up at the surface where the devices are located.
BIT LINE
WORDLINE • Refresh time of several msec
requires a generation lifetime of
1- - - -- --- -- -- STORED
CHARGEON MOS
CAPACITOR
ACCESSMOS
TRANSISTOR
sec 1001R μ≈
⋅⋅=
tth Nvστ (2)
• This requires Nt ≈ 1012 cm-3 or
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ7
q t≈ 0.02 ppb (see text).
Manufacturing, Cleaning, Gettering - Chapter 4
Example: The Role of Surface Cl i i P iCleaning in Processing
Oxide thickness
[Å]
Residual contaminants, layers affect kinetics of processes.
S f ff tSurface effects are very important (MORE) in scaled down devices
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ
Manufacturing, Cleaning, Gettering - Chapter 4
2003 ITRS Front End Processes
Year of Production 1998 2000 2002 2004 2007 2010 2013 2016 2018
Technology N ode (half pi tch) 250 nm 180 nm 130 nm 90 nm 65 nm 45 nm 32 nm 22 nm 18 nm
MPU Printed Gate Length 100 nm 70 nm 53 nm 35 nm 25 nm 18 nm 13 nm 10 nm
DRAM Bits/Chip (Sampli ng) 256M 512M 1G 4G 16G 32G 64G 128G 128G
MPU Trans istors/Chip (x106) 550 1100 2200 4400 8800 14,000
Critical D efec t Si ze 125 nm 90 nm 90 nm 90 nm 90 nm 90 nm 65 nm 45 nm 45 nm
Starting Wafe r Particle s (cm-2 ) <0.35 <0.18 <0.09 <0.09 <0.05 <0.05
Starting Wafe r Total Bulk Fe (cm-3 ) 3x1010 1x1010 1x1010 1x1010 1x1010 1x1010 1x1010 1x1010 1x1010
Metal Atom s on Wafer Surface
After Clean ing (cm-2 )
5x109 1x1010 1x1010 1x1010 1x1010 1x1010 1x1010 1x1010 1x1010
Particles on Wafe r Surface After 75 80 86 195 106 168
Clean ing (#/ wa fer)
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ9
Manufacturing, Cleaning, Gettering - Chapter 4
ITRS 2007 Ed. Front End Processes
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ10
Manufacturing, Cleaning, Gettering - Chapter 4
Clean Factories – Wafer Fab Facility• The wafer fabrication area “Clean Rooms” must
particle free.– Sources of particles
• Air (normal presence of particles) and Water• Machinery – particularly due to friction, metals• People – 5 to 10 million particles per minute, organic• Supplies – brought in to room for use• Processing
– Rooms and People• Clean room limited access, finger wall machine access• Bunny suits, gloves, air showers, covered faces/facemask• Glove box and robots
– Air handling• Positive air pressure, HEPA filters
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ
p ,
11
Manufacturing, Cleaning, Gettering - Chapter 4
Stanford’s CIS Clean Room• The 10,500 square feet clean room is vibration-
isolated from the rest of the building. Support equipment, such as chilled water, vacuum pumps,
i d id t t liair compressors, and acid waste neutralizers, are located in the basement. Corrosive and toxic gases are in a monitored gas area. Liquid gas storage tanks, emergency power generators, and a de-ionized water plant are outdoors.p
• Every 8 seconds, the entire air volume of this room is circulated out to large air ducts, like the meter-wide duct just behind the glass. From there the air flows up to the third floor, down through filters on the
d fl d th b k i t th lsecond floor, and then back into the clean room through the ceiling.
• People inside the room wear bunny suits to minimize the release of human particles into the air. The result is that each cubic foot of air has fewerThe result is that each cubic foot of air has fewer than 100 particles of size 500 nanometers (Class 100), much cleaner than the air in a hospital operating room.
• http://www-cis.stanford.edu/~marigros/CleanRoom.html
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ
• http://snf.stanford.edu/Education/VirtualTour.html
12
Manufacturing, Cleaning, Gettering - Chapter 4
Level 1 Contamination Reduction: Clean Factories
• Air quality is measured by the “class” of the facility.
o Less than X total particleso Less than X total particles greater than 0.5 um per cubic foot of air.
