Reliability of Heavy Gage Aluminum Wire-bonds under High Temperature Aging

Aluminum (Al) wires are widely used in semiconductor industry to electrically connect the integrated circuits to the microelectronic package. Heavy gage Al wires (10 mils diameter or above) finds most use in analog power packages TO, SO, IGBT etc. due to its ease of use, low cost and excellent thermal and electrical performance [1]–[3]. Ultrasonic wire bonding process is typically used to form thick Al wedge bonds in the power package. It is an advantageous method of interconnect formation as it does not require external heat application and can be performed at room temperature. Ultrasonic bonding is performed at room temperature with the use of ultrasonic energy in the form of vibration which forms a cold weld joint between the Al wire and the substrate. The major factors that determine a good wire bond formation includes, ultrasonic power which signifies the energy applied for bond formation, bond time which signifies the amount of time the energy is applied and bonding force which signifies how much force is used to push the wire against the substrates for bonding. During wedge bond formation a wedge shaped capillary flattens the wire and to form a weld between the Al wire and the underlying substrate [4]. This method has high risk of wire over-deformation which leads to a weak bond, and, therefore, demands extreme optimization. If the process is not fully optimized, the bonds will be susceptible to cracking during reliability test stressing.

Automotive applications demand very high reliable microelectronic package. The reliability stresses required to be performed to qualify a product for automotive requirements are very stringent and are outlined in Automotive Engineering Council (AEC) Q-100 document [5]. Wire bonds are the weakest component in the package and are most prone to reliability risks. High thermal aging stresses like High Temperature Storage Life (HTSL) and High Temperature Operation Life (HTOL) pose great risk to the reliability of the wire bond and require deeper investigation of the failure mode and mechanisms to develop robust and reliability automotive grade packages.

This paper investigates the reliability of Al wedge bonds on bare copper (Cu) and nickel (Ni) plated substrates under isothermal aging conditions as AEC Q-100 Grade 1 requirements. The results shown can be used to design highly reliable automotive grade packages.

The test vehicle chosen to study the reliability performance of Al wires on bare Cu substrates was a transistor outline (TO-220) type package. The package was 10 mm × 9.2 mm × 4.4 mm in dimension with 5 leads at 1.7 mm pitch. Leadframe thickness of 1.31 mm was used. The silicon test vehicle used was 4.7 mm × 4.0 mm × 0.36 mm which had Al metal cap on the die bond pads. High lead (Pb) solder was used to attach the die to the heatsink of the package. Heavy gage 99.99% Al wires were used to form wedge bonds on the silicon and leadframe side. Transfer molding was performed to encapsulate the package with a standard molding compound.

To assess the high temperature aging reliability performance reliability testing was performed as per requirements outlined in AEC Q100. The product was tested for Grade 1 qualification requirements which involves testing for HTSL at 150 C up to 1000 and 2000 hrs. (extended read point) and HTOL up to 1500 hrs. with junction temperature Tj = 174 C and ambient temperature Ta = 150 C.

Electrical testing was performed at zero hour and post exposure to reliability stress to monitor the electrical performance of the package as a function of stress exposure. The parameter monitored was resistance between drain and source (RdsON). RdsON values typically drift when there is an increase in resistance to the current flow due to degradation in the current flow path. The samples that were exposed to HTOL started to show drift in RdsON values after 1000 hrs. of exposure to stress, however the units were still within the RdsON spec of 65 mOhm. Further increase in HTOL exposure time caused significant drift in RdsON where several units started to show out of spec values as shown in Fig 1. Units subjected to HTSL up to 2000 hrs. did not show significant drift in the RdsON values (data not shown).

Failure analysis was performed on units that failed post HTOL for RdsON drift. Scanning acoustic microscopy (CSAM) was performed on the failing and passing units to check for any delamination close to the wire bonds which could have impacted the integrity of the wirebond. Results shown in Fig 2 show no significant difference in the delamination performance between failing and passing units. Both show no delamination around the wire bonds on the lead posts and the heatsink, thereby eliminating delamination as the root cause for the RdsON drift failures. X-ray analysis was performed to check for the voiding/degradation in the solder die attach area and results are presented in Fig 3. No significant difference in solder die attach quality was noted as both units show similar voiding and degradation performance. This suggests that the die attach quality is not the root cause of the RdsON failures observed after HTOL.

The packages were decapsulated to check for the integrity of the wire bonds. The results are shown in Fig 4. Image 4a shows an overview of the decapsulated part. Fig 4b shows the lifted wirebond with its imprint on the lead post. Fig 4c and 4d show the wedge imprint on the leadframe and backside of the lifted wire, respectively.

Fig 4e and 4f show SEM image of the backside of the wire showing the Cu/Al intermetallic rich zone and Al rich zone, indicating that the failure occurred in the intermetallic layer. Fig 4g and 4h show integrity of the wedge bond prior to HTOL stress and post HTOL stress, respectively. Before stress, the wire was well bonded to the leadframe but after high temperature exposure the wire lifted off once the molding compound was removed. Interesting thing to note was that both HTOL and HTSL stresses units showed wedge bond lifts post stress, even though HTSL stress did not lead to RdsON drift failures. This implies that the integrity of the wirebond is impacted due to high thermal temperature exposure and the impact is more significant for HTOL stress condition as compared to HTSL condition. Wedge shear testing was performed to verify this. Results from wedge shear tests are shown in Fig 5. On the die side (Fig 5a) the wedges have very high strength even after high thermal exposure but on the leadframe side (Fig 5b) the wedges were shown to degrade in strength significantly. Both HTOL and HTSL stressed samples showed a drastic drop in wedge shear strength. The wedge shear test failure modes are shown in Table I. On the die side all units showed “Bond Shear” mode while on the leadframe side the failure mode was “Bond Lift” for thermally aged samples.

