Characterization of Surface Metals on Silicon Wafers by ...

1

Characterization of Surface Metals on Silicon Wafers by SME-ICP-MS

Featuring the Agilent Technologies7500s ICP-MS

2

Presentation Outline

l Why is Surface Metal Extraction ICP-MS important?

l Techniqueu Wafer surface sampling

u Fundamentals of ICP-MS

l Typical analytical figures of meritu Detection Limits (DLs)

u Background Equivalent Concentrations

u Spike recoveries

l Validation of SME-ICP-MS methodologyu Matrix effects / silicon suppression

u Intentional contamination / extraction efficiency

l Conclusions / References / Website

3

Semiconductor Industry Trends

Metals contamination has become more critical as semiconductor devices move toward:

1) Higher operation speeds

2) Smaller feature and device sizes

3) A larger scale of integration

4) Copper interconnect technology

5) 300 mm wafer technology

4

Silicon Wafer and IC Device Manufacturing

l Semiconductor wafer and device manufacturing consist of a repetitive

series of chemical and physical

process steps

l Since most processes are chemical

in nature, many possibilities exist to

transfer and deposit chemical contaminants on silicon wafers, in

thin films, at interconnects, or within

the layers of an IC device

5

Effects of Metal Contaminants

l They diffuse on the surface, in the substrate, and at interfaces leading

to electrical defects and yield losses

l They can short out conductor lines

and circuit interconnects, cause surface conduction, and decrease

minority carrier lifetime

l Contamination accounts for over 50 % of the yield losses in

semiconductor manufacturing1

1Handbook of Semiconductor Wafer CleaningTechnology by Werner Kern. Chapter 2, section 2.3, page 8.

6

Challenges in Silicon Wafer Surface Characterization

l Sampling Challengesu Reproducible sampling of the wafer surface

u Collection of the resulting sample

u Metals extraction efficiency

u Contamination control

l Measurement Challengesu Small sample volume (250 uL)

u High silicon matrix and aggressive reagents

u Requires ultra-trace detection limits

u Potential for interferences

7

How is the Analysis Done?

l Surface Metal Extraction (SME) techniques are used to concentrate

contaminants present in either the

native oxide or the thermal oxide

layer of the silicon wafer surface

l Inductively Coupled Plasma Mass

Spectrometry (ICP-MS) is then used to determine up to 34 elements in a

single 250 uL extraction droplet

l Typically a 20 minute turnaround for

wafer sampling and analysis

8

SME Chemistry - Oxide Layer

l The primary chemical reaction occurs in an exposure chamber that is filled with hydrofluoric acid vapor. The reaction involves the decomposition of the oxide layer by the acid vapor and the formation of soluble metal fluorides as shown in Equations 1 and 2.

Equation 1

SiO2 (s) + 4HF (aq) SiF4 (aq) + 2 H2O (l)

Equation 2

CuO (s) + 2 HF (aq) CuF2 (aq) + H2O (l)

9

SME Chemistry – Challenge

l Even with the acidic (HF) character of the SME droplet, metals more electronegative than silicon (Au, Cu, etc.) can “plate” out onto the wafer surface:

2CuF2 (aq) + Si SiF4 (aq) + 2Cu (s)

l Metal re-deposition to the wafer surface can be minimized by keeping the SME droplet chemistry highly oxidative. In the presence of hydrofluoric acid and hydrogen peroxide, these metals react to form a soluble fluoride.

Cu(s) + 2 HF(aq) + H2O2 (aq) CuF2(aq) + 2H2O(l)

10

SME-ICP-MS Technique

l SME efficiently concentrates

contaminants from the wafer surface

l ICP-MS provides useful information

on the type, source, and the levels of

metallic contamination at virtually every processing step in

semiconductor manufacturing

l Features:Õ Sub ppt detection limits

Õ The ability to accurately calibrate and quantitate samples for up to 34 elements in a single extraction droplet

Õ Enables real time wafer monitoring for quality control and process contamination control

11

Surface Metal Extraction (SME) Sample Preparation

l The wafer is exposed to HF vapor for 10 minutes to dissolve the SiO2 layer

on the wafer surface and 20 minutes

for the thermal oxide

l The wafer is then placed on a wafer

scanner where a 250 uL extraction

droplet collects the contents of the

dissolved SiO2 layer

l The extraction droplet is deposited in

an auto-sampler and analyzed by

ICP-MS

12

What is ICP-MS?

