  |
|
|||||
|
||||||
    
|
|
|||||
ISHM '87 Proceedings
USE OF ARGON PLASMA FOR CLEANING HYBRID CIRCUITS PRIOR TO WIRE BONDING
Susan L. Buckles
March Instruments, Inc.
ABSTRACT
Processing hybrid circuits with argon plasma is shown to remove organic contaminants
from wire bonding areas. Data from three independent laboratories is presented
which shows a significant reduction in wire bond failures when the hybrid
circuits are cleaned with argon plasma prior to wire bonding. The reduction
in the number of wire bond failures ranged from 51% to 70%. Data is also presented
which shows an increase in the bond pull strength of 25%. It is concluded
from these data that plasma cleaning results in a significant improvement
in yield. Argon plasma cleaning is also compared with oxygen plasma cleaning.
INTRODUCTION
Hybrid circuit failures frequently can be traced to failures in wire bonding
caused by organic contamination of the bonding surface. This contamination
can be due to solvent residue on the device, residual photoresist on bonding
pads, epoxy bleedout, or other organic contaminants.
Plasma cleaning has been described as a technique for removing organic contaminates, thus improving bondability.1,2,3 In addition to cleaning prior to bonding, plasma cleaning can be used to clean bare alumina substrates before metallization, to clean before the die is attached to the substrate, and to clean before final sealing. Additionally, plasma cleaning will aesthetically improve the substrate.
MECHANISMS OF PLASMA CLEANING
Both argon and oxygen plasma have been used to clean hybrid devices prior
to bonding. Oxygen plasma uses a chemical process in which oxygen radicals
are formed and organic contaminates are chemically oxidized.
O2 + e ---›
O• + O• + e organics›
CO2+H2O+e
Argon plasma, on the other hand, is a physical process. Argon is ionized and
the ionized gas mechanically dislodges the organic contaminants.
Ar + e ---› Ar+ + 2e
COMPARISON OF ARGON AND OXYGEN CLEANING
Table 1 compares oxygen plasma with argon plasma for the purpose of cleaning
hybrids. while argon plasma is somewhat slower, it has the advantages of running
at lower temperatures and will not oxidize exposed metal components or silver
filled epoxy. For the purpose of cleaning hybrids, the less damaging argon
process is preferred.
EFFECT OF PLASMA CLEANING WITH
ARGON
Figures 1a and 1b are SEM photographs of a transistor epoxy bonded to a gold
surface. Figure 1a shows the epoxy bleedout. Figure 1b is the same transistor
after 3 minutes of cleaning with argon plasma. Epoxy bleedout is formed when
epoxy solvent separates from the epoxy matrix and leaches onto the substrate.
This solvent is not easily removed and if a bond is placed in the bleedout
area, it is likely to fail or be of low quality. In addition, if the bleedout
is not removed, it may continue to migrate inside of the sealed package and
eventually cause failure.
Figures 2a and 2b are photographs
of bonding pads prior to plasma cleaning and after 3 minutes of cleaning with
argon plasma. Residual baked on photoresist is not easily removed in the freon
degreaser processes used in hybrid manufacturing. The argon plasma procedure
removes the photoresist at a rate of approximately 300 angstroms per minute.
TABLE 1: OXYGEN vs ARGON CLEANING
Oxygen Argon Process: Chemical Physical Temperature: 200ºC 70-100ºC Etch Rate: 800-1000 Ä/min 500 Ä/min Pressure: 1 Torr 0.2 Torr Gas Flow: 5cc/min 2cc/min Benefits: Faster Less damage to device
EFFECT OF PLASMA CLEANING ON WIRE BOND YIELD
Experiment 1
Experiment Procedure: The data shown in Table 2 was generated by two hybrid
manufacturers. In each test, homogeneous samples were removed from assembly
areas and divided into two groups. The first group was plasma cleaned with
argon for 5 minutes in a Plasmod®
Plasma System using the following parameters: 0.2 torr pressure; 75 watts
power; 113 lpm vacuum. All devices were wire bonded using the same wire bonding
machine and the same operator. All devices were then subjected to pull tests
and inspected for failures.
Results: The results of the
experiments are outlined in Table 2. In laboratory 1, the wire bond failure
rate was reduced from 0.73% to 0.43% for 1.5 mil wires and from 24.50% to
11.00% for 1.0 mil wires. In laboratory 2, there was a reduction of wire bond
failures from 1.89% to 0.58%.
