SI CHINA     SI JAPAN
Login  |  Register          Free Newsletter Subscription
Subscribe
Email
Print
Reprint
Learn RSS

Factors for Successful Wafer-Level Solder Ball Placement

Chip-scale packages are one of the technologies demanding rapid increases in capacity for placing solder balls onto wafers, strips and substrates. Identifying the critical success factors for suitable processes helps to establish a set of best practices to ensure throughput, yield and cost-per-placement targets are met.

David Foggie and Jens Katschke, DEK International, Weymouth, UK -- Semiconductor International, 12/1/2007

Rapid increases in the performance and density of processors, memories, digital signal processors and a host of application-specific devices are being driven by market demands placed on handheld, personal computing and Internet devices. Advanced packaging technologies, such as wafer-level chip-scale package (WL-CSP) and package-on-package (PoP) technologies, are key to meeting these performance and density targets.

New solder ball placement techniques deliver speed, precision and repeatability for advanced packages.
Area-array interconnects created by attaching solder spheres directly to the substrate or wafer are fundamental to these technologies. Steadily increasing I/O density is driving up the number of solder spheres per device, and market data suggests that assembling a typical mobile processor package will require more than 1000 sphere attachments before 2010. Package assemblers need cost-effective, fast and accurate techniques for solder sphere attachment to satisfy this sheer volume of placements.

High-accuracy mass-imaging processes can place solder balls from 750 μm for conventional ball grid array (BGA) packages down to 250 μm for WL-CSP, at pitches as fine as 0.4 mm. Mass imaging can be used to place solder spheres onto wafers up to 300 mm, micro-BGA strips or singulated substrates, covering the entire range of chip-scale packages (CSPs) available today.

A globally installed base of active solder sphere attachment processes serves leading-edge diameters and pitch dimensions. Critical success factors governing process performance can now be identified from practical experience in high-volume assembly. Factors are identified and associated with the placement platform, solder ball transfer head, wafer or substrate support mechanism and compatibility with the preferred wafer or substrate transport medium, including FOUPs, cassettes or Auer boats. For example, the loading and unloading of wafers may require individual optimization to ensure interoperability with a chosen brand of FOUP or type of cassette. Correct optimization of associated tooling, including the emulsion-screen/stencil combination and precision shims for sphere placement onto wafers, affects the performance of the process.

Ball placement

Solder sphere attachment using high-accuracy mass imaging is a two-stage process. A preliminary fluxing process, performed using a precision screen-printing platform, deposits flux at the interconnect sites via an emulsion screen. Following screen-printing practices used in today's most advanced print processes, the required volume of flux can be deposited at each interconnect site with uniform thickness. The emulsion screen maintains a tight seal against the surface of the wafer or substrate to prevent smearing of the flux. Hundreds of cycles can be completed without cleaning the screen, allowing a high beat rate to be maintained for maximum productivity.

Fluxing is followed by attachment of solder spheres. The solder ball placement platform is similar in operation to the fluxing machine, and uses a dual-layer tool (metal stencil) to deposit a sphere at each fluxed under-bump metallization (UBM) site. A purpose-designed solder ball transfer head then directs spheres continuously to the surface of the stencil through internally machined channels. The transfer head maintains a reservoir of spheres that are usually sufficient for an hour of continuous operation. The transfer head is driven at a constant speed across the active area of the printer in direct contact with its surface. Positive placement forces a single solder sphere through each aperture. High-volume processes have demonstrated high yield rates, with <0.01% unpopulated sites or damaged spheres.

When completed, the wafer or substrate is unloaded from the screen printer; this may be performed manually or using a custom automated handling solution. Subsequent inspection, followed by reflowing of the solder spheres, completes the attachment process. Figure 1 shows a sample configuration of an automated fluxing, solder ball placement, inspection and reflow sequence.

1. The use of inline mass-imaging platforms for fluxing and solder ball placement, combined with integrated handling equipment, allows the automated production flow to be managed over industry-standard SMEMA protocols.

Critical factors for ball placement

The machine must first be set up correctly to ensure accurate execution of the process, meeting the optimum throughput and yield. This may seem obvious, but particular attention must be paid to the machine to ensure it is fully and correctly calibrated and meets the necessary tolerances of the process. During the machine build process, each element is checked to ensure parallelism of the transport rails, stencil and ball placement head throughout the complete cycle. This prevents solder balls from escaping at the boundaries between the stencil and ball transfer head. Solder ball escape is the most common failure mode among modern ball-attachment processes.

