1Department of Mechanical Engineering, Hong Kong University of Science & Technology. Clear Water Bay ... namely the electrolytic gold plating, immersion gold plating ..... 97-105. 4. L. J. Durney, Electroplating engineering handbook, 4th.
Wire Bondability of Au/Ni Bond Pads: Effects of Metallisation Schemes and Processing Conditions Yu Hin CHAN1, Jang-Kyo KIM1, Deming LIU2, Peter C.K. LIU2, Yiu Ming CHEUNG2, and Ming Wai NG2 1 Department of Mechanical Engineering, Hong Kong University of Science & Technology Clear Water Bay, Kowloon, Hong Kong 2 ASM Assembly Automation Ltd. 4/F., Watson Centre, 16 Kung Yip St., Kwai Chung, Hong Kong
Abstract The wire bondability and surface characteristics were studied of Au/Ni bond pads on organic FR-4 PCBs with different gold plating schemes and gold thickness. Three different plating techniques were used to deposit gold layers, namely the electrolytic gold plating, immersion gold plating and immersion gold plating followed by electrolytic gold plating. Process windows for the individual metallisation schemes were established to optimize the wire bond force and bond power. Bond performance after wire bonding was evaluated based on the wire pull strength and wire deformation ratio. Various elemental and surface characterization techniques, such as SEM, Optical profiler, XPS, nano-hardness test and surface energy analysis, were employed to characterize the bond pad surfaces, which were then correlated to the wire bond performance. The effects of plasma treatment using Ar and O2 gases were also studied. Introduction Wire bonding is a popular first-level interconnection technology used between the die and packages terminals. Gold wire bonding has the advantages of fast bonding process, higher productivity, better electrical performance, higher heat conductivity and better corrosion resistance. It has been widely used in various electronic packages, such as chip-onboard (COB), chip scale package (CSP) and ball grid array (BGA). The PCBs used for COB, CSP and BGA are usually made from glass-reinforced epoxy laminates to provide a low cost, light weight and low profile solution. However, the glass transition temperatures, Tg of organic substrates are low. The Tg of FR-4 laminates is typically 120 oC. This limits the wire bonding process temperature to be lower than the Tg in order to avoid softening of materials, which may degrade the wire bonding quality. For the bond pads for gold thermosonic bonding, it requires a 1-1.25µm thick electrolytic Au over a 45µm thick electrolytic Ni layer in traditional practices. However, the fabrication process of electrolytic plating may contaminate the plated gold surface. Also, much more Au may be plated than is actually necessary for wire bond pads and hence increasing the cost 1. Immersion Au over electroless Ni plating is an alternative solution. The fabrication process could minimize the contamination in the Au layer which is beneficial to the wire bonding process. Also, it could provide an Au thickness of 0.01-0.28µm which reduces cost. The quality of wire bond in low temperature thermosonic bonding depends on the bond pads metallisation surface characteristics and the wire bonding conditions. The present study focuses on optimizing the wire bond process (such as
bond power and bond force) and the metallisation (such as different Au/Ni plating schemes and thickness of Au) on the bondability of an Au-Au thermosonic bonding system. In addition to bond pad metallisation and wire bonding process, contaminants on bond pads have been known to affect both the bondability and the reliability of wire bonds. Organic contamination usually occurs in the form of an invisible film which makes the bond pad surfaces hydrophobic. A method for removing organic films is by use of plasma cleaning. Plasma is formed when an intensive radio frequency (RF) electromagnetic wave is applied to argon and oxygen gases, it will react with the hydrocarbon chains forming volatile gases that are carried away by a vacuum pump2. Bondability could be restored without damaging the surfaces. However, if the plasma process is not optimized, the end result can be more worse than that with no plasma cleaning at all3. The effect of plasma cleaning on the bondability of wire bonding is also included in the present study. Experiments Material and specimens The PCBs with copper cladding was 0.8mm thick FR-4 type glass fabric/epoxy laminate with a Tg of about 120oC. The thickness of Cu was about 25.4µm. On top of the copper layer, nickel was electroplated with nickel sulfate (NiSO4) solution for electrolytic Ni plating and Ni(H2PO2)2 solution for electroless Ni plating. On top of the Ni layer, a gold layer was plated based on several methods, including electrolytic gold plating, immersion gold plating and immersion gold plating followed by electrolytic gold plating. The Au thickness was measured by means of an X-ray film thickness gauge provided by a vendor. The details of the plating condition and gold thickness obtained are summarized in Table I. Table I. Metallisation combinations of plating method and thickness. Plating Method Sample Thickness (µm) Ni Au Ni Au Sample 1 0.01–0.1 4-8 Sample 2 Electrolytic Electrolytic 0.11-0.25 Sample 3 0.4 0.7 Sample 4 Sample 5 Immersion 0.1 Electroless 2.5 - 5 Sample 6 Immersion 0.25 +electrolytic Plasma Treatment
The bond pads were subjected to a plasma treatment to remove any contamination on the surface using a Plasma Clearer P100E. The process condition was optimized in terms of the gases used, pressure, time, and power. It was found that the Ar gas alone was not adequate to throughly clean the surface, and O2 gas must be added. The parameters used in the plasma treatment process are summarized in Table II. Both the Ar and O2 gases were used initially, followed by the Ar gas treatment after 60s for another 60s.
