Professional Documents
Culture Documents
Today, submicron feature size at the die level is driving package feature size down to the design-
rule level of the early silicon transistors. At the same time, electronic equipment designers are
shrinking their products, increasing complexity, setting higher expectations for performance, and
focusing strongly on reducing cost. To meet these demands, package technology must deliver
higher lead counts, reduced pitch, reduced footprint area, provide overall volume reduction, aid in
system partitioning, and be cost effective.
Circuit performance is only as good as the weakest link. Therefore, a significant challenge for
packaging is to insure it does not gate device performance. While packaging cannot add to the
theoretical performance of the device design, it can have adverse effects if not optimized. Package
performance, therefore, is the best compromise of electrical, thermal, and mechanical attributes, as
well as the form factor or physical outline, to meet product specific applications, reliability and
cost objectives.
The continuing demand for higher performance products is requiring levels of package
performance unattainable by the molded plastic and ceramic packages of the past decade. These
factors have driven a variety of major innovations in Intel packaging. Intel had in past years
introduced organic packaging with copper interconnect for improved electrical characteristics.
Intel has recently introduced flip chip between die and package as an interconnect approach to
further improve performance and offer very compact packaging. This has resulted in new classes
of technology using organic substrates for both surface mount (Organic Land Grid Array - OLGA)
and thru-hole (Flip Chip Pin Grid Array - FCPGA). The microPGA (µPGA) was introduced to
combine flip chip interconnect with a very small form factor and socketability for compact and
portable systems. While these packages differ in form factor, all can provide the required electrical
and/or thermal performance needed by our advanced products.
Chip scale packaging for memory applications has also been a focus of packaging innovations,
with new CSP form factors including stacked die packaging. Portability is expected to continue as
a strong driver of new packaging approaches.
Fit, form, and function tend to be market specific. Certain Intel devices serve more than one market
need but may require different package attributes. Therefore, "one size fits all" is not a practical
approach to device packaging. Packaging technology is not a single technology, but instead
consists of more than 20 industry proven combinations of core technologies or core technology sets
that can be categorized by package families.
In support of the growing number of Intel devices and to meet the industry demand for package-
specific applications, Intel’s package portfolio has more than doubled during the past ten years.
Ceramic Packages
Socket Mount
CPGA
(Ceramic Pin Grid Array)
(bottom view)
Insertion
or Socket Mount
C-DIP
(Ceramic Dual In-Line Package)
(Side-braze)
A5582-02
A5600-02
Glass-Sealed Packages
CERDIP
(Ceramic Dual In-Line Package)
(Insertion Mount; UV Window)
240817-2 A5603-01
1.3.4 Modules
Modules
SIMM
(Single In-Line Leadless
Memory Module)
(Top View)
SIP
(Single In-Line Leaded
Memory Module)
(Top View)
240817-2 A5686-01
Plastic Packages
- Surface Mount
PSOP
(Plastic Small Outline Package)
(Gull-Wing)
SSOP
(Shrink Small Outline Package)
Dual Row
(Gull-Wing)
Small Outline
Packages SOJ
(SOP) (Small Outline Package)
(J-Lead)
TSOP
(Thin Small Outline Package)
(Gull-Wing)
PLCC
(Plastic Leaded Chip Carrier)
PQFP
(Plastic Quad Flatpack)
Quad Row
QFP
(Quad Flatpack)
FLATPACK
MICRO BGA
(bottom view)
240813-3 A5684-01
Plastic Packages
Insertion Mount (wave solder or socketed)
P-DIP
(Plastic Dual In-Line Package)
Dual Row
SHRINK DIP
(Shrink Dual In-Line Package)
Plastic Packages
Socket Mount
Socket Mount
PPGA
(Plastic Pin Grid Array)
(bottom view)
Type I
Connector
WPS
Type II
Connector
Battery
WPS
240817-6 A5586-01
(front view)
001013 A5734-01
A6175-01
Comments / Footnotes
Comments / Some pin counts available in: Half lead, Wide body, Wide Body, and Standard Type P
Footnotes
Desiccant Pack x
Comments / Footnotes Gull Wing Lead Configuration
Desiccant Pack x x x x
Comments / Footnotes TSOP is “Gull Wing” Configuration
Desiccant Pack x
Comments / Footnotes Gull Wing Lead Configuration
Package attributes for Plastic Ball Grid Array can be found in Chapter 14. Package attributes for
Micro Ball Grid Array can be found in Chapter 15.
28 C
32 C
40 C
48 C
Ceramic Leadless Chip Carrier (LCC), 0.050” 18 R
Pitch, Socket or Surface Mount 20 R
32, 44, and 68 LCCs Available with EPROM or 28 R
Solid Lid
32 R
44 R
68 R
Ceramic Pin Grid Array (CPGA), 0.100” Pitch for 68 A
68L - 208L, 0.100/0.50” for 264L- 387L 88 A
88 A Cavity Down
Socket or Insertion Mount
132 A Cavity Down
68L and 88L “Cavity Up” Available with EPROM or 168 A Cavity Down
Solid Lid
208 A Cavity Down
240-280 A Cavity Down
272-320 A Cavity Down
387 Cavity Down
Ceramic Quad Flatpack (CQFP), 68L Available 68 Q Flat Leads
in 0.050” Pitch. 164L and 196L Available in 0.025” 164 K Flat Leads
Pitch, Socket or Surface Mount 196 K Flat Leads
Ceramic Dual-In-Line Package (CERDIP), 0.100” 16 D
Pitch 18 D
20 D
Socket or Insertion Mount 22 D
24 DP .300”
24 D
28 DP
28 D .300”
32 D
40 D
42 D
Plastic Dual-In-Line Package (PDIP); 0.100” Pitch 16 P
18 P
20 P
64L “Shrink DIP” has a 0.070” Pitch 24 P
Socket and Insertion Mount 24 PD .300”
28 P
28 PD .300”
32 P
40 P
48 P
64 U Shrink
68 N Sq.
84 N Sq.
Plastic Quad Flatpack (PQFP), 0.025” Pitch, 84 KD
Surface Mount
100 KD, KU, NG
Some Packages Available in a Variety of Options: 132 KD, KU, NG
Die UP, Die Down, and Die Down with Heat 164 KU
spreader
196 KU
Quad Flatpack (QFP), Variable Lead Pitch Surface 44 S
Mount 48 S
Quad Flatpack (QFP), Surface Mount, Copper 64 S
Lead Frame 80 SB, S Sq./Rect.
A5581-02
2.1.1 Symbol List for Ceramic Side Braze Dual In-Line Family
Letter or
Description of Dimensions
Symbol
α Angular spacing between minimum and maximum lead positions measured at the gauge
plane
A Distance from seating plane to highest point of body (lid)
A1 Distance between seating plane and base plane
A2 Distance from base plane to highest point of body (lid)
A3 Base body thickness
B Width of terminal leads
B1 Width of terminal lead shoulder which locates seating plane (standoff geometry optional)
C Thickness of terminal leads
D Largest overall package dimension of length
D2 A body length dimension, end lead center to end lead center
E Largest overall package width dimension outside of lead
E1 Body width dimensions not including leads
e1 Linear spacing between centerlines of body standoffs (terminal leads)
eA Linear spacing of true minimum lead position center line to center line
eB Linear spacing between true lead position outside of lead to outside of lead
L Distance from seating plane to end of lead
N The total number of potentially usable lead positions
S Distance from true position centerline of No. 1 lead position to the extremity of the body
S1 Distance from outer end lead edge positions to the extremity of the body
NOTES:
1. Controlling dimension: millimeter.
2. Dimension “e1” (“e”) is non-cumulative.
3. Seating plane (standoff) is defined by P. C. board hole size: 0.0415-0.0430 inch.
E1 E
Pin #1
Indicator Area
D
S S1
Base Plane
Seating Plane
L
A1 A3 A A2
C
B1 e1
B eA
D2 eB
1369-02 A5438-01
231369-2
Millimeters Inches
Symbol
Min Max Notes Min Max Notes
α 0° 10 ° 0° 10 °
A 3.30 5.51 Solid Lid 0.130 0.217 Solid Lid
A 4.04 6.58 EPROM Lid 0.159 0.259 EPROM Lid
A1 1.02 1.52 0.040 0.060
A2 2.29 3.99 Solid Lid 0.090 0.157 Solid Lid
A2 3.02 4.88 EPROM Lid 0.119 0.190 EPROM Lid
A3 2.03 3.66 0.080 0.144
B 0.38 0.56 0.015 0.022
B1 1.27 Typical 0.050 Typical
C 0.23 0.30 Typical 0.009 0.012 Typical
D 50.29 51.31 1.980 2.020
D2 48.26 Reference 1.900 Reference
E 15.24 15.75 0.600 0.620
E1 14.86 15.37 0.585 0.605
e1 2.29 2.79 0.090 0.110
eA 14.99 Reference 0.590 Reference
eB 15.24 17.15 0.600 0.675
L 3.18 4.06 0.125 0.160
N 40 40
S 0.76 1.78 0.030 0.070
S1 0.13 0.005
A Thickness of body
A1 Total package height
A2 Distance from top of base to highest point of body lid
B Width of terminal lead pin
D Largest overall package dimension of length
D1, E1 A body length dimension, corner cutout to corner cutout or end lead center to end lead
center
D2, E2 A body length dimension, end lead center to end lead center
D3, E3 A body length dimension, corner cutout to index corner cutout
D4, E4 Ceramic body fixture
E Largest overall package dimension of width
e Linear spacing
e1 Linear spacing between edges of true lead positions (corner terminal lead pads) lead
corner to lead corner
h Depth of major index feature
j Width of minor index feature
L Distance from package edge to end of effective pad
N The total number of potentially usable lead positions
R1 Inner notch radius
NOTES:
1. Controlling dimension: millimeter.
2. Dimension “e1” (“e”) is non-cumulative.
3. Seating plane (standoff) is defined by P. C. board hole size: 0.0415-0.0430 inch.
4. Dimensions “B”, “B1”, and “C” are nominal.
5. Corner configuration optional.
D
D1 A
A1
D2
E2 B E3 E4
E1
E
N L
R1 D4 Plane 2 Plane 1
(Index Corner)
D3 Terminal #1 Seating Plane
231369-10 A5448-01
Millimeters Inches
Symbol
Min Max Notes Min Max Notes
D
D1 A
A1
D2
e Terminal #1
E2 E4
E1 E3
E N
B
L
Plane 1 Plane 2
R1 D3 45˚ Chamfer
(index corner) Seating Plane
231369-11 A5446-02
Millimeters Inches
Symbol
Min Max Notes Min Max Notes
2.3.1 Symbol List for Square Ceramic Pin Grid Array Family
Letter or Symbol Description of Dimensions
D Seating Plane
A
D1 A3
S1 L Seating
01.65
Ref. Plane
0B (all pins)
e1
D
Pin B2
Swaged Pin
Detail
A1
Swaged Pin
2.29 (4 PL) 1.02 Ref. A2
Ref.
1.52 0.25 Base Plane
45˚ Chamfer 45˚ Chamfer
(index corner) (3 PL)
A5501-01
Millimeters Inches
Symbol
Min Max Notes Min Max Notes
D Seating Plane
A
D1 A3
01.65 S1 L Seating
Ref. Plane
e1 0B (all pins)
D
Pin B2
Swaged Pin
Detail
A1
Swaged Pin
2.29 (4 PL) 1.02 Ref. A2
Ref.
1.52 0.25 Base Plane
45˚ Chamfer 45˚ Chamfer
(index corner) (3 PL)
A5503-01
Millimeters Inches
Symbol
Min Max Notes Min Max Notes
Seating Plane
D A
A3
D1
01.65 S1 L
Ref. Seating
Plane
e1
0B (all pins)
D
Pin C3
Swaged Pin
Detail
A1
Swaged Pin
2.29 (4 PL) 1.02 Ref. A2
Ref.
1.52 0.25 Base Plane
45˚ Chamfer 45˚ Chamfer
(index corner) (3 PL)
A5504-01
231369-16
Millimeters Inches
Symbol
Min Max Notes Min Max Notes
Seating Plane
D A
A3
D1
01.65 S1 L
Ref.
Seating
e1 Plane
Swaged Pin
Detail
Swaged Pin A1
2.29
Ref. (4 PL) A2
1.52
45˚ Chamfer Base Plane
(index corner)
A5505-01
Millimeters Inches
Symbol
Min Max Notes Min Max Notes
Seating Plane
D A
D1 A3
01.65 S1 L
Ref.
e1 Seating
Plane
0B (all pins)
D
Pin D4
Swaged Pin
Detail
A1
2.29 Ref. Pin E5 Swaged Pin
1.52 (optional) A2
(4 PL)
45˚ Chamfer Base Plane
(index corner)
A5506-01
231369-19
Millimeters Inches
Symbol
Min Max Notes Min Max Notes
D
Seating Plane
D1
L
01.65 S1
Ref.
e1
D
Pin E5
(optional)
0B
Pin D4 A2
2.29 Ref. A1
1.52 A
45˚ Chamfer Base Plane
(index corner)
231369-86 A5508-01
Millimeters Inches
Symbol
Min Max Notes Min Max Notes
D
D1 Seating Plane
S1 L
01.65
Ref.
e1
0B
A2
2.29 Ref. Pin D4 A1
1.52
45˚ Chamfer A
(index corner) Base Plane
231369-91 A5510-01
Millimeters Inches
Symbol
Min Max Notes Min Max Notes
D
D1 Seating Plane
S1 L
A4
01.65
Ref.
e1
D
D2
0B
A
2.29 Ref. Pin D4 A1
1.52
45˚ Chamfer A2
(index corner) Base Plane
231369-92 A5511-01
Millimeters Inches
Symbol
Min Max Notes Min Max Notes
D
Seating Plane
D1 L
S1 A4
01.40
Ref.
e1
D D2
0B
A
Pin B2 A1
2.29 Ref.
1.52 A2
45˚ Chamfer
(index corner) Base Plane
231369-93 A5509-01
Millimeters Inches
Symbol
Min Max Notes Min Max Notes
D
D1 Seating Plane
S1 L
01.65
Ref. S1
e1
D1 D
0B
A
Pin C3 A1
2.29 Ref.
1.52 A2
45˚ Chamfer
(index corner)
A5512-01
231369-A0
Millimeters Inches
Symbol
Min Max Notes Min Max Notes
D Seating Plane
D1 L Spreader
A5 S1 A4 D2
01.40
Ref.
E1
E2
E
e1
0B
F1
e2
F2
A F4 Chip Capacitor
2.29 Ref. Lid
Pin B2 A1 F3 Clearance Area
1.52 A2 A
45˚ Chamfer Base Plane
(index corner)
F5
A5513-01
Millimeters Inches
Symbol
Min Max Notes Min Max Notes
A 3.56 4.19 0.140 0.165
A1 — 1.19 Solid Lid — 0.047 Solid Lid
A2 — 5.46 Solid Lid — 0.215 Solid Lid
A4 0.97 1.22 0.038 0.048
A5 1.07 — 0.042 —
B 0.43 0.51 0.017 0.020
D 62.23 62.74 2.450 2.470
D1 58.29 58.55 2.295 2.305
D2 30.48 35.56 1.200 1.400
E 67.31 67.82 2.650 2.670
E1 63.37 63.63 2.495 2.505
E2 56.26 56.77 2.215 2.235
e1 2.29 2.79 0.090 0.110
e2 1.02 1.52 0.040 0.060
F1 — 26.04 — 1.025
F2 9.65 0.290 —
F3 9.65 — 0.380 —
F4 — 4.95 — 0.195
F5 — 1.22 — 0.048
L 3.05 3.30 0.120 0.130
N 387 387
S1 1.52 2.54 0.060 0.100
Category Cerdip
Acronym Cerdip
Lead Configuration Dual-In-Line
Lead Counts 20, 28, 40
Lead Finish Hot Solder Dip
Lead Pitch 0.100”
Board Assembly Type Socket and Surface Mount
NOTES:
1. Alloy 42 Leads.
2. Pressed Ceramic Body.
3. UV Window is available for reprogramming.
E1 E
Pin #1
Indicator Area
D
S S1
Base Plane
Seating Plane
L
A1 A3 A A2
C
B1 e1 eA
B
D2 eB
A5522-01
Millimeters Inches
Symbol
Min Max Notes Min Max Notes
α 0° 10 ° 0° 10 °
A 5.08 0.200
A1 0.38 0.015
A2 3.56 4.44 0.140 0.175
A3 3.56 4.44 0.140 0.175
B 0.41 0.51 0.016 0.020
B1 1.52 Typical 0.060 Typical
C 0.23 0.30 Typical 0.009 0.012 Typical
D 24.38 25.27 0.960 0.995
D2 22.86 Reference 0.900 Reference
E 7.62 8.13 0.300 0.320
E1 7.11 7.90 0.280 0.311
e1 2.29 2.79 0.090 0.110
eA 7.87 Reference 0.310 Reference
eB 8.13 10.16 0.320 0.400
L 3.18 3.81 0.125 0.150
N 20 ½ Leads 20 ½ Leads
S 0.38 1.78 0.015 0.070
S1 0.13 0.005
E1 E
Pin #1
Indicator Area
D
S S1
Base Plane
Seating Plane
L
A1 A3 A A2
C
B1 e1 eA
B
D2 eB
A5522-01
Millimeters Inches
Symbol
Min Max Notes Min Max Notes
α 0° 10 ° 0° 10 °
A 5.72 0.225
A1 0.38 0.015
A2 3.56 4.95 0.140 0.195
A3 3.56 4.70 0.140 0.185
B 0.41 0.51 0.016 0.020
B1 1.52 Typical 0.060 Typical
C 0.23 0.30 Typical 0.009 0.012 Typical
D 36.58 37.72 1.440 1.485
D2 33.02 Reference 1.300 Reference
E 15.24 15.75 0.600 0.620
E1 13.08 15.37 0.515 0.605
e1 2.29 2.79 0.090 0.110
eA 15.49 Reference 0.610 Reference
eB 15.75 17.78 0.620 0.700
L 3.18 4.32 0.125 0.170
N 28 28
S 1.40 2.29 0.055 0.090
S1 0.13 0.005
E1 E
Pin #1
Indicator Area
D
S S1
Base Plane
Seating Plane
L
A1 A3 A A2
C
B1 e1 eA
B
D2 eB
A5522-01
Millimeters Inches
Symbol
Min Max Notes Min Max Notes
α 0° 10 ° 0° 10 °
A 5.72 0.225
A1 0.38 0.015
A2 3.56 4.95 0.140 0.195
A3 3.56 4.70 0.140 0.185
B 0.41 0.51 0.016 0.020
B1 1.52 Typical 0.060 Typical
C 0.23 0.30 0.009 0.012
D 51.69 52.96 2.035 2.085
D2 48.26 Reference 1.900 Reference
E 15.24 15.75 0.600 0.620
E1 13.08 15.37 0.515 0.605
e1 2.29 2.79 0.090 0.110
eA 15.49 Reference 0.610 Reference
eB 15.75 17.78 0.620 0.700
L 3.18 4.32 0.125 0.170
N 40 40
S 1.40 2.29 0.055 0.090
S1 0.13 0.005
α Angular spacing between minimum and maximum lead positions measured at the gauge plane
A Distance from seating plane to highest point of body (lid)
A1 Distance between seating plane and base plane
A2 Distance from base plane to highest point of body (lid)
A3 Base body thickness
B Width of terminal leads
B1 Width of terminal lead shoulder which locates seating plane (standoff geometry optional)
C Thickness of terminal leads
D Largest overall package dimension of length
D2 A body length dimension, end lead center to end lead center
E Largest overall package width dimension outside of lead
E1 Body width dimensions not including leads
eA Linear spacing of true minimum lead position center line to center line
eB Linear spacing between true lead position outside of lead to outside of lead
e1 Linear spacing between centerlines of body standoffs (terminal leads)
L Distance from seating plane to end of lead
N The total number of potentially usable lead positions
S Distance from true position centerline of No. 1 lead position to the extremity of the body
S1 Distance from outer end lead edge positions to the extremity of the body
NOTES:
1. Controlling dimension: millimeter.
2. Dimension “e1” (“e”) is non-cumulative.
3. Seating plane (standoff) is defined by P.C. board hole size: 0.0415 - 0.0430 inch.
4. Dimension “B1” is normal.
5. Details of Pin 1 identifier are optional.
E1 E
Pin #1
Indicator Area
D A2
S S1
Base Plane
Seating Plane
L
A1 A
B1 e1 C
B eA
D2 eB
A5534-01
Millimeters Inches
Symbol
Min Max Notes Min Max Notes
α 0° 10 ° 0° 10 °
A 5.08 0.200
A1 0.38 0.015
A2 3.68 4.06 0.145 0.160
B 0.41 0.51 0.016 0.020
B1 1.52 Typical 0.060 Typical
C 0.23 0.30 Typical 0.009 0.012 Typical
D 31.37 32.00 1.235 1.260
D2 27.94 Reference 1.100 Reference
E 15.75 0.620
E1 13.59 13.84 0.535 0.545
e1 2.29 2.79 0.090 0.110
eA 14.99 Reference 0.590 Reference
eB 15.24 17.78 0.600 0.700
L 3.18 3.68 0.125 0.145
N 24 600 MIL 24 600 MIL
S 1.52 2.03 0.060 0.080
S1 0.74 0.029
E1 E
Pin #1
Indicator Area
D A2
S S1
Base Plane
Seating Plane
L
A1 A
B1 e1 C
B eA
D2 eB
A5534-01
Millimeters Inches
Symbol
Min Max Notes Min Max Notes
α 0° 10 ° 0° 10 °
A 5.08 0.200
A1 0.38 0.015
A2 3.68 4.06 0.145 0.160
B 0.36 0.56 0.014 0.022
B1 1.52 Typical 0.060 Typical
C 0.23 0.30 0.009 0.012
D 36.70 37.34 1.445 1.470
D2 33.02 Reference 1.300 Reference
E 15.75 0.620
E1 13.59 14.00 0.535 0.551
e1 2.29 2.79 0.090 0.110
eA 14.99 Reference 0.590 Reference
eB 15.24 17.78 0.600 0.700
L 3.18 3.68 0.125 0.145
N 28 600 MIL 28 600 MIL
S 1.65 2.16 0.065 0.085
S1 0.86 0.034
E1 E
Pin #1
Indicator Area
D A2
S S1
Base Plane
Seating Plane
L
A1 A
B1 e1 C
B eA
D2 eB
A5534-01
Millimeters Inches
Symbol
Min Max Notes Min Max Notes
α 0° 15 ° 0° 15 °
A 4.83 0.190
A1 0.38 0.015
A2 3.81 Typical 0.150 Typical
B 0.41 0.51 0.016 0.020
B1 1.14 1.40 0.045 0.055
C 0.20 0.30 0.008 0.012
D 41.78 42.04 1.645 1.655
D2 38.10 Reference 1.500 Reference
E 15.24 15.88 0.600 0.625
E1 13.46 13.97 0.530 0.550
e1 2.54 Reference 0.100 Reference
eA 15.24 Reference 0.600 Reference
eB 15.24 17.78 0.600 0.700
L 3.18 3.43 0.125 0.135
N 32 600 MIL 32 600 MIL
S 1.78 2.03 0.070 0.080
S1 1.14 0.045
E1 E
Pin #1
Indicator Area
D A2
S S1
Base Plane
Seating Plane
L
A1 A
B1 e1 C
B eA
D2 eB
A5534-01
Millimeters Inches
Symbol
Min Max Notes Min Max Notes
α 0° 10 ° 0° 10 °
A 5.08 0.200
A1 0.38 0.015
A2 3.94 4.19 0.155 0.165
B 0.41 0.51 0.016 0.020
B1 1.27 Typical 0.050 Typical
C 0.23 0.30 0.009 0.012
D 51.94 52.58 2.045 2.070
D2 48.26 Reference 1.900 Reference
E 15.75 0.620
E1 13.59 13.84 0.535 0.545
e1 2.29 2.79 0.090 0.110
eA 14.99 Reference 0.590 Reference
eB 15.24 17.78 0.600 0.700
L 3.18 3.68 0.125 0.145
N 40 40
S 1.65 2.16 0.065 0.085
S1 0.99 0.039
E1 E
Pin #1
Indicator Area
D A2
S S1
Base Plane
Seating Plane
L
A1 A
B1 e1 C
B eA
D2 eB
A5532-01
Millimeters Inches
Symbol
Min Max Notes Min Max Notes
α 0° 10 ° 0° 10 °
A 4.55 0.179
A1 0.38 0.015
A2 3.75 3.95 Nom 3.85 0.147 0.155
B 0.35 0.51 0.014 0.020
B1 1.40 1.60 0.055 0.063
C 0.25 0.45 Nom 0.35 0.009 0.018
D 52.30 2.06
D2
E 13.50 13.09 Nom 13.70 0.531 0.547
E1
e1 2.29 2.79 Nom 2.54 0.090 0.110
eA 14.74 15.74 Nom 15.24 0.580 0.619
eB
L 3.00 3.60 Nom 3.30 0.118 0.141
N 40 ½ Lead 40 ½ Lead
S 1.65 2.16 0.065 0.080
S1
E1
Pin #1
Indicator Area
D A2
Base Plane
Seating Plane
L
A1 A
B1 C
B e1 eA
D2 eB
A5533-01
Millimeters Inches
Symbol
Min Max Notes Min Max Notes
α 0° 15 ° 0° 15 °
A 5.65 0.22
A1 0.51 0.020
A2 4.15 4.35 0.163 0.171
B 0.35 0.55 0.014 0.022
B1 1.00 Typical 0.040 Typical
C 0.20 0.30 0.008 0.012
D 57.80 58.20 2.28 2.29
D2 55.12 Reference 2.170 Reference
E
E1 16.80 17.20 0.661 0.677
e1 1.60 1.96 0.063 0.077
eA 19.05 0.750
eB 19.50 21.00 0.767 0.826
L 3.00 3.60 0.118 0.142
N 64 64
Pin #1
1.02 (0.04) Indicator
45˚ Chamfer
(4 PL)
A5535-01
Millimeters Inches
Symbol
Min Max Notes Min Max Notes
28 Lead 32 Lead
28 Lead 32 Lead
Symbol
Min Max Notes Min Max Notes
D
0.18 (.007) S C A S -B S D S
3 3
–A– –B–
E1
E
Ne
8 A
0.18 (.007) S C A S -B S
mm (Inch)
Seating Plane –C–
A5539-01
0.18 (.007) S C A S -B S D S
.002mm/mm (In/In) A–B
0.74 (.029) x 30˚
0.58 (.023)
D1 5
0.38 (.015) Min
1.22 (.048) 6
1.07 (.042)
3.81 (0.150)
Max Typ
5 E1
A5537-01
0.81 (.032)
1.27 (.050) N PLCS
0.66 (.026)
4 Sides
0.18 (.007) S C A S -B S D S
–H–
Seating
Plane A1
CP
–C– See Detail L
D2 See Detail J
0.38 (.015) S C A S -B S D S 4
A5541-01
1.27 (.050)
–H–
Seating
Plane A1
CP
–C– LT
See detail J for
E2 Measurement Zone
0.38 (.015) S C A S -B S 4
A5540-01
0.91 (.036)
0.71 (.028)
1.57 (.062)
1.37 (.054)
1.09 (.043)
0.89 (.035)
10˚
45˚
0.46 (.018)
0.36 (.014)
mm (Inch)
A5542-01
0.81 (.032)
0.66 (.026)
0.64 (.025)
Min Bottom of Shoulder
Seating
Plane 0.64 (.025) Measurement Zone for LT
–C– Min
Bottom of Plastic Body
231369-43 A5543-02
–H– –H–
231369-44 A5544-02
20 Lead 44 Lead
Symbol
Min Max Notes Min Max Notes
20 Lead 44 Lead
Symbol
Min Max Notes Min Max Notes
D
0.18 (.007) S B D-E S
Tweezing Surface
–A– 2 –H– Datum Plane
D1 TCP 11
(all four sides)
–D– 3 A1
3 3
–F– –G–
E1
E
–B –
A
3 –E–
0.18 (.007) S A F-G S
Seating Plane –C–
A5545-01
1.22 (.048)
–A– 5
1.22 (.048) D1 1.07 (.042)
1.07 (.042)
6 0.51 (.020) Min
2 PLCS
3.81 (0.150)
Max Typ
E1
5 –B–
A5546-01
0.81 (.032)
1.27 (.050) N PLCS
0.66 (.026)
4 Sides
0.18 (.007) S B A S
4
–H– See Detail L
Seating
Plane A1
CP
–C– See Detail J For
LT
Measurement Zone
See Detail J
D2
0.38 (.015) D-E S 4
mm (Inch)
E2
0.38 (.015) F-G S 4
231369-47 A5547-01
0.91 (.036)
0.71 (.028)
1.57 (.062)
1.37 (.054)
1.09 (.043)
0.89 (.035)
10˚
45˚
0.46 (.018)
0.36 (.014)
mm (Inch)
231369-48 A5548-01
0.89 (.035)
(Typ)
0.64 (.025)
5˚ Typ
0.66 (.026)
0.41 (.016)
5˚
45˚
0.51 (.020)
0.25 (.010)
0.81(.032)
0.66(.026)
Bottom of Shoulder
231369-50 A5552-02
–H– –H–
Detail L
231369-51 A5553-01
A Package height
A1 Standoff
D, E Terminal dimension
D1, E1 Package body
D1, E1 Foot print
D2, E2 Bumper distance
D2, E2 Foot radius location
L1 Foot length
N Leadcount
NOTES:
1. All dimensions and tolerances conform to ANSI Y 14.5M-1982.
2. Datum plane -H- located at top of mold parting line and coincident with top of lead, where lead exits plastic body.
3. Data A-B and -D- to be determined where center leads exit plastic body at datum plane -H-.
4. Controlling dimension, inch.
5. Dimensions D1, D2, E1 and E2 are measured at the mold parting line. D1 and E1 do not include an allowable mold pro-
trusion of 0.25 mm (0.010 in) per side. D2 and E2 do not include a total allowable mold protrusion of 0.25 mm (0.010 in)
at maximum package size.
6. Pin 1 identifier is located within one of the two zones indicated.
7. Measured at datum plane -H-.
8. Measured at seating plane datum -C-.
Symbol Description Min Max Min Max Min Max Min Max Min Max
A Package Height 0.160 0.180 0.160 0.180 0.160 0.180 0.160 0.180 0.160 0.180
A1 Standoff 0.020 0.040 0.020 0.040 0.020 0.040 0.020 0.040 0.020 0.040
D,E Terminal 0.770 0.790 0.870 0.890 1.070 1.090 1.270 1.290 1.470 1.490
Dimension
D1, E1 Package Body 0.647 0.653 0.747 0.753 0.947 0.953 1.147 1.153 1.347 1.353
D2, E2 Bumper Distance 0.797 0.803 0.897 0.903 1.097 1.103 1.297 1.303 1.497 1.503
D3, E3 Lead Dimension 0.500 REF 0.600 REF 0.800 REF 1.000 REF 1.200 REF
D4, E4 Foot Radius 0.723 0.737 0.823 0.837 1.023 1.037 1.223 1.237 1.423 1.437
Location .
L1 Foot Length 0.020 0.030 0.020 0.030 0.020 0.030 0.020 0.030 0.020 0.030
N Leadcount 84 100 132 164 196
Controlling Dimensions in Inches
Symbol Description Min Max Min Max Min Max Min Max Min Max
A Package Height 4.06 4.57 4.06 4.57 4.06 4.57 4.06 4.57 4.06 4.57
A1 Standoff 0.51 1.02 0.51 1.02 0.51 1.02 0.51 1.02 0.51 1.02
D,E Terminal 19.56 20.07 22.01 22.61 27.18 27.69 32.26 32.77 37.34 37.85
Dimension
D1, E1 Package Body 16.43 16.59 18.97 19.13 24.05 24.21 29.13 29.29 34.21 34.37
D2, E2 Bumper Distance 20.24 20.39 22.78 22.93 27.86 28.01 32.94 33.09 38.02 38.18
D3, E3 Lead Dimension 12.70 REF 15.24 REF 20.32 REF 25.40 REF 30.48 REF
D4, E4 Foot Radius 18.36 18.71 20.90 21.25 25.89 26.33 31.06 31.41 36.14 36.49
Location
L1 Foot Length 0.51 0.76 0.51 0.76 0.51 0.76 0.51 0.76 0.51 0.76
N Leadcount 84 100 132 164 196
Controlling Dimensions in mm
D2
D
0.20 (.008) M C A S –B S D S
D1 Base Plane
2 –H–
–D– 3
A1
3 3
–A– –B–
E2 E E1
A
0.20 (.008) M C A S –B S D S –C– Seating Plane
0.10 (.004)
A5554-01
D2 0.25 (.010) M C A S –B S D S S
.002 MM/MM (IN/IN) A–B
D1 0.25 (.010) M C A S –B S D S S
.002 MM/MM (IN/IN) A–B
6
E2 E1
See Detail M
0.25 (.010) M C A S –B S D S
.002 MM/MM (IN/IN) D
0.25 (.010) M C A S –B S D S S
.002 MM/MM (IN/IN) D
A5555-01
0.635 (0.025)
See Detail L
See Detail J
D3 / E3
D4 / E4
D/E
A5557-01
0.13 (.005) M C A S -B S D S 7
0.41 (.016)
0.20 (.008)
2 –H–
0.20 (.008)
0.14 (.005)
A1 –C–
0.31 (.012) D4 / E4
0.20 (.008) L1 8 Deg.
0 Deg.
0.20 (.008) M C A S -B S D S 8
Detail J Detail L
A5556-01
2.8.1.5 Detail M
1.32 (.052)
1.22 (.048) 0.90 (.035) Min.
