This page contains a Flash digital edition of a book.
Interconnect Bonding


to pitch dimensions below approximately 20 µm due to slippage during bonding [8]. For Cu/Sn-Cu dice with a mechanical key,


the effect of bonding temperature was studied on samples at 10µm pitch. The bonding process was performed using a pressure of 5*106 kg/m2 at a temperature of 275 and 300°C for 180s


under N2 purging. Electrical testing showed high channel yields were achieved at both 275°C and 300°C bonding temperature. Average channel yield (each channel contains 1272 bonds) was 93% and did not depend strongly on bonding temperature. The parts bonded at 300°C had similar median resistance (103 mΩ/bond, with standard deviation of 8mΩ) to those bonded at 275°C (96 mΩ/bond, with standard deviation of 15mΩ). Resistances include the wiring and pad structures. The two results are within one standard deviation.


Cu/Sn-Cu Bonding Qualification Run To demonstrate the repeatability and test the robustness of the Cu/Sn-Cu bonding process with mechanical keying, a qualification of 22 bonding runs at 10µm pitch was done. This run used standard bonding conditions: in situ coining of the Sn, followed by bonding at 275°C for 180s at a pressure 5x106 kg/m2 (32.2kgf) over an array of pads 512 x 640 on a 10µm pitch.


After bonding, electrical testing on the 22 bonded pairs was done and the median channel yield was 98.4% and the median channel resistance was 65Ω. Each channel contains 1272 interconnections in a daisy chain format. The average resistance of each interconnect including the wiring and pad structures was <100 mΩ.


The wiring and pad structures were modeled and their calculated resistance was subtracted from the raw resistance value to give approximately 20mΩ per bond and a contact resistance of 5x10-8 Ohm-cm2. This result is compared with Cu-Cu bonding in the Comparison of Electrical Results Section. A yield map of six consecutively bonded Cu/Sn-Cu 10µm pitch dice is shown in Fig. 3. The blue lines mark the channels that were electrically open after bonding. The location of defects is not random, with most defects clustered near the device edges. The opens are


most likely due to handling damage. Aside from the edge defects, assuming that each open channel is caused by a single open bond, then the individual bond yield is >99.99%. Samples were cross-sectioned for SEM and


energy dispersive spectroscopy (EDS) analysis. SEM analysis of cross-sections was done with a backscattered electron detector. Quantitative EDS was performed to identify intermetallic phases at the bondline ( Fig. 8). The IMC is the


Cu3Sn phase. This phase is thermodynamically stable up to 650°C. There was no Cu6Sn5 or unreacted Sn. The BCB mechanical key was


important for achieving high yield by constraining slippage during bonding. The presence of the Sn layer was important for achieving high yield by accommodating the non-planarity of the bonding surfaces through in situ coining prior to bonding.


Cu-Cu Bonding (CMP’d Cu to CMP’d Cu)


CMP’d Cu to CMP’d Cu bonding was done with varied mechanical pressure and temperature. Typical bonding conditions were a temperature of 325°C and a pressure of 32.2 kgf, equivalent to 5x106 kg/m2. As with the Cu/Sn-Cu test vehicle, this vehicle had 512 x 640 array of pads on 10µm pitch for an interconnect density of 106/cm2.


Electrical measurement of bond chains gave a resistance of 95.9 mΩ and 44.5% channel yield


Fig. 4. SEM cross section thermocompression bonded sample with CMP Cu top die and CMP Cu bottom die


15


Fig. 5. Schematic and SEM cross-section of CMP Cu to Plated Cu metal bonding at 15µm pitch


www.euroasiasemiconductor.com  Issue III 2011


Page 1  |  Page 2  |  Page 3  |  Page 4  |  Page 5  |  Page 6  |  Page 7  |  Page 8  |  Page 9  |  Page 10  |  Page 11  |  Page 12  |  Page 13  |  Page 14  |  Page 15  |  Page 16  |  Page 17  |  Page 18  |  Page 19  |  Page 20  |  Page 21  |  Page 22  |  Page 23  |  Page 24  |  Page 25  |  Page 26  |  Page 27  |  Page 28  |  Page 29  |  Page 30  |  Page 31  |  Page 32