Best working practices established for setup and maintenance of the ball transfer head also bear out this experience. Among established procedures for replenishing solder spheres and managing ball placement equipment generally, training for technicians emphasizes the importance of regular inspection of the head's skirts and internally positioned distribution pad. Figure 2 shows a cross-section of the transfer head and illustrates these components.

2. The purpose-designed solder ball placement head directs solder balls to the stencil apertures and applies a controlled placement force; maintaining the Nylacast skirt in good condition prevents ball escapes.

Correct calibration of the stencil proximity sensors, which are embedded in the transfer head (Fig. 3), is also critical. These are inductive sensors that monitor proximity of the stencil to ensure constant and continuous contact between the head and stencil. Their calibration must be performed accurately at the time the equipment and process are commissioned, then routinely verified throughout the life of the process. Regular inspection of the condition of the transfer head, during normal operation of the process, is advisable to identify any damage to the head's skirt or internal distribution device that may promote the escape of solder spheres.

3. Integrated solder ball escape sensors use inductive technologies to ensure uniform proximity between the solder ball transfer head and stencil surface.

Optimized stencil set

Fluxing, prior to solder ball placement, serves two functions by providing a way to retain the solder ball in position immediately after placement and ensure formation of a satisfactory solder joint during post-placement reflow. A precision emulsion screen is the optimal vehicle for flux deposition. Manufactured to correspond with coordinates captured from CAD or Gerber data, ideally under cleanroom conditions, the screen has high dimensional stability throughout its lifetime. High flux-volume repeatability and positional accuracy allow fluxing of UBMs for solder balls as small as 250 μm on a grid of 400 μm pitch.

The ball placement stencil is a two-layer composite component comprising a stainless steel or nickel stencil with a laminated hold-off layer that prevents contamination of the pre-deposited flux. The hold-off layer is created from standard photoresist, etched at the stencil aperture sites. Figure 4 shows a simplified cross-section.

4. The combined thickness of the stencil's metal layer and laminated stand-off layer is closely related to the solder ball diameter to optimize the process for maximum yield.

Numerous techniques are available for producing the metal stencil layer; selection depends on the solder ball dimensions and the number of balls to be deposited in a given area. For most applications, the stencil is cut from precision-rolled stainless steel stock using a CAD-driven laser. Advantages include high accuracy and stability in terms of the stencil thickness, as well as short production cycle times using laser cutting.

On the other hand, an alternative process may deliver a more desirable result when large numbers of apertures are required (up to 100-150,000 spheres/wafer). A typical threshold is roughly 15,000 apertures for a wafer-level stencil, above which an electro-formed or chemical-etched stencil may be preferable, depending on solder sphere diameter. We generated a stencil and screen design tool as an aid to optimal stencil technology selection and aperture design. This embeds several years' experience of creating stencils for solder ball placement to automatically calculate aperture shapes and dimensions for the types of deposit typically required. More than 40 parameters are evaluated in the design of a solder ball placement stencil. The calculator is used in conjunction with CAD conversion software, which converts the electronic GDS-II, .dxf or .dwg file to Gerber format.

When placing solder spheres onto wafers, the stencil is used in combination with a shim to maintain stability throughout the excursion of the transfer head. The shim is inserted beneath the stencil and surrounds the wafer. This enables package assemblers to maintain one shim per wafer type that can be changed when switching to a different wafer diameter or thickness. This approach allows one wafer pallet to be reused with a large number of wafers of varying diameters and thickness. Figure 5 illustrates the wafer pallet and corresponding shim.

5. The precision shim (left) carries an aperture corresponding to the wafer diameter, while the vacuum pallet (right) features concentric wafer vacuum channels, moveable pins to raise the wafer, and an outer vacuum channel to secure the shim.

Shims are manufactured in a range of standard gauges, increasing in 50 µm steps, and may be laser-cut from precision-rolled stainless steel stock or electro-formed in pure nickel. Shims may also be produced using a chemical etching process. The optimum shim for a given wafer type is defined according to the nearest standard thickness below the minimum thickness of the wafer. The shim aperture is created to match the diameter of the wafer being processed.