E2
1.32 (.052)
1.22 (.048)
2.03 (.080)
0.90 (.035) Min. 1.93 (.076)
2.03 (.080)
1.93 (.076)
D2
A5558-01
A Overall Height
A1 Standoff
AAA Lead True Position
b Lead Width
c Lead Thickness
D Terminal Dimension
D1 Body Package
E Terminal Dimension
E1 Body Package
e1 Lead Pitch
L1 Foot Length
N Leadcount
T Lead Angle
Y Coplanarity
NOTE: RECTANGLE PACKAGE
1. Not all packages are available with all products. Contact local Intel Representative for further package information.
D
D1 C
E1 E
P S AB C b
B
D2*
C
A6044-01
e1
A
T
A1
L1
YA
A6045-01
N Lead Count 44 48
A Overall Height 2.35 2.55
A1 Stand Off 0.05 0.05 0.25
b Lead Width 0.20 0.30 .040 0.25 0.30 0.40
c Lead Thickness 0.10 0.15 0.20 0.11 0.15 0.20
D Terminal Dimension 12.0 12.4 12.8 15.1 15.3 15.5
D1 Package Body 10.0 11.9 12.0 12.1
E Terminal Dimension 12.0 12.4 12.8 15.1 15.3 15.5
E1 Package Body 10.0 11.9 12.0 12.1
e1 Lead Pitch 0.65 0.80 0.95 0.70 0.80 0.90
L1 Foot Length 0.38 0.58 0.78 0.65 0.85 1.05
T Lead Angle 0.0° 10.0° 0.0° 7.0°
Y Coplanarity 0.10 0.10
Symbol Description Min Nom Max Min Nom Max Min Nom Max Min Max
Figure 2-1. Principle Dimensions and Data for QFP (Rectangular) Packages
Seating
D
Plane
D1
A1
E1 E
A
e1 b
C
Detail A
Seating
Plane
Y T
e1
See Detail A L1
A5549-01
A Overall Height
A2 Distance from Base Plane to Highest Point of Body (Lid)
B Width of Terminal Leads
B1 Width of Terminal Lead Shoulder Which Locate Seating Plane (Standoff Geometry
Optional)
D Largest Overall Package Dimension of Length
E Largest Over Package Width Dimension Outside of Leads
E1 Body Width Dimension Not Including Leads
e1 Linear Spacing Between Center Line of Body Terminal Leads (Standoffs)
eA Linear Spacing of True Minimum Lead Position Center Line to Center Line
N Total Number of Potentially Usable Lead Positions
E1 E eA
A2
D A
A2 A
e1
B B1
231369-61 A5563-01
Millimeters Inches
Symbol
Min Max Notes Min Max Notes
E1
Pin 1 E
A2
A
eA
e1 Seating Plane
231369-62 A5564-01
Millimeters Inches
Symbol
Min Max Notes Min Max Notes
A Overall Height
A1 Standoff
A2 Package Body Thickness
A3 Lead Height
b Width of Terminal Leads
c Thickness of Terminal Leads
D1 Plastic Body Length
E Package Body Width
e Lead Pitch
D Terminal Dimension
L Lead Tip Length
N Total Number of Potentially Usable Lead Positions
Y Seating Plane Coplanarity
Z Lead to Package Offset
Ø Lead Tip Angle
44 23
A2
C
E D 0
L
Detail A
1 22
D1
A
A1 A
Y
e b See Detail A
[231369-80]
A5565-01
Millimeters Inches
Symbol
Min Nom Max Notes Min Nom Max Notes
A 2.95 0.116
A1 0.050 0.020
A2 2.20 2.30 2.40 0.087 0.091 0.094
b 0.35 0.40 0.50 0.014 0.016 0.020
c 0.13 0.150 0.20 0.005 0.006 0.008
D1 28.00 28.20 28.40 3 1.102 1.110 1.118 3
E 13.10 13.30 13.50 3 0.516 0.524 0.531 3
e 1.27 0.050
D 15.75 16.00 16.25 0.620 0.630 0.640
L 0.75 0.80 0.85 0.030 0.031 0.033
Y 0.10 0.004
Ø 8° 8°
D
5.50 5.50 4-7˚±1˚
0.381
0.250
A2
A
Bottom E-Pin
0.85±0.125
(Cavity#) Mark
0.5x45˚
E E1
2.00
Bottom E-Pin
(Japan) Mark A1 L
Detail A
4.R0.2
1.20 0.381
(Application Laser Mark)
Surface Roughness: Top 3-5µm b
Bottom 5.5 ~ 9.5µm(RZ) 0.12 M C A-B S DS
4.R0.2 4-7˚±1˚
c c1
0.623 B
e
Detail B
A8069-01
Millimeters
Symbol
Min Nom Max
A2
A3 0
E D
Detail A L
D1
A
e A1
Y
See Detail A
b e
[231369-99]
A5566-01
Millimeters Inches
Symbol
Min Nom Max Notes Min Nom Max Notes
A Overall Height
A1 Standoff
A2 Package Body Thickness
A3 Lead Height
b Width of Terminal Leads
c Thickness of Terminal Leads
D Terminal Dimension
D1 Plastic Body Length
E Package Body Width
e Lead Pitch
L Lead Foot Length
N Total Number of Potentially Usable Lead Positions
Y Seating Plane Coplanarity
Z Lead to Package Offset
Ø Lead Tip Angle
Z
See Note 1,3 and 4 A2
See Note 2 See Detail B
Pin 1
E
e
Y
D1 A1
D
Seating
Plane
See Detail A
Detail B Detail A
A2
C
b
0
L
A5567-02
Millimeters Inches
Symbol
Min Nom Max Notes Min Nom Max Notes
A 1.200 0.047
A1 0.050 0.002
A2 0.965 0.995 1.025 0.038 0.039 0.040
B 0.150 0.200 0.300 0.006 0.008 0.012
C 0.115 0.125 0.135 0.004 0.0049 0.0053
D 19.800 20.000 20.200 0.780 0.787 0.795
D1 18.200 18.400 18.600 4 0.717 0.724 0.732 4
E 7.800 8.000 8.200 4 0.307 0.315 0.323 4
e 0.500 0.0197
L 0.500 0.600 0.700 0.020 0.024 0.028
N 32 32
Ø 0° 3° 5° 0° 3° 5°
Y 0.100 0.004
Z 0.150 0.250 0.350 0.006 0.010 0.014
NOTES:
1. One dimple on package denotes Pin 1.
2. If two dimples, then the larger dimple denotes Pin 1.
3. Pin 1 will always be in the upper left corner of the package, in reference to the product mark.
4. Package/Tray orientation: Package pin 1 will always be orientated towards the chamfer tray side (Dimension Z) as per
JEDEC standard JEP-95, CS-008.
Z
See Note 2 A2
See Note 1,3 and 4
Pin 1
e
See Detail B
E
Y
D1 A1
D Seating
See Detail A Plane
Detail B Detail A
0
b
L
A5570-02
Millimeters Inches
Symbol
Min Nom Max Notes Min Nom Max Notes
A 1.200 0.047
A1 0.050 0.002
A2 0.965 0.995 1.025 0.038 0.039 0.040
b 0.150 0.200 0.300 0.006 0.008 0.012
c 0.115 0.125 0.135 0.0045 0.0049 0.0053
D 19.800 20.00 20.200 0.780 0.787 0.795
D1 18.200 18.400 18.600 4 0.717 0.724 0.732 4
E 9.800 10.000 10.200 4 0.386 0.394 0.402 4
e 0.500 0.0197
L 0.500 0.600 0.700 0.020 0.024 0.028
N 40 40
Ø 0° 3° 5° 0° 3° 5°
Y 0.100 0.004
Z 0.150 0.250 0.350 0.006 0.010 0.014
NOTES:
1. One dimple on package denotes Pin 1.
2. If two dimples, then the larger dimple denotes Pin 1.
3. Pin 1 will always be in the upper left corner of the package, in reference to the product mark.
4. Package/Tray orientation: Package pin 1 will always be orientated towards the chamfer tray side (Dimension Z) as per
JEDEC standard JEP-95, CS-008
E See Detail B
D1 A1
D Seating
Plane
See Detail A
Detail A
Detail B
b 0
L
A5568-02
Millimeters Inches
Symbol
Min Nom Max Notes Min Nom Max Notes
A 1.200 0.047
A1 0.050 0.002
A2 0.950 1.000 1.050 0.037 0.039 0.041
b 0.150 0.200 0.300 0.006 0.008 0.012
c 0.100 0.150 0.200 0.004 0.006 0.008
D 19.800 20.000 20.200 0.780 0.787 0.795
D1 18.200 18.400 18.600 4 0.717 0.724 0.732 4
E 11.800 12.000 12.200 4 0.465 0.472 0.480 4
e 0.500 0.0197
L 0.500 0.600 0.700 0.020 0.024 0.028
N 48 48
Ø 0° 3° 5° 0° 3° 5°
Y 0.100 0.004
Z 0.150 0.250 0.350 0.006 0.010 0.014
NOTES:
1. One dimple on package denotes Pin 1.
2. If two dimples, then the larger dimple denotes Pin 1.
3. Pin 1 will always be in the upper left corner of the package, in reference to the product mark.
4. Package/Tray orientation: Package pin 1 will always be orientated towards the chamfer tray side (Dimension Z) as per
JEDEC standard JEP-95, CS-008
E See Detail B
D1 A1
D Seating
Plane
See Detail A
Detail A
Detail B
0
b
L
A5569-02
Millimeters Inches
Symbol
Min Nom Max Notes Min Nom Max Notes
A 1.200 0.047
A1 0.050 0.002
A2 0.965 0.995 1.025 0.038 0.039 0.040
b 0.150 0.200 0.300 0.006 0.008 0.012
c 0.115 0.125 0.135 0.0045 0.0049 0.0053
D 19.800 20.00 20.200 0.780 0.787 0.795
D1 18.200 18.400 18.600 4 0.717 0.724 0.732 4
E 13.800 14.000 14.200 4 0.543 0.551 0.559 4
e 0.500 0.0197
L 0.500 0.600 0.700 0.020 0.024 0.028
N 40 40
Ø 0° 3° 5° 0° 3° 5°
Y 0.100 0.004
Z 0.150 0.250 0.350 0.006 0.010 0.014
NOTES:
1. One dimple on package denotes Pin 1.
2. If two dimples, then the larger dimple denotes Pin 1.
3. Pin 1 will always be in the upper left corner of the package, in reference to the product mark.
4. Package/Tray orientation: Package pin 1 will always be orientated towards the chamfer tray side (Dimension Z) as per
JEDEC standard JEP-95, CS-008
A2 Overall Height
B Width of Terminal Leads
C Thickness of Terminal Leads
D Largest Overall Package Dimension of Length
E Largest Overall Package Width Dimension Outside of Leads
e1 Linear Spacing between Centerline of body Terminal Leads (Standoffs)
L Distance from Seating Plane to End of Lead
D E
A2
e1 B
Seating
Plane
A5571-02
Millimeters Inches
Symbol
Min Max Notes Min Max Notes
0 3.18 ± 0.05
20.45 Max
10.16 ± 0.13
6.35 ± 0.13
e1 B
3.38
Top View
C
E
Side View
231369-68 A5572-01
Millimeters Inches
Symbol
Min Max Notes Min Max Notes
1 80
Front View
B e1
1 80
0.7"
Back View
1 Mbit
32-lead PLCC
(550 x 450 MILS)
0.05" ± 0.004/-0.003
Side View
231369-69 A5573-01
Millimeters Inches
Symbol
Min Max Notes Min Max Notes
2.17.4.1 Type I
Connector
WPS
231369-76 A5574-01
2.17.4.2 Type II
Connector
Battery
WPS
231369-77 A5575-01
3 .378 (85.80)
3.362 (85.40)
Inch (mm)
Connector
2.130 (54.10)
2.122 (53.90) 2X 0.067 (1.70)
0.063 (1.60)
0.041 (1.04) 0.041 (1.04)
0.037 (0.94) 0.037 (0.94)
#34 #1
#68 #35
0.065 (1.65) 0.041 (1.04)
0.061 (1.55) 0.037 (0.94)
231369-78 A5576-01
Battery
Protect
Substrate Area
3 .378 (85.80)
3.362 (85.40)
mm (Inch)
Interconnect Area
Connector
2.130 (54.10)
2.122 (53.90)
0.041 (1.04) 0.041 (1.04) 2X 0.067 (1.70)
0.037 (0.94) 0.037 (0.94) 0.063 (1.60)
#34 #1
#68 #35
0.065 (1.65) 0.041 (1.04)
0.061 (1.55) 0.037 (0.94)
231369-82 A5577-01
2.17.8 Host Connector Pin Configuration for Both Type I and Type
II
Pin Type Pin Length (L) Pin #
2.17.9 Host Connector Pin Configuration for Both Type I and Type
II
110˚
0.018 (0.46)
0.016 (0.42)
0.018 (0.46)
Engagement Area
15˚
10˚
Inch (mm)
0.024 (0.60)
0.02 (0.5) Max 0.016 (0.40)
0.098 (2.50) Min
L
231369-83 A5578-01
mm (Inch)
Pin Insertion
231369-84 A5579-01
0.037 (0.95)
0.142 (3.60)
0.033 (0.85)
0.136 (3.45) 0.057 (1.45)
#1 #34 0.053 (1.35)
3.1 Introduction
The packaging technologies used to manufacture or assemble three basic types of component
packages are summarized in this chapter.
The package families, described in Chapter 1, provide the functional specialization and diversity
required by device and product applications. Material and construction attributes of individual
family members are provided by the following package technologies: (1) fired ceramic, (2) pressed
ceramic, and (3) molded plastic. Intel’s packaging technology using organic substrates will be
discussed in chapters 13, 14, and 15. Cartridge packaging assembly will be discussed in Chapter
16.
Each of the three package families described in this chapter have some similar process steps but,
the packaging materials and the form factors are uniquely different.
The assembly core technology process steps (die attach, wire bond, lid seal, finish) are most
commonly used in the industry today. However, several form factor modifications, driven on one
hand by the advent of “Surface Mount Technology” (Quad Flat Pack packages and Ball Grid
Array) and on the other hand by area array package socketing requirements (Pin Grid Array) are
now the more commonly used form factors for microprocessors.
This chapter will review in detail those core packaging technologies that are common to most of
the standard IC package family types, i.e. DIPs, QFPs & Ceramic PGAs.
Silver-Filled Epoxy/
Silver-Filled
Au-Si Cyanate Ester
Glass
Wafer Backside Metallization for Die Required Not Required Not Required
Attach
Wafer Backside Metallization for Not Required Required Required
Ohmic Contact
Thermal Dissipation Good Good Fair
Electrical Conductivity Good Good Fair
Lead Frame Compatibility N/A N/A Good
Substrate Metallization Compatibility
(a) Gold
(b) Silver Good Poor Good
(c) A12O3 Good Good Good
N/A Good N/A
Figure 3-1 through Figure 3-4 are schematic cross-sections through each of the different die attach
systems currently in use at Intel, for example, gold/silicon eutectic, silver-filled glass, and silver
filled organic adhesives, epoxy, and cyanate ester. The components of each system are identified.
Silicon
}
240818-5
A5588-01
Silver-Filled Glass
Silicon
Al203
240818-6
A5589-01
Silicon
Al203
240818-35 A5590-01
Polymide, Epoxy or
Silicon Cyanate Ester Ag-Filled Adhesive
Silver
Nickel
Copper
240818-7 A5591-01
Like the wafer backside, substrate metallization provides a readily wetted surface for hard
soldering. Metallizations used in hermetic package technologies, both gold and silver, are readily
wetted by the liquid solder at die attach temperatures and actually react with the solder to provide a
high-integrity metallurgical bond to the substrate. Both substrate metallizations resist oxidation,
which can impede wetting, and are electrically conductive for those devices that require electrical
contact.
Plastic packages use a plated silver metallization to adhere to organic adhesives, which are
electrically conductive. Silver-filled glass does not require a metallization, though it will adhere
readily to silver.
Die attach media serve several purposes other than the obvious one of attaching the silicon to the
substrate. They also provide a means of making an electrical connection to the die backside for
those devices requiring it, as well as a path for the conduction of heat from the die to the ambient.
For these reasons, the die attach media used at Intel exhibit good thermal and electrical
conductivity.
The incoming quality of die attach materials is monitored through a series of specifications unique
to each of the die attach media. Tests are performed to measure those specific characteristics
necessary to ensure that materials meet the requirements of die attach applications.
There are also limits to the bond line thickness, both thin and thick, that constrain the process. In
addition, it is imperative to maintain a nearly void-free die attachment. SFG is limited to (1) non-
gold wafer backsides and substrates due to poor adherence to gold, and (2) inert or oxidizing
processing ambients.
Like Au-Si eutectic, SFG requires close process control. Once processed, the SFG material is
stable to extremely high temperatures.
3.3.3 Processes
3.3.3.1 Au-Si
Figure 3-5 is a representation of the gold-silicon phase diagram. In this process, a pure gold
preform is placed into a preheated ceramic package under a heated inert gas. The die is placed onto
the preform and allowed to reach the preset die attach temperature.
oC
Atomic Percentage Silicon
10 10 60 70 80 90 95 97 99
1600
2800 oF 1414˚
1400 L
2400 oF
1200
2000 oF
1064.43˚
1000
1600 oF
800
200 oF
600
1000 oF
363˚
400
600 oF 2.85
200
200 oF
(Au) (Si)
0
0 10 20 30 40 50 60 70 80 90 100
J.C. Chonston
Weight Percentage Silicon
240818-8A A5592-01
As the temperature is raised, silicon from the die begins to diffuse through the diffusion barrier on
the die backside, and at 363 ° C it forms eutectic composition liquid. Once liquid formation occurs,
the reactions proceed rapidly by liquid phase diffusion. As the temperature increases beyond the
eutectic temperature, more silicon is dissolved from the die backside until the equilibrium volume
of silicon is reached. The liquid also begins to dissolve the gold or silver substrate. The amount of
gold or silver that dissolves depends on the temperature and the time of die attach.
Once the liquid is evenly distributed across the silicon backside to ensure intimate contact with all
areas, the package and die are cooled. During cooling, silicon begins to precipitate from the
saturated gold-silicon liquid. These precipitates grow epitaxially from the silicon die backside.
Analysis using a transmission electron microscope has confirmed that the epitaxial region is
continuous with the bulk silicon crystal structure. When the package again reaches the eutectic
temperature, the solder solidifies with a characteristic eutectic-type microstructure. There is no
solid solubility of silicon in gold or vice versa, as evidenced by the phase diagram for the system.
The joint obtained upon cooling is metuallurgically continuous from the substrate to the die.
At higher temperatures, the glass begins to soften, and the silver particles begin to sinter together
into a cohesive mass. Considerable shrinkage accompanies the reactions at higher temperatures,
and care must be taken to ensure that the minimum bond line thickness requirements are
maintained after shrinkage. At the highest temperature, the glass wets the silver particles, silicon
surface, and ceramic substrate, creating a strong chemical bond between the silicon and the
ceramic. Once the material has reached the maximum density, the package is cooled and readied
for subsequent assembly operations.
To prevent premature oxidation of the metallization, a diffusion barrier is necessary. The barrier is
required to adhere to both silicon and gold, yet not interfere with the integrity of the final joint. The
barrier kinetics also must allow rapid diffusion of silicon at the die attach temperatures. Because of
this, it is necessary to control the thickness of the barrier metal. The gold thickness must also be
controlled to prevent oxidation of the barrier metal.
• Au-Si. Larger dies can be attached with this technique. Dies as large as 1.8 x 1.8 cm (0.7 x 0.7
in.) have been successfully attached by this technique.
• Silver-Filled Glass. Within Intel it has been demonstrated that dies as large as 1.0 x 1.0 cm
(0.4 x 0.4 in.) can be successfully attached by this technique.
• Silver-Filled Epoxy/Cyanate Ester. Because there are no limitations on solvent or water
vapor evolution for this type of paste, there are no known die size limitations. Dies as large as
1.3 x 1.3 cm (0.6 x 0.6 in.) have been successfully attached using this die attach material.
Typical failure modes related to the die attach process are die cracking and cohesive failure at die
attach interface. Intel’s die attach process is characterized to prevent these failures. A round tip die
ejector die attach pick up process is used to avoid die backside damage. Intel’s die attach nozzle
and dispensing setup is designed for optimum adhesive coverage with minimum die attach void.
Table 3-3. Typical Tensile Bond Strengths of Various Die Attach Media
Tensile Strength Tensile Strength
Die Attach Media (kg/cm2) (psi)
To a achieve a high-quality wire bond, bonding parameters, piecepart material property and bond
pad metallization integrity must be well understood and controlled.
Intel works closely with wafer fab process development teams and wire and lead frame/ceramic
package vendors to obtain required characteristics for wire, leadframe/package and bond pad
metallization.
Critical parameters for wires are diameter, tensile strength, elongation, chemical composition and
surface contamination.
Key bond pad characteristics include surface hardness, roughness, cleanliness (freedom from glass
residues, oxide and silicon dust), and metallization integrity. The consistency of piecepart materials
properties is essential for a high yield and reliable bond process.
• Visual inspection
• Bond pull
• Bond shear test
• Bond etching
• Electrical testing
• Baking test
• Thermal cycling stress
• Surface analysis
The composite matrix material used in Intel packages is an epoxy cresol novolac polymer. This
crosslinked material is dimensionally stable, ionically clean, and resistant to assembly process and
field-use temperatures. The composite’s largest component by weight is silica filler, added to
provide control of thermal expansion coefficient, thermal conductivity and enhance the material
toughness. The molding compound was also designed for moisture resistance and
manufacturability. Intel optimizes the silica filler size and shape to give better flow property, die
tool wear resistance and moisture resistance property.
The molding compound consists of elastomeric toughening fillers, flame retardants, coupling
agents to improve adhesion between matrix and filler, and release agents to allow removal of the
product from the mold. Figure 3-6 is a scanning electron micrograph showing the epoxy molding
compound structure.
Epoxy Matrix
(Dark Background) Silica Fillers
Toughening Particles
240818-32 A5687-01
mold. A schematic plot of molding compound viscosity as a function of time is shown in Figure
3-7, where T2 is greater than T1. Figure 3-8 outlines the primary steps in the transfer molding
process.
Viscosity
T2
Note:
T2 > T1
T1
Time
240818-33 A5593-01
Molding Clamp
Pressure Applied
Pre-heated Molding
Molding Compund Pellets Compound
Loaded into Mold Die Transfer
Molding Compound
Hardens During Cure
Parts Degated
Runners Removed
240818-34 A5594-01
Molding process parameters such as cure temperature, compound transfer rate, and cure time are
controlled to ensure quality and reliability of a molded package. Of equal importance are the
designs of the mold runner and gate systems, the path through which the molding compounds enter
the package cavity. Potential defects seen if encapsulation process is not optimized such as
incomplete encap fill, encap void and wire sweep. Intel uses design of experiments (DOE)
methodologies that include these variables throughout the development and optimization of the
molding process.
Each of these components has a unique set of physical properties, and mismatches of these
properties, such as the coefficient of thermal expansion, elastical modulus will present a challenge
to maintain the mechanical integrity of the bulk materials and the interfaces between them.
Because of different rates of expansion and contraction, stresses concentrate at the interfaces
between materials during the temperature cycle as seen in accelerated reliability testing and circuit
board mounting. When these stresses exceed the interfacial strength between materials, package
delamination or cracking can occur. The presence of absorbed moisture in the molding compound
exacerbates this phenomenon during surface mount process.
Precautions are taken to prevent failure of the encapsulant during processing, testing, and
application ambient. From the material perspective, the supplier of the molding compound
synthesizes the polymer to exhibit minimum moisture absorption. In addition, coupling agents are
used to maximize adhesion between the epoxy matrix and silica filler to limit moisture ingression.
Optimization of other material properties, such as flexural strength, modulus, and toughness,
ensures that the material can perform under severe moisture and temperature-cycling conditions
without the occurrence of delamination or cracking, which can lead to the ingress of corrosion-
causing contaminants. The molding compound contains elastomeric toughening agents that curtail
the growth of cracks should they occur. These “low-stress” materials also provide protection to the
fragile chip surface by preventing cracking and shear deformation of the thin-film structures that
make up the circuitry.
Careful package and process designs also ensure integrity of the molded package. Package
engineers employ design features that enhance component robustness by creating mechanical
interlocks between molding compound and lead frame. Assembly process engineers design
material flows that maximize adhesion and limit exposure to moisture during manufacturing,
shipping, and board mounting.
• Gold Plate
• Solder Coat
• Solder Plate
Each type of lead finish offers advantages for specific applications and for internal processing.
Previously, gold was considered a universally superior lead finish because of its solderability and
electrical and nonoxidizing properties. Over the past several years, however, it has been
determined that soldering gold-plated component leads directly into a PC board has disadvantages.
Excess gold in the solder joints can result in the formation of a brittle alloy, causing the joints to
fail over time in high-vibration or board-flexing environments.
For example, in military applications requiring leadless chip carrier mounting, such considerations
become critical. Currently, military requirements demand a solder coated part, replacing gold in
direct soldering used for both leadless and leaded components. Gold plating remains the lead finish
of choice for socketed units.
The composition of solder coat is the eutectic alloy of 63% tin and 37% lead. Intel normally applies
the coating in the following sequence: (1) cleaning the leads, (2) applying a flux, (3) dipping the
leads into molten solder, and (4) finishing with a hot water rinse. Great care is taken to minimize
any thermal shock to the package and die during solder coat processing and cleaning of the unit
after coating. Controlling the temperature profile is important to minimize the thermal stress build-
up in the package.
Intel provides solder coat lead finish mostly on plastic and ceramic DIPs for commercial products.
The customer can use this lead finish in a variety of PC board assembly processes, including wave
soldering, infrared, and vapor phase.
The solder uses co-deposited elements from an alloy composition of 85% tin and 15% lead. As
shown in Figure 3-9, the plating on the packages with copper lead frames, with a minimum
thickness of 200 microinches, provides full coverage of the copper without exposing any formed
intermetallics to the air. At the same time, solder plate produces a very solderable finish with a
melting point of approximately 215 ° C. Like tin-plated leads, those plated with tin-lead alloy can be
directly soldered using infrared, vapor phase or socket form.
Plating Region
A5595-02
3.6.4.1 Gold
Packages with gold lead finish are purchased from the package vendor as raw piece parts.
Package Assembly
Fixturing
Process Step
Pretreatment
Mfg. Quality Gate
Rinse
Plate
Rinse
Dry
Quality Assurance
24081810 A5596-01
Package Assembly
Fixturing
Process Step
Pretreatment
Mfg. Quality Gate
Rinse
Dry
Fluxing
Preheat
Solder Coat
Cool
Rinse
Dry
Quality Assurance
24081811 A5597-01
• Physical arrangements and condition of the plating anodes and fixtures, as well as distance to
the part to be plated.
• Chemical analysis and control of solutions in pre-treatment, plating and rinsing baths.
• Plating parameters such as temperatures, rinse flows, voltage, current density and process
times.
• Output quality such as solder thickness, composition, solderability and appearance quality.
The solder coat process is controlled and monitored for:
Clamp Clamp
Shoulders Shoulders
240818-12 240818-13
Gull-Wing Forming
Gull Wing
is formed
Cut 3 45
Tie Bars Degree
Bend
240818-28 240818-21
Lead Cutting/Bending
A5598-01
,,
Figure 3-13. Formation of Gull-Wing Leads (continued)
,,
,
Clamp
Leads
,,
Lead Tip Burrs
From Cutting
240818-14 240818-15
,,
Lead Length Spanking
Cutting
, ,
240818-31
Cut Leads
To Length
240818-30
Flatten
Lead Tips
A5599-01
As part of the lead-forming process, dambars and excess molding compound or flash that has
flowed between the leads and out to the dambars are removed simultaneously from the leads. This
step electrically isolates the leads. Dambar removal and lead forming processes can either be done
on two separate machines or an integrated machine.
Finally, individual units are separated or singulated from the lead frame and carefully transferred
into tubes or trays before being sent to the testing process. These transport media such as tubes and
trays are designed to protect the leads.
Permeability (gm/cm-s-Torr)
10-4 10 -8 10-10 10 -12 10 -14 10 -16
10 1
10 0 Silicones
Epoxies
10-1
Thickness (cm)
Fluorocarbons
10-2
Glasses
10-3
Metals
10-4
MIN H DAY MO YR 10 100
YR YR
A5602-01
3.7.2 Materials
1100 ˚C 10 20 30 40 50 60 70 80 90
1064.43˚
˚
1900 F
1000
˚
1700 F
900
800 L
(Au)
˚
1400 F
700
˚
1200 F
600
˚
1000 F
490˚
500
86.2 96.8
418˚
˚
800 F
400
700˚F 309˚ 280˚
300
231,968 252˚
˚
500 F
217˚
200 03 10
(Sn)
β γ δ
˚
300 F
100
Sn 10 20 30 40 50 60 70 80 90 Au
240818-22 A5604-01
Figure 3-17. Schematic Cross-Section of a Metal Lid Seal Package (Thicknesses Not to Scale)
Gold Plating
Nickel Plating
Tungsten
Cofired Thick Film
AI 20 3 AI 20 3
2400818-24 A5828-01
The gold-tin eutectic seal is made to a seal ring on the surface of the alumina ceramic. The seal ring
is composed of a tungsten thick film that is co-fired to the ceramic material and plated with a thin
nickel diffusion barrier and gold overplate. The gold overplate prevents oxidation and provides a
wettable surface for the solder. Figure 3-17 is a schematic of the seal area.
The lid material is gold-plated Alloy 42 (a nickel-iron alloy with low thermal expansion) to which
is attached an 80% gold, 20% tin alloy (eutectic composition) preform. The seal preform is
attached by spot welding. The lid material was chosen for its stiffness and because the thermal
expansion of Alloy 42 is quite close to that of aluminum oxide ceramic. The stiffness prevents
damage during testing, and the low coefficient of thermal expansion (CTE) difference minimizes
stress at the seal due to thermal expansion mismatch.
The gold thickness of the plating used in the seal ring area is important to the success of sealing this
system. Gold thickness must be controlled to prevent premature nickel diffusion and oxidation at
the gold surface, which can create a poorly wetted surface that will result in seal defects. The nickel
diffusion barrier prevents the interdiffusion of tungsten and gold that can also result in seal defects.
The seal preform width and thickness are also specified and controlled to ensure that an adequate
amount of solder is present to create a continuous seal.
The sealing process takes place in a forming gas environment which removes oxides from the
sealant surface. Oxides, when present, inhibit solder flow and cause non-wetting which results in a
non-hermetic seal. The pattern and solder dimensions are optimized to ensure maximum seal
quality and reliability performance.
Figure 3-18. Schematic Cross-Section of a Ceramic Lid Seal Package (Not to Scale)
Ceramic Lid
Metallization
Layer
Lead-Based
Solder
Gold Plating
Nickel Plating
Tungsten
Cofired Thick Film
AI 20 3 AI 20 3
2400818-36 A5606-01
Zn0
B 20 3 Pb0
240818-25 A5607-01
The glasses used for package sealing have high thermal expansions relative to the ceramic. To
reduce thermally induced package stresses, it is necessary to reduce the thermal expansion of the
glasses by adding low-thermal-expansion fillers. These fillers are chosen to be compatible with the
lead glass and do not react during processing.
Because the glasses have similar compositions, they have similar strengths. Table 3-5 lists the
measured bending strengths for several different sealing glasses commonly used. The glass used by
Intel can be seen to have bending strengths equivalent to other commonly used glasses. The
fracture toughness of Intel’s sealing glass is also found to be quite similar. It is expected that the
mechanical performance of Intel’s sealing glass will be equivalent to other commonly used glasses.
Table 3-5. Bending Strengths and Fracture Toughness of Several Lead-Based Sealing
Glasses
Four Point Bending Strength Fracture Toughness
Glass (MPa) (MPa m )
Lead
Frame
Base
Glass
Base
Ceramic
240818-27 A5608-01
3.7.3 Processes
Sealing can be accomplished in either an inert (nitrogen) or reducing atmosphere (forming gas
N2H2). The seals obtained with either atmosphere are equivalent in performance. The reducing
atmosphere is used to build additional margin into the sealing process.
The sealing process is controlled by monitoring the seal temperatures with a thermocouple
embedded in a package that is passed through a fully loaded furnace. This ensures that the furnace
profile obtained is representative of that seen by actual product. Water vapor content of the sealing
atmosphere is also controlled to ensure that the final internal-cavity water vapor levels meet
industry requirements.
As with the metal-sealed packages, the process is controlled by profiling a fully loaded furnace by
means of a thermocouple embedded in a CERDIP-type package, to get an accurate representation
of the actual sealing conditions. Water vapor content and flow rates of the atmosphere are
controlled to ensure that the internal-cavity water vapor requirements are met.
3.7.4 Performance
Intel’s hermetic packages can be tested for hermeticity using two tests: fine and gross leak. Two
tests are required, since neither test can adequately detect the entire range of leak sizes. Fine leak
uses helium and a mass spectrograph to identify fine leaks. Larger leaks do not show up, as the
gases leak out too quickly for the detectors to identify. The industry-acceptable leak rate for
hermetic packages is less than 5 X 10-8 Std. CC atm/min.
Gross leak testing, which identifies larger leaks that fine leak testing cannot detect, employs a high-
vapor pressure fluorocarbon liquid as the detection medium. The unit is pressurized in a container
of the fluorocarbon to force the liquid into potential leakage paths. After pressurization, the unit is
immersed in a hot bath, which vaporizes the fluorocarbon trapped in the leak. A steady stream of
bubbles indicates the location of the leak. The gross leak test cannot be used for small leaks, as the
liquid will not penetrate leaks smaller than a certain size.
In simple terms, a package electrical model translates the physical properties of a package into
electrical characteristics that are usually combined into a circuit representation. The typical
electrical circuit characteristics that are reported are DC resistance (R), inductance (L), capacitance
(C), and characteristic impedance (Z_o) of various structures in the package. A package model
consists of two parts, both of which are necessary for fully understanding the electrical
performance effects of the package environment on Intel’s microprocessors.
The first is an I/O lead model that describes the signal path from the die to the board. Depending
upon the complexity of the model required for simulation purposes, the I/O lead model can take the
form of a simple lumped circuit model, a distributed lumped circuit model, a single-conductor
transmission-line model, or a multiple-conductor transmission-line model. While lumped models
can adequately model simple effects, such as DC resistive voltage drop, more sophisticated models
like the multiple-conductor transmission-line model include effects such as time delay and
crosstalk.
The second part of a package model is a power-distribution network that describes the power
scheme of the package. Like the I/O lead model, the sophistication of the power-distribution
network can vary from a simple distributed lumped model to a complex circuit network called a
PEEC (partial-element equivalent circuit) network. The simpler models can describe gross
electrical characteristics of the power-distribution network, such as DC resistive drop for the entire
package, whereas the more complex models enable the analysis of the effects of the power-
distribution topology.
The following sections provide an overview of basic package modeling terminology and
methodology, an overview of experimental characterization, and modeled data for the packages
that Intel uses for its most advanced microprocessors. These products are housed in packages
representative of a broad spectrum of package technologies, including CPGA (ceramic pin-grid
array), PPGA (plastic pin-grid array), H-PBGA (high thermal plastic ball grid array), TCP (tape
carrier package), OLGA (organic land-grid array), and FC-PGA (flip-chip pin-grid array). For the
sake of completeness, package parasitics data for older package technologies are included in the
final part of this section. The package types included are multilayer molded (MM-PQFP), ceramic
quad flatpack (CQFP), plastic leaded chip carrier (PLCC), quad flatpack (QFP, SQFP, TQFP), and
small outline packages (TSOP, PSOP). These packaging technologies are no longer used for Intel’s
leading-edge microprocessors but are still used for other products.
Since the packages used for Intel’s microprocessors are custom designed for each product, the
parameters given in the following sections may not reflect the actual values for a particular
product. The actual parameters can be obtained by contacting a local Intel sales office. For
electrical parameters of packages not listed, please contact your local Intel field sales office.
4.1.1 Terminology
Equation 4-1.
ρL
R = ------
A
Ceramic packages have relatively high resistance because of the high resistivity of the tungsten
alloy metallization used with ceramic technology. Plastic/organic packages have much lower
resistance because the metallization used is either copper or a copper alloy. The resistivity of
copper or copper alloys is approximately a factor of 6-12 lower than that of tungsten alloys.
Equation 4-2.
εA
C = ------
t
where A is the area of one of the plates, ε is the permitivity of the material separating the plates,
and t is the thickness of the material.
Capacitances which are important to package electrical performance are “loading” capacitance,
“lead-to-lead” capacitance, and “decoupling” capacitance. The loading capacitance is the total
capacitance of a lead with respect to all surrounding conductors. The lead-to-lead capacitance is the
mutual capacitance between the two leads. The loading capacitances are the diagonal terms in the
so-called “short-circuit” capacitance matrix, and the lead-to-lead capacitances are the off-diagonal
terms. The lead-to-lead capacitance and the mutual inductance determine the extent of
electromagnetic coupling between the two leads. Decoupling capacitance is the total capacitance
between the power leads and the ground leads. In a ceramic PGA with power/ground planes, the
decoupling capacitance is due to the capacitance between power and ground planes, as well as
added discrete capacitors to the package. In plastic PGA packages, decoupling capacitance is
usually provided by adding discrete capacitors to the package. Decoupling capacitance serves as a
reservoir which provides part of the energy required when buffers switch. This reduces the AC
voltage drop, also called the switching noise or the ground bounce, of the power/ground path.
4.1.1.3 Inductance
A simple definition of inductance (L) is the property of a conductor that describes the
proportionality between current change and induced voltage. An inductor is any conductor across
which there is a voltage drop when there is a time-varying current present. This aspect of a package
is important in determining the extent of the effects of crosstalk and simultaneous switching noise.
The classical definition of inductance implies that the inductance is that of a current loop, however,
a loop can be segmented, and partial-self and partial-mutual inductances can be attributed to each
segment. This is a useful concept for analyzing package AC noise and is widely used today. For
example, a pin inductance is a partial-self inductance.
Another useful term is the “open-loop” inductance which is the inductance of a loop with gaps at
two ends. We can visualize a segment of a transmission line as an open-loop and the total
inductance of that segment as the open-loop inductance. Inductances used in analyzing behavior
such as propagation delay and crosstalk of electrical interconnects are, in general, open-loop
inductances.
When partial inductances are used in package electrical performance analysis, it is essential to
understand the current direction in each segment. Wrong assumptions on current directions can
lead to erroneous results. The term “current return path” is widely used to stress the importance of
understanding the current direction in each segment involved.