TABLE 2: EFFECT OF PLASMA CLEANING ON WIRE BOND YIELD
|
|
# of Devices |
# of Wires |
Wire Size (mils) |
Pull Test |
# of Bond Failures |
Failure Rate |
|
Lab#1 |
|
|
|
|
|
|
|
Plasma Cleaned |
25 |
1380 |
1.5 |
5g |
6 |
0.43% |
|
Plasma Cleaned |
|
100 |
1.0 |
3g |
11 |
11% |
|
|
|
|
|
|
|
|
|
Control |
25 |
1378 |
1.5 |
5g |
10 |
0.73% |
|
Control |
|
94 |
1.0 |
3g |
23 |
24.5% |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Lab#2 |
|
|
|
|
|
|
|
Plasma Cleaned |
50 |
1375 |
|
3.5g |
8 |
0.58% |
|
Control |
50 |
1375 |
|
3.5g |
26 |
1.89% |
TABLE 3: EFFECT OF PLASMA CLEANING ON WIRE BOND YIELD
|
|
# of Devices |
# Bonds |
Missed Bonds |
Failure Rate |
|
Lab#3 |
|
|
|
|
|
Plasma Cleaned |
10 |
840 |
1 |
0.12% |
|
Control |
10 |
840 |
29 |
3.45% |
TABLE 4: EFFECT OF PLASMA CLEANING ON BOND STRENGTH
|
|
PULL STRENGTH |
FAILURE MODE |
|
|
4.0 |
2L |
|
|
5.0 |
2L |
|
|
4.7 |
N1 |
|
|
6.5 |
N1 |
|
|
6.5 |
N1 |
|
|
7.8 |
N2 |
|
|
8.3 |
N2 |
|
|
8.0 |
N1 |
|
|
6.7 |
N1 |
|
|
6.2 |
N1 |
|
|
4.5 |
2L |
|
|
2.9 |
2L |
|
|
5.1 |
N2 |
|
|
4.4 |
2L |
|
|
2.2 |
2L |
|
|
3.2 |
2L |
|
|
|
|
|
Avg. Std. Deviation |
5.3 |
|
|
|
1.89 |
|
|
|
|
|
|
|
PULL STRENGTH |
FAILURE MODE |
|
PLASMA CLEANED |
|
|
|
|
7.7 |
N1 |
|
|
5.8 |
N1 |
|
|
3.9 |
N1 |
|
|
8.2 |
WB |
|
|
6.7 |
WB |
|
|
4.6 |
N1 |
|
|
5.3 |
WB |
|
|
4.7 |
N1 |
|
|
6.4 |
N1 |
|
|
7.1 |
N1 |
|
|
8.3 |
WB |
|
|
9.3 |
N1 |
|
|
8.0 |
N1 |
|
|
7.6 |
WB |
|
|
6.1 |
N1 |
|
Avg. Std. Deviation |
6.65 |
|
|
|
1.57 |
|
Note:
N1: Breakage at neck on 1st bond
WB: Wire breakage (bond strength exceeded wire strength)
2L: 2nd wire bond lifting
N2: Breakage at neck on 2nd bond
EFFECT OF PLASMA CLEANING ON BOND STRENGTH
Experimental Procedure: A group of 30 samples were removed from the assembly area and split into two groups. The first group was plasma cleaned with argon for 10 minutes in a Plasmod® Plasma System using the following parameters: 100 watts power; 0.2 torr pressure; 113 lpm vacuum. The second group served as a control. All samples were then subjected to pull tests. The pull strength and failure mode were determined.
Results: The results of the test are outlined in Table 4 above. The plasma cleaned group showed an average pull strength of 6.65 grams with a standard deviation of 1.57. The control group showed an average pull strength of 5.3 grams with a standard deviation of 1.89. In addition, the mode of failure for 8 samples within the control group was wire bond lifting. Of the samples which were plasma cleaned, either the wire broke or breakage occurred at the neck down during testing. This indicates that the bond strength exceeded the strength of the wire.
CONCLUSION
It is concluded from the data presented above that plasma cleaning prior to
wire bonding can reduce the number of bond failures and reduce the necessary
bonding power on automatic bonders. the quality of bonds is improved, thus
improving reliability. If bonding failure is due to contamination from residual
solvents, baked on photoresist, epoxy bleedout, or other organic contaminants,
argon plasma will effectively remove the contamination and improve the bonding.
REFERENCES
1. Graves, John F., Proceedings of the 1983 International Symposium on
Microelectronics, 147, 1983.
2. Bonham, H.B. and Plunkett, P.V., Electronic Packaging and Production,
42, 1979.
3. McKee, Jan L. Janssen, Toth, William D., and Fath, Perry M., Proceedings
of the 1986 International Symposium on Microelectronics ,
259, 1986.
Home
| News | Applications
| Lab | Plasma
1 | Service | Contact
| Site Map
©2007 March Plasma Systems / A Nordson Company 800-326-1151