![]() July 1998 eMail the Editor |
µBGA® Lead Bonding Process Requirements and CapabilitiesBecause the processes differ significantly, traditional wire bonding equipment must be adapted for µBGA lead bonding.By Daniel Bolliger, Christian Brändli and Daniel Zanetti, ESEC SA, Cham, Switzerland The µBGA format1 is the most promising of the different up-and-coming chip-scale packages. The first assembly houses have begun mass production and packaged devices are now available on a commercial basis.2
The main processing steps in the µBGA packaging flow are die attach, lead bonding, edge sealing, ball attach and device singulation. The lead bonding process has a major impact on the reliability of the entire package. Due to the different thermal expansion coefficients of the die and circuit panel, leads must flex along the lead direction during power and heat cycling of the device without failure.3 The lead has to be shaped so that no fatigue effects are caused in it. In addition, a high performance, long-term, stable electrical interconnection is established between the bond pad and the signal trace on the µBGA flex material. Lead and Wire Bonding DifferencesThe lead bonding process differs in many points from traditional wire bonding. Standard wire bonding equipment, therefore, must be adapted to meet all the requirements for lead bonding.Common characteristics of both processes are material handling, use of thermosonic bonding principles and an extremely accurate, high-speed motion of the bondhead. The main differences are determined by the different materials and geometry which are employed in wire and lead bonding. In lead bonding, gold wire is replaced by the lead, which is an integral part of the tape material. For lead material, a compound of gold-plated copper or pure gold is used. Additionally, the ceramic wire bonding capillary has been replaced by a non-symmetrical bonding tool made of a different composition. This change causes a strong impact to the ultrasonic sound propagation and influences energy dissipation during the bonding process. The chip itself is mounted below the package substrate material, similar to the lead-on-chip process in wire bonding. A rigid material is no longer used for the package substrate. The organic µBGA tape material is about 2 mils thick and is, therefore, very flexible. Tape strips containing multiple die can be mounted on a rigid metal tape frame carrier which allows material handling similar to that employed for leadframes. Lead Bonding Process OverviewThe lead bonding process can be divided into three steps: cut, shape and bond. The first two differ completely from wire bonding and have to be specially designed for lead bonding:
Ultrasonic energy is applied to establish an intermetallic bond formation between the lead and the underlying bond pad.
The bonding scheme is completed with a simple complementary process where the bond head moves to the starting point of the next lead cycle. In contrast to wire bonding, the electrode flame-off and the entire wire handling process have been eliminated. AlignmentThe basic tape material for µBGA packages consists of a thin, flexible organic layer. Strips containing several die are attached to a metal tape frame carrier which can be handled like regular leadframes. Sprocket holes on the carrier refer to the relative placement position of the tape or of the die on the tape. The sprocket holes are used for coarse alignment of the tape on the lead bonder indexer and must follow the same design rules as leadframes. Alternatively, an optical leadframe centering method can be used which refers to the tape structure itself.The position of the single leads is related to the tape position. On the other hand, the position of the bond pads is linked to chip placement. Therefore, separate reference systems must be used for each die in a multi-device bonding area, with independent tape alignment references used for each die site. These references must be visible for the pattern recognition system and must not be hidden by non-transparent tape material or leads. For lead alignment, the position of the tape has to be known for each device. Tape references allow die bonding equipment alignment from the lower side. In contrast, the lead bonder uses the same references from the opposite side of the tape for alignment. Preferred references are made of metal structures with a very good placement accuracy measured relative to the lead positions. If the metal layer of the tape is placed between non-transparent tape material and die, a hole in the organic tape material must be opened for each reference. The metal references on the tape are also visible from above for the lead bonding pattern recognition system. Lead CuttingThe lead bonding tool approaches the lead with a perpendicular motion to the chip surface, slows down and begins touching the lead. The tool shape causes self-centering and gripping of the lead to the bonding tool. As the vertical motion continues, the lead is both elastically and plastically deformed until it breaks at a certain height (the cutting height) above the chip surface.The lead deflects by moving perpendicular with the tool towards the chip surface shown in Figure 1. L is the total lead length in initial state, h is the clearance between lower lead side and chip surface, z is the vertical lead displacement, T is the tool width and d is the lead edge to tool edge distance. A and B indicate where the tool is placed during lead cutting or bond formation and a a denotes the shaping angle . The vector from cut position A to bond position B is a key parameter for the lead bonding process. It determines the shaping angle a, cut height h-z and the excess bonded lead length which is essential for the final lead shape. In addition, the total lead elongation e at the break point has a major influence on the cut height. The overall measured elongation e can be as large as a factor of 1.4 at cut height. The lead of the length L can be stretched plastically and elastically to a length