The inductance values are determined by the lead length and cross-sectional dimensions, the
spacing between leads, the spacing between the power or ground plane, the permeability of the
conductor, and the number of leads involved. A general rule of thumb is that the smaller the entire
current loop involved the lower the inductance. This is an important concept to be aware of when
designing packages and systems, in general. Each signal needs to have a nearby, well-defined, and
continuous return path. There are few simple formulas for inductance because the inductance is
dependent upon both the physical geometry of the structure and the current return path. A few
classic problems have been solved in closed form. Software codes that use electromagnetic
analysis techniques are usually used to compute the inductance of the complex structures in
packages.
Equation 4-3.
L
Z_o = ----
C
L and C are per-unit-length values of the inductance and capacitance of the line, so Z_o is
independent of line length.
Z_o is an important factor in determining the amount of signal reflection that will occur at the
boundary between the die and the package and at the boundary between the package and the board.
The amount of reflection and, thus signal distortion, is directly proportional to the amount of
mismatch between Z_o’s at these boundaries. As an example, the Z_o of traces in FR-4 boards
typically used for motherboards is around 50 ohms, so typical packaging technologies attempt to
match this impedance as closely as possible. For example, a typical CPGA package has a 42-ohm
trace impedance and a typical PPGA package has a 47-ohm impedance.
4.1.1.5 Crosstalk
Crosstalk refers to unwanted signal coupling between lines. It is a complex function of the driver
and receiver characteristics, trace characteristics, and switching patterns. Some generalities
regarding package design and its relationship to crosstalk can be made. Crosstalk in a package is
dependent upon the stack-up, just as is characteristic impedance; however, crosstalk is primarily
affected by the distance between traces and the amount of parallelism between them. The longer
the parallel distance between two or more lines, the higher the crosstalk between them. Also, the
closer the lines are to one another, the higher the crosstalk. The proximity of power/ground planes
to the traces can help reduce crosstalk. That is, crosstalk is directly proportional to the distance
between the planes above and below the trace. There are no simple formulas for predicting
crosstalk. Models of the traces, with the mutual inductance and capacitance calculated, must be
generated and simulations run using buffer models and models representing the system loads to
correctly determine the amount of crosstalk in a package design or system.
The key to providing an accurate I/O signal lead model is to identify the return current paths. For
example, in analyzing the I/O signal lead bondwire inductance, the closest current return path is the
nearest power or ground bondwire. If the effect of this bondwire is not included in the analysis,
then the calculated I/O signal lead bondwire inductance will be higher than it actually is because
the mutual effects of nearby wires have not been included. In analyzing the signal trace, it is
necessary to identify the nearest current-carrying package planes that provide a current return path.
In effect, signal traces are usually modeled as microstripline or stripline structures, where a
microstripline structure is defined as a trace with one current-carrying plane in close proximity and
a stripline structure is defined as a trace sandwiched between two current-carrying planes.
Unless a highly accurate model for a particular critical signal is required, the I/O signal lead model
for a package is usually based upon the typical lead geometry. Crosstalk parameters are usually
based upon the worst-case crosstalk scenario, i.e., the minimum trace spacing. Typically, the
individual components comprising the signal lead are modeled individually using two and three
dimensional solutions obtained using electromagnetic field-solving software. The primary
parasitics of interest are the line characteristic impedance (Z_o), the line inductance (L), the line
resistance (R), the line capacitance (C), and cross-coupling L and C matrices for crosstalk analysis.
SSO models usually include mutual effects between signal lines and include the power distribution
network effects. To obtain the level of accuracy and complexity required for SSO modeling, three-
dimensional fully coupled models are necessary.
complex model is usually required for making power distribution design decisions, such as
determining the quantity and location of power and ground pins and bondwires and the quantity
and location of decoupling capacitors.
A simple distributed model usually consists of a single resistor/inductor element to represent each
major structure in the package. For example, all the power bondwires are considered to be parallel,
equivalent resistive/inductive structures, so their parasitics are lumped into one resistor/inductor
element. Similarly, all the pins and vias are lumped together, and each plane is represented as a
single resistor and inductor. This is a highly simplistic model which assumes equal current flow
through each pin. Although this is not a very accurate model, it is quite useful for obtaining an
approximation for the total package resistance and inductance. To more accurately model the
intricacies of the power distribution in the package, all elements must be represented by circuits
that contain both self and mutual inductance and capacitance terms. This type of representation is
typically called a PEEC (partial-element equivalent circuit) network and must be generated using
software tools that solve electromagnetic field equations. The drawback of this type of model is
that it consists of a large, complex circuit with many individual elements. The effects of various
elements are not intuitively obvious, so circuit simulations must be run. Since the network is large,
much computer memory and simulation time is required for this type of analysis.
Inherent in this development is the decision of how much loss in accuracy is acceptable in order to
provide a convenient and usable model. This is a decision that is best made after experimentation to
test for convergence of a model and, ideally, after comparison to measured data. In general, a
single I/O lead model can be quite accurately constructed using two-dimensional approximations
for the trace itself and even for the bondwire. Three-dimensional modeling is usually required for
the pin and via. All these structures can be easily created and accurately analyzed using
commercial software tools for solving electromagnetic field equations. Because of its complexity,
there are more engineering approximations that must be made in constructing a model for the
power supply loop. For example, a power plane may have a hundred vias connected to it; however,
inputting this complex geometry into a software tool would be a time-consuming task. In addition,
analyzing this type of geometry would require large quantities of computer memory and time. The
accuracy lost in grouping vias together may be very small, i.e., one via could be used to represent
ten vias in close proximity. After some practice and experimentation, one can learn to recognize
opportunities for model simplification which will not overly compromise accuracy.
Equation 4-4.
t r < 2t o l
where 2t o l is the round-trip delay in the medium in which the signal path lies and t r is the risetime
of the signal. As an example, a typical package trace in a ceramic package is around 1 inch long.
Using a dielectric constant of 10 for ceramic, the round-trip delay is:
Equation 4-5.
εr
2t o l = 2 × 10
ε r ε o µ o × l = 2 × --------- × l = 2 × --------------------------------
- × 2.54 cm/in. × 1 in. = 0.54 ns
c 10
3 × 10 cm/s
This is the order of the risetimes for signals in today’s microprocessors. Therefore, using a
transmission-line model for traces in a package is an appropriate modeling decision.
Typically, a transmission-line model should be used for the trace and lumped models for the
bondwires, pins, and vias. Most of the time delay in the package is due to the trace delay, and a
transmission-line model for the trace will appropriately account for the time delay. A lumped
model cannot account for time delay and should only be used when the delay through a structure is
not a significant portion of the overall signal propagation delay. As risetimes decrease and
propagation delays through the package become more critical, transmission-line models will be
necessary for many structures in the package.
It is usually adequate to simply model the DC resistance of a conductor in a package and to use
high-frequency, or quasi-static inductance calculations to model the inductance. Because signal
risetimes are decreasing with each new generation of microprocessor, however, the spectral
content of typical signal waveforms is increasing and frequency-dependent effects are becoming
more critical to accurate package analysis. For this reason, future package models should include
both frequency-dependent resistance and inductance values. These parameters must be calculated
using software tools which accurately perform a full-wave solution to Maxwell’s equations, the
equations which describe electromagnetic field interactions.
are two types of decoupling: low and high-frequency. Low-frequency decoupling is used on the
board to typically control the noise from the power supply entering the board. High-frequency
decoupling is used on the package and chip to control simultaneous switching noise.
When decoupling capacitors are placed on the package, they should be placed as close as possible
to the die to minimize stray inductance associated with the connection between the capacitor and
the die. It is important in using any type of decoupling capacitor to try to minimize the stray
resistance and inductance associated with the interconnect and the capacitor itself. This is more
important for high-frequency decoupling than low-frequency. In general, low-frequency
decoupling schemes require large values of capacitance. Larger inductances are usually associated
with larger capacitors. High-frequency capacitance values should be small to reduce the associated
stray inductance.
The placement of decoupling capacitors affects the overall effective inductance; therefore, a fairly
complex package model that allows analysis of the effects of capacitor placement is usually
necessary for power-distribution design. The package model for this use, therefore, should take the
form of a complex distributed model, or partial-element equivalent circuit (PEEC) model.
Most of these tools are based upon numerical solutions of electromagnetic field equations, i.e.,
Maxwell’s equations. The techniques used vary according to the software vendor. Some of the
most common solution techniques include moment method, finite-element method, finite-
difference time-domain technique, and boundary-element method. Although one need not be an
expert, using commercially available software tools which use these techniques usually requires
that one be somewhat knowledgeable about the basic theory behind the technique because
additional user input is usually required to accurately solve for the parasitics. For example, the user
is usually required to specify the meshing scheme used to divide the geometry for numerical
analysis. One should also be aware of the limitations and requirements of the tool and technique
used for solving for package parasitics or one could inadvertently produce data that is not useful or
accurate. Some of the issues that should be understood are the limits on the types of geometries that
can be analyzed, the assumptions concerning return-current paths, the assumptions concerning
ground planes and ground conductors, and the underlying analysis algorithms.
4.1.3.1 Equipment
The equipment necessary for characterizing the parasitics of a package falls into three categories.
The first is measurement equipment, which is standard and available from several companies
which specialize in the design and manufacture of this equipment. A well equipped laboratory
should contain a D.C.-ohmmeter, an impedance analyzer, a network analyzer for frequency-
domain characterization, and a time-domain reflectometry (TDR) set-up for time-domain
characterization. The second category is test fixtures. These must be specially designed for the
package and structure being characterized. Methods for de-embedding the test fixture from the
measurement must be devised. The third category is probes. Most package measurements involve
probing very small structures with fixed pitches. Probes and calibration standards for these probes
are available from companies which specialize in this area.
The following sections outline some of the basics for measuring I/O signal lead and power loop
parasitics. It is impossible to cover the intricacies of measurement technique adequately in such a
small space. Measurement methodology is continually evolving, and packaging engineers should
remain in close contact with measurement equipment representatives and should keep up with
conference proceedings and technical journals so that they are well informed concerning the latest
and most accurate techniques. Regardless of the measurement technique, the packaging engineer
should take care to model and measure the same scenarios if reliable validation data is to be
obtained.
4.1.3.2.1 DC Resistance
The resistance for a given CPGA package is measured from the tip of the bond finger to the pin
braze pad. The lead resistance for plastic packages is measured from tip to tip. Figure 4-1 shows
the four-probe setup used to measure resistance values. Two sets of readings are taken by reversing
the direction of current to eliminate the contact potential. The average value is considered the
resistance of the sample.
Voltmeter
R δ
Contact Potential
Power Supply
With Switch Ammeter
240819-1 A5609-01
4.1.3.2.2 Capacitance
Loading capacitances are measured using an impedance analyzer. The setup for measuring loading
capacitance is shown symbolically in Figure 4-2. In this setup, all leads except the lead of interest
are grounded. The capacitance is measured between this lead and ground. Decoupling capacitance
is measured between a power lead and a ground lead.
I = I ej t
o
V = Voe j ( t + c )
_
240819-2 V A5610-01
4.1.3.2.3 Inductance
Like capacitances, inductances are measured using an impedance analyzer. The inductance
measurement is a rather complicated process which involves sophisticated fixturing and de-
embedding techniques. Typically, measuring inductance is quite difficult because the test method
and fixture can affect the outcome of the measurement. One method of indirectly extracting the
inductance is to measure the characteristic impedance (Z_o) and self-capacitance, which can be
more accurately measured, and to determine the inductance using:
Equation 4-6.
2
L = C×Z
0
Figure 4-3. Package I/O Signal Lead Model for Multilayer Packages
Vccp
Bondwire Trace Pin / Land
Board
Tiebar L, R, C
Vssp
A5681-01
Table 4-1. Summary of Package I/O Lead Electrical Parasitics for Multilayer Packages
Wirebond Package Type Flip-chip Package Type
Electrical Parameter
CPGA PPGA H-PBGA OLGA FC-PGA
Bondwire/Die bump R (mohms) 126 - 165 136 - 188 114 - 158 2 0.06
Bondwire/Die bumpL (nH) 2.3 - 4.1 2.5 - 4.6 2.1 - 4.1 0.02 0.013
Trace R (mohms/cm) 1200 66 66 590 120
Trace L (nH/cm) 4.32 3.42 3.42 3.07 2.329
Trace C (pF/cm) 2.47 1.53 1.53 1.66 1.707
Trace Z_0 (ohms) 42 47 47 43 38.5
Pin/Land R (mohms) 20 20 0 8 20
Pin/Land L (nH) 4.5 4.5 4.0 0.75 2.9
Plating Trace R (mohms/cm) 1200 66 66 N/A N/A
Plating Trace L (nH/cm) 4.32 3.42 3.42 N/A N/A
Plating Trace C (pF/cm) 2.47 1.53 1.53 N/A N/A
Plating Trace Z_0 (ohms) 42 47 47 N/A N/A
Trace Length Range (mm) 8.83 - 26.25 6.60 - 42.64 4.41 - 22.24 3.0 - 18.0 10.0 - 42.6
Plating Trace Length Range (mm) 1.91 - 10.50 1.91 - 16.46 0.930 - 8.03 N/A N/A
Because the geometry of TCP packages is very different from that of multilayer packages, another
circuit representation for the package model must be used. This is shown in Figure 4-4. The model
consists of three transmission lines with different parasitics that represent different portions of the
signal path through the package. Typical TCP parasitics are given in Table 4-2.
Figure 4-4. Package I/O Signal Lead Model for TCP Packages
Vccp
Inner Lead Trace Fan-Out Lead Trace Outer Lead Trace
Segment Segment Segment
Vssp
A5682-01
Table 4-2. Summary of Package I/O Lead Electrical Parasitics for TCP Packages
Trace Segment
Electrical Parameter
Inner Lead Fan-Out Lead Outer Lead
Segment Segment Segment
L_Vcc R_Vcc
Die Board
L_Vss R_Vss
A5680-01
2 1 3
A5688-01
involving mature silicon technologies, older, more mature package technologies are the
appropriate choice. Table 4-6 through Table 4-12 give the package parasitics for some of the older
packages. Although they are not used for Intel’s leading-edge microprocessors, they are routinely
used for other Intel products. As with all the packages discussed in this chapter, the parameters
given in the following sections may not reflect the actual values for a particular product. These are
typical values only. The actual parameters for a particular product can be obtained by contacting a
local Intel sales office. For electrical parameters of packages not listed, contact your local Intel
field sales office.
I/O
RWire + Lead (mΩ) 81 106 83 115
LWire + Lead (nH) 6.6 8.3 7.6 10.2
CLoad (pF) 0.5 1.3 0.7 2.2
Vss and Vcc
RVss db Wire (mΩ) 34 34 34 34
LVss db Wire (nH) 1.1 1.1 1.1 1.1
LVss Plane (nH) 0.2 0.2 0.3 0.3
CPlane (nF) 0.090 0.090 0.210 0.210
RVcc db Wire (mΩ) 55 55 55 55
LVcc db Wire (nH) 1.9 1.9 1.9 1.9
LVcc Plane (nH) 0.2 0.2 0.3 0.3
RLead to Pin (mΩ) 9 10 10 12
LLead to Pin (mΩ) 3.8 4.2 4.7 5.3
NOTE: db = Down bond
Electrical Parameter 84 Lead PQFP 100 Lead PQFP 132 Lead PQFP 164 Lead PQFP
RWire + Lead (mΩ) 62.9 106.4 64.9 108.8 63.2 112.8 65.8 116.9
LWire + Lead (nH) 5.3 9.8 5.9 10.6 5.4 11.9 6.2 13.2
CLoad (pF) 0.2 0.6 0.3 0.7 0.2 0.9 0.3 1.0
Electrical Parameter 28 Lead PLCC 32 Lead PLCC 44 Lead PLCC 68 Lead PLCC
RWire + Lead (mΩ) 56.7 74.0 56.9 74.3 57.7 75.6 57.7 78.4
LWire + Lead (nH) 4.1 7.1 4.2 7.3 5.0 8.4 5.0 10.3
CLoad (pF) 0.2 0.6 0.2 0.7 0.3 0.8 0.3 1.2
Min Max Min Max Min Max Min Max Min Max
RWire + Lead (mW) 61.5 82.6 61.8 84.6 69.5 98.1 69.5 103.7 69.5 122.6
LWire + Lead (nH) 4.0 6.4 4.1 6.9 5.2 8.7 5.2 9.8 6.3 12.6
CLoad (pF) 0.2 0.4 0.2 0.5 0.3 0.7 0.3 0.8 0.4 1.0
L (nH) 6.86 - 7.84 5.05 - 5.95 3.62 - 4.28 4.44 - 5.39 6.57 - 11.05
Pin C (pF) 8 - 12
Lead-to-Lead C (pF) 0.80 - 0.94 0.59 - 0.73 0.38 - 0.40 0.45 - 0.50 2.09 - 3.32
Intel optimizes it’s packages through design, construction, material selection, and processes to
insure that mechanical characteristics are acceptable. Packages then go through extensive testing
and qualification.
E1, α1, ν1
E2, α2, ν3 α 3 > α1
>183o C
E3, α3, ν3
<183o C
+
Stress distribution Stress distribution Re-distributed
due to the force due to the moment stresses
A6002-01
Consider the simplified case of attaching a silicon die to an organic substrate using a thin layer of
eutectic lead tin solder, as depicted in Figure 4-7. The melting temperature of eutectic lead tin is
about 183°C. Therefore, at 183°C or higher, the individual materials are the free to expand
independently. While cooling the assembly down (at temperatures below 183°C), the solder layer
solidifies, adhering the silicon and the organic substrate rigidly at the mating surfaces. Relative
motion is thereby prevented between the mating surface of the silicon and the substrate, forcing
them to contract together. However, the organic substrate which has a larger CTE than the silicon,
would want to contract more. Compressive forces are therefore, induced on the die and tensile
forces on the substrate. These opposing forces constitute a moment forcing the assembly to bend in
a convex shape when viewed from the top. This bending has a significant effect on the distribution
of stresses in the attached layers. Due to the bending, tensile stresses are introduced on the top of
the die and compressive stresses on the bottom of the substrate. For elastic layers, the stress
distribution can easily be calculated. The location along the thickness where the direction of the
stress changes would depend upon the relative magnitudes to the two components of the stress. In
some cases, there are actually more than one neutral axis along the thickness. The actual stress
distribution in the assembly may vary from those calculated from an elastic model due to the
viscoelastic nature of the adhesive layer. Further die edge shear stresses are present to balance the
forces on the center regions of the assembly. As a result high stresses occur locally at the die
edges.
In general, there are three primary stresses that exist at the interfaces of the assembly after reflow-
normal stresses, shear stresses, and peeling stresses. All these stresses vary along the length of the
interface, and their magnitudes depend upon the stiffnesses of the individual components being
attached. Normal stresses have a maximum value almost over the entire center regions of the
assembly, and drop to zero at the edges. Shear stresses have a maximum magnitude at the edges of
the die. Peeling stresses change directions along the length of the die, and have a maximum
magnitude close to the edge of the die. There are a number of analytical formulations based on
Timoshenko’s theory of bi-metallic thermostats, that predict the stress distribution for a tri-material
assembly. Most of these formulations (though some account for limited adhesive non-linearity) are
derived for elastic material behavior. However, these analytical relations are very useful to
understand the fundamentals of thermal stresses in a tri-material assembly, and the relative
influences of different design parameters and material properties on their stress state.
One set of analytical relations developed by Suhir E. [1], for a tri-material assembly when the
thickness or the modulus of the adhesive material is small are of the form:
Equation 4-7.
Fdie ,max 6M die ,max ∆α∆T h D die
- = – --------------- 1 + 3 --------- ----------
σ die = -------------------- + ------------------------ χ( x)
h die h
2 λh die h die D
die
τ ∆α∆T
sol = κ --------------- χ’( x )
λ
where:
Fdie, Fsub, Mdie and Msub are the forces and moments acting on the die and the substrate due to the
CTE mismatch between them.
E, G, and v, are the elastic modulus, shear modulus, and the poisson’s ratio of the three materials.
∆ α is the CTE mismatch between the die and the substrate (αsub - αdie).
hdie, hsub, and h are thickness of die, substrate and the assembly (hdie + hsol + hsub).
Equation 4-8.
v v 2 3 3
1 – die 1 – sub h Edie h die E sub h sub
λ = -------------------- + --------------------- + -------, D = -------------------------------
- + --------------------------------
Edie h die E sub h sub 4D 2
12 ( 1 – v die ) 12 ( 1 – v sub )
2
Here, D is the flexural rigidity of the assembly and k is the interfacial compliance of the assembly
Equation 4-9.
k is a parameter of the assembly stiffness which is a function of the axial and the interfacial
compliances.
h 2h h
die sol sub
κ = -------------
3G
- + ------------- + ---------------
3G 3G
die sol sub
Equation 4-10.
λ
k = ---
k
Figure 4-8. Comparison of analytically and numerically calculated stresses along the length
of the die.
40
30
20
10
Die (FEA)
0
Sub (FEA)
Stress (MPa)
a b Solder (FEA)
-10
Die (anal)
a b Sub (anal)
-20
Solder (anal)
-30
-40
-50
-60
0.03
0.33
0.63
0.93
1.23
1.53
1.83
2.13
2.43
2.73
3.03
3.33
3.63
3.92
4.22
4.52
4.82
Figure 4-8 compares the normal stress in the die and the substrate, and the shear stress in the solder
joint, obtained using these relations to those obtained numerically using a finite element model.
The material properties and the design dimensions used for this comparison are listed in Table
4-13.
Table 4-13.
Note that for the most part, the results match well.
The following general conclusions can be made from the analytical equations for the stresses in the
assembly.
Figure 4-9. Variation of the maximum shear stress and normal stress with
increasing die size
0.9
0.8
X' (max)
X' (max), X(max)
0.7
X(max)
0.6
0.5
0.4
0.3
0.2
0.1
0
0 2 4 6 8 10 12
kl
A6001-01
• The factors χ’( x ) and χ ( x ) describes the longitudinal distribution of shear and normal
stresses along the length of the interface. The maximum shear stresses occur at the end
(at x = l) where the function χ’( x ) has the value:
sinh ( kl )
χ’( l ) = χ’( max ) = --------------------- = tanh ( kl )
cosh ( kl )
and the maximum normal stress occurs at the center where:
1
χ ( 0 ) = χ ( max ) = 1 – ---------------------
cosh ( kl )
Figure 4-9 shows the variation of these two factors for increasing values of kl. From Figure
4-9, it is evident that increases with kl, for values of kl less than 3.5, and increases with kl for
values of kl less than about 7.5. This indicates that for the die/solder/substrate assembly
considered, σdie and σsubstrate increases with increasing die size for kl< 7.5 (ie 2/ < 19.5 mm),
and tsolder increases with die size for kl< 3.5 (ie 2/ < 5.5mm). For die sizes greater than 19.5
mm square (and 5.5 mm square for shear stress), the normal stress in the die and the substrate,
and the shear stress in the solder layer will remain unchanged. For die and substrates of
different thicknesses and properties than that considered here, the size of the die beyond which
the stresses will remain unchanged, will be different.
• The coefficient of thermal expansion of the attachment material does not enter into the
relations for the stresses in the assembly. This means that the CTE of the solder material does
not affect the thermally induced stresses in the assembly, as long as the thickness and/or the
modulus of the adhesive are small compared to that of the adherents.
However, the modulus and the melting temperature of the adhesive layer plays an important part in
the magnitude and distribution of the stresses. Lower the modulus of the solder, lower will be the
amount of stresses transmitted to the attached components. Also, lower melting temperature
solders decrease the stresses in the assembly by lowering the temperature differential between
reflow and room temperature. However, low melting temperature solders are at high homologous
temperatures (T/Tmelting in the absolute scale) during the range of operating temperatures of an
electronic package, leading to inelastic strain accumulation in the solder material. This
accumulation of inelastic strain in the material could lead to fatigue failure of the material during
operation.
Eutectic lead tin solder melts at 183°C, and therefore, is at about 65% (0.65TM)of its melting
temperature in the absolute scale, at room temperature. In general, creep becomes significant in
materials at homologous temperatures above 0.5TM . Therefore, inelastic strain gets accumulated
in the eutectic lead-tin solder after reflow and during subsequent thermal cycles in operation.
In addition to these primary failure mechanisms, manufacturing processes induce numerous defects
in a package, which typically becomes the preferred site for failure initiation. For example, sawing
a die from a wafer introduces numerous micro-cracks at the edges of the die, which can propagate
due to the stresses induced in the die during operation.
To investigate the reliability of a microprocessor package during the intended life of the package,
they are subjected to temperature and power cycling tests.
In recent years, extremely sensitive, full-field optical interference techniques have been
extensively used to calibrate and compare numerical predictions to actual behavior. Some of the
commonly employed "opto-mechanical tools" that produce high-resolution, full-field contour maps
of thermo-mechanical deformation within an electronic package are Moiré Interferometry, Infrared
(IR) Fizeau Interferometry and Shadow Moiré [3-5]. While Moire interferometry measures the in-
plane deformation, Fizeau interferometry and Shadow moire are used to map the out of plane
deformation (warpage) of packages. In addition to model validation, these tools by itself can be
used to study the effect of design changes on the behavior of the package quickly.
Figure 4-10. Finite element model and out of plane displacement of a die C4 attached to a
substrate
A6007-01
A6008-01
However, before this model can be used for stress predictions in microscopic regions of the
package, this model has to be further validated. Figure 4-12 and Figure 4-13 compare the inplane
displacements in the fillet region of the underfill. The fringe constant for the fringe patterns shown
in these figures is 0.417 mm/fringe. For comparison purposes, an image analysis software has been
used to depict the displacements in a selected region of the fillet as a displacement contour pattern
having the same scale as the numerical contour pattern. Note that the results match well. This
validated model is now used for stress and life predictions. Typically, the properties of materials
used in these packages are not well characterized. Therefore, a number of iterations of model
calibration using the experimental results are required before the model can be used for stress
predictions.
Figure 4-12. Comparison of model prediction and moire results of horizontal displacement in
the fillet region
A6009-01
Figure 4-13. Comparison of model prediction and moire results of verticle displacement in the
fillet region
A6010-01
Besides it’s use as model validation tools, these interferometric techniques are also used to
compare materials and design options, due to its relatively short time-to-data. Two epoxy
samples proposed to be used as the encapsulant material in PLGA (Plastic Land Grid Array)
packages were compared using Moire interferometry. In PLGA packages, the silicon die is
covered with the encapsulated. This epoxy not only covers the active surface of the silicon, but
also the wire-to-pad interconnects. Therefore, these two epoxy samples were compared from a
wire bond reliability perspective.
Figure 4-14 compares the Moiré U-field displacement patterns (horizontal fields) of the wire heel
region (at the wire to die interface) using the two different encapsulant materials. Fringe gradients
in any direction would indicate relative motion between the encapsulant and the die in that
direction, increasing the likelihood of failure during temperature excursions. V-field images
(vertical field) revealed almost no fringe gradients in the wire heel region indicating negligible
relative motion between the die and the encapsulant in that direction.
Figure 4-14. Comparison of horizontal displacement fields in the wirebond region of PLGA.
A6011-01
Figure 4-15. Comparison of free expansions of the two encapsulant materials used in the
study
Sample A Sample B
A6012-01
However as shown in Figure 4-14, the horizontal displacement of the the encapsulant in wire heel
area of the sample using encapsulant B is twice that in the case of the package using sample A.
This difference in behavior between encapsulant A and B was confounding since both encapsulants
had nearly identical bulk CTE’s in the horizontal direction. The encapsulant material in the heel
area of both packages were seperated and moire performed. Figure 4-15 shows the horizontal
displacement fields obtained from both encapsulant materials. Figure 4-15 indicates that
encapsulant B has a much larger CTE at the interface that encapsulant A. This was later
determined to be due to a lower filler content at the interface.
During encapsulation, wire sweep may occur if wire lengths are too great or if drag during molding
flow is excessive. Current mold gate designs provide low drag flow patterns, and X-ray monitors
are used to confirm process stability. Wire diameter and length design rules are strictly enforced to
ensure that there is adequate mechanical stability of the bond arch to resist drag forces.
Development of new package designs includes analysis and experiment to ensure that wire sweep
is minimized.
In plastic packages, delamination at the interface between molding compound and silicon or lead
frame can cause high stresses at the first or second bond location, because the differential thermal
expansions must be accommodated across the bond itself. The delamination generally results from
temperature cycling of packages with trapped moisture. To suppress the adverse effects of
delamination, selection of materials with enhanced interfacial adhesion and the design of lead
frames that feature additional mold compound locking characteristics have proven to be effective.
Finite element models are used to guide and optimize designs. For packages that are particularly
moisture-sensitive, shipment of prebaked product in sealed moisture-proof bags with desiccant
ensures that delamination is minimized and operation of the assembled part will be reliable.
240819-20
A5614-01
240818-21
A5615-01
Units are mounted on both sides of the board using production-level processes and specifications
for each package. The board is mounted vertically in a tensile test set-up such as a materials test
system (MTS), and the narrow sides of the slot are cut, separating the two ends of the package. The
MTS is fitted with a 200 lb. load cell and a 6 inch displacement actuator. This allows the straddle
board to be tested with a 0.020 inch displacement, 0.010 inch in the tensile cycle and 0.010 inch in
the compressive cycle. These values were chosen to span the lead displacement experienced during
temperature excursions from –65° C to +150° C (MIL–STD–883C T/C [C]).
A total cycle time of five seconds was used with a maximum load range of 0.8-8 lbs., depending on
the package type and lead count. As Figure 4-18 shows, leads are loaded in both lateral and
transverse directions (i.e. in the plane and normal to the plane of the lead bend). Twenty-five
samples per lead count per direction were evaluated. Once the board was mounted, three full cycles
were run, and the force versus deflection curve was recorded. The linear portions of the loading
and unloading curves were used to determine the lead stiffness. A typical hysteresis curve is shown
in Figure 4-19.
Force Force
240819-62 A5616-01
4.0
3.0
2.0
1.0
Load (LSB)
-0.010* -0.005* 0.005* 0.010*
-1.0
-2.0
-3.0
-4.0
Designing leads with adequate compliance is now recognized as an important element of overall
product design, and mechanical modeling has proven to be a useful design tool. Lateral and
transverse lead stiffnesses were calculated for PLCC, and PQFP packages using the ANSYS 3–D
elastic beam element option. The calculated stiffness for each type of package was matched to the
value obtained from the straddle board measurements by locating the point on the lead where
agreement was reached. In all cases, the point lay within the solder joint, in agreement with
physical expectations.
Of particular interest were the boundary conditions imposed by the solder joint on the foot of the
lead. In the lateral direction, the solder volume is adequate for the joint to provide built-in support
to the lead foot (zero rotation). However, in the transverse direction for the PLCC and cerquad
packages, the J-bend lead stiffnesses are large enough to produce inelastic deformations in the
solder joint. As a result, the joint acts like a pinned support (free rotation). In all cases, the PQFP
gull-wing leads are sufficiently compliant for the solder joint to act as a built-in support. This result
suggests that reducing lead stiffness will significantly reduce solder joint stresses and associated
creep and fatigue. Experimental data recently reported in supports this hypothesis.
Because of the contribution of lead stiffness to the level of stress in the solder joint when there is
thermal mismatch between package and board, a comprehensive list of in situ lead stiffness values
is provided in Figure 4-21. All surface mounted packages currently used for Intel products (TSOP,
PLCC, PQFP and CQFP) are listed in Table 4-15. For these packages both translational and
rotational stiffness components are given. The translational stiffness is defined as the force in
pounds created by a displacement in inches applied at the solder joint in the direction of interest,
the remaining force and rotation components being zero. These stiffness components are important
when the thermal mismatch creates differential displacements which must be accommodated along
the leads. The rotational stiffness is defined as the moment in inch-pounds created by a rotation in
radians applied at the solder joint about the axis of interest, all other moments and displacement
components being zero. These stiffness components are important when thermally induced
differential rotations are imposed at the ends of the leads.
The magnitude of the stiffness components are useful in making comparisons of the level of stress
or strain induced in the solder joint by various package types; high lead stiffness will generate high
solder stresses, etc. Experience to date indicates that in situ translational lead stiffnesses on the
order of 100 lb./in. or less produce excellent joint reliability. Even lead stiffnesses above this level
are acceptable although they have less margin than the more compliant leads. With reference to
Figure 4-21, the x– and y–components of the gull-wing leads on 0.025 in. centers (PQFP and
CQFP) fall into this category. The z-components are always large and demonstrate the importance
of minimizing board flexing, which generates these z-components.
To assist in visualizing the stiffness and stress entries in Table 4-15, Figure 4-21 has been prepared.
The gull-wing lead of the PQFP has been used for reference. In all entries, the center line
represents the lead profile in the plane of interest. The heavy line represents the displacement
profile associated with the displacement or rotation component indicated at the point on the lead
connected to the solder. The line connected by shading to the undeflected profile shows the
distribution of the bending stress along the lead. In general, the maximum value is at or near the
solder connection point; however, for displacements along (or rotations about) the z-axis, the
bending stress maximizes at the package body. As mentioned previously, z-component
displacements may generate high lead stresses. Figure 4-21 suggests that the package body is the
site of maximum stress and damage in the lead.
Table 4-15. Surface Mount Package Lead Stiffness
Lead Stiffness Components Maximum Lead Stresses
Under certain conditions, stresses in the lead may be large enough to cause reliability problems in
the lead as well as in the solder joint. The figures of merit presented in Table 4-16 extracted from
Table 4-15, provides indicators of lead compliance and lead stress levels for current Intel package
types. All the lead compliance and stress of various packages and lead designs are normalized upon
the lead configuration of a 68L PLCC package. The first two columns give the level of lead
compliance along the lateral and transverse directions as compared to that of a 68L PLCC package.
The lead stress is estimated by applying a unit displacement (1 mil) along the lateral or the
transverse direction, the third and fourth columns provide the level of lead stress produced by the
proposed lead displacement. A correlation between the lead stiffness and the lead stress is
observed; the greater the lead stiffness, the higher the lead stress. A plot of log [lead stress] verses
log [lead stiffness] is shown in Figure 4-20. The trend of the lead stress dependence on lead
stiffness is linear on a log-log plot for most of the Intel packages except the TSOP I and TSOP II
packages which show much stiffer lead response because of their low profile and short lead design.
In general, a stiffer lead will introduce higher stress at the solder joint during temperature cycle (T/
C) stressing and subsequently degrades the solder joint fatigue performance. To assess solder joint
reliability, a methodology is described in the following paragraphs to investigate the solder joint
fatigue performance at various lead displacement amplitudes and frequencies.
Table 4-16. Figures of Merit for Lead Compliance and Lead Stress Package
Compliance FOM 1 Stress FOM 2
Figure 4-20. Lead Stress Dependence on Lead Stiffness for Various Lead Designs
3
CQ : CQFP Package, (L) : Lateral, (T) : Transverse.
TSII (T)
2.8 PQ : PQFP Package, (L) : Lateral, (T) : Transverse. TSI (T)
Cd : Cerquad package, (L) Lateral, (T) : Transverse. TSI (L) TSII (L)
2.6
PL68 : 68L PLCC Package, (L) : Lateral, (T) : Transverse.
2.4
PL44 : 44L PLCC Package, (L) : Lateral, (T) : Transverse.
2.2 TSI : TSOP (I) Package, (L) : Lateral, (T) : Transverse.
Cd (T)
2 TSII : TSOP (II) Package, (L) : Lateral, (T) : Transverse.
PL68 (T)
240819-48 A5618-01
Z Z
Y X X Y
Sigmax
Sigmax UY
UX
Z
Z
Y X
X Y
Sigmax
UZ
Rotx
Sigmax
240819-44 A5619-01
X Y
Sigmax
Z
Rotz
Roty
Sigmax
240819-46 A5620-01
A quantitative picture of the dependence of the total strain energy dissipation, W, per cycle on
displacement amplitude and frequency is shown in Figure 4-23 and re-mapped to Figure 4-24.
