Automation and Advanced Packaging Assembly
IMPROVING THROUGHPUT, QUALITY, & YIELD
BY BRUCE W. HUENERS
The advantages of automation in microelectronic production are well-known: increased throughput, lower costs, and higher quality. Yet to many, the thought of taking a known process, an existing manual manufacturing process, and evaluating the challenges to accommodate all the requirements of automation can be daunting. While change for the better is rarely without some discomfort, the shift to automation can be minimized with the right mindset, the right tools, an understanding of the process, advanced planning, and the right partners (Figure 1).
Figure 1. Assembly area before and after automation.
Automation of precision microelectronic assembly processes eliminates many of the human-induced process variations. In the optimized automated process, process inputs (materials) are placed into the assembly flow and finished products are output with little or no human intervention. This sounds simple enough; however, there are often significant material and process changes, together with a high degree of operator training that will need to take place to accommodate automation.
Although aids are available to assist the operator with accurate placements, any manual manufacturing process is inherently constrained by human limitations. Even with the aid of high-powered microscopes, optical grids, and video targeting, the assisted human hand induces process variations that can make device performance unpredictable, especially at higher frequencies. And with human operators, there is always the possibility of mental error due to inattention or fatigue.
Under the manual assembly model, operators performing a eutectic die attach process learn to place components that are extremely small, on the order of 25 mils wide by 150 mils long, and roughly 10-mils thick (by comparison, a human hair is approximately 3-mils thick). Aspect ratios on these components can be as high as 10:1. For example, die placement is the process step upstream to wire bonding. Any process variations in die placement will require adjustments at the wire-bonding step. If die are placed further apart in one package, a manual wire bonder operator can make a longer wire so that components are connected electrically. Manual wire bonder operators are trained and often encouraged to make these types of process adjustments to compensate for inconsistencies in upstream processes. Although these adjustments are well-intentioned, and in most cases necessary, they will lead to inconsistent device performance from part to part.
As frequencies increase above 2 GHz, each individual wire increasingly becomes an individual circuit subject to tuning. Wires connecting die, capacitors, and package are formed into loop shapes that may critically affect electrical characteristics, such as output linearity, bandwidth, and impedance. With these wires being placed as close as 5 mils from center to center, it is imperative that they are straight and parallel because in these small dimensions, cross-coupling of the electric and magnetic fields to adjacent wires is far more pronounced and modifies each wire’s performance (Figure 2).
Figure 2. Wires made on an automated wire bonder show consistent loop heights and lengths.
Process engineers not schooled in automated processes need to become conversant with the precise demands of implementing a mechanized microlectronic assembly process. Some process variations routinely accommodated by human operators, such as variations in substrate metallization, alternate die types, and bond wire placement, need to become near constants in automation. In addition, some inputs to the process may not be suited for automation. For example, in the die placement process, a human operator or an automated bonder must be shown where to place each component. In the manual process, almost any metallization pattern on the substrate could be used to guide the operator to the placement location. In the automated process, however, accurate die placement usually requires a repeatable fiducial or etch pattern that guides an automatic bonder’s vision system to lock onto a position and place all components according to its programmed sequence, with each placement relative to a registration or datum point marked by the fiducial. It may be necessary to revise the package design to include these repeatable registration points or fiducials on the substrate surface. These added fiducials in the package may add to its cost; however, this is one of the key elements in “designing for automation,” and nearly every step in the process - including die attach, wire bonding, testing, and capping - will benefit in terms of increased throughput and yield. In other words, by designing for automation at the incipient steps in a process, benefits accrue to all follow-on processes in the form of more predictable results.
Phasing in automation offers more flexibility on the part of the user, and enables gradual redeployability of operators and other resources. Automated microelectronic assembly equipment is complex and requires highly skilled operators and maintenance personnel who understand both process and equipment. Suppliers should have a support infrastructure capable of maintaining 24/7 operations worldwide.
Once the decision to automate has been made, considerations for the design of an automated process should include how the package is presently being manufactured, what processes need to be automated, and an analysis of present needs and future plans. In addition, a detailed description of material types, properties, presentation modes, yielded throughput targets, and capacity, together with utilization metrics, are required up front.
Precise goals should be established for the quality level and repeatability of the mechanized process. Process control indexes Cp and Cpk, as shown in Table 1, are useful tools to establish, improve, and monitor process quality and variation.
Table 1. Process capability indexes.
Design for manufacturability and design for automation are highly interrelated disciplines. Many of the component design rules are the same for both: avoid interlocking parts, avoid nesting of piece parts, use repeatable registration and fiducial marks, use standard fixtures, evaluate cumulative tolerance stack-up, and design for the assembly process to proceed from highest to lowest interconnect temperature. Other considerations include how the package is presently built, the methods, yields and factors affecting the package’s performance, the type of package, the components and materials being used, and the process flow.
Besides basic design and manufacturing considerations, there are also assembly considerations.
