SMT in Line
www.smtinline.com

Technology Drives Process Solutions
BY Brian J. Lewis and Paul Houston

Ultra-small footprint passives, such as 0201 components, are a hot topic in the electronics industry. Existing as a compliment to high input/output (I/O) devices, such as chip scale packages (CSP) and flip chip technologies1, these components are needed for electronic package miniaturization. Figure 1 references the size of an 0201 as compared to an 0805, 0603, an ant and a matchstick. Dimensions of 0.02 x 0.01" make these components ideal for high-density packaging when used in conjunction with other technologies. This article will cover a comprehensive view of the work that has been published, highlights aspects of board design guidelines and defines process windows for printing, placing and reflow. It also covers a project investigating circuit board design parameters, process limitations, and process guidelines to produce a robust process window and board design layout. Project aspects are discussed and tentative data is given, but as the project is ongoing, final data compilation is yet to be published.

Driving Forces
The consumer electronics industry, driven by the growing number of people carrying cell phones, pagers and personal assistants, recently exploded. The need to become smaller, faster and cheaper drives a never-ending need to research technologies that promote miniaturization. Most cell phone related manufacturers implement 0201 into their latest designs and, in the near future, other industry areas will adopt the technology. Wireless communication products in the automotive industry use 0201 technology in global positioning systems (GPS), sensors and communication devices. Additionally, companies use 0201 technology in multichip modules (MCM) to reduce overall package size. With these MCM devices, 0201 technology has moved closer to the semiconductor industry by being packaged directly with bare die and molded into packages for level-2 board assembly. Much research must be done to define pad designs and the printing, placement, and reflow process windows to achieve high first-pass yield and high throughput before full 0201 implementation can occur.

Board Design Guidelines
There have been several studies investigating board design guidelines for 0201 passives. Most looked at designs with varying pad size, pad geometry, pad-to-pad pitch and chip-to-chip pitch. Important design aspects included defect minimization and increasing component density while shrinking overall printed circuit board (PCB) dimensions. The following are the main defects that can be affected by pad design:

1. Tombstoning — This defect occurs when a component raises on end because of forces generated during solder reflow. Generally, tombstoning occurs when a component is placed unevenly across its corresponding pads and the solder surface energy on one end is greater than the other. This difference in surface energy causes more torque on one side that pulls the opposite side up and off the bond pad. Components that are smaller than an 0603 have more potential to tombstone than larger passive components. For 0402 and 0201 components, the pad design can decrease or even prevent tombstoning. Extending the pad laterally and reducing it longitudinally can reduce longitudinal forces that cause tombstoning. The reflow process also can affect tombstoning defects. By having too large a ramp rate, a passive's leading side entering the reflow zone may liquefy before the other and pull the component up.
2. Solder beading — Spherical solder balls are a process indicator that are attached to the passive component by the flux used in the solder paste and usually are located at the component base. Solder beads, commonly seen when no-clean solder paste is used because of flux residue and the lack of a cleaning step normally used with other paste types, indicates that the process is drifting out of the process window. Typically, beading occurs because of pads located too closely together, excessively large bond pads and excessive solder paste deposits printed on individual pads. Placing 0201 passives at high speeds can cause paste to splatter out of the solder brick. This splattered paste reflows around the passive, causing solder balls that would be defined as a defect by IPC 610. These are the most common defects that occur with ultra-small passive components. As mentioned above, design guidelines can be used to control these types of defects, as well as understanding the process window. Some recommend that 0201 pads be designed to limit the solder paste contact angle on the passive's longitudinal ends, while elongating the pad's lateral dimensions, allowing for a greater contact angle.^2,3,4 Forces associated with this pad design will tend to act toward the sides of the component, allowing for more self-alignment while decreasing the forces that cause tombstoning.

Pad spacing also may control solder balling defects. Studies have shown that pad center-to-center should be between 0.020 and 0.022" with an edge-to-edge spacing of approximately 0.008 to 0.010". The pad should be designed for placement tool accuracy. Additional studies have shown that with passives, restoration forces are greater along the longitudinal axis, but if the component is placed with a longitudinal offset, the component must be in contact with both bond pads, ensuring two distinct forces for self-alignment. Therefore, if the placement machine only has an accuracy of 0.006" and places the 0201 with too large an offset, the components will not self-align. Pad sizes and designs recommended for decreasing tombstoning and solder beading are listed in Table 1.