These iso-strain energy dissipation contours per cycle show strong dependence on the displacement
amplitude and weaker dependence on the cyclic frequency. The use of the cumulated total energy
could provide reasonably good prediction on the solder joint fatigue life. The use of this
methodology to predict a 68 L PLCC solder joint reliability is described in the next section.
Figure 4-23. Dependence of Total Strain Energy Dissipation per Cycle on Displacement
Amplitude and Frequency
35 f (Min--1)
Where Wt = Inelastic Strain Energy 0.0625
f = Fatigue Cycle Frequency 30 0.125
D = Displacement 0.25
0.5
25 1.0
2.0
Wf (Ib.-in/in 3 )
20
15
10
0
0 0.2 0.4 0.6 0.8 1.0 1.2
240818-49 D (10-3in.) A5621-01
0.8 160
D (10 In.)
-3
0.6
80
40
0.4 20
10
0.2
0
0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 1.8 2.0
f (Min-1)
240819-50 A5622-01
Considering the solder joint fatigue life is governed by the total inelastic energy dissipation during
fatigue stressing, estimations of its life expectancy may be conservative. Even though solder
fatigue life predictions based on inelastic strain energy density buildup to a critical level alone
might be conservative, it is still useful to examine the consequences of postulating a critical strain
energy density buildup level and examining its impact on the life prediction. For example, if 100
cycles count is the cycle count for a critical strain energy density buildup for the displacement of 5
mils and a period of 30 minutes, then based on the data in Figure 4-25, the corresponding cycle
count for various amplitude and period combinations to reach the same level of energy dissipation
(damage) could be calculated (in a case of fatigue displacement of 5 mils and period of 30 minutes,
if 100 cycles is the critical count to fracture, then, with data given in Figure 4-25, the corresponding
critical cycle counts for various fatigue conditions to fail could be estimated). The plot of various
cycle count in the form of log (d) versus log (N/100), where N is the total cycle counts for solder
joint fatigue failures, is given in Figure 4-26. Common practice when displaying fatigue curves
whose appearance is similar to Figure 4-26 is to obtain the slope for a fit to the Coffin-Manson
relation. In this case, the exponent varies from -0.667 for t = 0 minutes to -0.786 for t = 30 minutes.
Hence, although each curve appears linear, the slopes show a dependence on the period (inverse
frequency) of the cycle. For the displacement amplitude ranges investigated, these values are close
to those reported from experimental investigations. The consistency between the model and the
experiments suggest that the critical strain energy density buildup criterion may be a useful
approach for the development of a comprehensive failure prediction methodology for solder.
However, it should be pointed out that for displacement amplitude larger than those shown in
Figure 4-26, it is likely that the curves merge. Moreover, at very low displacement amplitude, the
curves should become horizontal (asymptotically approach the cycle to failure axis) since
negligible strain energy density buildup is possible and the solder should survive almost unlimited
cycling.
140
D = 5 Mils
D = 2.5 Mils 120
D = 1.25 Mils
100
80
W (Ib - in / in -3 )
60
40
20
0
0 5 10 15 20 25 30
240819-51 Period (Min.) A5623-01
Figure 4-26. Dependence of Log of Displacement Amplitude on Log of (N/100) for Different
Displacement Periods
0.7
Tau = 0 Min.
Tau = 1 Min.
Tau = 2.5 Min 0.6
Tau = 30 Min.
0.5
0.4
Log D (Mils)
0.3
0.2
0.1
0
0 0.2 0.4 0.6 0.8 1 1.2
240819-52 Log N / 100 A5624-01
Thermal management has two primary objectives. The first is to ensure that the temperature of
each component is maintained within both its functional and maximum allowable limit. The
functional temperature limit defines the maximum temperature up to which the electrical circuits
may be expected to meet their specified performance targets. Operation of the circuits at
temperatures higher than the functional limit may result in performance degradation or logic errors.
The maximum allowable temperature limit is the highest temperature to which a component or part
thereof may be safely exposed. Operation of the component at temperatures higher than the
maximum allowable temperature limit may cause irreversible changes in its operating
characteristics or may even cause physical destruction of the component.
The second objective of thermal management is to ensure that the temperature distribution in each
component satisfies reliability objectives. Failure mechanisms encountered in electronic
components are kinetic in nature and depend exponentially on the device operating temperature.
The exact relationship between the failure rate and temperature depends upon the thermophysical
properties of the packaging materials and the failure mechanism in operation. The relationship
between the normalized failure rate and temperature, for changes in the device operating
characteristics resulting from chemical or diffusive processes, can be defined by an Arrhenius
equation as follows:
Equation 4-11.
θT E
θ n = --------- = Exp ------A- ------ – ---
1 1
θ Tr k Tr T
Where:
θ = Failure rate
θ = Normalized failure rate
n
T = Absolute junction temperature (K)
Tr = Reference temperature (K)
E = Activation energy (eV)
A
k = Boltzmann’s constant: 8.616 X 10-5 (eV/K)
Figure 4-27 shows a plot of Equation 4-11 for a reference temperature of 100 °C and activation
energies of 0.4 eV to 1.0 eV. The figure shows that for activation energies between 0.6 eV to 0.8
eV, a 25 °C increase in the operating temperature, above the reference temperature, results in an
approximately five to six fold increase in the failure rate. Thus, precise control of component
operating temperatures is absolutely essential to ensure product reliability.
100
eV
= 1
Normalized Failure Rate
AE
eV
= 0.8
AE
V
6 e
10 = 0.
AE
V
0.4 e
AE =
4.3.2.1 Conduction
Conduction is a mode of heat transfer in which heat flows from a region of higher temperature to
one of lower temperature within a medium (solid, liquid, or gaseous) or media in direct physical
contact. In conductive heat flow, the energy is transmitted by direct molecular communication
without appreciable displacement of the molecules.
In a one-dimensional system (see Figure 4-22), conductive heat transfer is governed by the
following relation:
Equation 4-12.
∆T T1 – T 2
q = -kA ------- = – KA -----------------
L L
where:
q = Heat flow rate (W)
k = Material thermal conductivity (W/mC)
A = Cross-sectional area (m2)
∆T = Temperature difference, T1-T2, between the hot and cold
regions (K or °C)
L = Linear distance between the locations of T1-T2 (m)
Equation 4-12 indicates that in conduction, the heat flow rate is directly proportional to material
thermal conductivity, temperature gradient, and cross-sectional area. Equation 4-12 can be written
as:
Equation 4-13.
∆T
q = --------------
L ⁄ kA
Using an electrical analogy, if q and ∆ T are analogous to current and voltage respectively, L/kA is
analogous to electrical resistance. According to Equation 4-13, thermal resistance can be expressed
in terms of material thermal conductivity and geometrical parameters and is independent of the
power dissipation.
240819-29 T1 T2 A5630-01
4.3.2.2 Convection
Convection is a mode of heat transport from a solid surface to a fluid and occurs due to the bulk
motion of the fluid. The basic relation that describes heat transfer by convection from a surface
presumes a linear dependence on the surface temperature rise over the ambient, and is referred to as
Newton’s law of cooling:
Equation 4-14.
qc = hc A ( T s – T a )
where:
qc = Convective heat flow rate from a surface to ambient (W)
A = Surface area (m2)
Ts = Surface temperature (C)
Ta = Average convective heat transfer coefficient
hc = Average convective heat transfer coefficient (W/m2C)
Equation 4-15.
Ts – Ta
qc = -----------------
-
1 ⁄ hc A
Comparing Equation 4-15 to Equation 4-14, it is apparent that the convective thermal resistance
can be defined as 1/hcA.
In forced convection, fluid flow is created by an external factor such as a fan. In free or natural
convection, fluid motion is induced by density variations resulting from temperature gradients in
the fluid. Under the influence of gravity or other body forces, these density differences give rise to
buoyancy forces that circulate the fluid and convect heat toward or away from surfaces wetted by
the fluid.
4.3.2.3 Radiation
Radiation heat transfer occurs as a result of radiant energy emitted from a body by virtue of its
temperature. Radiation heat transport occurs without the aid of any intervening medium. Radiant
energy is sometimes envisioned to be transported by electromagnetic waves, at other times by
photons. Neither viewpoint completely describes the nature of all observed phenomena.
The amount of heat transferred by radiation, between two surfaces at temperatures T1 and T2
respectively, is governed by the following expression:
Equation 4-16.
q = ∈ σ A ( T 1 4 – T 2 4 ) F 12
where:
q = Amount of heat transfer by radiation (W)
∈ = Emissivity (0 < ∈ < 1)
σ = Stefan-Boltzmann constant, 5.67 X 10-8 (W/m2 K4)
A = Area (m2)
F12 = Shape factor between surfaces 1 and 2 (A fraction of surface 1 radiation seen by
surface 2)
T1, T2 = T1, T2 = Surface temperatures (K)
Note that the temperatures T1 and T2 in Equation 4-16 are absolute temperatures.
For radiation to make a rather significant contribution compared to either natural convection or
forced convection mechanisms, a relatively large temperature difference must exist between T1 and
T2. In the case of most low-power electronic applications, these temperature differences are
relatively small and, therefore, radiation effects are normally neglected. But for power application,
heat transfer by radiation should be considered. To compare radiative and convective effects, a
radiation heat transfer coefficient is defined as:
Equation 4-17.
h r = ∈σF 12 ( T 1 2 – T 2 2 ) ( T 1 + T 2 )
Equation 4-18.
Tj – T c
θ j C = -----------------
P
Tc – Ta
θ ca = ------------------
P
θ ja = θ jc + θ ca
Where:
θja = Junction-to-ambient thermal resistance (C/W)
θjc = Junction-to-case thermal resistance (C/W)
θca = Case-to-ambient thermal resistance (C/W)
Tj = Average die temperature (C)
Tc = Case temperature at a predefined location (C)
P = Device Power dissipation (W)
Ta = Ambient temperature (C)
The junction-to-case thermal resistance, θjc, is a measure of the internal thermal resistance of the
package from the silicon die to the package exterior. θjc is strongly dependent on the thermal
properties (i.e. thermal conductivities) of the packaging materials and on the package geometry.
The junction-to-ambient thermal resistance, θja, includes not only the package internal thermal
resistance, but also the conductive and convective thermal resistance from package exterior to the
ambient. θja values depend on material thermal conductivities, package geometry as well as
ambient conditions such as coolant flow rates and the thermophysical properties of the coolant.
To guarantee component functionality and long-term reliability, the maximum device operating
temperature is bounded by setting constraints on either the ambient temperature or the package
exterior temperature measured at predefined locations. Ambient temperature is most often
measured at an undisturbed location at a certain distance away from the package. As defined in
Intel experiments, the case temperature is measured at the center of the top surface of the package.
Depending on the environment (ambient and board temperatures) within the computer system,
thermal enhancements such as fins or forced air cooling may be necessary to meet requirements on
the package case temperature, "Tc".
heat transfer area. Figure 4-29 through Figure 4-31 show typical junction-to-ambient thermal
resistance values as a function of lead count for different package families. According to this
figure, thermal resistance is lower for packages with higher lead counts (i.e. larger package size)
within the same package family.
Figure 4-29. Effect of Package Size on Thermal Resistance of PLCC, PQFP, and PGA
Packages
70
30
20
10
0
0 50 100 150 200 250 300
Package Lead Count
240819-30 A5631-01
Figure 4-30. Effect of Package Size on Thermal Resistance of Plastic and Ceramic
Dual-In-Line Packages
160
Side - Brazed Ceramic
Cerdip
140 Plastic / Alloy 42
Thermal Resistance (C / W)
Plastic / Copper
Junction - To - Ambient
120
100
80
60
40
20
0
10 20 30 40 50
Lead Count
240819-31 A5632-01
Figure 4-31. Effect of Package Size on Thermal Resistance of Leadless Ceramic Chip Carrier
100
90
Thermal Resistance (C / W)
80 LCCC
Junction - To - Ambient
70
60
50
40
30
20
10
0
30 40 50 60 70
Lead Count
240819-32 A5633-01
70
60
CQFP
Thermal Resistance (C / W)
PQFP
Junction - To - Ambient
50
40
30
20
10
0
50 100 150 200
Package Lead Count
240819-33 A5634-01
Figure 4-33. Effect of Heat Spreader and Heat Slug on Thermal Performance
40 208L PQFP
30
(C / W)
20
JA
10
0
Standard Spreader Slug
240819-54 A5635-01
Figure 4-34. Reduction of θja by Heat Spreader and Heat Slug when Compared to Standard
Package
2.5
Heat Slug Ceramic PGA
2.0
∆ Theta ja (C / W)
1.5
Heat Spreader
1.0
0.5
0.0
300 400 500 600 700 800 900
40
68 - Pin PGA
168 - Pin PGA
Thermal Resistance (C / W)
Junction - To - Ambient
30
20
10
0.00 0.05 0.10 0.15 0.20
240819-34 Die Size (IN 2) A5637-01
40
ja
30 jc
Thermal Resistance (C / W)
20
10
0
0 2 4 6 8
Power Dissipation (W)
240819-35 A5638-01
It should be noted that the error in θja also increases at lower power levels, because both
temperature and power measurements are less accurate.
In the case of forced convection, the heat transfer coefficient does not depend on temperature
explicitly. Dependency on temperature is only due to changes in material properties. The material
properties for air do not vary significantly within the temperature ranges normally encountered.
However, at low flow rates, natural and force convection may have effects that are of the same
order of magnitude (mixed convection). A small dependency on power may be observed in mixed
convection heat transfer. However, as the flow rate increases the power dependency disappears.
Equation 4-19.
m
θ ca = K ⁄ V
Where:
K = Constant depending on air properties as well as package geometry
Figure 4-37 shows how θja varies as a function of air flow rate for 168-lead PGA with and without
heat fins. At lower air flow rates, there is a strong dependency of θja on the air flow rate, but as the
air flow rate increases, θja values become less sensitive to the changes in the flow rate.
m
θ ja = θ jc + K ⁄ V
Figure 4-37. Effect of Air Flow Rate on Thermal Resistance of 168-Lead PGA Package
20
15
jc, With Fin
10
0
0 200 400 600 800 1000
Another factor that impacts the thermal resistance from board to ambient is board thermal
conductivity. As board thermal conductivity or board thickness increases, the spreading resistance
for lateral conduction of heat within the board decreases. As a result a larger board area becomes
available for heat transfer to the ambient.
Figure 4-38. Effect of PC Board Material and Size on Thermal Resistance of 132-Lead PQFP
60
K = 0.64 (w / m c)
Thermal Resistance (C / W)
Junction - To - Ambient K = 2.2 (w / m c)
50
40
0 10 20
240819-37 A5640-01
PC Board Area / Package Area
In almost all practical applications, printed circuit boards are populated by many components, and
heating of one package influences the thermal performance of other packages. In order to get a
relatively good estimate of system thermal resistance, θja,s and θja values must be correlated
through another parameter. As it turns out, this parameter is Tb. A simple reistor network shown in
Figure 4-39 can be used to obtain this correlation. In this model, it is assumed that the heat flows
from the junction to the package boundary, and from there a portion of the heat is transferred
directly to the ambient, while the rest goes into the board and is distributed and transferred to the
ambient.
TJ
RC RC
R CA RB
TB
TA
R BA
TA
240819-38 A5641-01
It should be noted that the model shown in Figure 4-39 is simplified, and the actual package
resistor network model can be more complex. The following expression shows the relation
between θja,s and θja and board temperature rise (Tb - Ta):
Equation 4-20.
Tb – T SR a R b
θ ja, s = θ ja + S --------------- -----------------------------------------
P ( R a + R b + R ba )
Ra
where: S = --------------------
Ra + Rb
Equation 4-20 shows a linear dependence of θja,s on board temperature rise; as the board
temperature increases or decreases, the system thermal resistance will increase or decrease. The
rate of these changes is dictated by sensitivity parameter S, which in turn depends on Ra/Rb. As Ra/
Rb increases, thermal resistance becomes more sensitive to board temperature rise; in other words,
the packages becomes more thermally coupled with other components on the board. Large values
of Ra/Rb indicate that the package depends more on the board for transfer of heat to the ambient
than on direct heat transfer to the ambient (larger packages vs. smaller packages). On the other
hand, as Ra/Rb decreases, S decreases. Smaller Ra/Rb means a higher degree of insulation between
the package and the board. For example, the thermal resistance of a package with a copper lead
frame is more sensitive to board temperature rise than its counterpart with an Alloy 42 lead frame,
as shown in Figure 4-40.
Figure 4-40. Effect of Board Temperature Rise on Thermal Resistance of 132-Lead PQFP with
Copper and Alloy 42-Lead Frames
120
100
Thermal Resistance (C / W)
Junction - To - Ambient
80
60
40 K = 0.64 (w / m c)
K = 2.2 (w / m c)
20
0
0 20 40 60 80 100
240819-39 A5642-01
Board Temperature Rise
Above Ambient (C)
To determine sensitivity factors and intercept in expression 12, the package system thermal
resistance can be measured under two different board temperature conditions ( θja is assumed to be
known).
Board
Chip Package
System
240819-56 A5643-01
Mother Board
Fan
Power
Supply
Add-In Add-In
Boards Boards
HDD
CPU
FDD
Fan
240819-57 A5644-01
Air Flow
Figure 4-43 shows the configuration of the Baby AT PC. Its typical size is 20” x 17” x 6”. Because
of the larger available space, more add-in boards can be inserted into the mother board vertically. A
typical system power is around 120 Watts.
Figure 4-44 shows the configuration of a notebook with a typical size of 11” x 9” x 2”. The CPU
typically consumes 4-8 Watts depending on the products used. Other components including the
mother board, HDD, FDD and screen use about 6-10 Watts. The total system power is around 12-
15 Watts.
Air Flow
1 Add-In 2 3 4
Boards
HDD
CPU
FDD
Fan
240819-58 A5645-01
Air Flow
Mother Board
CPU Battery
Hard
Floppy Disk
Disk Keyboard
Drive
Drive
240819-59 A5646-01
CPU power: The use of the maximum CPU power is recommended for system thermal design
purposes. The power dissipation values for the Intel family of processors are the maximum power
levels for the corresponding processors.
Component power: The thermal impact of components such as the hard disk drive, video and
memory cards, and PCMCIA must be taken into account for system level thermal design. In
prototype testing, load boards with simulated components can be used to generate the maximum
component heat dissipation to investigate the thermal performance of the system.
Ambient temperature: Although most electronic systems work at room temperatures (25 °C), the
thermal design must consider the extreme environments which the system may experience.
Normally an ambient temperature of 35 °C to 40 °C is used by most system designers.
Budgeting: The total junction-to-ambient thermal resistance must be distributed among the
various sections of the thermal path; from chip-to-package, package-to-heat-sink and heat-sink-to-
inside-ambient and inside-ambient-to-outside ambient. The thermal resistance in each section
must then be managed to meet the assigned thermal budget.
Internal thermal management: This involves selection of thermal interface materials, package
type, bonding techniques, and via design. The design must also meet constraints of cost, available
technologies, reliability, manufacturing processes, and yield.
External thermal management. This involves selection of the cooling mode (i.e. conduction,
natural or forced convection, or radiation), heat sink design, and heat sink attachment process.
Figure 4-45 shows the typical convective thermal resistance values obtained using various coolants
and cooling modes.
When the component is cooled directly by contact with a gas or liquid, the resistance to the
convective heat removal from the surface is inversely proportional to the product of the heat
transfer coefficient and the wetted area, 1/(hA), where h is the convective heat transfer coefficient
and A is the wetted area. As shown in Figure 4-45, the convective coefficients range from 100 K/W
for natural convection of air, to 33 K/W of forced air convection, to 1 K/W in flurochemical liquid
forced convection.
When direct cooling of the package surface is inadequate to maintain the chip temperature below
desired levels, a heat sink can be attached to the package. A heat sink provides a significantly
larger wetted area for the heat transfer. In addition, the extended surfaces of the heat sink help to
spread the heat. However, the presence of the heat sink may increase the overall pressure drop in
the system, and the thermal interface between the package and the heat sink will introduce
additional thermal conductive resistance. Proper management of the heat sink design and the heat
sink attach method is required to achieve maximum thermal performance benefits from the use of a
heat sink. Typical air-cooled heat sinks can reduce the external thermal resistance to values less
than 15 K/W in natural convection and as low as 5 K/W for moderate forced convection. Liquid
cooled heat sinks can further reduce the external resistance to below 1 K/W. Once the cooling
mode has been selected, heat sinks, fans, and the heat sink attachment process must be optimized
for the system level thermal solution.
Figure 4-45. Typical Convective Thermal Resistances for Various Coolants and Cooling
Modes
Forced
Fluorochemical Liquids
Convection
Wetted
Water Area=10 cm 2
K/W
Heat slug/spreader: In order for the heat to spread efficiently from the silicon die to the package,
heat slugs or spreaders are used in the thermal design of many packages. The primary goal is to
allow the heat to conduct from the die and spread into the heat spreader or slug. Since the heat
transfer area provided by the slug is larger, it helps to reduce the thermal resistance. However, the
use of heat slugs and spreaders present several design challenges; thermally induced stresses
resulting from a mismatch between the coefficient of thermal expansion between the package and
the heat slug or spreader, reliability of the attachment between the die and the slug, and
manufacturability issues.
Thermal via and board design: Thermal vias are often used to reduce the thermal resistance of
materials with low thermal conductivity like printed circuit boards or substrates. Via designs are
divided into two configurations: stacked and staggered. In a stacked via design, successive via
layers are stacked with vias on top of each other. In the staggered configuration, the vias are not
stacked directly on top of vias in another layer.
Figure 4-46 shows an example of the stacked via designs on PCB. The via has an outside diameter
D, height H and is plated with copper of thickness Tw. The thermal resistance of the n vias can be
approximately estimated using one dimensional heat conduction analysis:
Equation 4-21.
2
Rv = H ⁄ Knπ ( DTw – T w)
where Rv is the total thermal resistance of the number (n) of vias and K is the thermal conductivity
of the via plating material. Vias reduce the thermal resistance but increase the electrical routing
difficulty. The density and size of the vias depends on many factors in the design criteria: electrical
routing, thermal performance requirements, cost, etc. The final design is usually a trade-off among
various considerations. For a fixed pitch to diameter (say, P/D = 2), the criterion favors vias with
small diameters. However, when the pitch is fixed, the minimum thermal resistance is obtained
with larger via diameters.
P D
Tw
240819-62 A5648-01
Heat sinks: Heat sinks vary in shape, size and material depending on applications. Figure 4-47
shows a regular cross cut pin fin heat sink and elliptical pin fin heat sink. The purpose of heat sinks
is to spread the heat and to increase the heat transfer area wetted by the coolant. Although heat
sinks help in heat spreading, the presence of the fins presents a blockage to the coolant flow
causing an increase in the system level pressure drop. Increased pressure drop would require a
larger fan to drive the required coolant flow through the computer chassis. The heat sink design
(i.e. number, shape and size of fins) must be optimized in order to maximize the heat transfer from
the heat sink with the smallest possible increase in the pressure drop.
Elliptical Pin
Cross-Cut
Square Pin
240819-64 A5649-01
Metal plate or rod: Heat spreading should be considered in the thermal design of the system to
minimize the temperature drop on the heat dissipating surfaces when: (1) concentrated heat must
be transferred to places where space is available for more enhancements like heat sinks or fans, or
simply where the thermal environment is more favorable; and (2) if a large surface area is required
for heat dissipation. Table 4-17 compares the temperature drops at the ends of a 10" long 1"
diameter rod with 1 Watt passing through the rod. Results show that the use of copper or aluminum
decreases the thermal resistance by nearly two orders of over the typical PCB. In portable
computer systems, flexible laminated copper foils called "flexible heat sinks" can be used in order
to satisfy space and weight constraints.
Heat pipes: An increase in the cross-sectional area of a spreader plate made out of copper or
aluminum can reduce the spreading thermal resistance. However, this approach is often
constrained by the space and weight limitations of the system. Under such circumstances, heat
pipes can be used for heat transfer. A typical heat pipe consists of a enclosed, partially evacuated
chamber containing a small amount of liquid. The liquid evaporates in the heat pipe section in
contact with a hot surface such as the processor package. The vapor travels to the colder sections
of the heat pipe and condenses. The internal surface of the heat pipe consists of a mesh or sintered
porous wick which transports the condensed liquid, via capillary fluid flow, to the hot section of
the heat pipe. Thus, the heat pipe provides an enclosed, phase-change based system for
transporting heat from hot to cold regions. Since heat removal in the hot section of the heat pipe,
and condensation in the cold sections of the heat pipe involve phase change, there is almost no
temperature gradient associated with transfering heat along the heat pipe. This is apparent from the
temperature drop value for a heat pipe listed in Table 4-17. Heat pipe designs utilizing water, freons
and dielectric fluorinert liquids are commercially available in a variety of configurations.
Fans: Fans provide forced air flow, which improves the convection coefficient significantly. Fans
are widely used in desktop computers. However, their use in notebooks will require the resolution
of issues typical of portables: power consumption (reducing battery life), space limitations, noise
and reliability. In spite of these constraints, the use of fans in high performance notebooks is
gaining acceptance because of the steady increase of CPU power and the fact that alternative non-
fan solutions are becoming increasingly complicated and expensive.
Fan-sinks. Fans are most effective when combined with heat sinks. Fan-sinks have emerged to
provide an integrated solution in improving the thermal performance while achieving smaller
overall size and higher efficiency.
Active thermal feedback (ATF). The worst case design methodology often creates products that
are “over-designed.” With the increasing complexity of thermal solutions, in many applications it
may not be economical to design products for the worst thermal case, which happens only to a very
small portion of the products. Rather ATF can be implemented to deal with those extreme
occasions. When the temperature of main components (say, CPU) reach certain threshold, ATF
will throttle the clock speed to reduce the power consumption of the system until the temperature
of the system decreases to a safe level.
It should be noted that when developing a thermal management strategy for the electronic system,
it is no longer sufficient to focus only on temperature. Rather, the thermal/packaging engineer has
to understand, control and ultimately eliminate the thermally induced failures present throughout
the electronic system. The attainment of that goal, accompanied by the severe electrical,
manufacturing, cost and reliability constraints, demands better collaboration of engineers with
various backgrounds to better the design of the system so that it can provide all the desired
functions reliably and inexpensively.
A single package is mounted either directly or via a socket on a thermal test board. Different test
board designs , are used for surface mount and through-hole mount packages( see Figure 4-48 and
Figure 4-49 respectively). The test chamber volume is 1ft3, and the ambient temperature is
measured 12 inches away from the package (see Figure 4-50). In order to obtain accurate
temperature measurements from the temperature sensing diodes/resistors, the diode output voltage
or sensor resistance must first be correlated to the temperature. This is done using a three point
calibration technique. For calibration purposes, the test board assembly is immersed in a constant,
uniform temperature dielectric fluid bath The temperature of the dielectric fluid in the bath is
measured using an RTD probe. The dielectric fluid in the bath is continuously stirred in order to
maintain temperature uniformity. If the temperature sensors are diodes, they are forward-biased
using a 100 mA constant current source and the voltage drop accross the diode is measured at 3
different bath temperatures. If the temperature sensors are resistors, a four wire resistance
measurement is obtained at 3 different bath temperatures. A linear best fit straight line correlation
is usually obtained to relate the diode voltage drop or sensor resistance to the measured bath
temperature
After calibration, the test structure is powered up to the desired power level. The on-die
temperature sensors are continuously monitored until no change is detected in the temperature
level, indicating an equilibrium state. At this stage, the chip surface temperatures, ambient and case
temperatures, and device voltage drop and current are recorded. The device voltage drop and
current are used to estimate the actual power dissipated in the test structure. When measuring
thermal resistance under forced convection conditions, the test board assembly is exposed to a
developed air flow. In this case, air velocity is measured at a location 12 inches upstream from the
leading edge of the test board, using a hot wire anemometer. Air temperature is also measured at
the same location.
Figure 4-48. Typical Thermal Test Board for Through Hole Mount Component
4.5 In.
6 In.
24019-25 A5626-01
Figure 4-49. Typical Thermal Test Board for Through Surface Mount Component
4.5 In.
132
84
5.5 In.
72 2
240819-26 A5627-01
Thermocouple
12
In.
Air Flow
Direction
240819-51 A5629-01
For validation purposes, the component and test board are instrumented with thermocouples to
measure the temperature distribution within and on the component and on the substrate.
Thermocouples are also used to measure case and ambient temperatures. The experiments are
conducted using a partial factorial design. A factorial design involves independently powering up
individual components. Powering up a single component allows the ability to monitor the heat
spreading inside the component and also within the substrate. This data is very useful to determine
the thermal properties for simplified component representations. Such experimental data can be
used to determine the thermal properties which when used in the numerical model would provide
temperature predictions to a reasonable degree of accuracy.
4.4 References
[1] Suhir E., "Calculated Thermally Induced Stresses in Adhesively Bonded and Soldered
Assemblies", ISHM International Symposium on Microelectronics, Atlanta, Georgia, Oct
1986.
[2] "Guidelines for Accelerated Reliability Testing of Surface Mount Solder Attachments", IPC-
SM-785, Nov 1992.
[3] Born, M., and Wolf, E., Principles of Optics, pp. 286-300, 1980, Pergamon Press, Oxford.
[4] Guo, Y., and Liu, S., "Development in Optical Methods for Reliability Analyses in
Electronic Packaging Applications," Experimental/Numerical Mechanics in Electronic
Packaging, vol. 2, pp. 10-21, R. Mahajan, B. Han, and D. Barker, ed., Society for Experi-
mental Mechanics, Bellevue, WA, June 1997.
[5] Post, D., Han, B., and Ifju, P., High Sensitivity Moiré, 1994, Springer-Verlag, Inc., New
York.
Electrical Resistivity Ω cm 1014 49 X 10-6 5 X 1012 >1011 <6 X 10-6 <2 X 16-6
Dielectric 7.9 - 10.0 NA < 5.0 11.5 NA NA
Constant (1 MHz)
Flammability Rating * inches 1/8
NOTE:
1. * UL-94V-0
Silver Silver
Silver
Filled Filled 99.99%
Filled
Property Units Glass Epoxy Au + 2% Si 99.99% Au
Adhesive
6.1 ESD
Several technical failure mechanisms associated with ESD cause damage to microelectronic
devices, including gate oxide breakdown, junction spiking, and latch-up.
• Gate oxide failure is a breakdown of the dielectric between the transistor gate and channel
resulting in excessive leakage or a functional failure.
• Junction spiking failure is a migration of the metallization through the source/drain junction of
MOS transistors causing leakage or a functional failure.
• Latch-up failure can be triggered by ESD, causing an internal feedback mechanism that gives
rise to temporary or permanent loss of circuit function.
When handling parts or their containers, ungrounded personnel can transfer high static charges.
Unless these static charges are slowly dissipated, ESD event can inflict damage to the devices.
Electrical fields can penetrate electrical devices. An ungrounded person handling a component or
computer board in a non-static shielding container can inadvertently transfer an electrical charge
through the container into the sensitive electronic device.
In-line film resistors between inputs and off-board connectors provide only marginal ESD
protection and are often damaged themselves.
Many vendors of ESD-protective equipment are willing to audit your facilities, recommend
appropriate procedures and materials, and assist in their implementation. You may choose to
consult with such a firm to determine your exact requirements; in the meantime, review this
chapter to gain an understanding of the basic components of a sound ESD prevention program.
ESD protective clothing/smocks. Street clothing must not come in contact with components or
computer boards since the various materials in clothing can generate high static charges. ESD
protective smocks, manufactured with conductive fibers, are recommended.
Electrostatic shielding containers or totes. These containers (bags, boxes, etc.) are made of
specially formulated materials which protect sensitive devices during transport and storage.
Antistatic or dissipative carriers. These provide ESD protection during component movement in
the manufacturing process. It must be noted that antistatic materials alone will not provide
complete protection. They must be used in conjunction with other methods such as totes or
electrostatic shielding bags.
Dissipative table mat. The mat should provide a controlled discharge of static voltages and must
be grounded. The surface resistance is designed such that sliding a computer board or component
across its surface will not generate more than 100 V.
Personal grounding. A wrist strap or ESD cuff is kept in constant contact with bare skin and has a
cable for attaching it to the ESD ground. The purpose of the wrist strap is to drain off the operator’s
static charge. The wrist strap cord has a current-limiting resistor for personnel safety. Wrist straps
must be tested frequently to ensure that they are undamaged and operating correctly. When a wrist
strap is impractical, special heel straps or shoes can be used. These items are effective only when
used in conjunction with a dissipative floor.
ESD protective floor or mat. The mat must be grounded through a current-limiting resistor. The
floor or mat dissipates the static charge of personnel approaching the work bench. Special
conductive tile or floor treatment can be used when mats are not practical or cause a safety hazard.
Chairs should be conductive or grounded with a drag chain to the flooring.
6.1.9 Summary
Intel uses a full range of electrostatic discharge prevention techniques. We invest in proper
employee training, purchase appropriate ESD protection equipment and supplies and adapt our
handling and manufacturing procedures for ESD prevention requirements. The same attention to
detail is required of all our suppliers, factories, repair centers, and field service staff.
Intel is committed to helping its partners — both suppliers and customers — to manage the ESD
problem. If we can be of further assistance in your efforts to eliminate electrostatic discharge
damage, please contact the Intel Components Quality Question Line: In the USA 1-800-628-8686
or 1-916-356-7599, or contact your local Intel Sales Office.
Junction Spiking
Aluminum
n+
241422-1
A5655-01
For long electrical pulse widths the most common failure modes are melted metallization and open
bond wires.
Bond Wire
Lead
Frame
Metal Interconnect
Silicon
Chip
Header
241422-2
A5656-01
Bond Wire
Lead
Frame
Metal Interconnect
Silicon
Chip
Header
241422-3
A5657-01
Conduct regular AC supply line monitoring and, if necessary, install EOS line control equipment
such as incoming line filtering and transient suppression circuits.
6.3 Reference
[1] Edward S. Yang, “Microelectronic Devices”, McGraw-Hill, 1988.
[2] Kohlhass, Phil, “Controlling Potential Static Charge Problems”, 3MNuclear Products Dept.,
St.Paul, MN.
[3] “Electrical Overstress/Electrostatic Discharge Symposium Proceedings”, The EOS/ESD
Association and ITT Research Institute, 1985 and 1986.
[4] DOD-HNBK-263,Electrostatic Discharge Control Handbook for Protection of Electrical and
Electronic Parts, Assemblies and Equipment”, 2 May, 1980.
[5] McFarland, W.Y., “The electronic benefits of an effective electrostatic discharge awareness
and control program—an empirical analysis”. 1981 Electrical Overstress/Electrostatic
Discharge Symposium Proceedings.
For more information regarding PGA insertion, request a copy of item # 8130 by calling the Intel
FAXBACK line U.S. 1-800-628-2283 or 1-916-356-3105 Europe 44-793-4960646
7.1 Introduction
Traditional through-hole Dual In-Line Package assemblies reached their limits in terms of
improvements in cost, weight, volume, and reliability at approximately 68L. SMT allows
production of more reliable assemblies with higher I/O, increased board density, and reduced
weight, volume, and cost. The weight of printed board assemblies (PBAs) using SMT is reduced
because surface mount components (SMCs) can weigh up to 10 times less than their conventional
counterparts and occupy about one-half to one-third the space on the printed board (PB) surface.
SMT also provides improved shock and vibration resistance due to the lower mass of components.
The smaller lead lengths of surface mount components reduce parasitic losses and provide more
effective decoupling
The smaller size of SMCs and the option of mounting them on either or both sides of the PB can
reduce board real estate by four times. A cost savings of 30% or better can also be realized through
a reduction in material and labor costs associated with automated assembly.
Type I is a full SMT board with parts on one or both sides of the board.
Type II is probably the most common type of SMT board. It has a combination of through-hole
components and SMT components. Often, surface mount chip components are located on the
secondary side of the Printed Board (PB). Active SMCs and DIPs are then found on the primary
side. Multiple soldering processes are required.
Type III assemblies are similar to Type II. They also use passive chip SMCs on the secondary side,
but on the primary side only DIPs are used.
PLCC Passive
Components SO
Solder Paste
Type I
SO Passive
Components PLCC
PLCC Passive
Component SO
Solder Paste
DIP
Type II
Passive
Components Only*
DIPs
Type III
Passive
Components Only*
NOTE: Intel does not recommend active devices be immersed in solder wave.