ICP - Inductively Coupled Plasmau high temperature ion source

u decomposes, atomizes and ionizes the sample

MS - Mass Spectrometeru featuring quadrupole mass

analyzer

u mass range from 7 to 250 amu (Li to U...)

Ý separates all elements in rapid sequential scan

u ions measured using dual mode detector

Ý ppt to ppm levels

Ý isotopic information available

An inorganic (elemental) analysis technique

ICP-MS has the detection limits of GFAA and the sample throughput of ICP-OES

13

Fundamentals of ICP-MS

14

Agilent 7500s ICP-MS System

Open Architecture Sample Introduction System

•Easy access

•MicroFlow nebulizer faci l i tates self

aspirat ion at 20 uL/min

Enhanced ShieldTorch System

•Long l i fe shield plate

•Self al igning shield mount

•Cool p lasma AutoTune

New Omega II Lens System

•Super ior Ion Transmission

•Lowest background ever !

I-AS Autosampler

•Ful ly integrated

•Dynamic r inse stat ion

•Covered for Ultra -trace

analysis

15

l ShieldTorch System and cool plasma conditions give effective interference removal and sub-ppt DLs for virtually all elements

l Soft Extraction mode reduces the background to give a 10 – 100 fold

increase in the signal to background ratio for most elements

l Use of a MicroFlow nebulizer, Peltier cooled spray chamber, wide bore

torch injector, and highly efficient 27.12 MHz generated plasma ensure

complete ionization of the sample matrix

l Flat mass-response curve of Omega II ion lens system is evidence of

efficient analyte ion transmission to the quadrupole mass filter with the

same high sensitivity across the entire mass range

Enhancing the Analytical Performance of the 7500s

16

Load coil

Shield plate

Plasma Torch

Agilent ShieldTorch System

l Developed commercially by

Agilent/Yokogawa

l For the first time, allowed all SEMI

elements to be determined at low or

sub-ppt levels

l Previously troublesome interferences

on K, Ca and Fe are completely

removed!

l Background spectrum is virtually free

from plasma-based peaks

17

What is “Soft Extraction”?

l Soft Extraction - a slight positive

voltage is applied to the first

extraction lens

l Advantages;

u Dramatically reduces the background across the entire mass range

u Lower backgrounds without any sacrifice in sensitivity

u Leads to a 10 to 100 fold increase in signal-to-background ratio for most elements

18

Agilent’s Multi-tune Software

l Switching between ShieldTorch System (cool plasma) conditions and Soft Extraction mode is completely automated using Agilent’s Multi-

tune software

l Multi-tune allows complete data collection in a single acquisitionÝ This saves t ime

Ý Saves sample volume

Ý Reduces potential for sample contamination

l Auto-sampler probe samples each vial only onceu Eliminates risk of contamination associated with multiple sampling

l All data is compiled in a single report

l Stabilization time in switching modes is minimal (20 seconds)

19

Element M a s s Mode of analysis M e a n(cps)

StdDev

% R S D

Lithium 7 Cool 75861.7 986.7 1.3

Beryll ium 9 Normal 2137.6 45.1 2.1

Sodium 23 Cool 52445.2 625.9 1.2Magnesium 24 Cool 24982.3 325.3 1.3

Aluminum 27 Cool 21280.6 317.7 1.5Potassium 39 Cool 26756.3 378.8 1.4

Calcium 40 Cool 8409.0 128.3 1.5

Ti tanium 48 Normal 6811.7 141.4 2.1Vanadium 51 Normal 9413.6 150.0 1.6

Chromium 52 Cool 10876.8 135.4 1.2Manganese 55 Cool 19922.3 273.7 1.4

Iron 56 Cool 13432.0 174.4 1.3

Nickel 58 Cool 10109.9 131.0 1.3Cobalt 59 Cool 14833.6 197.7 1.3

Nickel 60 Normal 2872.5 69.6 2.4

Copper 63 Cool 9837.0 112.6 1.1

Analysis of 500ppt Standard Over 8 Hours with Automatic Switching of Plasma Conditions

20

Analysis of 500ppt Standard Over 8 Hours with Automatic Switching of Plasma Conditions