Unfortunately, little published data exists to explain where the process window lies with respect to other board design variables, especially with respect to chip-to-chip pitch limitations. Chip pitch can be affected by various factors, such as board placement and 0201 component placement. To understand the design guideline process window, an extremely comprehensive study currently is underway.* The board used for this study is shown in Figure 2. The design includes various pad sizes, chip orientation (0°, 90° and ±45°), chip-to-chip pitch (0.004, 0.005, 0.006, 0.008, 0.010 and 0.012"), and trace thickness into the bond pad (0.003, 0.004 and 0.005"). The nominal pad size for the 0201 is 0.012 x 0.013" and is altered by ±10, 20 and 30 percent. Pad-to-pad spacing is 0.022". 0201 components are placed next to other 0201s, 0402s, 0603s, 0805s and 1206s with the chip pitchs mentioned above. The trace feed thickness is altered and there are areas where one of the two pads, for both 0201s and 0402s, are located on a ground plane. This investigates heat-sinking effects on the passive component yield.

Printing
Many of the same problems and rules that exist with printing advanced technologies, such as CSP, microBGA and flip chip, are equally important with 0201 component printing. With apertures that are many times smaller than other board devices, printing with thicker stencils and some pastes nearly is impossible. Commonly asked questions concerning 0201 processing involve stencil thickness, aperture size, paste type and required aperture geometry.

Currently, work is being done to understand how paste releases from various aperture sizes and geometries using different stencil thickness. One major aspect the project investigates is the importance of area ratio in determining robust printing windows. Area ratios are calculated by dividing the aperture cross-sectional area by the aperture wall area. Previous research5 indicates that area ratios provide a much higher resolution in determining robust process windows than stencil-wide aperture reduction methods, such as aspect ratios. The research determined that area ratios of approximately 0.6 and higher deposited solder volumes that closely matched the total aperture volume.

Tentative screening data from the ongoing study showed that for a 0.005" stencil, a pad size of 0.096 x 0.0104" with a 0.49 area ratio provided poor paste release from the stencil for the Type 4 eutectic Sn/Pb solder paste. However, an aperture size of 0.0108 x 0.0117" with an area ratio of 0.56 provided considerably better solder volume and paste release. While there is not much published data in regard to the screen printing process for 0201s, what does exist is vague, stating only that thinner stencils are better and Type 4 solder paste (smaller particle size than Type 3) works better in terms of paste release. However, because Type 4 is not as thick as Type 3, using Type 4 may not be viable for printing other SMT components because of material slumping. The study was established to understand the effects of printing parameters, stencil parameters and paste dependent variables. These variables used for the screen printing process are detailed in Table 2.

Understanding and correlating variables such as paste deposition volume and solder beading defects require many placements and iterations. Understanding how paste behaves with apertures that may be as small as 0.010" is important in controlling and implementing the 0201 print process.

Placement Placement may be considered the most important cell in the entire 0201 process. Because the placement system extracts 0201 passives from a feeding medium, visions and places the components accurately, care must be taken when setting up this process. Essentially, four separate operations are involved with the 0201 placement sequence:

• The first is picking components from a feeding medium. Most commonly, 0201 passives are packaged in paper tape-and-reel. These 0201 tape-and-reel devices typically are 8 mm wide with pockets buried in the tape for the components. Figure 3 details how the components are picked from the tape. Attention to detail is necessary when setting up the pickup process. Because 0201 has been an active part of the SMT process only since 1999, tolerance problems in creating the components and feeding tape remain an issue. Although they seem to be tightly packed in the tape, at the micron level, there actually is a considerable looseness. With nozzles that are almost the same size as the components, the chance for missed picks can be high. For this reason, nozzles usually are fabricated slightly larger than the component.
• Once components are picked up, a vacuum check determines component presence or absence. This is an important aspect of the check because if a component is not present, the placement head must dispose of a defective 0201 or repick the component. Pickup errors generally do not directly affect the actual process, but do affect the overall processing time and throughput. The current study also evaluates the differences in tape-and-reel, surf tape and, eventually, bulk-case feeding.
• Once component presence is confirmed via vacuum check, the vision system aligns the component to the board. Sophisticated vision systems can perform an outline measurement of the device or vision the two component leads. To do this, the vision system determines if the component is attached incorrectly to the nozzle or if it is out of the tolerance established for reliable components and placement. If the component is outside the tolerances, it is rejected.
• The final sequence is 0201 placement into the bond pad solder. Although this process must be done quickly, it also must be done accurately to ensure that the component is placed completely on the individual pads. If the component is not placed accurately, the chance for defects, such as tombstoning or solder bridging of components placed in a close array, increases dramatically. Placement system accuracy also should be taken into account when considering the minimum chip pitch allowed for boards designed with 0201 components. Figure 4 shows the minimum pitch that should be used based on placement accuracy. For example, if a placement system has an accuracy of ±45 µm, a minimum pitch of approximately 90 µm should be set.

Placement force and speed also are important placement process aspects. Because every machine is different, a characterization must be done to ensure the speed is high enough to keep solder from splattering from the paste brick and the force used does not overly press the component into the paste. If too high a force or too fast a speed is used, possibilities for solder balling or component skew increase.

The placement project evaluates speed, force and accuracy limits. Generally, speed and force will be machine dependent, but with accuracy, the physics behind solder wetting and self-alignment forces are not machine dependent and, therefore, will be consistent from platform to platform. Data shows that if the pad designs mentioned earlier are used, placement offsets in the longitudinal direction will allow for more self-alignment than offsets in the lateral direction. Excessive offsets in the longitudinal direction produce more defects than those in the lateral direction. Post-reflow component skewing is a more common defect caused by lateral offsets.

Reflow
Reflowing 0201 passives is not much different than reflowing slightly larger 0402s; however, ramp rates used in the 0201 reflow process must be watched carefully. Processing 0201 passives with a high preheat slope can increase the chance of tombstoning. Slopes greater than 2°C per minute can cause solder paste on one end of a component to reflow slightly faster than the other end. If one side of the component reflows first, unequal forces will act on the component, lifting it in the direction of the first reflowed pad because of the solder surface tension.

Another reflow process aspect is air vs. nitrogen use. Sources have shown that using air can reduce most defects.^4,5 Because solder wets better in nitrogen than in air, using an air environment reduces wetting and allows time for both sides of the component to wet more uniformly. The study takes the variables shown in Table 3 and parametrically changes them over three levels to understand their effects on solder joint quality and other process related defects.

Conclusion
Entirely understanding the 0201 process is unlikely. In upcoming years, there will be countless studies conducted examining the 0201 process with the goal of understanding all possible process and design considerations. Because there are so many variables in each process cell that are dependent on other factors, it is difficult to find one answer that will satisfy all possible process questions. As SMT processing advances, new innovations will require research to find process windows and appropriate board designs.

*Joint project conducted by Siemens and Georgia Institute of Technology.

REFERENCES
1. J.H. Lau, Flip Chip Technologies, McGraw-Hill, 1995.
2. Internal Siemens preliminary research results.
3. M. Yuen, H. Benedict, K. Havlovitz and T. Pitsch, Plexus Electronic Assembly Corp., and A.C. Mackie, Praxair Inc., "Tombstoning of 0402 and 0201 Components: A Study Examining the Effects of Various Process and Design Parameters on Ultra-small Passive Devices," referenced www.plexus.com.
4. J.H. Adriance and J.D. Schake, "Mass Reflow Assembly of 0201 Components," APEX 2000 Proceedings, P-EQ4/3-1, April 2000.
5. G.D. Rodriguez, "Analysis of the Solder Paste Release in Fine-pitch Stencil Printing Processes," Georgia Institute of Technology Master's Thesis, June 1998.

BRIAN J. LEWIS may be contacted at Siemens EAE, 2875 Northwoods Pkwy., Norcross, GA 30072; (404) 894-6351; Fax: (404) 875-0668; E-mail: brian.lewis@eae.siemens.com. Paul Houston may be contacted at the Manufacturing Research Center at the Georgia Institute of Technology; (404) 894-2793; E-mail: paul.houston@marc.gatech.edu.