deformation ratio implies that the bonding power was too high or most of the energy was consumed to deform the wire. Hook
1st bond
2nd bond 0.33mm
Table II. Plasma Cleaning process. Gas Flow rate Time Power (SCCM) (s) (W) Argon 5 120 100 Oxygen 5 60 100
120 120
Wire bonding, wire pull test and deformation ratio measurement Thermosonic gold wire wedge bonding was preformed on an ASM model AB559A wedge bonder equipped with a Uthle model 61.2kHz transducer. The wedge tool was an SPT model FP308-TI-1820-L-CGM w=0.0028. The gold wire was 25µm in diameter from Tanaka. It has an elongation of 0.53% and its tensile strength is 12-17gf. The distance between the 1st and 2nd bonds was 1.33mm and the average loop height was 0.328mm. In this experiment, no dies were mounted on the PCBs. Both the 1st and 2nd bonds were on the PCBs. Wire bonding was carried out after preheating at 120oC ± 4 oC with a bonding frequency of 61.2kHz and the nominal bonding time was 20ms. The bond force was selected at two levels, 20gf (low bond force) and 24gf (high bond force). The bond power ranged from 65mW to 163mW. The wire bonding process window for each metallisation schemes was established by wire bonding with various combinations of bond force and bond power. Twenty wires were bonded for each combination. The wire pull tests were used to measure the quality of wire bonding using a Royal Instrument System 550 wire pull tester. The focus of the wire pull test was made on the reliability of the 2nd bond. Therefore, a modified pull test was carried out by placing the testing hook closer to the 2nd bond, as schematically shown in Fig. 1. A load was applied to pull the hook upwards until the wire broke or the 2nd bond was lifted off the bond pad, and the corresponding force was recorded as the wire pull strength. The criteria of the bonding parameters that achieved successful bonds was that all failures should be wire neck broken at the 2nd bond. In addition to wire pull strength, the deformed wire widths of the 2nd bond were also measured using a profile projector. The wire deformation ratio was calculated, which is defined as the ratio of the wire width to the original wire: Deformation ratio =
W D
Glass\epoxy Laminate
Pressure (mTorr)
(1)
Where W is the width of a bond point and D is the original diameter of the wire. The deformation ratio reflects the energy consumed during the wire bonding process. A high
Fig. 1. Wire pull test configuration.
