TCB Process Options to Achieve the Lowest Cost

Abstract

Thermocompression bonding enables the next generation fine pitch, 2.5D and 3D assembly technologies using Cu pillar interconnects, but to achieve widespread adoption the cost of TCB must become competitive with mass reflow processes. Stacked memory products drive the commercial volume today using TSV structures and TCB since it is the only technology able to achieve the desired stacked die construction and improved performance, but reducing the cost of assembly is still a key goal for those suppliers. In non-memory applications the choice of TCB can be driven by the bump pitch of the device or the requirement to control warpage of large die on laminate during assembly, but cost is still a key factor in the decision. The cost of a TCB process is largely driven by the UPH of the process where cost calculations are based on the cost per unit of material produced. As the UPH of a TCB process approaches 1400, the differential cost of the TCB process as compared to mass reflow becomes negligible. In the choice of a potential TCB process, special attention must be given to those processes that enable the highest UPH and the lowest cost. Processes used for TCB today can be grouped into two main categories; processes that use a pre-applied underfill and those that apply underfill after the bonding process. Underfills applied prior to bonding can be in the form of a non-conductive paste (TC-NCP) applied to a substrate or a non-conductive film applied to the wafer before dicing (TC-NCF). If underfill is applied after the bonding process, it is done as a Capillary Underfill (TC-CUF). In this case the die is underfilled in much the same way as in standard flip chip processes, but the process can be more challenging because of flux cleaning requirements and the narrower bondline of a typical TCB device. UPH is primarily driven by two factors; the range of temperature required by the bond head and the temperature ramp rate of the bond head. A process with less temperature range will have higher UPH and bond heads designed for the fastest cooling and heating rates will provide higher UPH processes. Two process options have been developed to minimize the temperature excursions required by the bond head and maximize the throughput. TC- NCF processes targeting stacked die and interposer products have been developed with throughputs approaching 2000 UPH. Substrate flux TC-CUF processes targeting assembly on laminate have been developed with throughputs that approach 2500 UPH. These two processes are expected to dominate TCB volume production moving forward as TCB enters mainstream production. This presentation will describe the methods used to achieve high throughput for both processes and the product application space appropriate for each one.