Factory environment is cleaned by:Factory environment is cleaned by:• HEPA filters (99.07% eff.)• Air recirculation (laminar >50 cm/sec)• Bunny suits for workers• Filtration of chemicals and gases• Manufacturing protocols
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ13
(Photo courtesy of Stanford Nanofabrication Facility.)
Manufacturing, Cleaning, Gettering - Chapter 4
Wafer Cleaning• Remove particles, films such as photoresist, and any
other trace contaminants• Distinct processes
– Silicon wafer clean (frontend processes)Post metalization clean (backend processes)– Post metalization clean (backend processes)
• Frontend Wafer Cleaning based on RCA process– RCA was originally the Radio Corporation of America– Werner Kern developed the basic procedure in 1965 while
working for RCA.
• The RCA clean involves the following :The RCA clean involves the following :1. Removal of the organic contaminants (Organic Clean)2. Removal of thin oxide layer (Oxide Strip)3 R l f i i t i ti (I i Cl )
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ
3. Removal of ionic contamination (Ionic Clean)
14
Manufacturing, Cleaning, Gettering - Chapter 4
The RCA Clean• The first step (called SC-1, where SC stands for Standard Clean) is
performed with a 1:1:5 solution of NH4OH + H2O2 + H2O at 70º to 80º C for 10 minutes.for 10 minutes.
– This treatment oxidizes organic films and complexes Group IB and IIB metals as well as Au, Ag, Cu, Ni, Zn, Cd, Co, and Cr. The solution dissolves and regrows a thin native oxide layer on the silicon.
Th t t i h t i i i 1 50 l ti f HF + H2O t• The next step is a short immersion in a 1:50 solution of HF + H2O at 25º C
– Remove the thin oxide layer and some fraction of ionic contaminants.• Perform a DI rinsePerform a DI rinse.• The next step (called SC-2) is performed with a 1:1:6 solution of HCl +
H2O2 + H2O at 70º to 80º C for 10 minutes. – This treatment removes alkali ions and cations (AL, Fe, Mg) that form ( g)
NH4OH insoluble hydroxides in solutions like SC-1. In SC-2 they for soluble complexes. This solution also completes the removal of metal contaminants.
• Perform a DI rinse
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ
Perform a DI rinse
15
Manufacturing, Cleaning, Gettering - Chapter 4
Level 2 Contamination Reduction: Wafer Cleaning
120 - 150ÞC Strips organicsH2SO4/H2O2 120 150ÞC 10 min
Strips organics especially photoresist
Room T 1 min
Strips chemical oxide
2 4 2 21:1 to 4:1
HF/H2O 1:10 to 1:50
DI H2O Rinse Room T
80 - 90ÞC 10 min
Strips organics, metals and particles
NH4OH/H2O2/H2O 1:1:5 to 0.05:1:5
SC-1
• Organic Clean
• RCA clean is “standard
DI H2O Rinse Room T
80 - 90ÞC Strips alkali ionsHCl/H2O2/H2O
process” used to remove organics, heavy metals and alkali ions.
80 90ÞC 10 min
Strips alkali ions and metals
DI H2O Rinse Room T
1:1:6 SC-2
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ16
• Ultrasonic agitation is used to dislodge particles.
DI H2O Rinse Room T
Manufacturing, Cleaning, Gettering - Chapter 4
De-Ionized (DI) Water• Clean water for silicon processing
• DI Water conductivity example:
DI water is necessary: H2O↔H++OH- with [H+]=[OH-]=6x10-13cm-3y 2 [ ] [ ]Diffusivity of : H+ ≈9.3x10-5cm2s-1 →µH+=qD/kT=3.59cm2V-1s-1
of : OH-≈5.3x10-5cm2s-1 → µOH-=qD/kT=2.04cm2V-1s-1
cmMOHHq OHH
Ω=+
=−+
−+ 5.18)][]([
1μμ
ρ
Distilled water is typically 10 uS/cm or 0.1 MΩ cm
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ17
Manufacturing, Cleaning, Gettering - Chapter 4
Gettering• Removing trace elements from active transistor
locations by causing them to combine with defects in the silicon.– Active devices cover a very small portion of the silicon
volume, typically near one surface.yp y– Metals and alkali ions have very high diffusivity.– They tend to be easily captured either in regions with
mechanical defects or in regions which chemically trap them.mechanical defects or in regions which chemically trap them.