Handling, support

A wafer's thickness has a powerful bearing on the design of the optimal wafer support technology, in addition to influencing shim design and selection. A proprietary wafer pallet featuring machined vacuum grooves is typically the wafer support solution of choice. Increasingly, as emerging package technologies, such as stacked die and PoP, demand reduced wafer thickness for a low total profile, wafer-backgrinding processes are presenting progressively thinner wafers for solder ball attachment. These wafers can be distorted when the vacuum is applied.

Wafer backgrinding to thicknesses <100 μm is now achievable. This demands that package assemblers take special care to verify and optimize the characteristics of the wafer pallet. Distortion in ultrathin wafers is a major challenge to all companies currently developing wafer-level processes. For solder ball placement using mass imaging, custom wafer pallets are a potential solution, and may involve the redesign of vacuum channels to minimize wafer distortion. Another alternative is to use a porous aluminum wafer pallet, which can be machined to minimize distortion of the wafer while ensuring flatness and coplanarity to achieve successful solder ball placement.

Support for singulated substrates is more standardized, and available solutions can be used directly with the ball placement platform. These include tooling capable of aligning multiple substrates simultaneously, raising all substrates from a standard carrier to the same Z height to be populated with solder balls. Other tooling solutions, for example, present a single substrate for placement, allowing the placement platform to perform a full fiducial alignment before populating the substrate individually.

Specialized tooling is used for solder ball placement with substrates in forms such as micro-BGA strips. Placement is more challenging, because coplanarity does not match that of a wafer-level process. The roadmap for solder ball placement with micro-BGA strips encompasses minimum sphere diameters of 200 μm, at minimum pitch 300 μm and between 10,000–20,000 solder spheres per strip.

Factors for maximum returns

Wafer-level packaging (WLP) trends indicate that total solder sphere attachment capacity must increase dramatically in the short-to-medium term. At the same time, relentless pressure to meet competitive price points for consumer products demands a low overall cost per placement. Semiconductor OEMs and non-captive packaging specialists have adopted high-accuracy mass imaging as a means of attaching large numbers of spheres at high speed, yield and repeatability. Suitable process-hosting equipment can be relatively quickly changed-over to perform other assembly processes at the component or wafer level, which allows companies to amortize capital expenditure costs and realize cost-of-ownership savings. Other applicable processes can include wafer bumping, wafer backside coating, no-flow underfilling or transfer molding.

Identifying the critical success factors for solder ball placement based on mass-imaging techniques helps to establish best practices and guides the development of training for operators and in-house process engineers. Experience to date shows that careful attention to these factors reaps the rewards on a robust process able to place solder spheres down to 250 μm on pitches as fine as 400 μm. Individual assistance and training supported by a knowledgeable equipment vendor is mandatory to achieve the best results at these dimensions.


Author Information
David Foggie became manager of the Semiconductor Process Technologies and Alternative Applications Group for DEK in August 2007. His role is to manage the acquisition of new competencies and delivery of innovative solutions. Foggie holds an HND in management, and is currently studying for an MBA.
Jens Katschke has a master's degree in elektrotechnik from the University of Applied Science (Darmstadt, Germany). He is a process engineer within DEK's Semiconductor Process Technologies and Alternative Applications Group, specializing in semiconductor applications for DEK within German-speaking territories.
Email
Print
Reprint
Learn RSS

Talkback

We would love your feedback!

Post a comment

» VIEW ALL TALKBACK THREADS

Related Content

Related Content

 

By This Author

There are no other articles written by this author.

SPONSORED LINKS



 
Advertisement
SPONSORED LINKS

More Content

  • Blogs
  • Podcasts
  • Videos

Blogs

Podcasts

Videos

Advertisements





NEWSLETTERS
Plug in and get the latest SI news, trends and industry updates delivered free, directly to your inbox!

SI NewsBreak and Special Reports (Weekdays)
Wafer Processing Report (Monthly)
Lithography Report (Monthly)
Metrology Report (Monthly)
Clean Processing Report (Monthly)
Packaging Report (Twice Monthly)
©2008 Reed Business Information, a division of Reed Elsevier Inc. All rights reserved.
Use of this Web site is subject to its Terms of Use | Privacy Policy
Please visit these other Reed Business sites