W
D
Fig. 2. Measurement of deformation ratio. Elemental composition, surface morphology, roughness and hardness of bond pads The surface characteristics of the bond pads were examined using an X-ray photoelectron spectroscopy (XPS, Surface analysis PHI5600), a scanning electron microscope (SEM, JEOL6300) and an optical profiler (WYKO NT3300). The XPS scan size was about 600µm in diameter and the nominal analyzing depth was 8nm. The non-contact optical profiler provides the 3D surface topography can measure surface roughness at a vertical resolution of 0.1 nm, similar to an atomic force microscope. The scan area was 62µm x 81µm and the root-mean-square roughness Rrms was obtained. The nano-indentation test of metallisation was conducted using a Triboindenter to measure the hardness of bond pads. Contact angle measurement The cleanliness of bond pads was quantified by contact angle measurement. Contact angle measurements was performed using a goniometer (KRÜSS G10) using two probing liquids, de-ionized (DI) water (a polar liquid) and diiodomethane (a non-polar liquid). Droplets of the liquids were dispensed on the bond pad surface and the images of the droplets were captured using an image analyzer to measure the contact angles. This measurement is done for Sample 4 and Sample 5 only. The surface free energies of the bond pads were calculated using the contact angle data based on the Young-Depre equation: 1
1
(1 + cosθ )γ LV = 2[(γ SV γ LV ) 2 + (γ SV γ LV ) 2 ] (2) P
where
P
d
d
γ SV and γ LV are the surface free energies of solid and
liquid, respectively. The superscripts p and d are the polar and
Experiment Results and Discussions Surface morphology of bond pads The SEM images of bond pads produced from different metallisation schemes are presented in Fig. 3, showing distinct surface morphologies depending on the electroplating methods used. The Samples 1 and 5 exhibited a nodular profile, reflecting the underlying Ni layer, due to the very thin gold plate. In particular, the grains and grain boundaries are clearly seen for Sample 5. Comparison of the images between Samples 1 and 5 indicates that the immersion plating method (Sample 5) could deposit a gold layer of more uniform thickness. Immersion gold plating is based on the exchange of electrons between the atoms of Ni and the ions of Au. The deposition process ceases when the gold layer is a few atoms thick4, which is more uniform in thickness than in electrolytic gold plating. A similar observation holds for Sample 6 when compared with Sample 2 with a gold layer of a similar thickness. The surface feature of the underlying Ni layer became less obvious as the gold plate thickness increased, as indicated by Samples 2 and 3 that had a thicker gold layer than Sample 1. With a further increase in gold plate thickness (Sample 4), the morphology of the underlying Ni layer is completely conceded. Surface roughness Basically similar conclusions can be drawn from the gold surface profiles obtained from an optical profiler, as illustrated in Fig. 5. The gold peak was more dense and smaller in Sample 5 (immersion plating) than in Samples 1 to 4 (electrolytic plating). The surface roughness values measured thereby are summarised in Fig. 4 in terms of Rrms. For the bond pads produced using the same electrolytic plating method, the surface roughness decreased as the gold thickness increased. The decrease in roughness was mainly due to masking of the underlying Ni layer feature and flattening of the gold layer, as seen from the SEM images. The roughness of the gold layer produced by immersion plating (Samples 5 and 6) was higher than the bond pads of a similar gold thickness produced by electrolytic plating. X-ray Photoelectron Spectroscopy The elements detected by XPS on all metallisation schemes were mainly C, O, and Au. Metals such as Na, Si, Ni and Cu were also detected in small quantities. This indicates that the bond pad surfaces contained organic compounds, arising from the adsorption of gases from air and organic contamination such as oily vapors and outgassing vapors from epoxies. For the metals detected other than the top layer gold, it is discovered that the ratio of Ni to Au contents in electrolytic gold plating was approximately inversely proportional to the gold layer thickness, as indicated by Fig. 6. The Ni concentration was high in the thin gold layer especially on the bond pad produced by immersion gold plating (Sample 5) which has a Au/Ni ratio of 0.072. To determine the nature of Ni, the binding energies of the peaks on each metallisation scheme were evaluated. The result
indicates that the detected Ni on the gold layer was nickel oxide (Ni2O3), except that for Sample 4 which was nickel metal (Ni). The nickel oxide in the gold bond pads has a detrimental effect on wire bondability and bond yield6.
Electrolytic Au (0.01 – 0.1 µm)
Electrolytic Au (0.11-0.25 µm)
Electrolytic Au (0.4 µm)
Electrolytic Au (0.7 µm)
Immersion/Electrolytic Au (0.25 µm)
Immersion Au (0.1 µm)
Fig. 3. Scanning electron micrograph of bond pad surfaces. (Au thickness in parenthesis) 300 Roughness (nm)
dispersive components, respectively. The total surface energy is the sum of these two components.