• For the alkali ions, gettering generally uses dielectric layers on the topside (PSG or barrier Si3N4 layers).
• For metal ions, gettering generally uses traps on the wafer backside (extrinsic) or in the wafer bulk (intrinsic).
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ
(intrinsic).
18
Manufacturing, Cleaning, Gettering - Chapter 4
Level 3 Contamination Reduction: Gettering
N bl
H 1.008
1
3 4Li Be B C N O F
He 4.003
Ne5 6 7 8 9 10
2
Period
1
2
I A
IIAV AIIIA IV A VIA VII A
Noble Gases
Deep Level Impurites in Silicon
Alkali Ions
11 12
19 20
Li 6.941
Be 9.012
Na 22.99
Mg 24.31
K 39 10
Ca 40 08
Sc 44 96
Ti 47 88
V 50 94
Cr 51 99
Mn 54 94
Fe 55 85
Co 58 93
Ni 58 69
Cu 63 55
Zn 65 39
21 22 23 24 25 26 27 28 29 30
B10.81
Al 26.98
Ga 69 72
C12.01
Si 28.09
Ge 72 59
N14.01
P 30.97
As 74 92
O16.00
S 32.06
Se 78 96
F19.00
Cl 35.45
Br 79 90
Ne20.18
Ar 39.95
Kr 83 80
13 14 15 16 17 18
31 32 33 34 35 36
2
3
4
IIIBIV
BI B IIBV
BVI
BVII
BVIII
Deep Level Impurites in Silicon
39.10 40.08
Rb 85.47
Cs 132.9
Sr 87.62
Ba 137.3
37 38
55 56
44.96 47.88 50.94 51.99 54.94 55.85 58.93 58.69 63.55 65.39
Y 88.91
Zr 91.22
Nb 92.91
Mo 95.94
Tc 98
Ru 101.1
Rh 102.9
Pd 106.4
Ag 107.9
Cd 112.4
39 40 41 42 43 44 45 46 47 48
La 138.9
Hf 178.5
Ta 180.8
W 183.9
Re 186.2
Os 190.2
Ir 192.2
Pt 195.1
Au 197.0
Hg 200.6
57 72 73 74 75 76 77 78 79 80
69.72
In 114.8
Tl 204.4
72.59
Sn 118.7
Pb 207.2
74.92
Sb 121.8
Bi 209.0
78.96
Te 127.6
Po 209
79.90
I 126.9
At 210
83.80
Xe 131.3
Rn 222
49 50 51 52 53 54
81 82 83 84 85 86
5
6
Fr 223
Ra 226
87 88 Ac 227.0
Unq 261
Unp 262
Unh 263
Uns 262
89 104 105 106 1077
allo
w D
onor
s
ow A
ccep
tors
lem
enta
l ic
ondu
ctor
sSh
a
Shal
lo ESe
mi
Elements for gettering:• Alkali Ions• Metals
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ19
Manufacturing, Cleaning, Gettering - Chapter 4
Gettering• Steps:
1. Free elements from current Devices in near
PSG Layer
sites, making them mobile2. Diffuse through silicon to
desired sites
Devices in near surface region
Denuded Zone or Epi Layer
10 - 20 µm
3. Become trapped.
• Aalkali ions: dielectric layers on the topside (PSG or
Intrinsic Gettering
Region 500+ µ m
on the topside (PSG or barrier Si3N4 layers).
• Heavy metal ions: traps on the wafer backside (extrinsic) or in the wafer bulk (intrinsic).