E l e m e n t M a s s M o d e o f a n a l y s i s M e a n (cps )

S t d D e v

% R S D

Z i n c 64 N o r m a l 4776 .8 78 .7 1 .6

C o p p e r 65 N o r m a l 3373 .8 86 .0 2 .5

Z i n c 66 N o r m a l 2719 .6 42 .2 1 .6

G a l l i u m 71 C o o l 7391 .6 94 .7 1 .3

G e r m a n i u m 72 N o r m a l 2861 .9 54 .5 1 .9

Arsen ic 75 N o r m a l 1536 .0 31 .0 2 .0

S t r o n t i u m 88 C o o l 4560 .3 107 .7 2 .4

Z i r c o n i u m 90 N o r m a l 7635 .6 202 .3 2 .6

N i o b i u m 93 N o r m a l 12759 .8 260 .2 2 .0

M o l y b d e n u m 98 N o r m a l 3629 .0 81 .1 2 .2

S i lver 107 C o o l 4560 .3 51 .0 1 .1

C a d m i u m 114 N o r m a l 4701 .3 128 .9 2 .7

A n t i m o n y 121 N o r m a l 5305 .1 128 .3 2 .4

B a r i u m 138 N o r m a l 14359 .3 332 .2 2 .3

T a n t a l u m 181 N o r m a l 22810 .2 489 .0 2 .1

P la t inum 195 N o r m a l 4073 .5 111 .3 2 .7

G o l d 197 N o r m a l 2394 .2 48 .9 2 .0

T h a l l i u m 205 N o r m a l 13654 .5 364 .3 2 .7

L e a d 208 N o r m a l 10125 .0 292 .8 2 .9

B i s m u t h 209 N o r m a l 15668 .3 384 .1 2 .5

21

New, High Performance Nebulizer Technology

l The Agilent 7500 is designed to operate at low flow rates and new nebulizer technology

matches this capability

l Very high efficiency nebulization means that

better sensitivity can be achieved from very low sample volumes, by operating at flow rates of

<100 uL/min

l The new Agilent Micro Flow all-PFA nebulizers are resistant to most acids and organic

solvents

22

0

5000

10000

15000

20000

25000

30000

35000

40000

45000

50000

0 100 200 300 400 500

Time (min)

Sig

nal In

ten

sity (co

un

ts)

7 Li (0.90%)24 Mg (2.67%)39 K (0.77%)40 Ca (2.64%)55 Mn (1.61%)59 Co (1.40%)

35 uL/min, self-aspirationAgilent Micro Flow nebulizer

Nine Hour Stability -10ppt spike in 0.1% HNO3

Analytical Capability

l By coupling a very high efficiency nebulizer (Agilent Micro Flow

nebulizer) to a very sensitive ICP-MS

system (Agilent 7500 operating under ShieldTorch Cool Plasma

conditions), the current and future

requirements of the semiconductor

industry can be achieved.

l Calibrations at sub-ppt levels

l Quantitation and spike recovery at

single ppt levels

l Limits of Detection at 10’s ppq

23

SME-ICP-MS Analytical Performance

* International Technology Roadmap for Semiconductors

<2.5 E96.1 E63.00.86Zn (68)

<2.5 E91.9 E60.140.25Cu (63)

<2.5 E95.9 E61.00.74Ni (60)

<2.5 E91.1 E70.581.4Co (59)

<2.5 E92.6 E60.950.3Fe (56)

<2.5 E91.7 E50.180.02Mn (55)

<2.5 E93.7 E60.810.04Cr (52)

<2.5 E93.1 E60.970.26Ca (40)

<2.5 E97.0 E63.30.57K (39)

<1.0 E118.9 E50.160.05Al (27)

<1.0 E113.4 E61.000.17Mg (24)

<2.5 E94.2 E60.270.20Na (23)

<1.0 E112.1 E60.020.03Li (7)

ITRS 2009 Requirements

450 mm Wafer DL (atoms/cm2)