250 200 150 100 50 0 0
1
2
3 4 Sample
5
6
7
Fig. 4. Roughness of each metallisation scheme. Hardness of bond pads The hardness values measured from the nano-indentation test are plotted as a function of gold thickness, as shown in Fig. 7. It seems that the hardness decreases with increasing
gold thickness, indicating ‘composite hardness’ that reflects Fig. 5. Surface profiles of bond pads. the hardness of both the gold and the underlying Ni layer. The high Ni content in the bond pads with a thin gold layer It is interesting to note that the bondability was very (Samples 1, 2 and 5) may also be partly responsible for the similar between the thick gold plate made by electrolytic high hardness values. In this regard, it is seen that the surface plating (Sample 4) and the very thin gold plate made by characteristics, including hardness, Ni content and the gold immersion plating (Sample 5). Both of them have a wide layer thickness are inter-related. process window and generally a high pull strength. This may suggest that while the gold thickness was an important Wire bonding and Pull test parameter, there were also other important factors controlling Fig. 8 summarizes the wire pull test results for all metallisation schemes with bond power ranging from 100 to the wire pull strength. A comparison of the results among electrolytic gold 163mW at a bond force 20gf. No data points at certain bond plating indicates that the thinner the gold layer, the more powers means no successful wire bonds. The results indicate power needed for successful bonding. The hardness of the that a thick electrolytic gold (Samples 4) and immersion gold bond pads may be responsible for this observation because a (Sample 5) can be bonded successfully at a low bond power. higher energy is needed to deform the gold wire and form a A higher bond power can produce a good bond for other successful bond. metallisation schemes except for a very thin electrolytic gold The bond pad produced by immersion gold plating could (Sample 1), no successful bond could be made even when a high bond power is applied. The wire pull strength in general be bonded successfully at a low bond power although it has a decreases as the bond power increases. This process indicates high hardness value. For electrolytic gold plating following that the energy supplied at the lowest bond power threshold is immersion gold plating, its process window was narrow and enough to make a successful bond. If the bond power is the pull strength was relatively low. The high surface further increased, this extra energy will be consumed to roughness may be responsible for the poor bondability and deform the wire and weaken the neck causing the pull low pull strength. The bond pad with high roughness surface the high surface strength to decrease. This also explains why the pull strength may require a higher energy to overcome 5 of Sample 2 of a higher threshold bond power was higher asperity before a bond could be formed . During the bonding process, the wire was highly deformed in order to allow the than that of Sample 4 at high bond powers. free electrons to move across the interface to form metallic bonds. 0.01
Ni/Au ratio
0.008 0.006 0.004 0.002 Electrolytic Au (0.01 – 0.1 µm)
Electrolytic Au (0.11 – 0.25 µm)
0 0
0.2
0.4
0.6
0.8
Au thickness (um)
Fig. 6. Ni/Au ratio of electrolytic gold as a function of gold thickness. 5.00
Electrolytic Au (0.4 µm)
Electrolytic Au (0.7 µm)
Hardness (GPa)
Electrolytic gold 4.00
Immersion gold
3.00
Electrolytic+immersion gold
2.00 1.00 0.00 0
Immersion Au (0.1 µm)
Immersion/Electrolytic Au (0.25 µm)
0.2
0.4 Gold thickness
0.6
Fig. 7. Nano-indentation test for each sample.
0.8
was pressed at a higher pressure, constraining the vibration of the wire against the bond pads. The higher the hardness of bond pad, the more energy required to deform the wire and thus transmit an adequate amount of energy to the wire-bond pad interface for bonding. For Sample 4 whose hardness was the lowest, its bond pad absorbs ultrasonic energy readily. A high bond force could constrain the wire deformation in this sample and restore its pull strength at a high bond power.
11.00 10.00
8.00 Sample 1 Sample 2 Sample 3 Sample 4 Sample 5 Sample 6
7.00 6.00 5.00
2.00 1.90
4.00 100
110
120
Bond force = 20gf
1.80
130
140
150
160
170
Bond power (mW)
Deformation ratio
Pull strength (gf)
9.00
Fig. 8. Wire pull strength as a function of bond power and bond force is at 20gf.