Backside Gettering
Region
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ
bulk (intrinsic).
20
Manufacturing, Cleaning, Gettering - Chapter 4
Gettering• Diffusivity of various elements
– Diffuse as interstitials (they don’t easily fit in the lattice)(through wafer during normal dopant drive-in)( g g p )
– Dopants diffuse as a substitution
C1200 1100 1000 900 800
T (ÞC)
10 -8
10 -6
CrPt
Fe
AuI
Cu
InterstitialDiffusers
I Diffusivity
10 -4
m2
sec
-1)
14
10 -12
10 -10
AuS
Diffusers
Ti
y (c
m2
sec-
1 )
sivity
Diff
usiv
ity (c
m
10 -18
10 -16
10 -14
SiAs
B, P
Dopants
Diff
usiv
ity
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ21
10 -20
0.65 0.7 0.75 0.8 0.85 0.9 0.95 11000/T (Kelvin)
p
Manufacturing, Cleaning, Gettering - Chapter 4
Traps1100Outdiffusion
Precipitation Stacking V I
900
Precipitation gFault
V I
OI
OI
OI SiO2
500
700T e m p e r a t u r e Þ C
Time
Nucleation
OI Diffusion
[OI]
SiO2
OIOI
Time OI Diffusion2
1253.2exp13.0 −⎟⎠⎞
⎜⎝⎛−⋅= seccmeVDO
• “Trap” sites can be created by SiO2 precipitates (intrinsic gettering), or by backside damage (extrinsic gettering)
p ⎟⎠
⎜⎝ kTO
(extrinsic gettering).
• In intrinsic gettering, CZ silicon is used and SiO2precipitates are formed in the wafer bulk throughSiO precipitates (white dots) in bulk
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ22
p p gtemperature cycling at the start of the process.
SiO2 precipitates (white dots) in bulkof wafer.
Manufacturing, Cleaning, Gettering - Chapter 4
Intrinsic Gettering: Oxide Precipitates
Precipitates (size) grow @ high T
Density of nucleation sites grow @ low T
Surface (top) has few defects
The largest & the most dense defects -> the most efficient gettering
Surface (top) has few defects.
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ
Manufacturing, Cleaning, Gettering - Chapter 4
MEASUREMENT METHODSMEASUREMENT METHODS
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ24
Manufacturing, Cleaning, Gettering - Chapter 4
Measurement Methods• Clean factories = particle control
– Detect concentrations < 10/wafer of particles smaller than 0.1 µm
• Unpatterened wafers (blank)C t ti l i i– Count particles in microscope
– Laser scanning systems → maps of particles down to ≈ 0.2 µm(see MEMC video on laser inspection from Chapter 3)
• Patterned wafers– Optical system compares a die with a “known good reference”
die (adjacent die, chip design - its appearance) ( j , p g pp )– Image processing identifies defects (SEM)– Test structure (not in high volume manufacturing)
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ
Manufacturing, Cleaning, Gettering - Chapter 4
Test Structures
Trapped charge
Resistance CapacitanceTrapped charge
QT VTH change
Dielectric breakdown due to particles, metals etc.