A5664-02
The process sequence for Type III SMT is shown in Figure 7-2. Leaded components are inserted,
usually by automatic equipment. The assembly is turned over, and adhesive is applied. Next,
passive SMCs are placed by a "pick-and-place" robot, the adhesive is cured, the assembly is turned
over, and the wave-soldering process is used to solder both leaded and passive SMCs in a single
operation. Finally, the assembly is cleaned (if needed), inspected, repaired if necessary, and tested.
For this type of board, the surface mount components used are chip components and small pin
count gull wing components.
The process sequence for Type I SMT is shown in Figure 7-3. For a single sided type I, solder
paste is printed onto the board and components are placed The assembly is reflow soldered and
cleaned (if needed). For double-sided Type I, the board is turned over, and the process sequence
just described is repeated.
Type II assemblies go through the process sequence of Type I SMT followed by the sequence for
Type III. In general practice, only passive chip components and low pin count gull wing
components are exposed to solder wave immersion.
Figure 7-2. Typical Process Flow for Underside Attachment (Type III SMT)
(if needed)
A5665-02
Figure 7-3. Typical Process Flow for Total Surface Mount (Type I SMT)
Clean Test
(if needed)
A5666-02
A fine-pitch package can be broadly defined as any package with a lead pitch finer than the
1.27mm pitch of standard surface mount packages like PLCCs and SOPs. Most common lead
pitches are .65mm and .5mm. There are even some now available in 0.4mm pitch. Devices with
these fine pitches and leads on all four sides are called Quad Flat Packs, (QFPs).
The assembly processes most dramatically affected by the fine-pitch package are paste printing and
component placement. Fine pitch printing requires high quality solder paste and unique stencil
aperture designs. Placement of any surface mount package with 25 mils or less of lead pitch must
be made with the assistance of a vision system for accurate alignment.
Placement vision systems typically consist of two cameras. The top camera system scans the
surface of the board and locates fiducial targets that are designed into the artwork of the board. The
placement system then offsets the coordinates in the computer for any variation in true board
location. The bottom camera system, located under the placement head, views the component
leads. Since the leads of fine-pitch components are too fragile to support mechanical centering of
the device, the vision system automatically offsets for variations in the X, Y, and theta dimensions.
This system also inspects for lead integrity problems, such as bent or missing leads.
1. Printing various amounts of solder paste on the 25-mil and 50-mil lands. One stencil thickness
will usually suffice. But stencils may be stepped down to a thinner amount for fine pitch
aperture areas to keep volumes lower to prevent bridging.
2. Cleaning adequately under and around package leads,
3. Baking of the packages to remove moisture,. Thin QFPs are susceptible to a problem known
as popcorning where moisture in the plastic can literally explode when heating in reflow or
rework and crack the plastic package.
4. Handling of the packages without damaging fragile leads.
These challenges are by no means insurmountable. Many equipment choices have already found
solutions to these issues.
Of all the issues in design for manufacturability, land pattern design and interpackage spacing are
the most important. Interpackage spacing controls cost-effectiveness of placement, soldering,
testing, inspection, and repair. A minimum interpackage spacing is required to satisfy all these
manufacturing requirements, and the more spacing that is provided, the better.
With the vast variety of components available today, it would be difficult to list or draw the space
requirements for every component combination. In general, most component spacing ranges from
0.040 in. to 0.060 in. The space is typically measured from pad to pad, lead to lead, or body to
body, whichever is closest. Smaller spacing (0.040 in) is generally used for low or thin profile
parts and small chip components. Taller parts such as PLCCs are usually spaced at 0.060 in. The
placement capability of each individual piece of equipment will partially dictate minimum
requirements. However, often the ability to rework or repair individual leads, or entire parts, will
have a stronger influence on the minimum spacing. Allowing enough space for rework nozzles or
soldering irons can save considerable cost by allowing repair of a few bad solder joints versus
scrapping the entire board. Thus, each user must set spacing requirements based on the equipment
set used.
The spacing between the pads of conventional and surface mount components may be as large as
0.100 in. so that auto-insertion equipment may used for conventional components. Clear spaces of
at least 0.050 in. should be allowed around all edges of the PC boards if the boards are tested off
the connector, or 0.100 in. if vacuum seal is used for testing, such as bed-of-nails.
Via holes are used to connect SMC lands to conductor layers. They may also be used as test targets
for bed-of-nails probes and/or rework ports. Via holes may be covered with solder mask material if
they are not required for node testing or rework. Such vias are called tented or capped vias.
Via holes may be placed under surface mount components. However, in Type II and Type III SMT
(mix-and-match surface mount), via holes under SMCs should be minimized or tented with solder
mask to prevent trapping of flux under the packages during wave soldering. For effective cleaning,
via holes should only be located beneath SMCs in Type I SMT assemblies (full surface mount) that
are not wave soldered.
The lack of standardization of surface mount packages has compounded the problem of
standardizing the land pattern. There are a variety of package types offered by the industry, and the
variations in a given package type can be numerous. For example, for the small outline package
(SOP), there are not only two lead types (gull-wing and J-lead), but there are multiple body types
such as narrow wide and thin. In addition, the tolerance on components varies significantly, adding
to the manufacturing problems for SMC users.
In this section, general guidelines are presented for land pattern designs that accommodate
reasonable tolerances in component packages, process, and equipment used in manufacturing.
These guidelines are based on manufacturability and environmental testing of different land pattern
designs for reliability.
To simplify the land pattern design guidelines, surface mount components are divided into four
different categories:
Again, with the large variety of SMT part available today, listing every pad size would create a
very long list. So, instead of providing specific pad sizes, the general formulas for the land pattern
designs are given for each of these four categories. There are several different approaches to
dimensioning pads. In addition to the guidelines below, IPC also publishes its own set of
guidelines. Each customer should study several options and decide which is best for their
application.
2. Derive the ID, using the standard pad for this pitch.
• Subtracting two standard pad lengths from the OD established in (1). The standard pad for this
pitch is:
Pitch = 0.050” Pad Size = 0.025"x0.075"
Comments:
• The outer (heel) fillet is the important one for J-leaded devices.
2. Derive the ID, using the standard pad for this pitch.
• Subtract two standard pad lengths from the OD established in (1). The standard pad for this
pitch is:
Pitch = 0.050” Pad Size = 0.025"x0.075"
Comments:
• The inner (heel) fillet is the important one for gullwing devices. Toe fillets are not required, as
they add little strength, and often don’t form anyway, due to lead trimming after plating (end of
toe may be bare copper or alloy 42).
2. Derive the OD (outside distance), using the standard pad for this pitch.
• Add two standard pad lengths to the ID established in (1). The standard pad for each pitch is:
If several vendors’ parts are proposed for the same pattern, be sure to consider them all when
extracting the above dimensions.
Warning: For parts smaller than 0805, the rounding down to the next 0.010" in the above step may result in a
gap that is too small. The formula has not yet been modified to consider these small parts.
Most companies use bed-of-nails in-circuit testing for conventional assemblies. Use of SMCs does
not impact testability if rules for testability of assemblies are strictly observed. These rules require
that (1) 0.050-in. and 0.100-in. test probes are used, (2) solder joints are not probed, and (3)
through-hole vias or test pads are used to allow electrical access to each test node during in-circuit
testing. If possible, this electrical access should be provided both at top and bottom, with the
bottom access being necessary. The main drawback of providing all the required test pads is that
the real estate savings offered by SMT are somewhat compromised. To retain these savings
requires development of some form of self-test or reliance upon functional tests only. However,
self-test requires considerable development effort and implementation time, and functional tests
lack the diagnostic capability of in-circuit tests.
Designing for manufacturability, test, and repair are very important for yield improvement and thus
cost reduction. The following sections address process issues in the manufacturing of surface
mount assemblies that play a critical role even when boards are designed for manufacturability.
In most cases, solder paste is applied on the solder pads before component placement by stenciling.
Stencils are etched stainless steel or brass sheets. A rubber or metal squeegee blade forces the
paste through stencil openings that precisely match the land patterns on the PB. Stencils are
essentially the industry standard for applying solder paste. Screens with emulsion masks can be
used but stencils provide more crisp and accurate print deposits.
The types of solder paste available fall under three main categories: Rosin Mildly Activated
(RMA), water-soluble Organic Acid (OA), and no-clean. Each of these has advantages and
disadvantages as listed in the Table 7-1, and choosing one over the others depends on the
application and the product type.
Surface mount components are available in various shipping media The most common is tape and
reel. It requires fewer machine reloads allowing more machine run time. Trays are also used,
generally for large packages such as QFPs. The EIA specification RS-481A has standardized reel
specifications for passive components and active components.
In-line placement equipment employs a series of fixed-position placement stations. Each station
places its respective component as the PB moves down the line. These machines can be very fast
by ganging several in sequence. Simultaneous placement equipment places an entire array of
components onto the PC board at the same time. Sequential placement equipment typically utilizes
a software-controlled X- Y moving table system. Components are individually placed on the PC
board in succession. These are currently the most common high speed machines used in the
industry. Sequential/simultaneous placement equipment features a software- controlled X-Y
moving table system. Components are individually placed on the PC board from multiple heads in
succession. Simultaneous firing of heads is possible.
Many models of auto-placement equipment are available in each of the four categories. Selection
criteria should consider such issues as the kind of parts are to be handled, whether they come in
tube, trays, or tape and reel, and whether the machine can accommodate future changes in other
shipping media. Selection and evaluation of tapes from various vendors for compatibility with the
selected machine is very important. Off-line programming, teach mode, and edit capability, as
well as CAD/CAM compatibility may be very desirable, especially if a company has already
developed a CAD/CAM database. Special features such as vision capability, adhesive application,
component testing, board handling, and capability for further expansion may be of interest for
many applications. Vision capability is especially helpful in accurate placement of fine- pitch
packages. Machine reliability, accuracy of placement, and easy maintenance are important to all
users.
7.7 Soldering
Like the selection of auto-placement machines, the type of soldering process required depends
upon the type of components to be soldered and whether surface mount and through-hole parts will
be combined. For example, if all components are surface mount types, the reflow method will be
used. However, for a combination of through-hole and surface mount components, reflow
soldering for surface mount components followed by wave soldering for through-hole mount
components is optimum.
The most widely accepted reflow is now "forced convection" reflow. It is considered more suitable
for SMT packages and has become the industry standard. The advantage of forced convection
reflow is better heat transfer from hot air that is constantly being replenished in large volume thus
supplying more consistent heating. While large mass devices on the PB will heat more slowly than
low mass devices, the deltas are small allowing all parts to see nearly the same heat cycle.
Component input into the placement system can be accomplished through many different format
types: molded carrier ring, singulated slide carriers, or matrix trays. In this process, no solder paste
is used. Only the solder plated on the PB lands form the solder joints. Hence, a PB vendor with
tight solder plating control is needed. First, liquid flux is applied to the lands at the mounting site.
No-clean "low solids" fluxes that can withstand the higher temperature of the thermodes
(approximately 260°C - 300°C) without carbonizing are recommended for this application. An
advanced machine vision system is used to perform component lead inspection, board fiducial
location and calculate placement location. After component placement, the hot bar blades are
brought down to "gang bond" all the leads simultaneously. The blades physically contact the top of
the component leads, holding them in place during reflow and cool down. This hold down process
results in fewer problems due to coplanarity. Heat is conducted through the leads and into the
solder deposit to form solder fillets. The blades are then allowed to cool to let the solder re-solidify
before lifting. Computer control manages the temperature and force profiles for each component
type. Different thermal masses make this essential.
With the introduction of Ball Grid Array (BGA) packages, high pin counts can be achieved on
larger pitches, typically 1.00mm-1.27mm. So the hot bar process with very tight tolerance
requirements is fading in use.
7.8 Cleaning
In general, cleaning of SMT assemblies is harder than that of conventional assemblies because of
smaller gaps between surface mount components and the PB surface. The smaller gap can entrap
flux, which can cause corrosion, which leads to reliability problems. Thus, the cleaning process
depends upon the spacing between component leads, spacing between component and substrate,
the source of flux residue, type of flux, and the soldering process. RMA cleaning requires
chemicals and has waste affuents to deal with. OA cleaning uses water that must flush down the
drain. However in this chemistry, lead is often found in the wastewater and creates an
environmental concern. No clean is generally becoming the preferred solder process since it
eliminates cleaning all together. This eliminates the environmental issues and saves in capital
costs.
One of the key issues in SMT has been to determine the cleanliness of SMT assemblies. The
Omega meter is a common tool originally used for DIP boards. For SMT, the industry also uses
Surface Insulation Resistance (SIR) surface mount boards. These boards check for ionic
contaminates left on the PB by measuring the electrical resistance between adjacent traces or
circuits.
7.9 Repair/Rework
Repair and rework of SMT assemblies is easier than that of conventional components. A number
of tools are available for removing components, including hot-air machines for removing active
surface mount components. As with any rework tool, a key issue in using hot-air machines is
preventing thermal damage to the component or adjacent components.
No matter which tool is used, all the controlling desoldering/soldering variables should be studied,
including the number of times a component can be removed and replaced, and desoldering
temperature and time. It is also helpful to preheat the board assembly to 150°F - 200°F for 15 to 20
minutes before rework to prevent thermal damage such as measling or white spots of the boards,
and to avoid pressure on pads during the rework operation. To prevent moisture induced damage,
SMT components may require bake-out prior to removal from the board. The guidelines outlined in
Chapter 8 should be followed.
7.10 Conclusion
The major technical considerations for implementing SMT include surface mount land pattern
design, PB design for manufacturability, solder paste printing, component placement, reflow
soldering, wave soldering, cleaning, and repair/rework. These areas must be studied and
thoroughly understood to achieve high quality, reliable surface mount products.
8.1 Introduction
This chapter examines surface mount assembly processes and establishes preconditioning flows
which encompass moisture absorption, thermal stress and chemical environments typical in the
variety of surface mount assembly methods currently in use. Also discussed are the standardized
moisture sensitivity levels which control the floor life of moisture/reflow sensitive PSMCs along
with the handling, packing and shipping requirements necessary to avoid moisture/reflow related
failures. Baking to reduce package moisture level and its potential effect on lead finish
solderability is described. In addition, drying, shipping, and storage procedures are included.
Note: No component body should ever be immersed directly in the solder during the wave solder
operation.
To ensure PSMC package integrity throughout the surface mount process, precautions must be
taken by both supplier and user to minimize the effects of reflow solder stress on the component.
Plastic molding compounds used for integrated circuit encapsulation are hygroscopic and absorb
moisture dependent on time and the storage environment. Absorbed moisture will vaporize during
rapid heating in the solder reflow process, generating pressure at various interfaces in the package,
which is followed by swelling, delamination and, in some cases, cracking of the plastic as
illustrated in Figure 8-1 and Figure 8-2. Cracks can propagate either through the body of the plastic
or along the lead frame (delamination). Subsequent high temperature and moisture exposure to the
package can induce the transport of ionic contaminants through these openings to the die surface
increasing the potential for circuit failure due to corrosion. Components that do not exhibit external
cracking can have internal delamination or cracking which impacts yield and reliability.
It should be noted that PSMC moisture sensitivity relates only to the risk associated with direct
exposure of components to reflow solder process stresses. No loss of package integrity is expected
for socketed parts or for through-hole mounted components not subjected to the solder reflow
environment. If through-hole components are exposed to SMT processing, then they can exhibit the
same moisture sensitivity as PSMCs. If through-hole devices are exposed to solder reflow
processes such as Convection, VPS, or IR, then they should be baked dry first, using the same
baking procedures described for SMT packages. Current data indicates that there is no negative
long term effects on reliability of PSMCs when package integrity is maintained through surface
mount processing.
The effect of moisture in PSMC packages and the critical moisture content which may result in
package damage or failure is a complex function of package design and material property
variables. These include: silicon die size, encapsulant thickness, encapsulant yield strength,
moisture diffusion properties of the encapsulant, and adhesive strength and thermal expansion
properties of the materials used in the package. The PSMC moisture sensitive phenomenon has
been identified as a contributor to delamination related package failure mechanisms including bond
lifting, wire necking and bond cratering, as well as die surface thin film cracking and other
problems. External package cracking is commonly treated as the most visible and severe form of
moisture sensitivity. It should be noted that internal cracking/delamination can be present even if
there is no evidence of external cracks. Intel has evaluated PSMC moisture sensitivity for its
current portfolio. Package moisture level has been measured as a function of temperature and
relative humidity. Critical moisture level limits to avoid cracking/delamination and other internal
damage have been determined and products susceptible to cracking/delamination have been
identified. Intel implemented handling procedures to ensure that these products are delivered to
users so that packages will not incur damage that could affect yield or reliability, during user solder
reflow processing. The user must take responsibility during storage, board mount assembly and
board rework to avoid package overexposure to moisture by following precautions recommended
in the following pages. These steps help to ensure that package integrity is maintained throughout
the surface mount process.
Moisture Absorption
During Storage Die
Minimum
Plastic Thickness
Lead
Frame
Plastic
Encapsulant
Note:
Moisture saturates the package to a level determined by storage RH, temperature, time and
plastic moisture equilibrium solubility.
241187-1 A5736-01
Plastic Stress
Fracture
Crack Collapsed
Void
Note:
Crack Forms and Pressure Dome Collapses, Emanating From Boundary of Delamination Area at
Frame Pad Edge. Remaining Void Area Acts to Concentrate Stresses in Subsequent Temperature
Cycling, Leading to Further Crack Propagation.
241187-3 A5737-01
Figure 8-3. Component Exposure to Wave Solder and Reflow Solder Environments
Through-Hole SMD
40˚ C 220˚ C
140˚ C 220˚ C
260˚ C 220˚ C
Solder Wave Solder Reflow
Temperatures
241187-4 A5738-01
Pure convection reflow processing uses convection heating to provide heat for soldering. Electric
heaters heat the atmosphere inside the furnace which then heats the boards traveling on the
conveyor. A gradual heating of the printed circuit board is necessary to drive off the volatile
constituents of the solder paste and ensures a controlled heating rate. After an appropriate preheat
time, the board is raised to the reflow temperature for soldering and then carefully cooled down
Vapor Phase soldering uses the latent heat of vapor condensation to provide heat for soldering.
Latent heat is transferred to the component as the vapor of the inert liquid condenses on the
component. The VPS temperature reaches its maximum possible value at the fluid boiling point
(215° C - 219° C). The maximum heating rate of the component on the board occurs when it is
initially immersed in the primary vapor, hence control of the heating rate for any component is
limited to preheating the part before immersion in the primary vapor zone. Because of very high
ramp rates and the high cost of fluids, VPS is used very little in production environments.
IR solder reflow processing uses radiant heating to provide heat for soldering. IR panels heat the
board traveling on the conveyor from top and/or bottom.
Hot Bar (Thermode): This is a relatively new process that is ideal for smaller to midsize volumes
of SMDs. The component's leads are held in direct contact to a heated clamp or holding device that
presses the component into the solder paste and heats the leads to the reflow temperature. The
advantage is that the plastic body does not experience the temperature extremes of a full reflow
process. This is useful for heat-sensitive components or low volume production that does not
warrant the purchase of production machinery and moisture sensitive storage equipment. (See
Chapter 9).
Note: No component body should ever be immersed directly in solder during the wave solder operation.
Intel has determined guidelines for mass reflow soldering and post solder reflow component
rework which (see Chapter 9) when followed, minimizes thermal shock to packages and meets
users' solder reflow requirements.
Because of the drive to eliminate CFC-containing materials in solder flux removal, alternative
cleaning methods have been under development. Terpene based materials have not shown to cause
any long term effects with PSMC components. A relatively recent development is the no-clean or
low clean fluxes. These require virtually no cleaning other than a water rinse following SMT
reflow.
With Ceramic or Hermetic packages (that have an internal cavity) higher power ultrasonic
equipment has been shown to cause damage to the internal bond wires and solder joints when
exposed to long cleaning times. Most of these problems have been overcome with the use of the
lower power equipment. Each component's user must evaluate their application in light of the
criteria in Table 8-1 to determine any reliability jeopardy to the boards.
8.3.1 Solderability
Table 8-1. Ultrasonic Cleaning
Power of Ultrasonic Cleaner < 30 Watts per Liter
Intel currently supplies PSMCs with copper lead frames and solder plated (tin/lead) finished leads.
A potential for lead finish solderability degradation can occur due to formation of Copper/Tin
intermetallics. To meet users' solderability requirements the formation of intermetallics must be
minimized. Intel has performed evaluations to determine solderability degradation of PSMC after
burn-in and baking, which is necessary to drive out package moisture prior to sealing sensitive
products in moisture barrier bags (MBBs). Based on the solderability work done at Intel it is
recommended that PSMCs with copper lead frames be baked at high temperature (125°C) for no
more than 48 hours by the user. Intel monitors outgoing solderability to ensure that product meets
user's solderability requirements.
8.3.2 Conclusion
Component susceptibility to moisture damage can manifest itself in many ways. Some of these can
be package cracking, bond lifts, die surface thin film cracking, bond cratering or delamination of
internal interfaces. Package cracking is one of the most severe forms of moisture sensitivity.
Package cracking has been correlated to be a function of die and package geometry and is
aggravated by moisture absorption in the plastic encapsulant. Moisture damage to crack-
susceptible components can be minimized through users’ processes if absorbed moisture is kept
below critical levels and surface mount process thermal limits are observed. Maximum temperature
profiles are recommended for these solder processes to minimize crack jeopardy due to thermal
stresses. Moisture absorption and desorption characteristics of these sensitive packages have been
characterized and moisture exposures sufficient to induce moisture damage have been determined.
Intel bakes level 2a through level 6 PSMC packages dry and seals them in bags with
desiccant before shipping*. Level 2 PSMC packages are not required to be baked prior to dry
pack. Recommended shelf life, storage conditions, floor life, maximum reflow temperatures,
redrying and handling procedures are described on the bag and in this Packaging Databook. The
user must limit exposure of moisture sensitive components to environmental moisture during SMT
assembly and rework processing to keep absorbed moisture below recommended limits and ensure
package integrity is maintained throughout the assembly process.
* 32-lead and 40-lead TSOP packages are not shipped bake and bag.
** 32-lead and 40-lead TSOP packages are the only exceptions to this practice.
• Shipping Box. The label on the shipping box indicates that desiccant packed material is
included (see Figure 8-4). This label indicates the seal date of the enclosed MBB and, thereby,
the remaining shelf life. Quantities of units shipped per box also differs to accommodate the
additional packing materials for shipments in tubes.
(V) SUPPLIER:
ASSEMBLED IN MALAYSIA
(1P) IPN: LEVEL HOURS
BAG SEAL DATE 23JAN95
(S) SPEC: (R) ROM CODE:R
241187-51 A5739-01
• Moisture Barrier Bag (MBB). Inside the shipping box is a MBB containing components. The
bag is strong, ESD-safe, and allows minimal moisture transmission. It is sealed at the factory
and should be handled carefully to avoid puncturing or tearing of the materials.
A Caution Label (Figure 8-5) and a Barcode label on the bag outlines precautions that should
be taken with desiccant packed units.
This bag protects the enclosed devices from moisture exposure and should not be opened until
the devices are ready to be board mounted. Section 8.5, Supporting Technical Information in
this document provides information on the technical aspects of the bag and characterization
information.
• Desiccant. Each MBB contains pouches of desiccant to absorb moisture that may be present in
the bag. The Humidity Indicator card (Figure 8-6) should be used as the primary method to
determine whether the enclosed parts have absorbed excessive moisture.
Do not bake or reuse the desiccant once it is removed from the MBB.
• Humidity Indicator Card (HIC). Along with the desiccant pouches, the MBB contains a
humidity indicator card (HIC). This card is a moisture indicator and is included to show the
user the approximate relative humidity level within the bag. A representation of the HIC is
shown in Figure 8-6. If the 20% dot on the card is pink and the 30% dot is not blue, then the
components have been exposed to moisture beyond the recommended limits for use in an SMT
process. If this should happen, then to use these units safely in a surface mount application, the
units should be baked dry (see Section 8.5.2 Rebaking of PSMCs). New cards being phased in
will indicate that rebake is needed if the humidity has exceeded 10%. The HIC is reversible
and can be reused. Recommendations to avoid expiration of the HIC and the need to rebake
units are included in Section 8.4.3.
• Labels. Labels relevant to this process are the “Barcode label”, “Caution label” and “ID label”
mentioned in the section on MBBs. The Barcode label (Figure 8-4) contains the date that the
bag was sealed (MM/DD/YY), the IPC/JEDEC J-STD-020 Moisture Sensitivity level and the
maximum floor life is attached to the outside of the box and on the MBB itself. The remaining
storage life of the units in the bag is determined from the seal date. All components are
guaranteed 12 months of shelf life starting from the seal date on this label. See Section 8.4.3.
The Caution label (Figure 8-5) is attached to the outside of the MBB and outlines precautions
that must be taken when handling desiccant packed units if they are to be kept dry. The ID
label is placed on the same end of the container as the barcode label.
Note: Starting in 2000 Barcode and Caution Labels that indicate the maximum reflow temperature
allowed will be phased in.
A. ID Label
UTION
CA
I VE
MO I S
IT
T S
UR
E S EN
B. (MBB) Caution Label
LEVEL
CAUTION
This bag Contains
MOISTURE-SENSITIVE DEVICES If blank see adjacent
bar code label
1. Calculated shelf life in sealed bag: 12 months at <40˚C and <90% relative
humidity (RH).
2. Peak package body temperature: ________________________˚C
(If blank, see adjacent bar code label)
3. After bag is opened, devices that will be subjected to reflow solder or other
high temperature process must be
a) Mounted within ________________________ hrs. of factory conditions
(If blank, see adjacent bar code label)
A5740-02
Humidity Indicator
Examine
Item 30
if Pink
Change
Desiccant 20
if Pink
Warning
if Pink 10
241187-6 A5741-01
• Tubes. Units shipped in tubes are packed with an additional precaution. Antistatic foam
protects the bag from the sharp edges of the tubes and tacks. Otherwise, the units shipped in
tubes are packed with materials as indicated in Figure 8-7 through Figure 8-11.
• Trays. Units shipped in injection-molded trays are packed with additional precaution.
Antistatic foam lids enclose the trays to protect the bag from the sharp edges of the trays. Trays
are packed with materials as indicated in Figure 8-12 through Figure 8-14. (See “Tray Recycle
Program” in Chapter 10.)
• Tape and Reel. Units shipped in tape and reel are packed as indicated in Figure 8-15 through
Figure 8-17.
Figure 8-7. Bag Packing for PLCC Full or Half Length Tubes
Desiccant
Pouches
Foam
Cavity
Humidity
Indicator
Card
241187-7 A5742-01
Bubble Wrap
241187-29 A5743-01
Desiccant Bag
Desiccant
Pouches
Humidity Indicator
241187-9 A5744-01
End Pads
ABCDEF ABCDEF
Top Portion
Folded Under
241187-10 A5745-01
#3 Box Desiccant
Bar Code Label
#3 Box
(Front Flap)
241187-11 A5746-01
Humidity
Indicator
Card
Desiccant Bag
Foam Lid
Pin 1 Corner
Chamfer
Desiccant Pouches
241187-12 A5747-01
To Box
Top Portion
Folded Under
241187-13 A5748-01
241187-14 A5749-01
Desiccant Bag
Desiccant Pouches
Reel
Humidity Indicator
Card
241187-15 A5750-01
ABCDEF
ABCDEF
Corners
Folded
Under
241187-16 A5751-01
Figure 8-17. Placement of Desiccant Included Label on Tape and Reel Box
ABCDEF
241187-17 A5752-01
8.4.3 Handling
The following information details handling procedures that should be used with PSMCs packed in
desiccant bags and intended for surface mount applications. Following these handling guidelines
will ensure that components maintain their as-shipped, dry state alleviating package cracking and
other moisture-related, stress-induced concerns that may be associated with the surface mount
process.
1. Incoming Inspection. Upon receipt, shipments should be inspected for a seal date within the
last six months. Bag integrity should also be verified. There should not be holes, gouges, tears,
or punctures of any kind that expose either the contents or an inner layer of the bag. The
barcode label can be reviewed for conformance to the purchase order, but the bag should not
be opened until the contents are ready to be used (either inspected or board-mounted). Please
see the following Manufacturing Conditions/Floor Life section for details of allowable
exposure times once the devices are removed from the bag or exposed to the ambient.
2. Storage Conditions/Shelf Life. The customer receives components in the sealed MBB
between 0 and 6 months after the seal date indicated on the Desiccant Barcode label. The
sealed bag and enclosed desiccant have been designed to provide a minimum of 12 months of
storage (Intel storage time + customer storage time) from the seal date in an environment as
extreme as 40° C and 90% relative humidity. The customer will have at least six months of
shelf life available on the components without the need to rebake them before use.
If the worst-case storage conditions (time, temperature, or relative humidity) are exceeded and
there is a need to verify whether inventory has been affected, then a bag can be opened and the
HIC can be checked for expiration. If the HIC has not expired, then new desiccant can be
added and the bag resealed. If the HIC has expired, then the devices should be 1) rebaked and
used in manufacturing within the guidelines outlined in the Rebaking section, 2) rebaked and
resealed in an MBB with fresh desiccant, or 3) rebaked and stored in an environment of ≤
10%RH before they are used in a surface mount process. Please see Rebaking section for
additional information.
3. Opening MBBs. To open a moisture barrier bag when the contents are ready to be used or
inspected, simply cut across the top of the bag as close to the seal as possible, being careful not
to damage the enclosed materials. By cutting close to the seal, you will allow as much room as
possible for resealing. Once the bag has been opened, please follow the guidelines for ambient
exposure time in the following section to ensure that the devices are maintained below the
critical moisture level.
Time Time
Time Conditions Conditions Conditions
(Hours) (Hours)
If the actual MET is less than 24 hours the soak time may be reduced. For soak conditions of
30 °C/60% RH the soak time is reduced by 1 hour for each hour the MET is less than 24 hours.
For soak conditions of 60 °C/60% RH, the soak time is reduced by 1 hour for each 5 hours the
MET is less than 24 hours.
If the actual MET is greater than 24 hours the soak time must be increased. If soak conditions
are 30 °C/60% RH, the soak time is increased 1 hour. for each hour that the actual MET
exceeds 24 hours. If soak conditions are 60 °C/60% RH, the soak time is increased 1 hour for
each 5 hours that the actual MET exceeds 24 hours.
Please contact your local Intel Sales office for specific handling questions. All of the same
restrictions for exposure time (outlined above) apply. Reference IPC/JEDEC J-STD-020
“MOISTURE/REFLOW SENSITIVITY CLASSIFICATION FOR NON-HERMETIC SOLID
STATE FURFACE MOUNT DEVICES” and IPC/JEDEC J-STD-033 "STANDARD FOR
HANDLING, PACKING, SHIPPING AND USE OF MOISTURE/REFLOW SENSITIVE
SURFACE MOUNT DEVICES".
Where exposure times and/or ambient conditions are difficult to control, Intel highly
recommends dry storage capability.
5. In-Process Storage. Intel highly recommends having dry storage capability available for units
that will not be used within the allowable exposure time. PLCCs and PQFPs (≤100 leads) can
be stored outside of the barrier bag for long periods of time if the ambient relative humidity is
less than 10%. This applies to both long term and in-process storage. A desiccator with dry
nitrogen or air (≤5%RH source) is suggested for such storage. Desiccator storage conditions
for larger PSMCs are currently being established.
6. Rebaking. PSMCs should be rebaked if and only if they have been exposed to excessive
moisture as indicated by exceeding the recommended ambient exposure time or by expiration
of the HIC.
In the event that the units should be rebaked, see Section 8.5.2.
7. Resealing Moisture Barrier Bags. If there is a need to reseal Moisture Barrier Bags for any
reason, then Intel recommends the following guidelines to ensure that the bag seal does not
allow moisture into the bag. The seal area must not exhibit any separation when subject to load
and temperature conditions specified in MIL-B-81705, and must be impermeable to moisture
according to MIL-B-81705. Intel uses a seal pressure to 60 psi–70 psi, and a seal time of 3-4
seconds at approximately 225 °C. The integrity of the seal is vital to the storage life of the
devices.
Intel ships moisture sensitive PSMCs which have been packed following a tightly controlled
process. This flow is shown in Figure 8-18.
Text
Mark
Bake
Package
Scrap
Visual
Reject
Ship Out
241187-18 A5753-01
Once the cracking jeopardy of surface mounted PSMCs was identified, Intel characterized
component absorption/desorption rates and saturation limits as a function of temperature and
relative humidity. The handling and shelf life guidelines outlined in the first section of this
document are based upon the experimental analysis of the desiccant pack materials. The water
vapor transmission rate of the moisture barrier bag (MBB), desiccant absorption rate and saturation
levels, and maintainable MBB internal relative humidity were all important factors in developing
the recommendations given.
Manufacturers using PSMCs in non-SMT applications may continue to use PSMCs without
altering their current process flow. Non-surface mounted PSMCs do not undergo the same
temperature excursions and thermal stresses associated with the Convection, VPS or IR solder
reflow processes and, therefore, do not have the same jeopardy related to them as unprotected
PSMCs used in SMT applications.
Molecular Sieve
1.0
Phosphorus Pentoxide
0.9
0.1
1 2 3 4
Time (hrs)
241187-22 A5757-01
22
20
Molecular Sieves
Percent Moisture Capacity
18
16
14
12
10
8
6
4
Silica Gel
2
0
0.1 1 10 20 40 100
The number of UNITS required per bag may be determined by the following equation:
Equation 8-1.
( 0.304xMxWVTRxA )
U = -----------------------------------------------------
D
Where:
U = Amount of desiccant in UNITS
M = Shelf life desired in months
WVTR = Water vapor transmission rate in grams/100 in2 in 24 hrs
A = Total surface area of the MBB in square inches
D = The amount of water in grams, that a UNIT of desiccant will absorb at 10% RH
Note: If materials such as trays, tubes, reels, etc., are placed in the bag without baking, additional
desiccant will be required to absorb the moisture contained in these materials.
• Shelf Life. Intel has determined the shelf life of bagged components based upon the bag
WVTR, the desiccant absorption rate, and the desiccant saturation limit. The total shelf life
(Intel + customer) for bagged components in worst case warehouse conditions of 40 °C/90%
RH is 12 months.
≤ 1.4 mm 2a 4 h. 5 days
3 7h. 11days
4 9 h. 13 days
5 10 h. 14 days
5a 14 h. 19 days
≤ 2.0 mm 2a 18 h. 67 days
3 24 h. 67 days
4 31 h. 68 days
5 37 h. 68 days
5a 48 h. 68 days
It is not advisable to store PSMC units at the low bake temperature longer than the time required to
dry out the units for use in the reflow. Lengthy storage times at elevated temperatures can lead to
problems with intermetallic formation between the lead frame material and the lead finish,
increased oxidation of the lead finish which can contribute to added reflow and wetting problems,
and extended elevated temperatures can cause the antistatic properties of the shipping media
(tubes/tape and reel) to deteriorate.
• Solderability Considerations/Number of Rebakes. Solderability tests performed on PSMCs
exposed to either bake cycle are the basis for Intel's recommendations and limits on the
number of allowable bake cycles. PSMCs should not be baked more than 48 hours by the
customer if using the high temperature bake of 125 °C for 48 hours. Following this guideline
will limit the formation of Cu6Sn5 intermetallic and therefore, not promote solderability
degradation. The low temperature bake of 40 °C does not require this restriction.
• Because components are baked in the shipping containers at 40 °C, possible outgassing
products as well as intermetallic formation impact on solderability were evaluated. Figure
8-21 is a "box plot" analysis of solderability measured by coverage of the leads in "number of
squares". There is overlap in the distributions of, 1) the control units stored in metal trays, 2)
units stored in plastic shipping containers and, 3) units baked in plastic shipping containers,
therefore, there is no statistical difference between the treatments. No difference in
solderability was observed after multiple rebakes at low temperature. Based on this data, there
is no restriction on the number of times devices can be rebaked at the recommended low
temperature before solderability is degraded beyond acceptable limits.