Bond force = 24gf
1.70 1.60 1.50 1.40 1.30 1.20 1.10 1.00
11.00
100
110
120
Pull strength (gf)
10.00
130 140 150 Bond power (mW)
160
170
Fig. 10. Deformation ratio as a function of bond power (Sample 4)
9.00 8.00 7.00
2.00
6.00
4.00 100
110
120
130
140
150
160
170
Bond power (mW)
Fig. 9. Wire pull strength as a function of bond power and bond force is at 24gf. Symbols as in Figure 8. Fig. 9 summarizes the pull test results for the wire bonds made at high bond force of 24gf. The result indicates that the bond pads with a thick electrolytic gold plate and those produced by immersion gold plating can be bonded successfully at a low bond power. For the samples with thinner gold pads produced from electrolytic plating (Samples 1, 2 and 3) and electrolytic plating following immersion gold plating (Sample 6), the bondability was worse than those corresponding to a low bond force (20gf). A higher bond power was required to achieve successful wire bonding. Sample 1 could not be bonded at the above bond power settings. In addition, the trend of wire pull strength with respect to the bond power was different between the different bond forces. The pull strength of Sample 4 increased whereas those for the other samples decreased with increasing bond power. The large difference in gold thickness in these samples indicates that the gold layer thickness played an important role, probably through its effect on the hardness of the metallisations. When a high bond force was applied, the wire
Deformation ratio
5.00
1.90
Bond force = 20gf
1.80
Bond force = 24gf
1.70 1.60 1.50 1.40 1.30 1.20 1.10 1.00 100
110
120
130
140
150
160
170
Bond power (mW)
Fig. 11. Deformation ratio as a function of bond power (Sample 5). The results for Sample 5 were rather different from those for Sample 4. At both low and high bond forces, the deformation ratio increased when the bond power increased. The deformation ratio was in general higher at a high bond force than at a low bond force, due probably to the relatively hard bond pad (see Fig. 11). For a hard bond pad, the ultrasonic energy tends to be consumed to deform the wire instead of welding the wire onto the bond pad. This is why the wire pull strength decreased with increasing bond power for both bond forces used (see Fig. 9).
Effect of Plasma Cleaning The effect of plasma cleaning on wire bondability of the bond pads with thin (Sample 2) and thick (Sample 4) electrolytic gold and immersion gold (Sample 5) was studied, and the major results are shown in Fig. 12. Significant improvements in both the process window and pull strength were noted for the samples with electrolytic gold layer. Both Samples 2 and 4 could be bonded successfully using lower bond powers. For Sample 2, the lowest bond power for successful bonds was 128mW, which was reduced to 75mW after plasma cleaning, whereas for Sample 4 the lowest bond power for successful bonds was 101mW and it was reduced to 65mW after plasma cleaning. This is because the ultrasonic energy required to push aside the contamination to allow direct contact between the wire and the gold pad surface was reduced. The corresponding wire pull strengths were also improved significantly after plasma cleaning. It is noted that after plasma cleaning the bondability of thin electrolytic gold plating was even better than immersion gold plating. 13.00
Pull strength (gf)
12.00 11.00 10.00 9.00 8.00 Sample 2 (PT) Sample 4 (PT) Sample 5 (PT)
7.00 6.00 5.00 60
80
Sample 2 (AR) Sample 4 (AR) Sample 5 (AR) 100 120 Bond power (mW)
140
Deformation ratio
Fig. 12. Pull strength as a function of bond power of Sample 2, 4 and 5 as-received (AR) and after plasma cleaning (PT). 2.00 1.90 1.80 1.70 1.60 1.50 1.40 1.30 1.20 1.10 1.00
As-received Plasma Cleaned
60
70
80
90
100
110
120
Bond power (mW)
Fig. 13. Deformation ratio as a function of bond power (Sample 4). For Sample 5 with immersion gold plating, the improvement was less significant. The pull strength increased slightly and the lowest bond power for successful bonding could not be further reduced. This indicates that the bond pad with immersion gold plating is not easily contaminated, or the wire bonding is not sensitive to the surface cleaning. The difference in wire bonding performance arise from the difference in the manufacturing processes. For example, a
