Upload
others
View
5
Download
0
Embed Size (px)
Citation preview
No Interfacial
IMC
Maureen Williams, NISTECTC, Reno, NV
May 29, 2007
1
NIST NIST SnSn Whisker TeamWhisker Team
Metallurgy Division Gaithersburg, Maryland
William J. Boettinger
Kil-Won Moon
Gery Stafford
Maureen Williams
2
Motivation for this experimentMotivation for this experimentMeasured Deflection Curves for 3 Alloys (16 µm) 2005 work
-160
-120
-80
-40
0
40
-40
-32
-24
-16
-8
0
8
1000 104 105 106
Can
tilev
er D
efle
ctio
n (µm
) Initial Film Stress (M
Pa)
time (s)
Sn
Sn-Pb
Sn-Cu
3
How to separate the IMC effect from the deposit?How to separate the IMC effect from the deposit?
• To eliminate interfacial IMC electrodeposit Sn on a non-reactive substrate.– Sn on Tungsten (W)
Phase diagram constructed by U. Kattner, NIST, 2005
- based on Binary Phase Diagram, Massalski
4
GoalGoal
• Electrodeposit Sn, SnCu, and SnPb on W, a substrate that does not form an interfacial intermetallic.
• Determine if IMC is needed for hillock & whisker growth.
• Measure– whisker growth
– Stress as a function of time
5
OvOverview of erview of ApproachApproach
• Part One – experiments done in 2006 - Williams M.E., et al., JEM (36) 3, pp.214-219 (2007)
– 1 mm thick pure tungsten (W) substrates– Vapor deposited 0.2 µm (5-9’s) Sn with e-beam– Electroplated 15 µm Sn (made with 18 Mohm cm water)
• Part Two – experiments done in 2007– 150 µm phosphor bronze (Cu) cantilever substrate– Vapor deposited 0.02 µm (4-9’s) W with e-beam on both sides– Vapor deposited 0.2 µm (5-9’s) Sn with e-beam– Electroplated 15 µm Sn (made with 18 Mohm cm water)
6
ResultsResults
Surface eruptions form without interfacial intermetallic (IMC) between Sn and substrate.
Both Cu & W substrates:
•Similar compressive stress values for Sn
•Sn deposits have similar hillock formation.•SnCu deposits have similar whisker formation.•SnPb deposits are clean (some plating bumps).
7
ResultsResults
Avg. Stress 15 minutes after electroplatingmeasured by cantilever beam deflection (2007)
Pure Sn Sn-3 wt% Cu Sn-2 wt% Pb
W - 28.2 MPa - 57.6 MPa No data
Cu -31.2 MPa No data No data
8
Surface Defects on Pure Surface Defects on Pure SnSn Deposits Deposits
W substrate (2006)
10 µm
100 µm
500 µm
141 days oldebeam W on Cu substrate (2007) Cu substrate (2006)
256 days old
500 µm
100 µm
10 µm
26 days old
SEM photos of deposit surface
9
Surface Defects on Surface Defects on SnSn-- 3 wt% Cu Deposits3 wt% Cu Deposits
20 µm
100 µm
500 µm
Cu substrate176 days old
ebeam W on Cu substrate (2007)21 days old
500 µm
100 µm
10 µm
W substrate (2006)
10 µm
500 µm
100 µm
133 days old
SEM photos of deposit surface
10
Experimental OutlineExperimental Outline
• Sample – preparation
– FIB cross sections of the deposit - substrate interface
• Stress Measurement– Deflection of cantilever as a function of time
• Sn and SnCu on W• Sn on Cu and W
11
For Sn electroplating, need to overcome oxide barrier on W substrate surfacePreparation of a Preparation of a NNonon--reactive reactive Cantilever BeamCantilever Beam
Electron beam vapor deposition – Plasma etched the surface of diamond polished Cu cantilever beams in high vacuum (10-8 torr)
– vapor deposited 0.02 um of Tungsten (W) in high vacuum (10-6 torr) to both sides of cantilever beam
– vapor deposited 0.2 um of Tin (Sn) in high vacuum (10-6 torr) to one side of cantilever beam
–Performed ASTM adherence test
Surface of e-beam deposited Sn (0.2 um) on
e-beam deposited W layer (0.02 um)
10 um5 um
Surface of e-beam deposited W layer (0.02 um)
on Cu substrate
12
Cantilever BeamsCantilever Beams
• Sample Mounting– Mask is cut and placed on substrate– Sample is attached to cathode
• Plating Conditions **(same as 2005 experiments)– 1 liter of electrolyte (made with 18 Mohm cm water)
• 100 rpm rotating cathode• 5-9s pure Sn anode• current density of 60 mA/cm2
** Boettinger W.J., et al., Acta Materialia 53, 5033 (2005)
13
Pure Pure SnSn Deposit / Substrate InterfaceDeposit / Substrate Interface
Top FIB photo courtesy of
A. Deal, Lehigh Univ.
Bottom FIB photo courtesyof
G. Galyon & M. Palmer, IBM
3.0 µmSn / Cu interfaceCu substrate
Sn deposit
Cu6Sn5
Sn / W interface
W substrate
Sn deposit
1.0 µm
Sn on W substrateCu6Sn5
Sn on Cu substrate
14
SnSn--3 wt% Cu Deposit / Substrate Interface3 wt% Cu Deposit / Substrate Interface
Top FIB photo courtesy of
A. Deal, Lehigh Univ.
2 µmSn-Cu / W interface
Cu6Sn5
W substrate
Sn-Cu deposit
Bottom FIB photo courtesy of
G. Galyon & M. Palmer, IBM
1.5 µm
Sn-Cu deposit
Cu substrateSn-Cu / Cu interface
Cu6Sn5
SnCu on W substrate
SnCu on Cu substrate
15
-70
-60
-50
-40
-30
-20
-10
0
0 5 10 15 20 25
Sn-Cu
Sn
Time (days)
Stress Calculated from Stress Calculated from Cantilever Beam DeflectionCantilever Beam Deflection
Stoney’s single layer equation is valid - No interfacial IMC
Stre
ss (M
Pa)
16
Comparison of 16 um thick Comparison of 16 um thick SnSn deposit on deposit on TungstenTungsten (W) and (W) and phosphor bronzephosphor bronze (Cu) substrates(Cu) substrates
100
-80
-60
-40
-20
0
0 5 10 15 20 25 30
Sn on P-Bronze
Sn on W
Time (days)
Bea
m D
efle
ctio
n (u
m)
-
Without IMC Sn deflectionrelaxes to constant value
With IMC reaction Sndeflection moves in
positive direction
17
SummarSummaryy• Successfully plated Sn on a non-reactive cantilever beam• Measured residual stress in deposits without interfacial IMC
formation– Sn-Cu deposit shows higher initial stress than pure Sn– Sn has hillocks, and Sn-Cu has filament whiskers– deposits maintain compressive stress initially and after
relaxation• Comparison of Sn deposits on Cu and W cantilever beams
– with IMC, Sn deflection moves in positive direction– without IMC, Sn deflection relaxes to constant compressive
value• Interfacial IMC is not required to initiate whisker growth