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ
Manufacturing, Cleaning, Gettering - Chapter 4
Monitoring the Wafer Cleaning EfficiencyConcentrations of impurities determined by surface analysis
works with SEM He+ 1-3 MeVRutherford Back-scattering
Secondary Ion Mass Spectro-metry: O+ or Cs+
•Excite•Identify (unique atomic signature)•Count concentrations emitted
sputtering and mass analyses
•Count concentrations
Primary beam – e - good lateral resolution
Detected beam – e – good depth resolution and surface sensitivity
X-ray poor depth resolution and poor surface sensitivity
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ
ions (SIMS) excellent
ions (RBS) good depth resolution, reasonable sensitivity (0.1 atomic%)
Manufacturing, Cleaning, Gettering - Chapter 4
Electron Beam Emissions (1)
Inelastic collision with target electrons, which are then emitted from the solid
(as in SEM)
are then emitted from the solidElastic collision of incoming electrons with atoms (reflected back) ~ the same energy as for the incoming electronsg
~ 5 eV
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ
Manufacturing, Cleaning, Gettering - Chapter 4
If X Ray is at the input:
Electron Beam Emissions (2)If X-Ray is at the input:
• el. Emitted= X-ray Photoelectron Spectroscopy (XPS)
• X-ray emitted= X-ray Fluorescence (XRF)
AES scheme is for lighter elements (Z=33 as is crossover between Auger /AES and X-Ray/XES) Several
( )
XPS usually more dominant for lighter elements, XRF for heavier
g y ) SevekeV
The core electron energy levels
Several keV3 3
kicked out a core electron
Electron Microprobe
energy levels
12
12
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ
X-Ray Electron SpectroscopyAuger Electron Spectroscopy
Manufacturing, Cleaning, Gettering - Chapter 4Monitoring of Gettering Through Device Properties and Dielectricp – n leakage, refresh time DRAM junction and dielectric breakdown, β of n-p-n
Material properties : τG(>>τR) in the bulk and on the surface emission↔capture
Device Properties and Dielectric
Photoconductive Decay Measurements
∆n=gopτG
•Carriers are generated due to light •Decrease resistivity•Recombine
)/exp()0()( Rtntn τ−Δ=Δ
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ
Manufacturing, Cleaning, Gettering - Chapter 4
Carrier Generation Lifetime
== +DL
inversion
t =10(NA /ni)τG
Deep Depletion - Return to Inversion via Carrier Generation (measure τG) and surface recombination (s)
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ
p p ( G) ( )
Zerbst technique:
ddt
( CoxC
)2 → τG
ddt
( CoxC
)2 → s if plotted vs. (Cmin/C)-1
s=f(NST, σ) σ=Capture cross section
Manufacturing, Cleaning, Gettering - Chapter 4Lifetime Measurements: Open Circuit Voltage Decay
VD (t) = VD (0) − kT ln(erfc t )
t=0
≈0.7V
Voltage Decay
τ R
Diode switched from ON VD when carriers recombine
off
for t/τ>4
τR =kT/q
dVD /dtMeasurements include surface and
for t/τ>4
bulk recombination
Use also DLTS: identifies traps (Et) and concentrationsThermal or photoexcitation processes in voltage modulable space-charge region (Schottky Diode, p-n junction, MOS Capacitor)
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ
Measured: capacitance, currents or conductance
Manufacturing, Cleaning, Gettering - Chapter 4
Deep Level Transient Spectroscopy• An experimental tool for studying electrically active
defects (known as charge carrier traps) in semiconductors. – DLTS enables to establish fundamental defect parameters
and measure their concentration in the material. – Some of the parameters are considered as defect “finger
prints” used for their identifications and analysis.– identifies traps (Et) and concentrationsidentifies traps (Et) and concentrations– http://en.wikipedia.org/wiki/DLTS
• Thermal or photoexcitation processes in voltage d l bl h imodulable space-charge region
(Schottky Diode, p-n junction, MOS Capacitor)– Measured: capacitance, currents or conductance
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ
p
13
Manufacturing, Cleaning, Gettering - Chapter 4
MODELINGMODELING
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ14
Manufacturing, Cleaning, Gettering - Chapter 4
Models and Simulation Goal
Computer Integrated Manufacturing (CIM)• Tools to monitor and control machines, to maintain
recipes, to control wafer throughput, and to improve operating efficiency (i.e. higher yields)
• Higher initial yields and shorter time to maturity.
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ
Manufacturing, Cleaning, Gettering - Chapter 4
Contamination and Yield• ≈ 75% of yield loss in modern VLSI fabs is due to particle
contamination.Yi ld d l d d i f ti b t th di t ib ti f• Yield models depend on information about the distribution of particles.