solder mask was applied after electroplating for the bond pads with electrolytic gold plate, while a solder mask was applied before electroplating for the bond pads with immersion gold plate. This may reduce the concentration of organic contaminants, such as outgassing of organic compound during solder mask curing. In addition, the etching process of electrolytic gold plating may cause the underlying Cu and Ni layers of the bond pad to be exposed to the ambient environment. In the immersion gold plating process, the underlying metal layers are always fully covered by gold because the etching process is not required. The deformation ratio was reduced after plasma treatment as shown in Fig. 13 for Sample 4. This result may indicate that plasma cleaning aids the transmission of energy to the interface between the wire and the bond pad. The wire deformation was reduced, giving rise to a strong wire neck with improved bondability and wire pull strength. Effect of exposure time after Plasma Cleaning The effect of exposure time after plasma cleaning was studied. The cleanliness of the bond pads after exposure to the ambient environment was quantified based on the contact angle measurement, surface energy calculation and wire pull strength. Thick electrolytic gold plating (Sample 4) and immersion gold plating (Sample 5) were selected for study. The contact angles of DI water on bond pad surfaces at different exposure time after plasma cleaning of Sample 4 and Sample 5 are illustrated in Fig. 14 and Fig.15 respectively. The initial contact angles of DI water after plasma cleaning of both samples were in the range of 20-30o, indicating a hydrophilic surface in nature. The contact angles increased sharply in the first 8 h due to contamination, and those after 8 h of exposure were about 60o and 50o respectively, for Samples 4 and 5. The increase in contact angle became almost saturated after about 24 h of exposure for Sample 4, whereas the contamination still continued after 24 h for Sample 5. Because most organic contaminants are hydrophobic in nature, the contact angles of diiodomethane remained relatively unchanged. The surface free energies of bond pads were calculated using the contact angle data based on Eq (2), and the results are shown in Figs. 16 and 17. The total surface energies for all samples decreased with exposure time, the decrease upon initial exposure being most significant. After 8 h of exposure, the surface energies remained almost unchanged until 24 h. The surface energies obtained for 24 h exposure were still slightly higher than those in the as-received condition. This indicates that a slightly longer exposure time would be necessary to match that of the as-received status. The sensitivity of surface energy drop with exposure time was similar for all metallisation schemes, indicating similar thermodynamic characteristics for all surfaces. The wire pull strength decreased with increasing exposure time. The drop in pull strength was most significant in the first few hours for both samples. After 8 h of exposure, the pull strength became almost saturated for Sample 4, whereas it continuously decreased even after 24 h of exposure for Sample 5. All these observations are in good agreement with the surface energy measurements. The wire pull strength became almost
equivalent to that for the as-received condition after about 24 h of exposure. From this result, it is recommended that the allowable dwell time be less than 8h.
Fig 16. Surface energies as a function of exposure time (Sample 4). 100 Surface Energy Surface Energy (mN/m)
100
Contact Angle
80 60 40 20
80
Polar Non-polar
60 40 20 0
DI Water
0
Diiodomethane
5
10
0
5
10
15
20
25
30 AR
Fig 14. Contact angles as a function of exposure time (Sample 4).
AR30
35
Fig. 17. Surface energies as a function of exposure time (Sample 5). 11.00
100
10.00 9.00 Pull strength (gf)
80 Contact Angle
25
35
Exposure Time (Hour)
60 40 20 0 0
5
10
15 20 25 Exposure time (Hour)
30 AR 35
Fig 15. Contact angles as a function of exposure time (Sample 5). 100
Surface Energy Polar
80
Non-polar 60
40
20
0 5
10
15 20 25 Exposure Time (Hour)
8.00 7.00 6.00
Sample 4
5.00
Sample 5
4.00
DI Water Diiodomethane
Surface energy (mN/m)
20
Exposure Time (Hour)
0
0
15
30 AR 35
0
5
10 15 20 25 Exposure T ime (Hour)
30AR
35
Fig. 18. Pull strength as a function of exposure time of Sample 4 and 5 after plasma cleaning. Concluding Remarks The wire bondability and surface characteristics were studied of Au/Ni bond pads that were produced using different plating methods. The following conclusions can be drawn from the experimental study. 1. The surface morphology of bond pads produced by different gold plating methods was different. The immersion gold plating produced a fine, thin gold layer. The surface roughness of bond pads created by electrolytic plating decreased as the gold layer thickness increased. High surface roughness is considered detrimental to wire bonding because more energy is required to overcome the surface asperity before bonding is made in the present study. 2. The thinner was the gold plating, the higher was the ‘composite hardness’. A high hardness is regarded detrimental to bondability as it requires a higher energy to form metallic bonds.