18
15
A B C
Control 22 40
Degrees Degrees
Treatment
241187-25 A5759-01
this evaluation are described in Figure 8-22 and show that package crack susceptibility for PLCCs
is dependent on the die attach pad dimensions and the thickness of plastic between die attach pad
and nearest external surface.
Figure 8-22. Crack Sensitive Packages: Package Die Attach Pad Area versus Package
Minimum Plastic Thickness
80
70
60
Plastic Package Minimum
Plastic Thickness (Mils)
50
Safe
40
30
Crack Sensitive
20
10
0
.04 .08 .12 .16
2
Die Attach Pad Area (Inches )
241187-26 A5761-01
1. Bake 10 units of each product for 48 hours at 125 °C to dry out any absorbed moisture
(preferably 5 units from each of 2 date codes).
2. Use acoustic microscopy to detect initial internal delamination and cracking. Record images
and analyst's observations. If these parts do not meet the acceptability criteria listed below,
then contact the vendor.
3. Saturate the units by soaking them in an unbiased Temperature/Humidity chamber for time
and temperature/humidity combinations required for the level of moisture sensitivity being
evaluated (see Table 8-2).
Note: Devices which exhibit package cracking after saturation at 85 °C/85% RH have an increased
probability of cracking during the SMT process if they are surface mounted after storage under
uncontrolled conditions. Such devices should be treated as moisture sensitive and only used in a
dry state for SMT applications. This dry state can be achieved either by baking the units prior to
surface mount or by receiving dry devices in desiccant pack from the vendor (as Intel currently
provides).
4. Run the units through three passes of vapor phase solder or convection reflow (infrared reflow
should not be used unless equivalence with VPS has been demonstrated) within 2 hours of
removal from the Temperature/Humidity chamber.
5. Use acoustic microscopy to detect post-reflow internal delamination and cracking. Record
images and analyst's observations. If these parts do not meet the acceptability criteria listed in
Section 8.7.2, then they fail the tested level.
8.7.2 Criteria
The general criteria defining moisture sensitivity are applied in a hierarchical manner.
1. If the components pass electrical test, there is no visual evidence of external cracks, and there
is no evidence of delamination or cracks observed by acoustic microscopy, then the component
is considered to pass that level of moisture sensitivity.
2. If the components pass electrical test and there is backside paddle or heatspreader
delamination, but there is no evidence of cracking or other delamination, then the component
is considered to pass that level of Moisture Sensitivity.
3. If internal cracks are observed by acoustic microscopy, then components will be cross-
sectioned and the cracks evaluated according to the following criteria.
— Cracks are not allowed to intersect the bond wire, ball bond, or wedge bond.
— Cracks are not allowed to extend from any lead finger to any other internal feature (lead
finger, chip, die attach paddle).
— Cracks are not allowed to extend more than two-thirds (2/3) of the distance from any
internal feature to the outside of the package.
— Failing components must be evaluated to the next level of moisture sensitivity.
Components with internal cracking that do not fail this criteria should be subjected to
temperature cycle, and tested to full function electrical end points.
— If acoustic microscopy shows any surface-breaking feature which is delaminated over its
entire length, the component must be tested to the next level of moisture sensitivity. A
surface-breaking feature includes: lead fingers, tie bars, heatspreader alignment features,
heat slugs, etc.
4. If acoustic microscopy scans exhibit any delamination which meets the following criteria, then
the components will require further evaluation using Environmental stresses.
— Measurable change in delamination on the top surface of the chip.
— Measurable change in delamination on any wire bonding surface of the leadframe/die
paddle.
— Measurable change in delamination along any polymeric film bridging any metallic
features which are designed to be isolated.
The method outlined here indicates whether or not a device is susceptible to internal delamination
or package cracking. To evaluate surface mount related failure rates over time, it is necessary to
stress the units. The following section (Section 8.7.3) describes a method using temperature
cycling and THB (temperature/humidity, biased) stressing to evaluate failure rates due to any kind
of surface mount-related, stress-induced failure.
Cross Section
Locations
241187-27 A5762-01
• Determination of failure rates and resulting comparisons should only be made after analysis of
failures has been completed. Invalid failures may result and should not be used in the final
failure rate assessment.
• Precondition 154 units per lot from three different lots of the same product (462 units total).
This flow simulates the conditions and chemical exposures a device typically sees during
board mount and rework as indicated. The component vendor should be contacted to
determine the preconditioning flow and parameters appropriate for the component under
evaluation.
• A reduced sample size of 90 units per lot (270) total can also be used. Sample sizes given are
based on LTPD charts given in MIL-STD 38510.
• Please note that the preconditioning mentioned below is determined by the level of moisture
sensitivity of the specific component.
• Following pre-conditioning, divide each of the three lots in half and subject them to the
following stresses:
THB (85° C/85% RH, Biased) Temp Cycle MIL STD Condition "B"
3 lots of 77 units each 3 lots of 77 units each
(45) (45)
Read-out at 168 hours Read-out at 200 cycles
500 hours 500 cycles
1000 hours 1000 cycles
NOTE: Numbers given in ( ) represent number of units if using reduced sample size.
8.7.4 Preconditioning
The purpose of a preconditioning step in the qualification and reliability stressing flow is to
simulate the actual board mounting process that the parts will see at the customer's site. By
completing this stress on the units before the reliability data is gathered, the data more accurately
reflects the life expectancy the units will experience in the field or customer's application.
To ensure that SMT process stressing is comprehended in component reliability evaluations, Intel
has established product qualification precondition flows to which all surface mountable plastic
products are subjected prior to standard component reliability stressing*. The effect of these
preconditioning stresses and their impact on long term package performance continues to be
quantified. User assembly processes not comprehended by this preconditioning flow should be
discussed with Intel engineers to verify that package integrity of Intel PSMCs are maintained in the
specific application.
* The flow is selected to match the level of moisture sensitivity classification of the specific component.
The information in this document is for reference only. Manufacturing processes are unique, and may
require unique solutions to ensure acceptable levels of quality, reliability, and manufacturing yield.
Due to differences in equipment and materials, customer-specific process parameter development and
validation is required.
Although there are a number of lead free (Pb-free) alloys, the most commonly used compositions
contain tin, silver, and copper, commonly expressed as SAC, for SnAgCu. Within SAC solders, by far
the most common usage is Sn/3Ag/0.5Cu, a near-eutectic which melts between 217°C and 220C.
Apertures sizes can be 1:1 with pad size, but certain parts may require reduced apertures to reduce
solder ball defects. Larger pads may benefit from crosshatched openings, to reduce the amount of
paste applied, and to control scavenging.
Pb-free solder paste may spread less during reflow than SnPb paste, potentially leaving extremities of
pads unsoldered. Although full pad coverage by solder is not a requirement of IPC-A-610, some
customers prefer to enlarge apertures to ensure that pads are covered.
Using a metal squeegee reduces scavenging and provides more consistent printed paste volume.
Equipment used to print SnPb paste can be used, without modification, to print Pb-free paste.
Process parameters (such as squeegee speed, pressure, and separation speed) need to be optimized for
the specific solder paste used.
Solder joint formation depends on temperature and time which are reflected in the reflow profile. In
leaded devices the volume of solder paste on the land is significantly greater than the plated solder
volume on the component lead and is the key contributor to joint formation. However in BGAs the
balls on the component are the main contributor to the solder volume of the joint. In both cases, the
paste volume applied is critical to the formation of the joint.
There is no one best reflow profile for all board assemblies. Ideally, a reflow profile must be
characterized for each board assembly using thermocouples at multiple locations on and around the
device. The solder paste type, component and board thermal sensitivity must be considered in reflow
profile development.
In addition to having higher reflow temperatures, Pb-free reflow soldering also requires a narrower
temperature range, in order to produce reliable joints, without damaging components. Maintaining this
narrower range could require new reflow ovens, depending on number of zones and degree of control
in ovens formerly used for SnPb soldering.
Because of additional oxidation that occurs at higher temperatures, an inert reflow atmosphere
(nitrogen) may be beneficial for Pb-free reflow soldering.
Of course, higher temperatures drive the need for all Pb-free components to be rated to higher
temperatures.
Finally, these temperatures can also cause greater warpage in PCBs, and in some cases, may require
alternate PCB materials, or carrier fixtures during reflow.
Each customer should develop their own reflow profile and oven settings, appropriate to their materials,
equipment, and products. As a starting point, this chapter contains considerations and
recommendations for reflow solder parameters. Because some reflow parameters differ with solder
paste formulation (even if they have the same metal composition), the profile envelope recommended
by the solder paste manufacturer should be considered.
Unless stated otherwise, all temperatures in this chapter are measured at solder joints, rather than at
components bodies, PCB surface, or air around components. This provides the best repeatability and
accuracy.
Thermocouples (TCs) for solder joints should be placed in joints expected to be the hottest and coolest,
so that the range of peak temperatures for all components on the board can be confirmed to be within
specifications. The hottest joints on a board are typically on small passive components, so one of these
should be monitored for peak temperature on the profile board. The coolest joints on a board are
typically large BGAs and sockets. A TC should be used in a joint at one corner of the component, and
in a joint at the center of the part, or as near to the center of the part as possible. Sockets with
actuating mechanisms may require an additional TC at a joint near the mechanism, if its mass could
make that area harder to heat.
In addition to solder joints, component body temperature, measured at top center or as close as possible,
may also need to be monitored, to avoid exceeding the body temperature spec of the part.
Here are examples of TC locations for reflow profiling on BGAs or sockets, for both fully populated
arrays and partially populated arrays (no balls in the center area of the part).
Components with other Pb-free ball alloys, such as SnAg or SAC105, may require higher minimum
peak temperatures to form reliable joints.
Components are typically rated as per J-STD-020C (or later), based on their package thickness and
volume. Although Intel BGAs are generally rated at 260°C, other components, especially large ones,
may be rated at 250°C or 245°C. This means that 250°C may not be usable as the max joint
temperature for some components; a lower temperature may be required. Since larger parts normally
reach lower maximum temperatures during reflow than smaller parts due to the physics of heat transfer,
keeping them below their ratings may not be difficult, as long as 250°C is used as the maximum for the
joints of smaller parts.
Time Above Liquidus, or TAL, is often used to describe this time. But technically speaking, 217°C is
the solidus temperature of SAC305, the point at which the solder becomes fully solid during cooling.
220°C is the liquidus temperature, the point at which it becomes fully liquid during heating. Between
these two temperatures, the solder is partially molten and partially solid.
Intel recommends that TA217 of 40-90 seconds be used for SAC305 or 405 solder paste and balls.
With large or massive boards, an exception may be required, allowing up to 120 seconds above 217°C.
During cooling from peak temperature down to 205°C, Intel also recommends that a minimum ramp
rate be used. Solder joints with cooling rates of 1°C/sec or greater are characterized by finer
microstructure features. Literature studies indicate that this is better for long term reliability. Faster
cooling rates also inhibit growth of intermetallic compounds in the bulk solder.
Producing and controlling a narrower temperature range, while maintaining production speeds,
typically relates to the number of heating zones in the oven. Assemblers with reflow equipment with
greater temperature control (i.e. greater numbers of zones) will be better positioned to meet the tighter
Pb-free process envelope requirements, particularly for larger, more complex boards.
However, other measures can often be taken to achieve similar results without using nitrogen at reflow.
Examples include:
• Solder paste selection.
• PCB surface finish selection.
• Printing solder paste on test pads rather than leaving them exposed during reflow.
• Test probe head style selection.
Warpage amount varies with PCB size/thickness/laminate, number of reflow cycles, and warpage
control method. In experiments at Intel, warpage of 1.0 and 1.2mm thick boards did not cause any
performance problems.
• Sample process applies to all types of SMT components on the board, not just the BGAs.
• All temperatures are measured with thermocouples inside solder joints, for better accuracy
• Max temp applies to the hottest joint on the board, typically a joint of a small passive device.
• Reference process applies to all PCBs with nominal thickness .040” to .077” (1.02 to 1.96mm), and
to PCBs .078” to .093” (2.0 to 2.36 mm) with large active devices on one side only. Thick PCBs
with thermally massive parts on both sides may require adjustments to Peak Temp and TA217.
217ºC
9.5 Rework
9.5.1 Pb-free vs SnPb Rework
Rework must provide higher temperatures for Pb-free solder. Greater temperature profile control is
required, which may require different nozzles or equipment replacement. Rework can be the most
difficult module to develop for Pb-free. Rework (both SMT and Through Hole) on thick PCBs is
especially challenging.
After developing the initial profile, place an additional TC at the topside location corresponding to the
TC with the hottest joint temperature.
• Because of the nature of hot air rework, and the variety of nozzle designs, there may be a
significant temperature gradient across the part during rework.
• Therefore, monitoring body temperature with a TC only in the center may not represent what
the rest of the body is exposed to.
• Use this topside TC to confirm that component body temp is not exceeding its max rating.
Adjust profile if needed.
Here are examples of TC locations for rework profiling on BGAs or sockets, for both fully populated
arrays and partially populated arrays (no balls in the center area of the part).
* * * * * = Locationjoint.
of TC in solder
TCs to measure joint temperatures are installed through holes in the board, using the same method
described earlier for reflow profile boards.
Pb-free hand solder may require soldering iron tips hotter than used for SnPb rework. Hotter tips allow
rework at a pace similar to SnPb rework. Without hotter tips, desoldering and resoldering is slower.
Intel Packaging Databook 9-11 Board Reflow Process Recommendations
Revised 12-2007
However, with hotter tips, caution must be used to prevent pad lift. If Pb-free tips are not available, Pb
may be purged from standard SnPb tips by repeatedly flooding with Pb-free solder and then cleaning.
Because of the difficulties of placing, aligning, and printing with a mini-stencil on the assembled PCB,
a method is available that prints paste directly onto the socket BGA balls, rather than onto the PCB, as
follows:
5. A mini squeegee prints paste onto the balls, and removes excess paste.
6. The jig is inverted once more, the clamping frame is opened, and the part is removed by the rework
machine’s vacuum pick for placement onto the board.
250º
245º
Socket max peak temp range 230-250ºC
150º
Falling Ramp Rate
Soak Time from 150 to 217ºC: < 100 sec
0.5-2.0ºC/sec
(varies with solder paste selection)
Start with solder After nozzle is lowered, Nozzle is down during Nozzle rises when
joint temp < 40°C. prior to peak reflow. peak reflow. joints go below 217°C.
Preheat with
bottom heater,
before nozzle
is lowered.
Target to exit Solder Joint BGA Solder Joint Temp: Peak Temp Range, Solder Joint
this step Temp: 200 to 220°C. and TA217 specs met. Temp < 80°.
125 – 150°C Socket Solder Joint Temp:
190 to 215°C.
Other specs Rising Rising Ramp Rate. Peak Temp Range. Falling Ramp Rate.
to check Ramp Rate. Critical Rising TA217.
during this step Ramp Rate. Component Max Body
Soak Time. Temp and Time.
Tube profiles are designed with minimum clearance over the maximum package
dimensions to reduce damage caused by movement of the device within the tube. For
some package types, tubes have “riding rails” on which the packages rest while in the
tube. The rails protect the fragile leads from touching anything in the tube. PVC tacks,
nylon tacks, or rubber plugs are used to retain the units. All tube wall thickness are 0.025
inches to 0.040 inches. Table 10-1 through Table 10-8 show tube dimensions and cross-
sections and quantity of packages per tube for most Intel package types. Additional
information on new packages should be requested through Intel Field Sales.
NOTES:
1. (P) = PDIP
2. (C) = Ceramic Sidbrazed
3. (D) = CERDIP
10.1.2 Carriers
Additional protection from lead damage is necessary for the fragile leads of flatpack packages,
which are shipped flat to be trimmed and formed at the customer site. Plastic carriers are used to
hold each unit, then the loaded carrier is placed in the tube. Carriers are either coated with antistatic
surface treatment are intrinsically static dissipative. Figure 10-1 through Figure 10-3 show a
variety of carrier types.
-E-
25.40 (1.000)
A
13.56 (0.534)
0.05 (0.002) M D E M F M
19.05 (0.750)
-F-
-C-
2X 10.16 (0.400)
(16X 1.27 (0.050)
A
4X 45˚ X 0.51 (0.020)
2X 10.16 (0.400)
[Dimensions in mm (inches)]
240822-1 A5784-01
1.995
3X 2.000
[Dimensions in Inches]
240822-2 A5785-01
1.995 ± 0.005
3X 0.183 ± 0.002
2.000
± 0.005
1.250 0.372
0.487
Section A-A
[Dimensions in Inches]
240822-3 A5786-01
10.1.3 Trays
Shipping trays are built in compliance with JEDEC thick and thin standard dimensions. Mid-
temperature trays can be baked to 140° C while low temperature trays can withstand a maximum
sustained temperature of 65° C. Trays are constructed in modified polysulfone (PS) or equivalent
for mid-temperature applications and polycarbonate (PC) for low temperature applications because
of their high deflection temperature, superior strength, and dimensional stability. All JEDEC trays
have the same “X” and “Y” dimensions and are easily stacked for storage and manufacturing.
PQFP 84LD, 100 LD, 132 LD, 164 LD, 196 LD thick mid-temperature
84 LD, 100 LD, 132 LD, 196 LD thin mid-temperature
132 LD, 196 LD single unit thick mid-temperature
PGA 68-84 LD 11 x 11, 88-100 LD 13 x 13, 132-139 LD 14 x 14, 149 LD 15 x 15, 168-208
LD 17 x 17, 240-296 LD 19 x 19, 273 LD 21 x 21, 325 LD 26 x 24 thick
low-temperature
PLCC 28 LD square, 28 LD rectangular, 44 LD square, 68 LD square, 84 LD square thick
high-temperature
TSOP 32 LD, 40 LD, 48 LD, 56 LD thick mid-temperature
32 LD, 40 LD, 56 LD thin mid-temperature
SSOP 48 LD, 56 LD thick mid-temperature
CQFP 132 LD formed, 164 LD flat, 196 LD formed, 196 LD flat thick high-temperature
MQFP 44 LD (10 x 10), 64 LD (12 x 12), 80/100 LD (14 x 20) thick and thin
mid-temperature
SQFP 80 LD (12 x 12), 100 LD (14 x 14), 208 LD (28 x 28) thick and thin mid-temperature,
208 LD (28 x 28), single unit thick mid-temperature
TQFP 144 LD (20 x 20), 176 LD (24 x 24) thick and thin mid-temperature
TCP carrier high-temperature
MSC 19 x 19, 0.880 high spacer low-temperature
TCP carrier high-temperature
BGA 27 x 27 and 35 x 35 thin mid-temperature
PPGA 296 LD thick low temperature
µBGA 40B (12x22) and 48B (9x18) thin mid-temperature
Illustrations of trays for various packages are shown on the following pages. Intel field sales
engineers can provide detailed drawings and specifications upon request.
5.35
-D-
M3
M
-B-
M1 M2 Detail B 3X 0.06 R
Package -C-
Pin-1
Orientation 12.25
Detail C -7 ˚Draft
Detail D 0.008 M E M Detail A
0.480 5X
1.00
-A-
0.030
1.35 5X
0.400 0.10
Y1
12.27
-7 ˚Draft
0.008 M E M
10.05
12.40
-E-
12.70
240822-6 A5787-01
0.15
0.10 0.05
0.11
0.065 0.18
Typ 60˚
Detail A
Scale 4:1
3˚
Typ 0.05
0.157
VENDOR REV
PART #
Tray Made in XXX
Detail D
Vendor Information Black Scale 2:1
5.20
-7 ˚Draft
0.008 M D M
0.12
Typ 0.62 Typ
4X R 0.015
0.50
5.22
-7 ˚Draft
0.008 M D M
End View
240822-7 A5788-01
12.70
12.40
Detail C -E- Detail D
5.35
-D-
M3
TMT PQFP 100
XXX˚C MAX
M
-B-
M1 M2 Detail B 3X R 0.06
Package -C-
Pin-1 Detail A
Orientation 12.25
-7 ˚Draft/Side 4X 1.000
0.008 M E M +2 ˚/Side
0.300
0.250
-A- 0.000
0.000
0.030
4X 0.100 8X 0.03R
2X 10.050
2X 1.350
0.000
0.060 Ref
12.27
-7 ˚Draft/Side
0.008 M E M
0.015 Corner Radius
240822-98 A5789-01
0.100
0.150
0.078
Detail A 30˚ Detail B 0.157
4:1 4:1
2X Detail R 0.187
0.138
0.050
5.20
Stacking Feature
0.008 M D M
0.015 Corner Radius
2X 0.500 2X 3.626
5.22
Stacking Feature
0.008 M D M
0.015 Corner Radius
End View
240822-99 A5790-01
2.500
-E-
M2
Detail A
2.500
-D-
M3
1.250 Vendor/Device
Information
Block
Vendor/Device
-B- Part #
XXXX XXXL
3X R 0.03
0.125 X 45˚ 1.250 3X R 0.06 Detail A
-C-
Revision Level
0.480
0.400
-A-
0.000
0.10
4X R 0.03
4X R 0.015
2X 2.380 ±0.010 -7˚/S TSC
0.008 M DE M
240822-A0 A5791-01
Table 10-9. Injection Molded Thick and Thin PQFP JEDEC Tray
PQFP Tray Dimensions
240/ 325/
Pocket Locations Symbol 68 L 88 L 132 L 168L 273 L
296L 387L
Pocket Cntr Location to Edge Y M 1.072 1.573 1.556 1.506 1.473 1.456 2.675
Pocket Cntr Location to Edge X M1 1.152 1.077 1.285 1.513 1.446 1.861 1.700
Pocket-Pocket Cntr Distance X M2 1.683 1.708 1.966 2.344 2.377 2.892 3.000
Pocket-Pocket Cntr Distance Y M3 1.603 2.204 2.237 2.337 2.404 2.438 N/A
# of Rows of Pockets Rows 3 2 2 2 2 2 1
# of Columns of Pockets Columns 7 7 6 5 5 4 4
Total # of Pockets Pockets 21 14 12 10 10 8 4
NOTE:
1. Dimensions are in inches.
Table 10-16. Injection Molded Thick and Thin MQFP JEDEC Tray
MQFP Thick and Thin Tray Dimensions
80/100 80/100
(14 x 20) (14 x 20)
Symbol 44 LD (10x 10) 64 LD (12 x 12)
Pocket Locations Thick Thin
Table 10-17. Injection Molded Thick and Thin SQFP JEDEC Tray
SQFP Thick and Thin Tray Dimensions
Pocket Locations Symbol 80 LD (12 x12) 100 LD (14 x 14) 208 LD (28 x 28)
28F160B3A
28F016B3A
28F160C18
28F800B3
28F008B3
28F160B3
28F016B3
28F160C3
28F320B3
28F320C3
28F008S3
28F016S3
28F160F3
28F640J5
Symbol
Pocket
Locatio
ns
All trays are subjected to a variety of inspections to ensure they meet Intel’s specifications prior to
reuse by an Intel factory. Non-Intel trays or trays that fail to meet Intel’s quality requirements are
sent to plastic reclamation vendors for utilization in other plastic applications. No trays are sent to
land fills.
For further details on Intel’s wafer and die sales procedures, refer to Chapter 18 of this Intel
Packaging Databook. .
Po A
E1
P1 A
T
Detail B
Section A-A
0.20 ± 0.05
R 0.75 Max
*R Detail B
1.00 Max
100.00
1.00 Max
250.00
40 Min.
Access Hole at
Slot Location
W2 (Measured at Hub)
A D
• Moisture Barrier Bag (MBB). Inside the shipping box is a moisture barrier bag containing
components. The opaque MMB is constructed of three layers: a conductive polyethylene inner
layer for sealing, an aluminum film mid-layer, and a tyvek outer layer. The bag meets MIL-
STD-81705 TYPE I for electrostatic discharge (ESD) and mechanical stability. The measured
water vapor transmission rate (WVTR) of the bag is better than the MIL-STD requirements for
moisture protection. A “warning” label on the bag outlines precautions that should be taken
with desiccant-packed units. A desiccant barcade label is also affixed to the bag.
• Desiccant. Each MBB contains one or more pouches of desiccant to absorb moisture that may
be present in the bag. The desiccant is supplied in one-unit pouches. The number of pouches
required is a function of the bag surface area. GelPaks use 3x4" desiccant cards.
• Humidity Indicator Card (HIC). Each MBB contains a humidity indicator card. This card is
a military-standard moisture indicator and is included to show the user the approximate
relative humidity (RH) level within the bag. The HIC is reversible and can be reused.
• Labels. The desiccant barcode label (shown in Figure 10-11) mentioned above in the section on
MBB, contains the date that the bag was sealed (MM/DD/YY). The remaining storage life of
the units in the bag is determined from this date. The “warning” label attached to the outside of
the MBB outlines precautions that must be taken when handling desiccant-packed units if they
are to be kept dry.
• Shipping Box. The barcode label on the shipping box indicates that desiccant-packed material
is included. This label indicates the seal date of the enclosed MBB, and thus, the remaining
shelf life.
10.5.2.1
Tube and Reel Labels. Tube labels with information on lot traceability, part and spec numbers, quantity of
parts, and ROM and PROM codes are available by special order. Reel labels are standard and provide the
same information. Customer part number references also can be included on either type of label by special
order.
• Box Labels. Bar-coded labels for each box are standard on Intel product shipments. Box labels provide all
the information on the tube labels, show order packing and shipping information, and allow more space to
define special requirements.
Figure 10-11
• The standard box and bag labels identify the parts as Pb-free (or with a 2nd Level interconnect that meets
the requirement to be ROHS compliant) with the type of interconnect material, the peak reflow
temperatures of the component, the quantity of parts in the box, the lot, and the pack dates.
• See chapter 17 of this Databook for further information on Pb-Free package marking and labeling.
Number Nomenclature
ED -7406A General rules for the preparation of outline drawings of integrated circuits small outline J-
lead packages (SOJ)
ED -7408A General rules for the preparation of outline drawings of integrated circuits pin grid array
packages (PGA)
ED -7414 General rules for the preparation of outline drawings of integrated circuits guarding quad
flat packages (GQFP)
ED -7415 General rules for the preparation of outline drawings of integrated circuits small outline
packages with heat sink (HSOP)
ED -7417 General rules for the preparation of outline drawings of integrated circuits bumpered
quad flat packages (BQFP)
ED -7418 General rules for the preparation of outline drawings of integrated circuits glass sealed
quad flat packages (QFP-G)
ED -7419 General rules for the preparation of outline drawings of integrated circuits glass sealed
dual in-line packages (DIP-G)
ED -7421 General rules for the preparation of outline drawings pf integrated circuits ceramic dual
in-line packages (DIP-C)
ED -7422 General rules for the preparation of outline drawings of integrated circuits glass sealed
quad flat J-leaded packages (QFJ-G)
ED -7423 General rules for the preparation of outline drawings of integrated circuits ceramic quad
flat J-leaded packages (QFJ-C)
ED -7431-1A Recommended outline drawings for carriers quad tape carrier packages (QTP carrier)
ED -7431A General rules for the preparation of outline drawings of integrated circuits quad tape
carrier packages (QTP)
ED -7432 General rules for the preparation of outline drawings of integrated circuits dual tape
carrier packages (type1) (DTP(1))
ED -7433 General rules for the preparation of outline drawings of integrated circuits dual tape
carrier packages (type 2)
ED -7441B Standard for the package of universal memory devices
ED -7500A Standards for the dimensions of semiconductor devices (discrete semiconductor devices)
EDR-7311 Design guideline of integrated circuits for quad Flat package (QFP)
EDR-7312 Design guideline of integrated circuits for thin small outline package (type1)
EDR-7313 Design guideline of integrated circuits for thin small outline package (type 2) (TSOP2)
EDR-7314 Design guideline of integrated circuits for shrink small outline package (SSOP)
EDR-7315A Design guideline of integrated circuits for ball grid array (BGA)
EDR-7316 Design guideline of integrated circuits for Fine-pitch Ball Grid Array and Fine-pitch Land
Grid Array (FBGA/ FLGA)
EDR-7317 Design guideline of integrated circuits for Surface Vertical Package (SVP)
EDR-7318 Design guideline of integrated circuits for plastic very thin small outline non-lead package
(P-VSON)
EDR-7319 Design guideline of integrated circuits for quad flat J-lead packages (QFJ)
EDR-7320 Design guideline of integrated circuits for small outline package (SOP)
EDR-7601 Guidance of embossed carrier taping for integrated circuits
The Tape Carrier Package (TCP) format is one way to meet the small outline and high leadcount
interconnection needs of high performance microprocessors. The TCP has been designed to offer
reduced pitch, thin package profiles, smaller footprint on the printed circuit board, without
compromising performance. Intel continues to provide packaging solutions which meet rigorous
criteria for quality and performance. The Tape Carrier Package is no exception. Key package
features include surface mount technology design, lead pitch of 0.25 mm, 48 mm tape format,
polyimide-up for pick and place, and slide carrier handling. Shipped flat in slide carriers, the leads
are designed to be formed into a “gull-wing” configuration and reflowed onto the PCB by one of
several methods. Intel has done extensive optimization of the hot bar reflow process and
suggestions for that process are included in this chapter. Satisfactory placement and rework
capability has been demonstrated by industry sources using the hot gas reflow process. Industry
data also exists which demonstrates process feasibility for laser reflow.
The TCP family has been characterized for thermal, electrical, and mechanical performance.
Component and system level thermal testing has shown the TCP package to be capable of meeting
system level thermal design needs. Additional potential board level enhancements have been
identified and characterized to provide the most flexible design choices. A full suite of component
and board level stress testing has been completed to ensure that the component meets Intel’s
reliability targets. Evaluations of solder joints by stress testing, lead stiffness studies, and finite
element modeling have demonstrated that the mounted component will meet field use conditions
and lifetimes. The TCP package is capable of meeting a wide variety of design and use
applications. Table 12-1 provides an overview of TCP package attributes.
The opposite side view—topside of tape, topside of die, and bottomside of the carrier is seen in
Figure 12-2.
The tape is in 48 mm format and includes test pads outboard of the OLB window (see Figure 12-3).
Pads are 0.5 mm x 0.65mm on 0.40 mm pitch on two rows.
A cross section view of the TCP package is illustrated in Figure 12-4. The “five-sided” encapsulant
covers the top surface of the device, the sides of the device, and the ILB area to the polyimide
carrier ring. The tape bow or offset across the polyimide carrier film is product specific. Contact
Intel Corporation for additional information. After forming and mounting to the PCB, the total
height of the component above the PCB is less than 0.75 mm.
Details of the tooling holes can be seen in Figure 12-5. Intel uses the tooling holes shown in the
upper left and lower right of Figure 12-1 for alignment during processing at Intel and should not be
used during board assembly. The other two tooling holes have been left pristine for use during
board assembly.
The OLB window has been designed to facilitate excise and form. The polyimide carrier film can
be cut and a narrow strip left in place at the outer edge of the OLB area to act as a “Keeper Bar” to
maintain lead coplanarity, and spacing, if so desired. The detail of the OLB Window area is shown
in Figure 12-6.
Pin # 1
Tooling
Holes
#1 And #3
DIE
Used For BACKSIDE 63.00 ± 0.15
Intel
Assembly
Process
Not For
Use In
Board
Assembly
63.00 ± 0.15
5.00 ± 0.05
272588-1 A5698-01
4X 01.925 ± 0.05
0.10 M A S BS Pin # 1
63.00
± 0.15
OLB Area
36.15 33.00
ENCAP ± 0.05
–B–
Polyimide Tape
33.00 ± 0.05
–A–
36.15
51.00
3X R2.50 ± 0.05
ø0.05 M AS BS
272588-2 A5699-01
(43.94)
36.15 ± 0.05
48.18 ± 0.12
DIE
21.90 ± 0.06 BACKSIDE 36.15 ± 0.05
272588-3 A5700-01
0.125 ± 0.015
(Polyimide)
0.130 ± 0.040
(Encap)
0.560
A2 Lead Thickness ± 0.050
(LT) (Die) 0.430
272588-4 A5701-01
2x 24.00 ± 0.10
28.50 20.57
± 0.10 ± 0.10
ENCAP
22.59
± 0.10
2x (30.00)
See Detail B
See Detail A
Top View
4 x 0 0.60 ± 0.03
272588-5 A5702-01
(Device Window)
1.06
0.10 ± 0.01
0.25 Ref
(Encap)
12.00 2.25
272588-6 A5703-01
Develop Expose
Anneal
Saw Wafer
Reel-To-Reel ILB
Encapsulate
Quality Control
Electrical Test
In parallel with the bumping process, TAB tape sites are manufactured in reel-to-reel format.
Polyimide carrier film with an adhesive in reel form, is punched to create the Inner Lead Bond
(ILB) and Outer Lead Bond (OLB) windows and tooling holes for subsequent processing steps.
Copper foil is laminated onto the polyimide and cured. Again, a photolithographic technique is
used to create the specific pattern of metal leads and test pads. Once the tape metal pattern has been
created, the exposed copper metal is plated with a Ni barrier metal and Au outer plating. Au outer
plating is used for the entire tape interconnect path; both ILB and OLB lead areas are plated with
gold. Intel has done extensive testing of the solder joint reliability of nickel-gold plated leads.
The silicon and the TAB tape are brought together at the TCP package process. After the wafers are
sawn into individual devices, the TAB tape sites are matched to the device. During Inner Lead
Bonding, the ILB area of the tape is aligned to the bumps on the device. They are brought into
contact and a gold-to-gold weld is formed. This process establishes the silicon to PCB
interconnection path. After ILB, the component sites are singulated from the reel, the tape plating
bars are “debussed” from the tape, and the individual component sites are loaded into ESD
protective slide carriers (see Figure 12-1 and Figure 12-2). Once in slide carriers, the devices are
encapsulated in a high temperature polymer coating. The coating covers the top and sides of the
silicon, the bumps, and the ILB area to the polyimide carrier ring. Complete coverage of the ILB
area provides mechanical support to the ultra-fine pitch leads, protecting them from handling
damage and thermomechanical stress induced damage. The thermoset polymer is cured to ensure a
high enough cross-link density to fully protect the device from environmental degradation. After
quality control checks and electrical test, the components are ink marked, packed, and shipped. The
components are shipped flat in slide carriers. This ensures that the outer lead area is undamaged at
the time of board mount processing.
2x 7.62
A
1.20 ± 0.13 A
294.00 ± 1.50
Side View
78.40 ± 0.25
70.80 ± 0.25
64.50 ± 0.50
25.50
± 0.50
2.75 Max
3.00 ± 0.30
78.40 ± 0.25
70.80 ± 0.25
64.50 ± 0.50
10 x 45˚ Cham.
272588-8 A5706-01
At this time Intel suggests hot bar and hot gas reflow processes that do not subject the component
body to reflow temperatures. There is no jeopardy with moisture sensitivity when using these
localized reflow processes. The TCP package has met all reliability requirements after exposure to
the hot bar mounting process. Therefore, the TCP components are shipped without desiccant
packing materials. There are no “out of bag” shelf life restrictions prior to component mount. For
additional information contact your Intel representative.
Parts Are Cut Out of The Carrier And Leads Are Formed
The Die Backside is Placed Into The Preapplied Die Attach Material, Attaching it to The Board.
Parts Are Immediately Soldered to The Board Using One of The Available Reflow Processes.
A5707-01
The land length should be long enough to allow a solder fillet to form at the toe and heel of the
lead. To prevent a starved joint or excess gold concentration in the solder joint it is suggested that
the total solder volume of the land and the resultant solder joint be greater than 0.004 mm3. The
land pattern terminal dimension is defined as the distance from outer land edge to land edge as
illustrated in a sample land pattern in Figure 12-13. This dimension is based on the lead toe
location from center, the incremental land length allowed for solder fillet formation, and the
tolerances. Generally, this is approximately 1.0 mm longer than the toe-to-toe dimension.
Additionally, it is suggested that trace connection to lands be “necked down” by 0.013mm to help
eliminate “solder thieving” during reflow.
125 ± 25 µm
Top
Cu Cu
100 µm
125 ± 25 µm Min. Bottom
Bottom Between Copper
Land Width Specification
This Specification Applies to the Copper Geometry Only
(Before Solder Application)
272588-18 A5708-01
Solder
Cu Cu
Figure 12-13. Sample Land Pattern for a TCP with 24 mm Body Size for Use with a Hot Bar
Reflow Process
2.50
Land Length
1.00 0 Typ
0.125
Fiducial Dia.