• Particles on the order of 0.1 - 0.3 µm are the most troublesome:– larger particles precipitate easily– smaller ones coagulate into larger particles
Den
sity
ity o
f Par
ticle
Y
ield
Los
s
0.6
0.8
1
If particle is in a critical areaDoes not go to zero
Part
icle
Prob
abili
Cau
sing
Y
0.2
0.4
critical areaDoes not go to zero
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ16
Particle Diameter- 0.1 - 0.3 µm
Manufacturing, Cleaning, Gettering - Chapter 4
Contamination and Yield Model• Yields are described by Poisson statistics in the
simplest case.DAY
where AC is the critical area and DO the defect density
OC DAY −= exp
density.• If some fraction of the wafer, G, always produces
zero yields (near the edge of the wafer) this is y ( g )modified as
( ) OC DAGY −⋅−= exp1
• This model assumes independent randomly distributed defects and often under predicts yields
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ
distributed defects and often under predicts yields.
17
Manufacturing, Cleaning, Gettering - Chapter 4
Alternate Models• Use of negative
binomial statistics COC DA
Y⎟⎞
⎜⎛
=
1
1(4)
eliminates these assumptions and is more accurate.
OC
C⎟⎠⎞
⎜⎝⎛ +1
1more accurate.
– where C is a measure of th ti l ti l
0.1Negative Binomial (C = 2)
the particle spatial distribution (clustering factor).
0 001
0.01
Poisson
Negative Binomial (C = 10)
DO = 1 cm -2
0.0001
0.001
0 2 4 6 8 10 120.0001
O
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ18
0 2 4 6 8 10 12Chip Area (cm 2)
Manufacturing, Cleaning, Gettering - Chapter 4
Negative Binomial Yield for C=2
1997 2006 2012
1Do = 0.01 cm-2
D = 0 1 cm-2
• Vertical lines are estimated chip sizes (from the ITRS).
0. 1
C h Y d
Do = 0.1 cm 2
• Note that defect densities will need to be extremely small in the future to achieve the high yields
i d f i ICip Y
ield
C h i p Y i e l d
Do = 1 cm-2
required for economic IC manufacturing.
• (See particle defect densities on
Chi
0 .0 10 5 10 15 20
Chip Area (cm2)
( pTable 4-1.)
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ19
Manufacturing, Cleaning, Gettering - Chapter 4
Overall Yield• The total wafer yield must be the product of the yields
for each process step.
∏=
−
⎟⎠⎞
⎜⎝⎛ +=
levels
i
CiOiCi
CiDAY
1
1
• Note that previous curves refer to chip yield– Table 4-1 and Figures 4-22 and 4-23– Table 4-1 and Figures 4-22 and 4-23
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ20
Manufacturing, Cleaning, Gettering - Chapter 4
Modeling Wafer Cleaning• Cleaning involves removing particles, organics
(photoresist) and metals from wafer surfaces.– Particles are largely removed by ultrasonic agitation during
cleaning.– Organics like photoresists are removed in an O2 plasma or in g p 2 p
H2SO4/H2O2 solutions.– The “RCA clean” is used to remove metals and any remaining
organics.g
• Metal cleaning can be understood in terms of:– Convert metal into ions soluble in the cleaning solution
Oxidation: the process that removes electrons from an atom ( )• Oxidation: the process that removes electrons from an atom (→)• Reduction: the process that gains and electron (←)
Si + 2H 2O ↔ SiO 2 + 4H + + 4e− (5)
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ21
M ↔ Mz+ + ze −
( )
(6)
Manufacturing, Cleaning, Gettering - Chapter 4
Modeling Wafer Cleaning
−+ ++↔+ eHSiOOHSi 442 22 (5)22
−+ +↔ zeMM z
( )
(6)
−+ ++↔ eHOHOH 222 (7)
• If we have a water solution with a Si wafer and metal atoms and ions, the stronger reaction will dominate
++↔ eHOHOH 222 222 (7)
the stronger reaction will dominate.• Generally (7) is driven to the left and (5) and (6) to the right so that
SiO2 is formed and Mz+ is a solution soluble ion.• Good cleaning solutions drive (6) to the right since Mz+ is soluble and• Good cleaning solutions drive (6) to the right since Mz is soluble and
will be desorbed from the wafer surface.