3. 4.
5.
The XPS result indicates that the thinner the gold layer, the higher the risk of Ni migration to the top gold layer. Plasma cleaning enhanced the wire bondability. The bond pads produced by electrolytic plating, especially with a thin gold layer, experienced significant improvements in process windows and wire pull strengths. It is recommended that the bond pads be wire bonded within 8h of exposure to air after plasma cleaning. The immersion gold plating is a low cost alternative choice for wire bonding. Its bondability is comparable to that of the bond pads with a thick electrolytic gold layer even before plasma cleaning.
Acknowledgments The authors wish to thank the Innovation and Technology Commission (ITC), Hong Kong SAR Government for the continuous support of this project through the Innovation and Technology Fund (UIT/32). Assistance with experiments rendered by the Materials Characterisation and Preparation Facilities (MCPF) at HKUST is also gratefully appreciated. References 1. C. Dunn, R.W. Johnson, and M. Bozack, “Thermosonic gold ball bonding to alternate plating finishes on laminate MCM substrates,” 1997 Int. Conf. Multichip Modules, 1997, pp. 170-175. 2. L. Yang, J.B. Bernstein, and K.C. Leong, “Effect of the plasma cleaning process on plastic ball grid array package assembly reliability,” IEEE Trans. Electron. Packag. Manufac., Vol. 25, (2002), pp. 91-99. 3. C. Lee, R.Gopalakrishnan, K. Nyunt, A. Wong, R.C.E. Tan, and J.W.L. Ong, “Plasma cleaning of plastic ball grid array package,” Microelectr. Reliab., 39, (1999), pp. 97-105. 4. L. J. Durney, Electroplating engineering handbook, 4th edition, Van Nostrand Reinhold (New York, 1984), pp. 421. 5. J.K. Kim, B. P.L. Au, “Effects of metallization characteristics on gold wire bondability of organic printed circuit boards,” J. Electr. Mater., 30, (2001), pp. 1001-1011. 6. G.G. Harman, Wire bonding in microelectronics: Materials, processes, reliability and yield, 2nd edition, McGraw Hill (New York, 1997), pp.173. 7. H. Haji, T. Morita, K. Arita, H. Nakashima, and H. Yoshinaga, “Bondability of gold wire to gold-plated electrodes,” Mater. Trans. JIM, Vol. 34, (1993), pp.960965. 8. Y.F. Chong, R. Gopalakrishnan, G.F. Tsang, G. Sarkar, S. Lim, and S. Tatti, “An investigation on the plasma treatment of integrated circuit bond pads,” Microelectr. Reliab., 40, (2000), pp.1199-1206. 9. V. Koeninger, H.H. Uchida, and E. Fromm, “Degradation of gold-aluminium ball bonds by aging and contamination,” IEEE Trans. CPMT-A, Vol. 18, (1995), pp. 835-841. 10. J. K. Nesheim, “The effects of ionic and organic contamination on wirebond reliability,” Proc. 1984. Int. Sym. Microelectr., 1984, pp.70-78.
11. D.W. Bushmire and P.Ho Holloway, “The correlation between bond reliability and solid phase bonding techniques for contaminated bonding surfaces,” Proc. 1975 Int. Sym. Mcroelectr., 1975, pp.402-407. 12. J. E. Krzanowski, “A transmission electron microscopy study of ultrasonic wire bonding,” IEEE Trans. CHMT, Vol.13, (1990), pp. 176-181. 13. R. D. Rust, D.A. Doane, “Improvements in wire bonding and solderability of surface mount components using plasma cleaning techniques,” IEEE Trans. CHMT, Vol. 14, (1991), pp. 573-579. 14. J. L. Jellison, “Kinetics of thermocompression bonding to organic contaminated gold surface,” Pro.c Electron. Comp. Conf., 1976, pp.92-97.