before it breaks, where this stretched length is constant. By some arithmetic calculations, the deflection z of the lead perpendicular to the chip surface can be calculated as a function of elongation e, lead length L, tool width T and the parameter d, which denotes the distance between the edge of the lead and the point where the tool edge is placed for cutting:
The maximum cut height h-z is preferred for an optimum lead shaping trajectory. Analyzing equation 2, z becomes minimum by minimizing L, e and d while maximizing T. To achieve optimum values for these parameters, the following tradeoffs must be considered:
Lead ShapingAfter cutting the lead at cut height h-z, the shaping process takes place. The tool moves with the search angle a from the cut position towards the bonding pad with an optimized shaping trajectory. With an air gap of a few microns between the chip surface and the lower side of the lead, a horizontal overtravel moves forwards and backwards (from A to B). This travelling is also called dragging. The reason for this dragging is to complete the bending of the lead and to release stress, induced by the shaping process in the lead.The bonded leads must have a lazy "S" shape3 to accommodate flexure in the longitudinal and transverse directions of the lead without fatigue during power and temperature cycling. Long-term reliability of the package is reported by Tessera in 1, 3 and 5. Kinks and sharp edges along the lead would provoke long-term stability failures. Therefore, the lead shape has to be as smooth as possible. A lead is shaped perfectly when the global minimum of the radius of curvature is maximized along the whole lead. Lead and Chip GeometryThe design of lead and chip geometry restricts the degree of freedom for shaping the lead. The lead has to be broken at the edge of the tool which is placed at point A (Figure 2). Then this lead part has to be guided to point B, where it is finally bonded. The lead must be gripped by the tool during the entire shaping trajectory. This can only be achieved by a specially-designed shaping trajectory and a suitable µBGA tool. If this fails, the lead slips and leaves behind scratch marks on the tail as illustrated in Figure 4. The length of the bonded lead between bond and elastomer becomes shorter if the tool slips with a drawback on lead shape and pull force.Cut point and the bond pad at should be designed in one line with the symmetry axis of the unbonded lead. Because of fabrication tolerances of the tape substrate material and placement variations of the die bonding process (3 <=12µm achievable on a sophisticated die bonder), the lead also must be corrected perpendicular to the lead direction to ensure accurate placement of the bond on the bond pad. Any perpendicular bending is carried out automatically by the lead bonder and is limited only by the long-term mechanical properties of the lead under thermal and power cycling stress.The shaping angle a (Figure 1) can be as large as 60 degrees, with horizontal travel distances one to two orders of magnitude shorter than conventional wire bonding. This shaping implies that an error in z height estimation will result in a twofold placement error in the xy plane. Therefore, the z height must be very accurately estimated and the trajectory has to be travelled with very high precision.4 Bond FormationAfter dragging, the lead is bonded onto the bonding pad. The metallurgical interconnection is established using the same thermosonic bonding principles employed in conventional wire bonding. The ultrasonic power is on the same order of magnitude as wire bonding. Therefore bond formation experience from sophisticated wire bonding equipment can be exploited.Optimum lead deformation is between and of the lead thickness, which guarantees that the neck of the bond is not over welded (Figure 2). Overwelding would result in "neck break" grading, either in pull tests or during heat and power cycling lifetime tests.5 The strategy for optimizing bond formation is to maximize the pull force together with a minimum lead deformation by adapting impact force, bond force, ultrasonic power and temperature. The bonded leads can be tested destructively by conventional pull testing. The pull values change with variations in the loop shape, resulting in different effective pull forces along the tested lead. This effect is well known from wire bonding, if pull forces are compared between high standard loops and flat loops with a small loop height.
ConclusionsDevice geometry and tape material layout lead to close boundary conditions in the lead bonding process, limiting the achievable shape of the bent lead. Relevant geometrical parameters are discussed and arguments are given which lead to an optimum tape design for lead bonding.Compared to conventional wire bonding, lead bonding deals with much shorter travel distances, requiring very small trajectory deviations, especially during looping. High precision bonding equipment, as well, as a controlled bonding process have to be used for satisfactory results. AcknowledgmentsWe would like to thank Hamid Eslampour and Flynn Carson of Tessera. We also thank the ESEC and Zevatech lead bonder team who made it possible to develop an ESEC bonder in a short period of time.References
Dr. Daniel Bolliger has been a Process Engineer in the process R&D wire bonder department at ESEC SA since 1996. In 1995, he received his Ph.D. in natural sciences from the Swiss Federal Institute of Technology.
Christian Brändli has been a Product Specialist in the product management wire bonder department at ESEC SA since 1995. He studied mechanical engineering, with majors in combustion engines and mechatronics, at the Swiss Federal Institute of Technology, Zurich, where he received his diploma. |
Chip Scale Review o 7291 Coronado Drive, Suite 8 o San Jose, CA 95129 o Email: editor@chipscalereview.com
| © 1998 ChipScale REVIEW |