0.59 ppm Silicon

7500 DL (ppt)

No Silicon

7500 DL (ppt)

Element / Mass

Agilent 7500 Performance Exceeds ITRS* Requirements for 2009

24

Comments on Detection Limit Results

l Exceptional performance for Ca, K and Fe

l A comparison of the detection limits measured in the calibration matrix and SME synthetic matrix show no significant differences

u This highlights the effectiveness of the ShieldTorch interface in removing matrix-based polyatomic interferences

l Analytes e.g. Zinc, with a high ionization potential are effectively ionized in the high silicon matrix

u This shows the effectiveness of the ShieldTorch System in eliminating matrix based interferences without loss of sensitivity

l 100 uL MicroFlow Nebulizer would further improve detection limitsu These results were obtained using a 20 uL/min sample uptake

25

Background Equivalent Concentrations in the SME Matrix

0.72

0.69

0.21

1.82

0.6

3.8

3.8

500 ppb Si BEC (ppt)

1.4

1.55

0.35

2.78

0.9

2.8

3

No Silicon BEC (ppt)

Pb (208)

Au (197)

Ta (181)

Sn (118)

Zr (90)

Zn (68)

Cu (63)

Element (Mass)

55.2Fe (56)

1.30.9Ni (60)

4.34Co (59)

2.52.9Mn (55)

2.52.7Cr (53)

3622Ca (40)

3625K (39)

11Al (27)

0.540.24Mg (24)

5.33.9Na (23)

0.060.07Li (7)

500 ppb Si BEC (ppt)

No Silicon BEC (ppt)

Element (Mass)

26

ICP-MS Analysis - K, Ca & Fe Calibration Curves in SME Extract

Elimination of Ar and ArO interferences on K, Ca and Fe respectively.

27

10 ppt Spike Recoveries in SME Matrix

10110.160Ni10010.059Co10010.055Mn949.453Cr999.951V

10810.847Ti969.640Ca999.939K

10310.327Al989.824Mg959.523Na959.511B

11011.09Be989.87Li

% RecoveryConc (ppt)MassElement

(SME matrix = 6% H202, 5% HF)

28

…continued

999.9209Bi

10010.0208Pb

10010.0205Tl969.6121Sb

10010.0118Sn

959.5111Cd10310.390Zr

10210.288Sr

919.175As10610.670Ge

999.969Ga

11511.568Zn

999.963Cu% RecoveryConc (ppt)MassElement

29

Comments on Spike Recovery Data

l Excellent spike recoveries for all elementsu Indicating negligible plasma ionization suppression from SME sample

matrix

l The excellent recoveries also indicate the absence of any nebulization or transport interferences

Note: All recoveries were determined without the use of internal standards therefore simplifying sample preparation and eliminating a potential source of contamination

30

Analytical Signal Suppression Experiments

l Purpose: Matrix effects resulting from the extraction sample matrix were evaluated to determine the extent of sample nebulization, transport, and plasma loading interferences. These results were also used to identify the concentration of silicon where suppression of the analytical signal begins.

l Method:u Prepare samples of extraction matrix with increasing levels of high purity

silicon

u Spike each level of silicon with the same concentration of trace metals standard

u Monitor analytical signal suppression in each level of silicon matrix

u Evaluate the use of internal standards for correction of signal suppression

31

Analytical Signal Suppression Results (100 to 1000 ppm Si) No Internal Standard (IS)

Ref: Huiling Lian, Beau Nicoley, Arnold J Howard and M A Radle, Silicon Wafer Thermal Oxide Metals Characterization by Surface Metals Extraction Inductively Coupled Plasma Mass Spectrometry (SME-ICP-MS), Semiconductor International, July 2000, Vol 34, p 217

0

20

40

60

80

100

120

100ppmSi 250ppmSi 500ppmSi 1000ppmSi

Silicon Matrix Concentration

% S

pik

e R

eco

very Lithium (5.39 eV)

Aluminum (5.98 eV)

Calcium (6.11 eV)

Iron (7.87 eV)

Tin (7.34 eV)

32

Signal Suppression Observations...

l Data exhibits excellent spike recoveries for the extraction matrix containing 100 ppm silicon indicating minimal interference from the chemical extraction matrix

l Native oxide extracts contain < 100 ppm silicon. This study proves that there are no suppression effects and no need to use internal standards when doing SME-ICP-MS on the native oxide

l The spike recovery data for silicon concentrations > 100 ppm, typical in extraction samples of the thermal oxide, exhibits signal suppression as indicated by the lower spike recoveries

l The lower spike recoveries indicate either a reduction in the effective ionization potential of the plasma, “plasma loading”, or a nebulization interference.