N d l i t b t th i l d t di f th h i t
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ
• No model exists, but there is a general understanding of the chemistry.
22
Manufacturing, Cleaning, Gettering - Chapter 4
Oxidation-Reduction ReactionsOxidant/
Reductant Standard Oxidation Potential
(volts)
Oxidation-Reduction Reaction
Mn2+/Mn 1.05 Mn ↔ Mn2 + + 2e−
SiO2/Si 0.84 Si + 2H2O ↔ SiO2 + 4H+ + 4e− Cr 3+ /Cr 0.71 Cr ↔ Cr3 + + 3e−
Ni 2 + /Ni 0 25 Ni ↔ Ni2 + + 2e−
Less dominant
Ni /Ni 0.25 Ni ↔ Ni + 2eFe 3+ /Fe 0.17 Fe ↔ Fe3+ + 3e−
H2SO 4/H2SO3 -0.20 H2O + H2SO3 ↔ H2SO4 + 2H + + 2e−
Cu 2+ /Cu -0.34 Cu ↔ Cu 2 + + 2e−
/ +O2 /H2O -1.23 2H2O ↔ O2 + 4H + + 2e−
Au3+ /Au -1.42 Au ↔ Au3+ + 3e−
H2O2 / H2O -1.77 2H2O ↔ H2O2 + 2H+ + 2e− O3/O2 -2.07 O2 + H2O ↔ O3 + 2H+ + 2e− Dominant
3 2 2 2 3
• The strongest oxidants are at the bottom (H2O2 and O3). These reactions
go to the left grabbing e-, dominating, and forcing (6) to the right.
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ23
• Fundamentally the RCA clean works by using H2O2 as a strong oxidant.
Manufacturing, Cleaning, Gettering - Chapter 4
Modeling Gettering• Gettering consists of
1. Making metal atoms mobile.2 Migration of these atoms to
Devices in nearsurface region
PSG Layer
2. Migration of these atoms to trapping sites.
3. Trapping of atoms.
surface regionDenuded Zoneor Epi Layer
10 - 20 µm1
2
Aus → Aui
• Step 1 generally happens by “kicking-out” the substitutional atom into an interstitial site. One possible reaction is:
IntrinsicGettering
Region 500+ µm
3
*Trapping
Diffusion
Au + I ↔ Auis:
• Step 2 usually happens easily once the metal is interstitial since most
t l diff idl i thi f3
Au S + I ↔ Au i
metals diffuse rapidly in this form.
• Step 3 happens because heavy metals segregate preferentially to damaged
BacksideGettering
Region *3
Trapping
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ24
segregate preferentially to damaged regions or to N+ regions or pair with effective getters like P (AuP pairs).
Manufacturing, Cleaning, Gettering - Chapter 4
Gettering
“I”
All metal atoms mobile (DMi > DMs 10x)
“I” are closer to wafer surface
DM>> DDopantss
Sol Sol M >>M (Cu Ni)
Except of Ti, Mo, etc.
Sol. Sol MI>>MS (Cu, Ni)Sol. Sol. MI<<MS (Au, Pt)
AuS+I Aui kick-out mechanism, then getterSAus Aui+V dissociative or Frank-Turnbull mech.
I increase improves gettering of AuV increase hinders gettering
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ
Ex. P diffusion, Ion Implant=damage, intrinsic gettering (=I ↑ )
Manufacturing, Cleaning, Gettering - Chapter 4
Metal Diffusion to the Gettering Sites
Initial constant concentration
Time increase
Au diffuses to the waferAu diffuses to the wafer back side and is trappedBack
side trap
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ
Manufacturing, Cleaning, Gettering - Chapter 4Trapping the Metal Atoms at the Gettering Sites (1)
• Trapped by: ion implantation, P diffusion, laser damage, poly-Si films, mechanical damage, etc. But HOW?