FA analysis presented in the previous section concluded that the weakness in the wedge bond is arising due to the brittle intermetallic (IMC) that forms between bare Al wire and Cu substrate during isothermal aging conditions at high temperature. Fig 6 shows Cu/Al IMC growth after 1500 hrs. HTOL stress. Severe cracking in the IMC layer is visible. Maximum IMC thickness measured was ~ 2 μm. Fig 7 shows Cu/Al IMC growth and wedge bond integrity for a HTSL 2000 hrs stressed sample. IMC growth up to 1.2 μm was visible, but no obvious signs of IMC cracking is visible.

Previous research has shown that Cu/Al IMC layer is high in hardness and electrical resitivity suspectible to cracking [6], [7]. Therefore, when IMC layer thickness exceeds a critical limit it will tend to crack under the stress that the mold compound exerts on the wedge bond during high thermal aging conditions. Due to high electrical resitivity, the thicker the IMC layer, yields a higher RdsON value. IMC thickness has been related to the exposure time and temperature through the following relationship [8]:

where δ is the IMC layer thickness, t is the exposure time and T is the exposure temperature. Using equation (1) theoretical IMC thickness could be calculated for HTOL 150 C, 2000 hrs. and HTOL Tj 174 C, 1500 hrs. exposure. Fig 8 shows the theoretical IMC thickness plotted against square root of t, which shows linear dependence.

Acceleration factor between 150 C and 175 C thermal exposure is calculated as 2.16, which makes 175 C thermal exposure twice as prone for IMC related failure as compared to 150 C aging for the same duration. Fig 9 shows comparison between maximum IMC growths that was measured at the Al/Cu interface in the TO-220 package. There is good agreement between the analytically predicted values and the actual measured IMC thickness. Higher temperature and higher exposure time leads to higher IMC growth and increases the probability of IMC related reliability failures. The effect will be more pronounced for higher temperature/time conditions, which explains why HTOL stress showed RdsON failures but HTSL stress did not.

To overcome the reliability issues encountered with thick Al wires under high temperature aging conditions, selective nickel (Ni) plating was applied on the bare Cu leadframes. Ni was plated in the areas where the wedge bonds were placed on the leadframe. Application of Ni on bare Cu substrate does not require any change in the wire bonding recipe. Application of Ni also increases the risk of delamination between the mold compound and leadframe as in general the mold compound adheres better to Cu than it does to Ni [9]. HTOL Tj 174 C, 1500 hrs. and HTSL 150 C, 2000 hrs. stresses were performed on Ni plated leadframes to ensure package and wedge bond reliability. Fig 10 shows CSAM results obtained, showing no delamination at the heatsink level and the lead post level showing no negative impact of Ni plating on the delamination performance after HTSL 2000 hrs.

Electrical testing performed showed no RdsON failures for both conditions, only a slight drift in RdsON was recorded. Wedge shear testing performed after HTSL and HTOL stresses showed very minor degradation the shear strength as shown in Fig 11. Fig 11a shows shear test data post HTSL stress and Fig 11b shows shear test data post HTOL tests. The degradation is very minor as compared to the degradation that was seen with bare Cu leadframes as shown in Fig 5. Table II shows the failure modes observed after HTSL and HTOL stresses. All failure modes observed were acceptable, bond shear mode. No bond lifts were observed as with bare Cu leadframes.

Representative images of the wedge bond after wedge shear testing are shown in Fig 12. Fig 12a shows a wedge bond lift observed on bare Cu leadframe after HTSL 2000 hrs. Fig 12b shows wedge shear observed on Ni plated leadframe after HTSL 2000 hrs., clearly showing that the integrity of the Al wedge bond has been significantly improved by applying Ni plating under the bonds.

Cross-section analysis was performed to check for any cracking and intermetallic growth at the wirebond/leadframe interface. The results obtained are shown in Fig 13. Fig 13a and 13b shows cross-section results after HTSL stresses. No significant IMC growth or cracking at the interface was observed. Similar results were observed after HTOL stress as shown in Fig 13c and 13d. Highly reliable Al wedge bonding was therefore realized by plating the leadframe with Ni to ensure no IMC growth which is detrimental to wirebond and package reliability during high thermal aging conditions.

Automotive grade microelectronic packages require careful selection of materials to ensure a robust package with no risk of reliability failure. Wire bond is a very important component of the package and is also the weakest link most susceptible to failure under reliability stresses performed during product qualification or during service in the field. Integrity and robustness of the wire bond should be carefully understood, analyzed and designed for during the package development stage. For automotive applications where the reliability stress and operating environment is very high, the use of bare Cu leadframes is not recommended due to brittle Al/Cu intermetallic that forms during high temperature exposure. By using Ni plating on the leadframe the brittle Al/Cu IMC growth can be restricted leading to a highly robust automotive package.

The authors would like to acknowledge Tom Battle, Goxe Julien and Alvin Youngblood for support with the failure analysis and package analysis.