Land Width
[Measurements in mm]
272588-9 A5710-01
12.4.7 Fiducials
The ultra fine pitch of these components may require that pattern recognition systems be used to
accurately locate the lead and the land. To facilitate the pattern recognition algorithms it is
recommended that PCB lands have fiducials associated with each TCP site. Each equipment type
will have specific requirements for fiducial configurations. For KME equipment, round fiducials of
0.3mm in diameter located on adjacent corners (same side of the site) are suggested. For Universal
Instruments or Zevatech equipment, Surface Mount Equipment Manufacturers Association
(SMEMA) Std. 3.1 compatible fiducials 1.0 mm (39 mil) in diameter should be placed at all 4
corners of the TCP site; at least 2 fiducials in opposite corners are required. Other equipment
manufacturers may have other requirements. Please verify fiducial design requirements with the
equipment supplier before finalizing board designs. A possible fiducial position is shown in Figure
12-13. They should be located within the terminal dimensions of the land pattern and equally
spaced from the centroid of the site. The contrast of these fiducials is critical to providing adequate
edge contrast. Therefore, the finish and finish morphology should not change during reflow
processes. For example, Au or Sn are acceptable. Solder mask should be pulled back from the edge
of the fiducial by 20 mils (SMEMA 3.1) to maximize Pattern Recognition System effectiveness.
Figure 12-14 shows the solder mask pull back and the fiducials for a TCP land pattern site.
Fiducials
Solder Mask
272588-10 A5711-01
Traces can be routed on the signal layers between the thermal vias, however, this area of the board
can get to near 100° C. This should be considered before routing critical traces through this area.
Figure 12-15. Hole Size and Location Illustration for a 24 mm Body Size TCP Lead Guard
0.554"
1.108
± 0.008"
(Max. True Position Tolerance)
272588-20 A5712-01
D/A Pad
Size Die Size + 7.5 mm (75 mil free area around periphery for
wet-out)
Via Diameter 13.5mils
Via Location Center of Vias should start 0.65 mm (25 mils) from the
edge of the pad.
Fiducials
Size (Diameter) 1 mm (39 mils) for Universal and Zevatech. 0.3 mm for
KME equipment.
Location Minimum 2 cross-diagonal corners of the TCP site located
within the terminal dimension of the land pattern for
Universal and Zevatech equipment. Minimum of 2
adjacent side corners of the TCP site for KME equipment.
Solder Mask
Clear area around entire Land Pattern and 1.25 mm + 0.125 mm pull back from the edges of the land
Fiducials. pattern and 0.23 + 0.125 mm from the ends of the lands.
Fiducials should be clear of Solder Mask.
NOTE:
1. Land Patterns should be dimensioned in metric.
TCP lead forming is a three step process which removes the component from the slide carrier and
excess carrier film, cuts the leads from the support structure, and bends the lead to specified
configuration and dimensional accuracy. Excise, or removal of the component from the excess tape
occurs immediately prior to fluxing and component placement. The recommended tool set should
cut the leads free of the tape and then bend them into a modified “gull-wing”. Although not
required, a “keeper bar” or strip of polyimide carrier ring can be used to maintain coplanarity and
lead spacing during fluxing and placement. The keeper bar is a narrow strip of the carrier tape
which is cut during the trim operation and remains in place on each row of leads after excise. The
excise operation, itself, removes the device area and leaves the test pad and sprocket hole portion
of the tape in the slide carrier. Figure 12-16 shows a recommended lead form configuration. Key
items are tabulated in Table 12-5.
12.74
13.64
14.32
14.22
0.52
(0.125) Polyimide
0.30
0
Lead Thickness (LT) 0.5
9
0.2
2x R0.175 +10˚
35 ˚ -05˚
0.37 ± 0.025
0.060
(0.90)
12˚ ± 12˚
272588-11 A5713-01
It is necessary that sufficient distance between the bottom of the silicon device and the die attach
pad be built into the leadform to allow for die attach material to attach the device to the board. A
thermally conductive die attach medium is used to ensure optimum performance of the device. A
suggested die attach material and processing flow are discussed in the next section.
Intel engineers have assessed several different lead form radii for shoulder, heel, and toe angles. To
eliminate cracks in the outer plating of the lead which expose the copper base metal, a radius of at
least 0.15 mm is suggested.
Table 12-5. Key Lead Form Dimensions
Controlled Dimension Recommended Range
* Other brands and names are the property of their respective owners.
Electrically conductive, this material was selected for its thermal conductivity and mechanical
performance characteristics.
Several die attach materials are available commercially for use as thermally conductive,
electrically non-conductive adhesives. These materials may be adequate substitutes for a thermally
and electrically conductive die attach material. For additional information on these materials
contact your local Intel representative and request the applications note “Thermally Conductive
Adhesives.”
Alternate materials which meet these criteria may be available but have not been characterized by
Intel.
Table 12-7. Recommended Thaw Times and Temperatures for Different Containers and
Container Sizes of Ablebond* 8380 (Courtesy Ablestik)
Container Container Recommended Thaw Recommended Thaw
Size Type Time (Hours) Temp (°C)
Containers of Ablebond* 8380 that appear to have separated should not be used. Separation is
visually observable as a band of color (yellow or amber) along the length or top of the container.
To maintain high quality performance, adhesive dispensed from an unstirred reservoir (10 cc and
smaller) must be completely used within a 24 hour period.
All pastes must be used in a 24 hour period from the time the syringe is opened. Any thawed
adhesive not required for production should be returned to the freezer immediately. Any thawed
adhesives not used (not opened in a 24 hour period) may be refrozen once. Contact Ablestik
directly for maximum recommended time between dispense, placement and cure.
Reference material dispense parameters for Ablestik Ablebond* 8380 are listed in Table 12-9, and
a reference pattern for a die size of 13.302mm x 12.235mm and a single needle time-pressure
dispense system is illustrated in Figure 12-17 and Table 12-10. Further process development may
be required to optimize the amount of material dispensed in order to minimize voids, control bond
line thickness, and control fillet height. Dispense patterns should be verified for each thermal via
pattern.
Figure 12-18 illustrates the low temperature cure profile used at Intel. This profile was set at 130°
C maximum temperature as a compromise between snap cure and the need to keep the profile
below the glass transition temperature of a majority of PCB materials.
Note: THE PARAMETERS IN TABLE 12-9 ARE EVALUATION SETTINGS ONLY. VALUES WILL
CHANGE DEPENDING ON SEVERAL FACTORS INCLUDING DIE PAD VIA PATTERN,
DIE PAD PLATING TYPE, EQUIPMENT SET AND EPOXY VISCOSITY.
Figure 12-17. Basic Dispense Pattern Locations for Die Attach Material
Die Outline
272588-12 A5714-01
Table 12-10. Sample Die Attach Medium Dispense Pattern: 24 mm Body Size Component and
13.302 x 12.235 mm Die Size Component
STEP TYPE GEO[1] [2] [3] [4] AMOUNT/SPEED
1 DOT 0 0 0 0 0.2
2 LINE -0.5 -0.5 -6.25 -6.75 5.5
3 DOT -4. 5 -1.61 0 0 0.2
4 LINE +0.5 -0.5 6.25 -6.75 5.5
5 DOT 4.5 1.61 0 0 0.2
6 LINE 0.5 0.5 6.25 6.75 5.5
7 DOT -1.66 4.-5 0 0 0.2
8 LINE -0.5 0.5 -6.25 6.75 5.5
9 DOT 1.61 -4.5 0 0 0.2
10 DOT 1.61 4.5 0 0 0.2
11 DOT -4.5 1.66 0 0 0.2
12 DOT -1.61 -4.5 0 0 0.2
13 DOT 4.5 -1.66 0 0 0.2
This pattern provides baseline information for process development. The pattern requires
verification by the board level assembly site.
Figure 12-18. Low-Temperature, Cure Profile Used on a Convection Belt Oven to Cure
Ablebond* 8380
135˚C 3 Status-3
130˚C 2
Temp=56
Batt-4.985
100˚C Pts=900
Acc=00020
00:00:01.0
60˚C
1
20˚C
272588-17 A5715-01
12.4.10.3 Fluxing
A Rosin Mildly Activated (RMA), halide free, no residue flux is suggested. Multi-core no-clean
X33-04 has been used successfully for the hot bar application for both SnPb and Au lead finish.
Optimum results have been obtained by immersing the leads of the TCP component in the flux.
Immersion should cover the entire surface of the foot, top and bottom up to the top of the heel
radius. The specific gravity of the flux controls the amount of flux which remains on the leads after
immersion and should be closely controlled at 0.80 to 0.81. The solids content of the flux should
remain in the range 1%-3%. Because this is a no-clean material the surface insulation resistance
should be monitored and kept at >109 Ω minimum between adjacent leads. Extractable ions have
been measured for this material at less than 100 ppm Cl-, Na+ and less than 50 ppm K+. The time
between application of flux and solder reflow should be minimized.
* Other brands and names are the property of their respective owners.
4. A gold plated copper lead identifier flag on pin one corner of the component.
Intel engineers have done extensive process development of hot bar reflow for 0.25 mm TCP
components. The reflow thermal and force profiles are shown in Figure 12-19. Blade design is
crucial for effective hot bar reflow. Blades should maintain flatness across the active surface of the
blade. Temperature variations across the blade should be less than 10° C. Four independent
ceramic blades with tungsten resistors are suggested. Each blade is used to reflow one side of a
TCP component. Mean blade-to-blade temperature differentials should be kept less than 5° C. The
blade width should be such that the contacted area of the TCP foot is less than the length of the flat
of the foot. A baseline thermal profile is shown below. This profile has been shown to yield
acceptable solder fillets. Different temperature/time profiles may be required for different thermal
densities.
Figure 12-19.
Time
Bond Force
Time
272588-13 A5716-01
Heel fillets should extend 1/3 to 1/2 the height of the heel radius. The solder wetting angle should
be positive. Reliability stress test data has shown that toe fillets are not required for acceptable joint
reliability after 1000 cycles of -55° C to 125° C. However, the presence of toe fillets may be a
quality indicator for the reflow process.
The pick-up head design can contribute to component alignment control and bond line thickness of
the die attach medium. Pick-up contact on the polyimide carrier ring area has been found to provide
a wide process window for alignment in some equipment. Pick-up tooling design should be
verified with your equipment supplier.
12.4.10.5.3 Removal Process Suggestions After Hot Bar Mount And Cure
Intel has developed the following removal process which can be used as a starting point for the
customer’s own development effort. The actual times and temperatures may differ depending on
the rework machine and equipment utilized. Please note that this process was developed on a 0.062
thick FR-4 PCB with Au die attach pad, using Intel’s suggested die attach material Ablebond*
8380. Other PCB assemblies with varying thickness, material and die attach, may require different
removal profiles.
TCP removal is a thermally profiled, stepped process, which removes the device from the PCB
assembly after die attach cure. This process is centered around breaking the bond between the die
attach material and the PCB and/or die, by using the coefficients of thermal expansion variations
associated with the die and PCB material. This characteristic of the removal process is based on the
thermoset properties associated with Intel’s suggested die attach material Ablebond* 8380.
The stepped process is derived from the underboard heating of the removal site, prior to top surface
heating. This process allows for the PCB to expand for a longer period of time than the die,
creating the forces necessary to break the thermoset die attach bond. Intel has developed the
removal process with Air-Vac’s* DRS 26 Semi-Automated Soldering and Desoldering Machine.
The removal time profile is approximately 135 seconds with the equipment settings as shown in
Table 12-11. These actual settings and time may differ depending on the actual configuration of the
PCB assembly the TCP is being removed from. The time profile may even be reduced by using
higher wattage heating elements.
Table 12-11. DRS-26 Settings
Air Pressure 85psi
Even though this process is centered around a manual method, the DRS-26 allows the user
programming capabilities to semi-automate the process to improve throughput and efficiency.
The nozzle assembly that Intel used during the development of this process is a center vacuum
ported design with the air flow directed to the outer perimeter of the nozzle. This allows for
peripheral heating from the top side of the TCP, which is an advantage when trying to break the
thermoset bond created during the curing process. This nozzle was equipped with a high
temperature O-ring to seal the vacuum for TCP pick up, around the inner tape perimeter. See
Figure 12-20 for the nozzle design.
0.042 R
0.060 Typ 0.078 R
0.188 Dia
0.691
0.689
1.061 0.619
SQ
1.059 0.617
SQ SQ
0.063 R
0.063 R
[Measurements in inches]
272588-21 A5717-01
The following outline for TCP removal is for a manual operation under the DRS-26 operating
environment.
Thermal vias in the die attach pad allow the heat from the die to be transferred and spread into the
board. Heat is also transferred through the board to the opposite side where it can be further spread
and transferred. Figure 12-21 shows thermal vias connecting to a heat spreading plane on the
opposite side of the board. A heat pipe is shown for illustration purposes. In the illustration, a
transfer block made from copper or aluminum is used to clear the component height on the
backside of the board. The other end of the heat pipe is connected to a heat spreading area such as
the keyboard plate or the bottom chassis.
Thermal vias can be arranged in several configurations within the die attach pad. It is suggested
that a full grid of 0.34 mm (13.5 mil) as-drilled thermal vias be placed on 1.27 mm (50 mil)
minimum centers across the die attach pad. Decreasing pitch (increasing via count under the die)
will further improve heat transfer into the board. The thermal vias should be connected without
thermal relief to the ground planes(s). A ground plane that mirrors the die attach pad should be
placed on the opposite side of the board from the TCP site to enhance system heat spreading
solutions.
Additional use of heat pipes on the opposite side of the board further enhances thermal
performance. Figure 12-21 illustrates a possible mounting. The advantages of mounting the heat
pipe to the board rather than the device include: coupling to the major thermal path for this device,
the ability to select the adhesive or mechanical attachment method, mechanical isolation of TCP
leads from the load of the heatsink especially in vibration, and the ability to utilize potential open
real estate on the back side of the board to increase the thermally active area.
Encapsulant Tape
Thermal Vias
Thermal Plane
Thermal Grease or Adhesive
Thermally Conductive
Material
Transfer Block
Heat Pipe
272588-14 A5718-01
For detailed information on system thermal design solutions please contact your local Intel sales
office.
The construction of TCP components is unique compared to traditional CPU packages such as
PGAs and PQFPs. Both the PGA and the PQFP use wire bond technology that connects the die to
the package. The package leads then provide the final connection to the printed circuit board. The
TCP package uses TAB (tape automated bonding) interconnect which provides a direct connection
from the die to the outside world. The result is a low inductance path from the die to the board
when compared to traditional PGA or PQFP packages.
Originally, the mobile processor required a thermally and electrically conductive path between the
board and the device. Additional tests have revealed that an electrically conductive path between
the board and the device is not required.
For more detail about the electrical performance of a specific product in the TCP package consult
the datasheet or call your local Intel sales office.
The TCP package lead has a nickel underplate and gold final plate in the outer lead bond area of the
package. The nickel acts as a barrier between the copper lead and gold surface, ensuring a
solderable lead at the customer site.
Intel has done extensive testing on the reliability of solder joints. Packages with Ni/Au were
assembled onto boards with various plating thicknesses which bracket Intel’s recommended lead
finish thickness. No failures were detected after the boards were subjected to vibration stress,
mechanical shock, and thermal cycling stresses performed in Figure 12-22.
2
Vibration 5 Hz--2000 Hz, 0.01g /Hz, (3 Axis, 15 Minutes/axis)
A5719-01
Intel has developed a light weight, low profile and low cost TCP lead guard. For additional
information on this design, please request a copy of the TCP Lead Guard Application Note from
your Intel representative.
Die Opening
2.0mm
Ref
Cross Section Detail
34.0 mm Ref
34.0 mm Ref
Top View
272588-22 A5720-01
Zakel, E., Azdasht, G., Kruppa, P., and Reichl, H., “Reliability Investigations of Different Tape
Metallizations for TAB Outer Lead Bonding”, in Proc., 4th ITAB, Feb 16-19, 1992, San Jose, CA,
pp. 97-120.
Lee, C.K., Wong, Y.M., Doherty, D., Tai., K. L., Lane, E., Bacon, D.D., and Baiocchi, F. “Study of
Ni as a barrier metal in AuSn soldering application for Laser chip/submount assembly”, in J. Appl.
Phys., 72 (8) Oct 1992, pp. 3808-3815.
Bai, P., Gittleman, B.D., Sun, X-Y., McDonald, J.F., and Lu, T-M., Costa, M.J., “Diffusion in Ni/
Cu bilayer films”, in Appl. Phys. Lett., 60 (15), April 1992, pp. 1824-1826.
Daebler, D.H., “An Overview of Gold Intermetallics in Solder Joints”, in Surface Mount
Technology, October 1991, pp. 43-46.
Thompkins, H.G. and Pinnel, M.R., “Relative Rates of Nickel and Copper Diffusion Through
Gold”, in J. of Appl. Phys., 48 (7) July 1977. pp. 3144-3146
Kotlowitz, R.W., “Comparative Compliance of Representative Lead Designs for Surface Mounted
Components”, IEEE Trans. CHMT, vol. 12, Dec. 1989, pp. 431-448.
Lau, J.H., “Stiffness of PQFP “Gull-wing” Lead and Its Effect On Solder Joint Reliability”, in
IEEE-CHMT Proc., 1988, pp. 131-132.
Englemaier, W., “Test Method Considerations for Smt Solder Joint Reliability”, in Proc. IEPS
Conf., Oct. 1984, pp. 360-369.
Sandor, B.I., “Life Prediction of Solder Joints: Engineering Mechanistic Methods”, In Solder
Mechanics--A State of the Art Assessment, D.R. Frear, W.B. Jones, and K. Kinsman, eds., TMS,
1990, pp. 363-419.
Morris, James, E. ed., Electronics Packaging Forum, Vol. 2, Van Nostrand Reinhold, NY, 1991.
IPC-SM-782A, “Surface Mount Design and.n Land Pattern Standard”, IPC, Chicago, Aug. 1993.
Takubo, C., Tazawa, H., Yoshida, A., Hirata, S., and Sudo, T., “A Remarkable Thermal Resistance
Reduction in a Tape Carrier Package on a Printed Circuit Board”, in Proc., 5th ITAP, Feb 2-5,
1993, San Jose, CA, pp. 44-51.
13.1 Introduction
As Intel microprocessors become faster, more complex and more powerful, the demand on
package performance increases. Improvements in microprocessor speed and functionality drive
package design improvements in electrical, thermal and mechanical performance. Package
electrical and thermal characteristics become attributes of component performance along with the
mechanical protection offered by the package.
To meet these requirements, Intel has introduced a variety of innovative package designs. The
development of the plastic pin grid array (PPGA) package, PPGA2, and Flip Chip Pin Grid Array
(FC-PGA) have provided an improvement path for enhanced power distribution and improved
thermal and electrical performance. While each consists of organic package materials, the primary
differences within the package is that PPGA utilizes wirebond interconnect technology while the
FC-PGA utilizes Flip Chip Interconnect. Externally, the PPGA thermal interface will be made to
an integral package heat slug while the FC-PGA thermal interface will be made directly to the die
backside. PPGA and FC-PGA are both socket compatible.Table 13-1 summarizes the key
attributes of the PPGA package. The following sections detail the physical structure, electrical
modeling and performance attributes of the PPGA package.
Physical
Appearance Circuit board, exposed pins
Package Body Material BT laminate, Ni plated Cu heat slug, epoxy
encapsulant, Au bond wires
Body Thickness 3.0 mm (overall, includes heat slug)
Weight 18 grams
Package Trace Metal Copper
External Heat Slug Yes
External Capacitors Yes
Performance
Thermal (θjc) with Heat Sink 0.30 - 0.50 °C/W
Power Distribution Cu traces and multiple planes enhance distribution
Package Trace Propagation Delay Cu traces have low resistance and reduce delay
Others
Thermal Interface Used for Heat Sink Attachment Thermally conductive and electrically non-conductive
grease, phase-change material, film, or tape
Board Mount PPGA socket
Caution: For PPGA packages, electrically conductive surfaces should not touch any part of the processor
except the heatslug. For example, an electrically conductive heat sink should not contact the exposed pins,
external capacitors, or the exposed metal on the side of the package.
Physical
Performance
Thermal (θjs) with Heat Sink 0.6 °C/W (12-15 lbf. Clip force)
Power Distribution Cu traces and multiple planes enhance distribution
Package Trace Propagation Delay Cu traces have low resistance and reduce delay
Other
Thermal Grease Used for Heat Sink Attachment Thermally conductive and electrically non-conductive
Board Mount Socket only
Warning: For FCPGA packages, electrically conductive surfaces should not touch any part of the
processor except the die. For example, an electrically conductive heat sink should not
contact the exposed pins, external capacitors, or the exposed metal on the side of the
package.
The Micro Pin Grid Array (µPGA) is the latest innovative packaging approach, developed as a
conveyance for Organic Land Grid Array (OLGA) package technology CPUs in the thin and light
configuration of mobile notebook computers. The µPGA is also known as an Interposer based
package.
The interposer is a pinned FR-4 carrier which affords the OLGA package to be surface mounted to
the interposer for future socketing by the OEMs. The development of the interposed package is a
result from the OEM’s requirement for a manufacturing alternative allowing flexibility in selecting
whether to surface mount an OLGA or socket onto the motherboard at final assembly.
This High I/O interposed package utilizes a one for one, pin to BGA ball connectivity arrangement.
Advantages to this packaging technique, is it’s minimally larger surface area than the OLGA CPU
itself, and that it attains a height reduction over the earlier Mobile Module used in notebook
computers.
The entry of the µPGA packaging technology continues the commitment to provide packaging
solutions that meet Intel’s rigorous criteria for quality and performance.
Table 13-3 summarizes the key attributes of the PPGA package. The following sections detail the
physical structure, electrical modeling and performance attributes of the PPGA package.
Table 13-3. µPGA Package Attributes
µPGA
Sq./Rect. R R
Pitch (mm) 1.27 1.27
Interposer Thickness (mm) nominal 1.0 1.0
Socketable pin length (mm) 1.25 1.25
nominal
Weight (grams)
Max. Footprint (mm) nominal 34.21 x 28.27w 37.51 x 35.56w
Shipping Media
Trays X X
Desiccant
Comments/Footnotes
NOTE: Interposer size are Die size dependent. Please contact Intel Technical support for latest specifications
13.2.1 PPGA
|
Kovar Pins
Solder Reset
BT Core
Copper
,,
zz
yy
Signal BT Pre Preg
Signal Slug Adhesive
Power
VSS
||
{{
Die
Copper Slug
Chip Cap
Solder Solder
A5769-01
A5770-01
Figure 13-3, Table 13-4 and Table 13-5 illustrate the package outline drawings and dimensions for
the 296 lead PPGA package. The package meets JEDEC outline spec MO-128 for pin count and
package size. The package is square and 1.95 inches on a side and 3 mm in total thickness. The
index corner has a 45° chamfer.
Seating Plane
L Solder Resist
D
D1 e1 A2 Chip Capacitor
S1
F1
F2
B
1.65 D2
(Ref)
D
A
2.29
1.52 Heat Slug
A1
Pin C3
˚
45 Chamfer
(Index Corner) measurements in mm
A5771-01
13.2.2 FC-PGA
FIBER-REINFORCED DIE
RESIN
NON-REINFORCED RESIN
COPPER
SOLDER RESIST
SOLDER
AU/NI-PLATED KOVAR
A7646-01
A7647-01
Figure 13-6, Table 13-7 and Table 13-8 illustrate the package outline drawings and dimensions for
the 370 lead FCPGA package. The package meets JEDEC outline spec MO-128 for pin count and
package size. The package is nominally 1.95 inches (49.53 mm) on each side and .076 inches (1.93
mm) in total thickness (not including the pins). There is no chamfered corner on this package, so
pin 1 is indicated by a gold triangle.
Capacitor
Placement
Area
G2 D B2 C2
Side View
Seating Plane
A1
A2 L
G3 Pin TP φP
A7648-01
13.2.3 µPGA
Because of the material likeness between the OLGA package and the Interposer, the assembled
package exhibits improved coefficient of thermal expansion (CTE) compliant package properties
in an assembled application.
The interposer has a die specific OLGA land pattern of .024" (0.609mm) diameter metal defined
lands on .050" (1.27mm) pitch. It correspondingly accommodates a .050" (1.27mm) pitch offset
pattern of pin lands. This pin land pattern is offset .025" (0.635mm) in the X and Y directions in
relation to the OLGA land pattern. The pin lands and BGA land pairs are connected to each other
with a .010" (0.254mm) wide trace. Figure 13-7 illustrates this concept.
Figure 13-7. Pin Spacing and BGA land offsett for .050” (1.27mm) pitch
.035 .050
.025
φ .031"
Pin Land
.025
.025 .0078
.025
φ .024"
BGA Land .019
.050
A7190-01
The pin base material is copper alloy (C19400) or Kovar, chosen because of its excellent electrical
characteristics and resistance to bending. During manufacturing, the pin wire is extruded through a
series of progressive dies to arrive at its configuration. The sharp edges are broken to .002"
(0.050mm) typical, and are plated with 80 microinches minimum, of nickel and then overplated
with 8 microinches, minimum of gold to ensure like metal contact with the socket contacts. This
configuration was specifically developed for the thin and light mobile application.
The pin configuration used with interposer -1 interposers are a contour shoulder type. This
configuration is illustrated in Figure 13-8.
These pins are nominally .012" (0.304mm) in diameter, have a .010" (0.254mm) maximum thick
and a .024 (0.609mm) nominal, diameter shoulder. The socketable area of the pin, as defined, is
from the bottom side of the interposer to the pin tip which is nominally, .049" (1.25mm) in length.
This includes a .010" (0.254mm) high X .030" (0.965mm) diameter zone which accommodates the
shoulder and solder fillet area when inserted into the socket.
After the pins are fixtured, 95% Sn 5% Sb solder is used to reflow the pins into the interposer. The
solder height is maintained at .002" (0.050mm) maximum, above the land array of the pins to
ensure no interference with the solderpaste stencil during solder application prior to OLGA
assembly.
Soldermask Opening
.024”
.014” + .002/ -.001
dia. Finished Hole .002” max
.031” Land Diameter Solder Protrusion
.041”
Land Diameter
Soldermask
Opening .030”
.049” Insertable
Length
A1
OLGA Offset to Interposer
A2
J
E El K
H
G
F
φB M L N
Dl
Seating Plane
D
A7191-01
A1
K
E El
H
G
F
φB M L N
Dl
Seating Plane
D
A7192-01
13.3 Applications
Both the PPGA and FC-PGA package has been developed for Intel’s advanced microprocessor
family of products. They have been designed for use in socketed applications, using socket
footprints which are compatible with the Intel ceramic Pin Grid Array package family. While the
PPGA package can be used in both Zero and Low Insertion Force sockets (ZIF/LIF), the FC-PGA
should only be used in a ZIF socket application.
As preparation for die attach, wafers are mounted on a pressure sensitive carrier tape and diced
with a high speed saw. The cut wafer is washed with a detergent solution to remove silicon dust.
A silver filled epoxy adhesive is applied to the package substrate at die attach. Dice are picked
from the wafer and placed on the adhesive. The adhesive is then cured.
The die are connected to the gold plated package leads by way of gold wedge wire bond
technology. Bond pad and package lead placement accuracy and wire pull strength monitors ensure
high integrity connections.
PPGA packages are encapsulated with silica filled liquid epoxy in contrast to the CPGA lid seal
process. The encapsulant provides mechanical and environmental protection for the die and wires.
Process trays are moved beneath a valve which fills the package cavity with epoxy and are
transferred to an in-line oven in which the epoxy is cured.
A matrix code containing the assembly date code information is marked on the top side of the
package using a laser. The mark is inspected for orientation and readability. The packages are then
ready for the testing, finishing and packing processes.
Wafer Mount
Die Attach
Wire Bond
Test
Mark
Pack
A5772-01
In preparation for chip attach, C4 wafer reflow process is to modify the shape and surface
composition of the Pb/Sn bumps from the as-plated state to one acceptable for chip join. The
wafers are then mounted on a pressure sensitive carrier tape and diced with a high speed saw. The
cut wafer is washed with a detergent solution to remove silicon dust.
Die are picked from the wafer and mounted to the package. The units are then reflowed in a furnace
to form C4 solder joints of the die and package.
The FCPGA units are then pre-baked in the oven to remove the moisture from the organic package
before epoxy underfill materials is dispensed. The liquid capillary flow pull the underfill materials
to fill the gaps around the solder joints in between the die and the package. The epoxy underfill
materials are then cured in the oven.
The units are then ready for testing and finishing process. The finishing process includs laser
marking the human readable product/assembly/test information along with pin inspection, final
visual inspection and pack.
Wafer Reflow
A7677-01
PPGA packages may be used in either low insertion force (LIF) or zero insertion force (ZIF)
sockets. 296 lead sockets are available for PPGA packaged components from many suppliers. The
socket design is a standard footprint on the PCB. Insertion and extraction forces were measured on
sockets from various vendors. For all LIF socket designs tested, the maximum insertion force
required was 80 lbs. Avoid uneven loading during insertion. The applied load overcomes the
frictional resistance applied to the package pins as they are inserted into the socket. Table 13-12
summarizes the average measured insertion and extraction force for different sockets.
If the manufacturing flow requires inserting the package after the heat sink is applied to the
component, then the same suggested force may be applied uniformly across the top surface of the
heatsink.
Shipping trays are no longer color coded by particular product. Color standardization by process
type will result in the new low-temperature no-bake trays meeting the requirements of the drawing
regardless of form factor. Shipping trays will meet ESD safe shipping requirements.
The effect of foil sizes, clip force and voiding on Thermalcote I Conductacoat* thermal
performance was also explored. If there is no grease on the Al foil, then voiding will exist in the
bond line. This results in higher θja values. While up to 30% grease voiding can be tolerated
without any impact on PPGA thermal performance, zero voiding is strongly recommended for any
heat sink attachment methods.
A larger foil size will cover a larger area with grease, but a smaller foil size is preferred for
handling. For example, thermal resistance θja, measured with a standard sized foil (1.0” x 1.0”) is
0.1 C/W to 0.2 C/W lower than the θja measured with smaller foil (0.7” x 0.7”). Note that the Al
foil is carrier specific for Thermalcote I Conductacoat* grease to aid volume manufacturing. The
decision to use Al foils should be based on the OEMs particular heat sink attachment methods and
assembly line(s).
Clip force determines the thermal grease bond line thickness and directly impacts thermal
performance. To determine the effect of clip force on thermal performance, θja and θcs values for
the heat sink assemblies were measured at different clip forces. The actual clip forces of these
modified clips were individually measured by using a Material Testing System (MTS) before
thermal resistance measurement. All clip forces were also verified by using MTS after thermal
resistance measurement. Figure 13-13 shows θja and θcs values versus clip force. If the clip force is
higher than five pounds (corresponding to ~ 5 psi), then there is no significant effect of clip force
on PPGA thermal performance.
Figure 13-13. Comparison of θcs and θja for Different Clip Forces
0.8 3.9
3.8
0CS [C/W] 0.6 3.7
0ja [C/W]
3.6
0.4
3.5
0.2 3.4
3.3
0.0 3.2
0 2 4 6 8 10 0 2 4 6 8 10
Clip Force [lb] Clip Force [lb]
NOTES:
0ja = Junction to Ambient
0cs = Case to Sink Thermal Resistance (˚CW)
A5562-01
Heat Sink
Clip PPGA Package Heat Sink Lip
Socket Lever
ZIF Socket
Board
A5773-01
Note: Components on the top, and exposed metal on the side of the PPGA package can be shorted by any
electrically conductive material including heat sinks, thermal grease (if electrically conductive) and
thermal grease foil carriers.
CPGA PPGA
....................
T
....................
Large Die
....................
T
....................
Small Die
θja of PPGA is about 1.1° C/W lower than that of CPGA. Table 13-13 to Table 13-18 detail the
thermal resistance values for the components in ceramic pin grid array and plastic pin grid array
packages.
Table 13-13. θca [°C/W] for Different Heat Sink Heights and Air Flow Rates (CPGA)
θca [oC/W] vs Air Flow Rate [LFM]
Heat Sink Height 0 100 200 400 600 800
Table 13-14. θjc [°C/W] for a CPGA Package with and without a Heat Sink (CPGA)
No Heat Sink With Heat Sink
Table 13-15. θa [°C/W] for Different Heat Sink Heights and Air Flow Rates (CPGA)
θja [oC/W] vs Air Flow Rate [LFM]
Heat Sink Height 0 100 200 400 600 800
Table 13-16. θca [°C/W] for Different Heat Sink Heights and Air Flow Rates (PPGA)
θca [oC/W] vs Air Flow Rate [LFM]
Heat Sink Height 0 100 200 400 600 800
Table 13-17. θjc [°C/W] for a PPGA CPGA Package with and without a Heat Sink (PPGA)
No Heat Sink With Heat Sink
Table 13-18. θja [°C/W] for Different Heat Sink Heights and Air Flow Rates (PPGA)
θja [oC/W] vs Air Flow Rate [LFM]
Heat Sink Height 0 100 200 400 600 800
R0 Lp
dV/dt C0 Cp
A5775-01
13.8.3 EMI
The Federal Communications Commission (FCC) has set limits on the maximum radiation from
electrical systems. Each component in a system should not exceed the level that is allocated to it.
Electromagnetic Interference (EMI) levels from components in CPGA and PPGA packages have
been measured with and without heat sinks attached. Measurements were performed up to a core
clock frequency of 280 MHz. EMI levels are well below critical levels at all clock speeds tested.
There is no significant EMI performance difference between CPGA and PPGA.
14.1 Introduction
The plastic ball grid array (PBGA) has become one of the most popular packaging alternatives for
high I/O devices in the industry. Its advantages over other high leadcount (greater than ~208 leads)
packages are many. Having no leads to bend, the PBGA has greatly reduced coplanarity problems
and minimized handling issues. During reflow the solder balls are self-centering (up to 50% off the
pad), thus reducing placement problems during surface mount. Normally, because of the larger ball
pitch (typically 1.27 mm) of a BGA over a QFP or PQFP, the overall package and board assembly
yields can be better. From a performance perspective, the thermal and electrical characteristics can
be better than that of conventional QFPs or PQFPs. The PBGA has an improved design-to-produc-
tion cycle time and can also be used in few-chip-package (FCPs) and multi-chip modules (MCMs)
configurations. BGAs are available in a variety of types, ranging from plastic overmolded BGAs
called PBGAs, to flex tape BGAs (TBGAs), high thermal metal top BGAs with low profiles (HL-
PBGAs), and high thermal BGAs (H-PBGAs).
The H-PBGA family includes Intel’s latest packaging technology - the Flip Chip (FC)-style, H-PB-
GA. The FC-style, H-PBGA component uses a Controlled Collapse Chip Connect die packaged in
an Organic Land Grid Array (OLGA) substrate. In addition to the typical advantages of PBGA pack-
ages, the FC-style H-PBGA provides multiple, low-inductance connections from chip to package,
as well as, die size and cost benefits. By providing multiple, low-inductance connections the FC-
style, HPBGA offers equivalent or better performance than an extra on-chip metal layer. The FC
technology also provides die-size benefits through the elimination of the bond pad ring and better
power bussing and metal utilization. The OLGA substrate results in a smaller package, since there
is no cavity, and thermal management benefits since the thermal solution can directly contact the
die.