33

Analytical Signal Suppression Test, (100 to 1000 ppm Si) Using Internal Standards (IS)

0

20

40

60

80

100

120

140

100 ppm Si 300 ppm Si 600 ppm Si 1000 ppm Si

Silicon Concentration

% S

pik

e R

eco

very

, 1

pp

b

Boron / 11

Arsenic / 75

Silver / 107

Zinc / 66

Titanium / 50

Cadmium / 111

34

Internal Standards in High Silicon Matrices

l The analytical signal suppression observed in the higher levels of silicon, typical of the thermal oxide layer, can be corrected for using internal standardization.

l Recommended technique for measuring the thermal oxide layer

l Contamination associated with the use of internal standards can be avoided by improved lab technique, the use of high concentrationinternal standards (1000 ppm), and small volume spikes (50 uL)

35

Summary of Signal Suppression Experiments

l The native oxide layer which contains < 100ppm silicon is routinely analyzed for trace metal contaminants.

u No signal suppression

u No internal standards

l The thermal oxide layer, which is less frequently monitored, contains > 100ppm silicon.

u Moderate suppression

u Correctable with internal standards

l Both applications can be routinely addressed by SME-ICP-MS

36

Intentional Contamination of Silicon Wafers

l Purpose: An intentional contamination study was done to assess the extraction efficiency of removing trace metals from the silicon wafer surface. A standard clean 1 (SC1) cleaning bath was used to deposit trace metals onto silicon wafer surfaces.

l Method: An SC1 formula was prepared at the following concentration; 1% NH4OH, 1 % H2O2, 98 % DI H2O

l The SC1 formula was then spiked with a multi-element standard at 200 ppt

l The SC1 was the poured onto the surface and the wafer exposed for five minutes and then blown dry with nitrogen

l Successive extractions were performed to asses extraction efficiency

37

Metal Extraction Efficiencies -Intentional Contamination

100.01190Ti (50)

100.0163V (51)

99.862740Al (27)

99.911830Be (9)

98.3181040B (11)

100.01130Ni (60)

99.01100Sr (88)98.1271430Fe (56)

100.01130Mn (55)

100.0173Cr (52)

100.01170Ca (40)100.01319K (39)

98.35297Na (23)

97.4274Li (7)

Extraction Efficiency (%)

Extraction Drop 2 Conc (ppt)

Extraction Drop 1 Conc (ppt)

Element (Mass)

38

Metal Extraction Efficiencies -Intentional Contamination Continued

96.5383Pd (105)

97.720840Cd (111)

100.0166As (75)

100.013900Zn (68)99.883510Ga (69)

99.883610Pb (208)

96.6384Ba (137)

97.7284Sb (121)96.4902410Sn (118)

94.9475Nb (93)

98.95443Zr (90)

98.7174Ge (72)

99.1282950Cu (63)

Extraction Efficiency (%)

Extraction Drop 2 Conc (ppt)

Extraction Drop 1 Conc (ppt)

Element (Mass)

39

Summary of Intentional Contamination Study

l The extraction efficiency demonstrates excellent recoveries from the wafer surface

l 33 Elements were spiked into the SC1 solution, all except Ag and Au were successfully added to the wafer surface. Some of the metalswere not reported (Sr, Mo, etc) due to space limitations

l All elemental extraction efficiencies were greater than 95 %

40

Conclusions

l SME-ICP-MS offers a sensitive and accurate method for the characterization of trace metals on silicon wafer surfaces

l Silicon wafers can be prepared and analyzed in less than 20 minutes for the native oxide layer, and 30 minutes for the thermal oxide layer.

l This technique provides real time data for manufacturing quality assessments

l Fully investigated and developed methodologies

41

…continued

l Potential physical interferences associated with the analysis of the SME droplet matrix by ICP-MS can be virtually eliminated using the Agilent ShieldTorch System

l Pioneered by Agilent Technologies and its industry partners, SME-ICP-MS has become the industry standard technique for the characterization of trace metal contaminants on silicon wafer surfaces

l Agilent 7500s exceeds the International Technology Roadmap requirements for 450 mm wafers to the year 2009

42

References

www.agilent.com/chem/semicon