at the Gettering Sites (1)
• Physical damage -> metal trapped at defect sites; binding energy Eg depends on T;
i d ( )Fraction Bound=(1-K1exp-Eg/kT)
• Segregation, related to solubility in the silicon perfect crystal and in the gettering regionand in the gettering region
CAu,Si=NSiexp(-EA1/kT) in silicon regionCAu,G=NGexp(-EA2/kT) in gettering region
• The segregation coefficient is defined asThe segregation coefficient is defined ask0=(CAu,G+CAu,Si)/CAu,Si= 1+K2exp[-(EA1-EA2)/kT]For the case of phosphorousk0=1+NG/5x1022exp(0.82eV/kT ) (fraction of Au bound in gettering )
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ
0 G p( ) ( g g )
Manufacturing, Cleaning, Gettering - Chapter 4Trapping the Metal Atoms at the Gettering Sites (2)
• Enhances sol.sol by high dopant concentrations: in “n” Au=acceptor, “p” - Au=donorA + - A - K [A -]/[A ][ -] t t
at the Gettering Sites (2)
Au+e- ↔Au-, Keq=[Au-]/[Au][e-]=constant,
[Au-i]/[Au]ni= [Au-
n]/[Au]n or [Au-n]/[Au-
i] = n/ni
Au acceptor ↑ in “n+” Si (100x if ni(1000°C)=7 14x1018 -> 1021cm-3 doping level)Au acceptor ↑ in n Si (100x if ni(1000°C) 7.14x10 > 10 cm doping level)
• Ion pairing model: AuP ↔ less strainIncreases with Phosphorous concentration
• Coulombic attraction Au+P → Au-P+
• Interaction with point defects V- ↑↑ in “n+” Aui+V- ↔Aus at the trapepd siteInteraction with point defects V ↑↑ in n Aui V Aus at the trapepd site
• Intrinsic gettering - trapping on dislocations and SF which surround precipitates. Dislocations have compressive and tensile stress - accommodate smaller and larger atoms
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ
atoms
Manufacturing, Cleaning, Gettering - Chapter 4Limits and Future Trends in Technology and Modeling (1)Technology and Modeling (1)
• Eliminate defects from wafers: particles, contaminants, clean room.– SMIF Box (Standard Mechanical Interface): limit clean room
size, transport between stations in a clean box
• Wafer cleaning (next slide)Wafer cleaning (next slide)• Gettering:
– intrinsic (less extrinsic), – control Oi, Cs, – use low T processing, – use modeling tool →point defects engineering, g p g g,– release, diffuse, entrap.
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ
Manufacturing, Cleaning, Gettering - Chapter 4Limits and Future Trends in Technology and Modeling (2)Technology and Modeling (2)
• Wafer cleaning in future ICsl h i l (li id– less chemicals (liquids, vapors), more diluted (disposal)N l i– New cleaning:
• Use ozone• Dry and vapor phase,
(Vapors Plasmas)(Vapors, Plasmas) environment! Cluster tools
• Low Energy Physical Processes (sputtering)
Watch for surface roughness
( p g)• Photochemically enhanced
clean
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ
Manufacturing, Cleaning, Gettering - Chapter 4
Summary of Key Ideas• A three-tiered approach is used to minimize contamination in wafer
processing.• Particle control wafer cleaning and gettering are some of the "nuts and• Particle control, wafer cleaning and gettering are some of the nuts and
bolts" of chip manufacturing.• The economic success (i.e. chip yields) of companies manufacturing
chips today depends on careful attention to these issues.c ps oday depe ds o ca e u a e o o ese ssues• Level 1 control - clean factories through air filtration and highly purified
chemicals and gases.• Level 2 control - wafer cleaning using basic chemistry to remove g g y
unwanted elements from wafer surfaces.• Level 3 control - gettering to collect metal atoms in regions of the wafer
far away from active devices.• The bottom line is chip yield. Since "bad" die are manufactured
alongside "good" die, increasing yield leads to better profitability in manufacturing chips.
SILICON VLSI TECHNOLOGYFundamentals, Practice and ModelingBy Plummer, Deal & Griffin
© 2000 by Prentice HallUpper Saddle River NJ31