Lead Count 196 208 241 256 256 304 324 421 468 492 544
(15mm) (23mm) (23mm) (17mm) (27mm) (31mm) (27mm) (31mm) (35mm) (35mm) (35mm)
Sq/Rect. S S S S S S S S S S S
Pitch (mm) 1.0 1.27 1.27 1.0 1.27 1.27 1.27 1.27 1.27 1.27 1.27
Package 1.61 2.33 2.38 1.56 2.13 2.33 2.13 2.38 2.38 2.38 2.38
Thickness (mm)
Weight (gm) .67 1.56 .70 3.46 2.86 3.87 5.06
Max. Footprint 15.20 23.20 23.20 17.20 27.20 31.20 27.20 31.20 35.20 35.20 35.20
(mm)
Shipping Media:
Tape & Reel X X X X X X X X
Trays X X X X X X X X X X X
Desiccant Pack X X X X X X X X X X X
Comments/
Footnotes
clad bismaleimide triazine (BT) laminate. Four-metal layer substrate designs generally contain ad-
ditional power and/or ground planes to improve electrical and thermal performance. The die and
bonds are protected and encapsulated with molding compound. Via holes drilled through the sub-
strate provide routing from the lead fingers to the respective eutectic (63/37 Sn/Pb) solder balls on
the underside. Thermal performance can be enhanced by adding heatsink fastened through mechan-
ical means using thermal grease or by using conductive epoxy.
The H-PBGA and HL-PBGA, however, are configured differently to provide for greater thermal and
if required, electrical performance. The thermal advantage provided by this design is based first
upon attaching the die to the bottom surface of a heatspeader or slug that also forms the topside of
the package. Secondly, because the copper heatspreader forms the top of the package, the thermal
resistance is extremely low and exposes the package surface to available air flow. If required, this
heatslug can be directly coupled to active or passive thermal management devices such as heat sinks
or heat pipes. Improved electrical performance is achieved through additional power and/or ground
planes.
The FC-style, H-PBGA package consists of a die reflowed onto an oraganic substrate. The substrate
consists of four to ten layers of copper with insulating materials in between. The copper layers are
connected by vias. BT (Bismaleimide Triazine) resin reinforced with glass fiber forms the core of
the organic substrate. Solder bumps (3% Sn, 97% Pb) on the die surface are joined with solder pads
(60% Sn, 40% Pb) on the organic substrate in a reflow furnace. These joints form the electrical/
mechanical connection between the FC die and the OLGA package. An epoxy underfill fills the gap
between die and the substrate. This underfill provides mechanical support and protection for the die-
to-package interconnects and also minimizes thermal stress on the die due to CTE (coefficient of
thermal expansion) mismatch with the substrate materials. The die backside is exposed allowing the
thermal solutions and thermal interface material to have direct contact with the die surface.
See Figure 14-1, Figure 14-2, and Figure 14-3 for description of PBGA, HL-PBGA, and FC-style
H-PBGA packages.
BT PCB
Die Up Design Mold Compound
Non-laminate
Solder Balls
A5764-01
A5765-01
Solder balls
A7428-01
Table 14-4. Symbol List for Plastic Ball Grid Array Family
Letter or Symbol Description of Dimensions
A Overall Height
A1 Stand Off
A2 Encapsulant Height
A3 Die Height with FC Bumps and Underfill
b Ball Diameter
c Substrate Thickness
D Package Body Length
D1 Encapsulant Length
E Package Body Width
E1 Encapsulant Width
F1 Die Width
F1 Die Length
e Ball Pitch
N Ball Count i.e. Lead Count
S1 Outer Ball Center to Short Edge of Body
S2 Outer Ball Center to Long Edge of Body
NOTE:
1. Controlling Dimensions: Millimeter
Pin #1 Pin #1
Corner Corner
PBGA 196 PBGA 208
Pin #1 Pin #1
Corner Corner
PBGA 241 PBGA 256 (17mm)
Pin #1 Pin #1
Corner Corner
PBGA 256 (27mm) PBGA 304
A5487-03
Pin #1
Corner Pin #1
PBGA 324 Corner
PBGA 421
Pin #1 Pin #1
PBGA 468 Corner PBGA 492 Corner
Pin #1
Corner
PBGA 544 (35mm)
A6124-02
15.00 ±0.20
Pin #1
13.00 ±0.25 Corner
Pin #1 Corner 14 12 10 8 6 4 2
10.57 Ref 0.60 13 11 9 7 5 3 1
0.40
A
B
C
13.00 D
±0.25 E
F
10.57 G
15.00 H
Ref ±0.20 J
1.00
K
L
M
N
1.00 Ref P
Notes:
1. All Dimensions are in Millimeters
A5829-01
D1
Pin #1 Corner Pin #1
b Corner
E1
A A2 ˚
30
Note:
1. All Dimensions are in Millimeters
C A1 Side View
Seating Plane
A5766-01
Min Max Min Max Min Max Min Max Min Max Min Max
Min Max Min Max Min Max Min Max Min Max
b 0.60 0.90 0.60 0.90 0.60 0.90 0.60 0.90 0.60 0.90
c .32/.55 .40/.67 .52/.55 .60/.67 .52/.55 .60/.67 .52/.55 .60/.67 .55 .67
(2L/4L)
Pin #1 Pin #1
HL-PBGA 304 Corner HL-PBGA 352 Corner
Pin #1 Pin #1
HL-PBGA 432 Corner H-PBGA 540 Corner
A5832-02
Pin #1 I.D. S1 e
1.0 Dia.
Top View Bottom View
A C
A1 Note:
Side View
Seating Plane 1. All Dimensions are in Millimeters
A5830-01
D1
b S1
E1
Top View
e
Pin #1 Pin #1
A2 Slug Corner Corner
Bottom View
C Note:
A1 Side View 1. All Dimensions are in Millimeters
Seating Plane
A5831-01
A 3.59 4.10
A1 0.40 0.70
A2 0.95 1.10
b 0.60 0.90
c 2.00 2.30
D 42.30 42.70
D1 - 27.70
E 42.30 42.70
E1 - 27.70
e 1.27
S1 1.56 REF
NOTE: Measurement in millimeters
FC-style, FC-style,
H-PBGA 495 Pin #1 H-PBGA 615 Pin #1
Corner Corner
A7458-01
E
F1 S2 e φb
AD
AC
AB
Substrate AA S1
Keepout W
Y
Outline V
U
T
R
D P
N
F2 M
L
K
J
H
Label G e
F
Mark E
D
C
B
A
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21
Die Pin #1
Top View Corner
Bottom View
C A3
A A1
Seating Plane
Side View
Notes:
1. All Dimensions are in Millimeters
A7459-01
E
F1 S2 e e φb
Substrate AF
AE
Keepout AD
Outline AC
AB
S1
AA
Z
W
V
U
T
R
P
D F2 N e
M
L
K
J
Label H
G
Mark F
E
D
C
B
A
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24
Pin #1
Die Top View Corner Bottom View
C A3
A A1
Seating Plane
Side View
Notes:
1. All Dimensions are in Millimeters
A7460-01
In general, the ball grid array features shorter electrical path lengths which reduce inductance. Me-
chanical problems such as fragile leads are absent. The larger spacing between solder lands provide
adequate tolerances for more reliable surface mounting. Some heat dissipation can be facilitated
through the substrate. These characteristics make the ball grid array package suitable for a wide va-
riety of devices: microprocessors/microcontrollers, ASICs, memory, PC chip sets, and other prod-
ucts. A thin profile and smaller footprint make the BGA an attractive option when board space is a
major concern. Small body size BGA packages come close to chip scale package size for use in
space constrained applications.
The FC-Style, H-PBGA also allows for Voltage Indentification through an open circuit or a short to
Vss on the processer substrate. These opens or shorts are achieved by selectively depopulating some
of the balls. Voltage Identification can be used to support automatic selection of power supply volt-
ages.
exposure. Maintaining proper control of moisture uptake in components is critical to the prevention
of "popcorning" of the package body or encapsulation material. BGA components, before shipping,
are baked dry and enclosed in a sealed desiccant bag with a desiccant pouch and a humidity indicator
card. Most BGA components are classified as a level 3 or level 4 for moisture sensitivity as per the
IPC/JEDEC Spec J-STD-020, “Moisture/Reflow Sensitivity Calculation of Plastic Surface Mount
Devices.”
With most surface mount components, if the units are allowed to absorb moisture beyond their out
of bag times for their moisture rating, damage may occur during the reflow process. Chapter 8 of
this data book provides an in-depth view of package preconditioning methods and moisture sensi-
tivity requirements. Please refer to Chapter 8 for more information regarding how moisture sensitive
components are classified.
Prior to opening the shipping bag and attempting solder reflow, the moisture sensitivity of the pack-
ages being used should be understood so proper precautions can be taken to insure that a minimal
out of bag time is maintained. This will insure that the highest possible package reliability is
achieved for the final product. If previously bagged product cannot be mounted before the elapsed
out of bag time for that product, the parts can be rebaked as per Chapter 8. Another option is to store
the opened units in a nitrogen cabinet or dry box until needed. Placing units in a dry box effectively
‘stops the clock.’
It should be understood that packages continue to gain moisture even after board mounting. Com-
ponents that need to be reworked must be completely processed throught all thermal exposures be-
fore the original out of bag limits are reached. If this is not possible, or the time allotment is not
ridgely followed, bake-out of the completed boards must be accompliished before subjecting the
components to the heat of the rework process. Products being removed from boards that have been
returned from the field for failure analysis, must be baked dry before heat expousure. If this step is
skipped, massive damage to the component will result, rendering useless any further efforts at de-
termining the cause of failure.
Solder Mask
Copper Pad Area
A5826-01
Solder Mask
Defined
Copper Pad Area
A5833-01
Using the routing scheme shown in Figure 14-16, the first two ball rows are routed on the top signal
layer and the inner two rows are routed on the bottom side of the package substrate. In this case vias
are required between the BGA pads. Using the routing scheme shown in Figure 14-10, the first three
ball rows are routed on the top signal layer and the inner row is routed on the boards bottom side. In
this case the vias are not required in between the BGA pads. Vias are discussed in Section 14.8.3.3.
50 mil pitch
50 mil pitch
Figure 14-14 shows the connection between the BGA ball pad and a via. This connection is often
referred to as the dogbone footprint.
Solder Mask
Must Cover The Via
10 Mil Trace
A5824-01
of the connection. Soderability problems on BGAs are often traced to oxidation of the pads on the
boards.
14.9.1 Fluxing
Most BGA assembly is done with a solder paste that contains flux, however, there are some compa-
nies reporting adequate results from mounting BGAs when using low residue, no clean or an aque-
ous clean flux. To obtain high yields and reliable joints this may require monitoring the surface
insulation resistance of the board to insure there are no foreign contaminates on the board or slder
ball surface that will show up as reliability problems later on.
Fluxing without paste does come with disadvantages; the ball reflowed with only flux will have a
smaller solder volume than the initially attached ball. Therefore, the package will result in a slightly
(usually 0.001-0.002mils) lower standoff height from the PCB. The thermal cycle reliability of BGA
component could be slightly compromised because of the reduced standoff height, especially if sol-
der balls are under the die area as in full array BGA types. The self centering ability decreases be-
cause of the smaller solder volume. The formation of the fillet between the ball and the board can
be hindered if the ball is mis-shapen or there is some minor imperfection with the pad area.
Since the BGA balls consist of solder, the flux activity on the ball surface is assured, so long as the
paste reaches the ball surface. The selection of paste is generally made to fit the entire component
mix being assembled, not driven by the use of BGA packages. It is necessary, however, to ensure
that all the thermal and environmental requirements of the paste can be met.
On the average, the BGA packages do not need any specialized solder paste. However, most solder
suppliers have developed a BGA paste that has minimal voiding during reflow. This paste may also
be used for non BGA components as well. It is advisable that a time limit of 45 minutes or less be
maintained from screen pasting to reflow (30 minutes is optimum) to avoid the paste drying out and
affecting solderablity or contribuing to voids in the solder balls. Some manufacturers ratings (of 2
to 6 hours) for air exposure is judged by using 500 gram jars, not small dots of solder sitting on a
board which dry out considerably faster.
Excessive volume of paste will have some negative effects on the self centering properties of the
BGA (see Section 14.9.4.1) and could cause yield or reliability issues (shorts or bridges). However,
adequate paste thickness is required to compensate for board warp, poor component coplanarity, and
to achieve acceptable board reliability. Another factor in yield or reliability is the presence of ade-
quate flux. Any situation, such as a plugged stencil hole, that causes flux or paste to be omitted or
severely limited on a pad can lead to an open connection after reflow. Print misregistration or ex-
cessive slumping that connects adjacent conductors (either pads or PTHs) is also a source of con-
cern, as a short may be the final result. Therefore, it, is usually beneficial to perform a paste
inspection step, especially during early manufacturing runs. Whether to use visual/microscope in-
spection, manual measurement, or invest in an automated system depends on the volume to be run
and the overall manufacturing philosophy.
Maintaining a diameter to stencil thickness ratio of at least 3 to 1 can provide good BGA print char-
acteristics, with larger openings providing better print quality. It is also beneficial to use an opening
at least as large as the mounting pad to give a wide placement window. A typical design might be a
0.028 opening in a 0.006 thick stencil, going over a 0.024 pad on the PCB with 30 mil solder balls.
The printing of small amounts of paste onto the solder mask surrounding the pad has not proven to
be a problem in either yield or reliability.
Squeegee
Direction of
squeegee travel Stencil side of Screen
14.9.4.1 Placement
BGA packages have shown excellent self-centering properties. Because of this, wide variation in
placement location is accommodated during reflow of the solder joints. The general rule for BGA
packages is that the placement be at least 50% on pad. This is illustrated graphically in Figure 14-16.
The self centering characteristics of the BGA are attributed to surface tension which will pull the
component onto the pad during peak reflow temperatures.
Mounting Pad
Solder Ball
A5825-01
Most common placement tools used for flatpack or other non-discrete packages have much better
accuracy than necessary, and placement variability is rarely an issue for BGAs.
14.9.4.2 Alignment
The pick and place accuracy governs the package placement and rotational (theta) alignment. This
is equipment/process dependent. Slightly misaligned parts (less than 50% off the pad) typically au-
tomatically self-align during reflow. Self centering on the pads is greatly reduced for grossly mis-
aligned packages (greater than 50% off the pad) and may develop electrical shorts, as a result of
solder bridges, if they are subjected to reflow.
Because the top surface of some of the BGA packages (like the HL-PBGA types) are highly reflec-
tive, some top side vision systems do better with their inspection process if they use a diffuse light-
ing source instead of polarized source. A round fluorescent tube type light near the package or a light
filter over the polarized fixture seems to enhance the operation of some systems.
Given that the allowable misalignment of the balls is 50% or less of the pad width (e.g.; 12mils for
a 24mil pad), a manufacturable process can be achieved using only the package edge (outline) and
alignment marks on the stencil for machine placement, see Figure 14-16.
A5827-01
When doing second pass wave solder of mixed technology components on the same board as BGAs,
insure that the solder wave profile is very tightly controlled. A high temperature on the wave solder
process can warp the board and break the joints on the topside of the BGA component.
Rework is the process of removing a component from a PCB, and replacing it with a new compo-
nent. The removed component is not immediately reusable. The shape and volume of the solder balls
will not be the same as a new package. If component reuse is desired, a separate process for replace-
ment of solder balls should be used.
To perform BGA rework, there are four basic steps: removal, site preparation, flux or solder paste
application, and replacement/reflow. The following subsections describe each.
The component is removed from the PCB by a vacuum nozzle within or integral to the hot gas head.
When profiling, shut the vacuum off so the component is not removed to avoid damage to the profile
card. Control the pressure of the head down onto the component during removal. If pressure is ap-
plied after the solder balls are melted, the solder is pressed between the “plates” of substrate and
PCB, resulting in bridging that must be removed manually. Two possible solutions are:
• Establish the head height prior to reflow and control the stroke.
• Place shims under the edges of the component to prevent collapse
Consider the effects on other components. If other SMT packages are near the package being re-
worked, monitor their solder joint temperature. If the temperature approaches reflow temperatures,
shielding may be necessary.
Preheating the PCB assembly is good practice when reworking BGA packages. Advantages are:
• It can reduce heating time required using the rework head. In cases where one PCB assembly
can be preheated while another is reworked, cycle time can be greatly reduced.
• More uniform profiles are achieved. One challenge of profiling is the temperature spread
between center and edge solder joints. Preheating reduces the spread by bringing the baseline
temperature closer to the reflow temperature.
• PCB warpage is minimized. Warpage is caused by the higher local temperature at the site than
the surrounding area. By raising the PCB assembly temperature, the mismatch is minimized
and warpage is reduced.
• It reduces problems with adjacent components, as it allows shorter gas heat times or lower gas
temperatures to be used, thus reducing risk of affecting neighboring solder joints.
d. Use of stenciled solder paste is optional (and as PBGA packages get larger, may be
required). For applications where a solder stencil is not possible or not desired, acceptable
results may be obtained by only applying flux to the pre-tinned pads. This is done by
using a non-metallic spreader and applying a no-clean flux paste to the pads on the board.
Be careful not to scratch the pads or the board.
e. Apply liquid flux to the solder balls of the new package. Once the liquid flux is applied,
within two minutes, place the package on the board and then reflow (be sure to use the
board’s alignment features fudicials and then place component using either a mechanical
or manual means).
f. Reflow solder balls using hot air directed at the edge and under the package body. It is
recommended that temperature experiments be conducted to determine optimum
conditions.
g. Remove the board from the heat source and allow to cool to a safe handling temperature.
h. Inspect the board and package to verify proper solder ball collapse and observe any
defects that may have been caused by the rework procedure.
In general design-related factors have greater thermal effect on the PBGA than material-related vari-
ables. A large die spreads heat easier, as does a larger package size and thus a higher ball count. Sub-
strate design features have tremendous effect on the package’s ability to dissipate heat. Vertical vias
running through the substrate help to transfer heat from the die to the solder balls. Four-layer designs
often incorporate conductive 2-ounce copper ground planes, which have a significant, positive effect
on thermal performance. The Enhanced PBGA has thermal balls under the die while H-PBGA and
the HL-PBGA utilize a heat spreader or slug across the top of the package to dissipate heat even
more efficiently. PBGAs with center thermal balls dissipate considerable heat into the board. A con-
siderable increase in thermal effectiveness of a BGA package can be obtained by using boards that
are thermally efficient, increasing the airflow, or providing thermal paths from the board. Remem-
ber, with PBGAs, the board is your primary heatsink.
Environmental conditions play a critical role in the thermal performance of PBGAs. Ambient con-
ditions, junction and case temperatures, the device’s placement and orientation on a board, in con-
junction with the volume and temperature of air flowing past the unit present a broad range of
possible thermal solutions and problems for IC packaging. Typically a package cannot be capable
of handling a given power requirement unless the environmental conditions allow heat to dissipate.
When environmental or geometric constraints limit a BGA’s ability to dissipate heat, a copper or
aluminum heatsink is often used to provide an additional method for heat transfer. As with other
types of packages, the heatsinks for BGAs vary in design and methods of attachment. Most appli-
cations recommend a maximum case temperature for the package. Various factors effect the case
temperature including ambient conditions and airflow. If the case temperature exceeds the recom-
mended rating, a heatsink may be required. Contact an Intel applications engineer for the product to
determine if a heatsink has been developed for the particular package or application.
15.1 Introduction
Since the introduction of Chip Scale Packages (CSP’s) only a few short years ago, they have
become one of the biggest packaging trends in recent history. There are currently over 50 different
types of CSP’s available throughout the industry and the numbers are increasing almost daily.
Intel Flash memory products began using CSP's in the µBGA* package a few years ago and have
expanded into multiple types of CSP's in order to meet the needs of new product functionality and
applications. Currently, the majority of Intel's CSP's are used for flash memory products. However,
other types of Intel products are beginning to take advantage of the benefits of CSP's as well.
CSP's are evolving so rapidly, that by the time you read this chapter, there will probably be new
package information and design considerations to take into account. Intel has attempted to include
as much as possible in this chapter, reviewing many different areas such as package information,
application considerations, printed circuit board (PCB) design and manufacturing tips and tools.
However, since CSP's are continually evolving, the contents of this chapter will continue to evolve.
Therefore, until new versions of this package guide are printed, new CSP information and
manufacturing considerations for Intel Flash Memory products will continue to be updated in the
Flash Memory CSP User's Guide on the WWW at:
http://developer.intel.com/design/flash/packtech/index.htm.
There are many reasons why CSPs have been so well accepted within the industry. One of the
biggest advantages of CSPs is the size reduction of the package (see figure 15.1) vs. more
traditional peripherally leaded packages. This is mainly due to the Ball Grid Array (BGA) design
of the package. By designing all interconnects under the package in the BGA style, you can
increase the number of interconnects while saving PCB routing space. Other manufacturing
advantages of CSPs include the self alignment characteristics during PCB assembly reflow and
lack of bent leads which cause coplainarity issues. Both of these CSP features increase PCB
assembly yields and lower manufacturing costs.
One of the barriers for new packages to be accepted in the industry is the lack of existing Surface
Mount Technology (SMT) infrastructure such as assembly and manufacturing processes and
equipment. This is not the case for CSPs which take advantage of existing infrastructure and in
most cases require no capital equipment investment to implement CSPs .
In the past, CSP's have been defined as a package that is 1.2X the size of the die. However, some
types of CSPs maintain their package size as the internal silicon die reduces in size as a result of the
fabrication lithography process gets smaller (die shrink). This effect changes the package to die
size ratio. As CSP's have evolved, the definition has changed to "near die size packages with a ball
pitch of 1mm or less".
As mentioned earlier, Intel has introduced several different types of CSP packages. This is because
each application has different requirements. Since almost every application varies, there are many
considerations to take into account when selecting the best package for the application. Please refer
to the "Package Usage" section of this chapter to review this in more detail.
A7583-01
http://www.intel.com/design/flash/packtech/index.htm
This section reviews CSP specific information such as various CSP construction, material sets,
attributes, and dimensional examples. It also explains the use and construction of various
mechanical samples referred to as Silicon Daisy Chain (SDC) samples to be used for mechanical
/process equipment set-up and evaluation.
The µBGA package (Figure 15.2) is a .75mm and .5mm ball pitch package and takes full
advantage of any reduction of silicon die size. This makes the µBGA package the smallest discrete
Intel flash memory package. Its unique construction utilizes a layer of elastomer which decouples
the stresses caused by the coefficient of thermal expansion (CTE) of the silicon die and the PCB
material during temperature variations.
Silicon Die
1.0mm
Elastomer
A7584-01
Since the size of the package equals the size of the die, as the die gets smaller due to fabrication
lithography process reductions (die shrinks), so does the package. At a certain point, the associated
ball pitch will get smaller as well, in order to accommodate the smaller size of the die. This
eventually leads to ball pitches as small as .5mm and below. Currently the majority of µbga
packages are in .75mm pitch.
S1 Ball A1
D Corner
8 7 6 5 4 3 2 1 S2
Ball A1 1 2 3 4 5 6 7 8
Indicator A A
B B
C C
E
D D
e
E E
F F
A1
A2 A
Seating
Plane
Y
Side View
A4805-01
Note: The µBGA package is die-size dependent and may vary. Actual products vary with different levels
of matrix ball depopulation. Refer to the µBGA* Package Mechanical and Shipping Media
Specifications for specific product/package dimensions/drawings and pinouts at:
http://www.intel.com/design/flash/packtech/index.htm
Ball (Lead) Width (all .75mm pitch) b 0.300 0.350 0.400 0.0118 0.0138 0.0157
Ball (Lead) Width (all .50mm pitch) b 0.259 0.309 0.359 0.0102 0.0122 0.0141
Pitch [e]
See µBGA Package Attribute Table
ball (Lead) Count N
GT28F016/160B3
.75 R 51 6x8 46 7.286 6.964 1.018 1.607 Y
GT28F160C3
Shipping Media All µBGA products are available in Tape & Reeel or Trays
Silicon Die
.8mm Pitch
A7585-01
A1
Index S2 A1
Mark e
S1
A
B
C
D
E
E
F
G
H b
1 2 3 4 5 6 7 8 9 10 11 12
D
A A2
A1
A7587-01
Note: Refer to the IntelÆ StackedCSP Package Mechanical and Shipping Media Specifications for
specific product/package dimensions/drawings and pinouts at:
http://www.intel.com/design/flash/packtech/index.htm
Table 15-3. Generic Intel® StackedCSP Dimensions
Millimeters Inches
Symbol
Min Nom Max Min Nom Max
Lead Count N 72 72
Shipping media All µBGA products are available in Tape & Reeel or Trays
Another advantage of the Easy BGA package is its constant package size/footprint in respect to
memory density upgrades and die shrinks. A key element of embedded applications is the need for
long product life cycles (5-7 years) that require the same package size/footprint. Not only does the
package size/footprint need to stay constant over time; it does not change as a result of memory
density upgrades or die process shrinks. This attribute is very beneficial because many embedded
applications increase in memory density over time in order to incorporate additional functionality.
Silicon Die
A7586-01
Many embedded applications require a high level of reliability due to the usage conditions of their
environments. The Easy BGA package has been constructed specifically to address these types of
requirements. The Easy BGA package construction incorporates many key features that
differentiate it from other CSP packages. Besides the wider ball pitch as previously discussed, the
Easy BGA uses large diameter eutectic solder balls and a thick BT laminate rigid substrate. The
combination of large solder balls and thick BT laminate substrate provide very good reliability by
buffering and maximizing the separation between the silicon die and PCB surface to minimize the
affects of CTE induced stresses.
A1
A2 A
Seating
Plane
Y
A7588-01
Corner to First Ball Alnog D S1 1.400 1.500 1.600 0.0551 0.0591 0.0630
Corner to First Ball Along E S2 2.900 3.000 3.100 0.1142 0.1181 0.1220
Desiccant Pack1 Refer to moisture barrier bag label for specific IPC level
1. Desiccant Pack levels relate to IPC Moisture Sensitivity Levels. Refer to the handling section of this guide
for the complete moisture level table.
2. SDC’s represent the mechanical samples available in the various package size/type equivalents
NOTE: All Dimensions in mm
Figure 15-8. Molded Matrix Array Package (MMAP) 144/225/256 ball count
Top View S1 e
Bottom View
A A1
A2
C
Seating Plane
Side View
Corner to First Ball Alnog D S1 0.90 1.00 1.10 1 0.0354 0.0394 0.0433
Corner to First Ball Along E S2 0.90 1.00 1.10 1 0.0354 0.0394 0.0433
NOTES:
1. The tolerances above indicate projected production accuracy. This product is in the design phase. The
minimal, nominal, and maximum package body width and length are subject to change dependent on final
die size. Actual die size could shift these values by ± 0.008 inches for the 28F320J5 product.
2. Some ball locations may not be populated on some products. See the specific product signal list for
complete details.
Corner to First Ball Alnog D S1 0.90 1.00 1.10 1 0.0354 0.0394 0.0433
Corner to First Ball Along E S2 0.90 1.00 1.10 1 0.0354 0.0394 0.0433
NOTES:
1. The tolerances above indicate projected production accuracy. This product is in the design phase. The
minimal, nominal, and maximum package body width and length are subject to change dependent on final
die size. Actual die size could shift these values by ± 0.008 inches for the 28F320J5 product.
2. Some ball locations may not be populated on some products. See the specific product signal list for
complete details.
Sq/rect R R R
If your application requires the smallest possible package, the µBGA* package and the Intel Æ
Stacked-CSP are the best package choice for your design. For the smallest size and highest
reliability, the µBGA package remains the best single-die CSP for your design. For applications
that use flash and SRAM, the Intel Æ Stacked-CSP adds value by integrating Flash and SRAM and
stacking both individual die into one package. This unique packaging approach provides the
ultimate in size reduction by eliminating a component from the board. These CSP's were designed
to meet the demands of handheld applications such as cellular phones, pagers, personal digital
assistants (PDA) and Global Positioning Systems (GPS) units.
For additional detailed information about Intel® Flash memory CSPs, refer to the Intel Flash
Memory CSP Users Guide located on the web at:
http://developer.intel.com/design/flash/packtech/index.htm
You will find extensive information on IntelÆFlash memory CSPs covering a wide variety in
topics such as:
• CSP assembly flow diagrams
• CSP package attributes
• CSP Materials, and packaging dimensions
• CSP shipping media and handling
• Printed Circuit Board (PCB) design considerations (trace/space, via, etc.)
• CSP to PCB assembly and manufacturing process recommendations
• CSP manufacturing support tools
• Intel’s quality criteria
• CSP solutions for diverse applications
Table 15-12. Solder Stencil Design for µBGA, Easy BGA, and Intel ® Stacked-CSP
.75mm mBGA*
Feature .5mm mBGA* CSP .8mm Stacked CSP 1.0mm Easy BGA
CSP
Top of Stencil
.279 (.011) .33 (.013) .33 (.013) .33 (.013)
Aperature
Bottom of Stencil
.30 (.012) .356 (.014) .356 (.014) .356 (.014)
Apperature
Stencil Thickness .127 (.005) .127 (.005) .127 (.005) .127 (.005)
NOTE: All Dimensions in mm (inches)
For additional detailed information about Intel ® Flash memory CSPs, refer to the Intel Flash
Memory CSP Users Guide located on the web at:
http://developer.intel.com/design/flash/packtech/index.htm
16.1 Introduction
Intel has introduced a variety of innovative package designs over the years: surface mount, small
out-line, very thin package, multilayer molded plastic quad flatpacks (PQFP), and the Tape Carrier
Package (TCP) format. Recent configurations for optimum microprocessor performance are the
Single Edge Contact (S.E.C.) cartridge and the Mobile Mini-Cartridge.
The S.E.C. cartridge typically combines an area array packaged processor core, L2 cache, other
components, and a thermal solution attach point. The entry of the S.E.C. cartridge technology
continues the commitment to provide packaging solutions which meet Intel’s rigorous criteria for
quality and performance.
The mobile mini-cartridge is a packaging technology aimed at the notebook PC market, where
minimum form factor is required. It is designed to protect the electronic components of the product
and enables a controlled thermal interface between the processor and the thermal solution in the
notebook system (see Figure 1). The mini-cartridge provides a socketable microprocessor
solution, i.e. it enables the OEM to mount the processor via a motherboard-mounted connector,
rather than by soldering it, and mechanically securing it with screws in a manner similar to other
notebook components such as hard disk drives.
Several times throughout this chapter you will be referred to Intel’s website. The URL for the Intel
website is http://www.intel.com. Intel’s developers website can also be accessed through this site.
There are several variations to the S.E.C. Cartridge form factor. They are the Single Edge Contact
Cartridge (S.E.C.C.) which has a cover and a thermal plate, the Single Edge Contact Cartridge 2
(S.E.C.C.2) which has a cover, but no thermal plate, and the Single Edge Processor Package
(S.E.P.P.) which has no cover or thermal plate. In implementations with no thermal plate, the
customer can attach a heatsink directly to the MP Package or die.
16.2.1 Terminology
The following terms are used in this document and are explained here for clarification.
S.E.C. cartridge — The processor packaging technology is called a "Single Edge Contact
cartridge."
S.E.P.P. — The processor packaging technology known as “Single Edge Processor Package.”
Processor substrate — The structure on which the components are mounted inside the S.E.C.
cartridge (with or without components attached).
Thermal plate — The surface used to connect a heatsink or other thermal solutions to the S.E.C.C
processor.
Cover — The processor casing on the opposite side of the processor core.
Front View
Back View
A6153-01
Front View
Back View
A6151-01
Figure 16-3. S.E.C. Cartridge (330 contacts)— Thermal Plate and Cover Side Views
A6137-01
A7487-01
Additional terms referred to in this and other related documentation are the Mechanical Support
Pieces (MSPs), which are used on the system to connect the processor to the system baseboard, and
are responsible for retention of the processor during system shock and vibration. The MSPs
represent one solution for retention of the processor in the SC 242 connector. This chapter focuses
on the use of these pieces:
SC 242 & SC 330 Contact Connector — The connectors that the S.E.C. cartridges uses.
Retention Mechanism — A mechanical piece which holds the cartridge in the SC 242 or SC 330
connector.
Retention Mechanism Attach Mount — An enabled mechanical piece which secures the
retention mechanism to the baseboard.
Dual Retention Mechanism — A mechanical piece which holds two S.E.C. cartridges in two SC
242 contact connectors for a 2-way SMP processor system.
Processor Substrate
with Core and Second
Level Cache
Cover
A6172-01
Plastic Enclosure
Thermal Plate
Retention Clips
Pin Fasteners
A6149-01
Figure 16-7 and Figure 16-8 shows an exploded view of the mobile mini-cartridge assembly. The
PCB substrate is populated on both sides with surface-mount passive and active components
consisting of the processor core and tag RAM memory on the primary side, and two L2 cache
memory devices and the connector on the secondary side. After the substrate is populated, it is
placed into the top cover and then the bottom cover is snapped into place and retained by nine snap
lances around its perimeter.
The connector plug is designed to mate with either of the two available receptacle versions. The
receptacle versions enable connector stack heights of approximately 3.4 mm and 4.2 mm. Final
connector stack height is dependent on the characteristics of the surface-mount process used to
place the receptacle onto the motherboard.
Mini-cartridge
Top Cover
Processor Core
Internal
PCB Substrate
Memory device
Upper View
A7408-01
Memory Devices
Temperature
Sensor
Product Label
BGA-Technology
Connector
A7409-01
BLT
H
SH PCB
P
S M
R
BC
A7406-01
Where Where
Required Required
Pack
A5802-01
Standard SMT processes are used to place components on the mini-cartridge substrate. The cover
is a two piece snap together assembly. A simplified process flow is shown in Figure 16-11.
A7407-01
This section will provide additional handling guidelines and information on the shipping media
used for the processors. The “datasheet cross reference” contains specific operational and storage
specifications for the processor.
The Static-Dissipative insert (see Figure 16-12) is recyclable. The inserts may be shipped to:
Richmond Technology
1897 E. Colton Avenue
Redland, CA. 92373
(909)794-2111
Static-Dissipative
Insert
Shipping Box
A5803-01
A5804-01
Clamshell
Load Unit Connector
End Down
A7410-01
Warning: Only handle the mini cartridge by holding at the edge of the unit and with clean personal protective
equipment (PPE) eg: vinyl ESD dissipative gloves. Do not touch the connector, die and flat metal
surface area.
Handle HERE
A7411-01
Intel has enabled a Mechanical Solution (MS) to support the S.E.C. cartridge and to ensure
retention of the processor into the SC 242 and SC 330 contact connectors. This solution is referred
to as the Mechanical Support Pieces (MSPs). These pieces represent only one solution to ensure
that the processor stays in the connector. Other methods and mechanical solutions will NOT be
covered by this document. This document only provides details on the use and operation of the
MSPs. While each OEM must perform actual validation for mechanical performance ensuring that
the processor stays in the connector, Intel has shown that, through shock and vibration test
validation, these pieces provide adequate mechanical support of the processor when used correctly.
The mechanical support pieces consist of the connector, retention mechanism, retention
mechanism attach mount (RMAM) and heatsink support (HSS). The connector provides the
electrical path between the processor and the other logic components on the baseboard. The
retention mechanism holds the processor into the connector during mechanical shock and
vibration. The RMAM attaches the retention mechanism to the baseboard. Figure 16-17 provides
illustrations of all MSPs and how they interact with the processor.
Figure 16-16. S.E.C Cartridge with All Mechanical Support Pieces, Full Assembly
2.038
± 0.010
1.00 ± 0.01
1.095 ± 0.015
A5735-01
Figure 16-17. Exploded View of S.E.C. Cartridge with All Mechanical Support Pieces
Retention Mechanism
Slot 1 Connector
Retention Mechanism
Attach Mount
A5806-03
Intel does not supply the MSPs or equipment detailed in this chapter. Information for suppliers is
provided in this document as applicable. An updated supplier’s guide can be located at the Intel
website by searching for the terms “suppliers of support components.”
B1 B121
73 Contact Pairs 48 Contact Pairs
Side View
001008 A5808-01
Pin B1
Top View
Pin B2 Pin B165
Side View
A6150-01
A5728-01
The mechanical force of the connector contacts provide some friction grabbing of the processor.
The retention mechanism is designed to allow a small amount of back-out of the connector before
stopping the travel of the processor. The vertical travel associated with this back-out is
comprehended in specifications for the processor (particularly the substrate), connector and the
retention mechanism.
Attach Heatsink
to Processor
A5810-02
Figure 16-22. Criteria to Ensure Correct Interaction of S.E.C. Cartridge and Mechanical
Support Pieces
0.015 inches
max tilt
Short Axis
Unacceptable - No lead protrusion
0.015 inches
max tilt
Connector
Long Axis
A5811-01
http://developer.intel.com/design/litcentr/index.htm