- How consistent is the incoming material? Does it match the equipment’s capabilities?
- What adaptations to the process will have to be made to accommodate the materials being used?
- Are there thermal constraints in the process?
- Will the process require temperature ramp-up or cool-down?
- Are there curing and outgassing issues?
- What is the source and format of product build data? Is it CAD download or teach?
Adapting automation requires a comprehensive review and evaluation of individual material suppliers and the entire supply chain and purchasing process.
Figure 3. An automated wire bonder with a machine vision system.
Automated assembly provides the opportunity to optimize a machine’s capability to “see” and locate devices. A high-resolution machine vision system enables accurate placement and alignment equating to improved yield in an automated environment without operation intervention. For example, a machine vision process can achieve 0.1 pixel resolution (~1.4 µm at standard magnification) with 256 gray-scale levels, enabling pre- and post-process inspection (Figure 3).
Automation offers flexibility in material presentation and handling. Piece-part presentation of components for die attach, for example, can be via waffle pack, gel-pack, tape-and-reel, or wafer. Package or substrate presentation can be via a single part that is manually loaded, a multi-up part tray that is batch-loaded by hand, or a process carrier with multi-up parts that is automatically conveyor-loaded and unloaded from a stack of parts in a magazine elevator.
If the plan is to integrate several processes together into a production line, especially if a relatively high mix of products will be manufactured, it is usually necessary to select a common carrier to hold parts as they are transported between process stations. A well-designed common carrier will accommodate all the relevant steps in an automated process flow. Standard magazines or cassettes are available to eliminate human handling of components and additionally facilitate line balancing. Process carriers are commercially available in industry-standard widths of 3.1, 4.3, or 5.4 in., all 11.95-in. long. Custom carriers can be designed; however, they may require specialized magazines, conveyors, and elevators. SMEMA-compatible equipment, conveying product through each process station at a standard height of ~37 in., enables flexibility and scalability of the automation process.
Table 2. RAM stats summary.
A key advantage that automation offers is the capability to “paperlessly” monitor and perform analysis on an assembly process in detail. Monitoring includes ID tracking of components’ use and location and machine reliability, availability, and maintainability (RAM) metrics. RAM statistics include cycle time and capacity data, providing a means for measuring the productivity and reliability of equipment using metrics such as Uptime, MTBF (mean time between failure), MTBA (mean time between assist), MTTA (mean time to assist), and failures. Table 2 shows a portion of the R.A.M. statistics as automatically tracked during an engineering evaluation.
Justifying Process Automation
Return on Investment (ROI). There are several different ways to present this value, and your company’s formula is the one that counts. The basic elements of this calculation are: the cost of the investment and expected payback. Creating this summary forces understanding of the real benefits of automation and helps justify the financial equation. Cost savings will come in many shapes and sizes besides the obvious three - increased throughput, better quality, and reduced labor costs. There are many others such as less rework, higher yield, simplified testing, better field reliability, higher customer satisfaction, increased sales (repeat and new customers), etc.
Table 3. Elements of cost-of-ownership calculation.
To understand the benefit, you will need to quantify the factory floor costs and the costs “above the factory floor” to get an accurate assessment for expected ROI. Table 3 shows some of the key elements in a cost-of-ownership calculation. In high-volume or high value-added production environments, the initial cost of automated equipment is often a relatively small part of the total cost-of-ownership.
Automation of microelectronic assembly processes, especially as these become more precise and complex, provide manufacturers one of the best tools to gain competitive leverage. Automation can improve throughput, increase quality (consistency), and improve yield.
Improved Quality → Higher Yield →
Lower costs → Scalability
Requirements for smaller, more complex parts in many cases render the manual assembly process obsolete. Typically, significant material and process changes and a high degree of operator training need to take place to accommodate automation. Process engineers not schooled in automated processes need to become conversant with the precise demands of implementing a mechanized microlectronic assembly process. Automated microelectronic assembly equipment is complex and requires highly skilled operators and maintenance personnel who understand both process and equipment. Suppliers should have a support infrastructure capable of maintaining 24/7 operations worldwide.
Precise goals should be established for the quality level and repeatability of the mechanized process. Design for manufacturability and design for automation are highly interrelated disciplines. Designing for automation at the incipient steps in a process, provide benefits that accrue to all follow-on processes in the form of more predictable results. Adapting automation requires a comprehensive review and evaluation of individual material suppliers and the entire supply chain and purchasing process.
Quantifying return on investment requires an accurate assessment of the factory floor costs and the costs “above the factory floor”. Equipment selection is a major part of automating a process as complex as precision microelectronic assembly. While change for the better is rarely without some discomfort, the shift to automation can be minimized with the right mindset, the right tools, an understanding of the process, advanced planning, understanding your company’s current manufacturing process, including the expectations for automation, and a careful selection of partners, are some of the important keys to success in process automation.
BRUCE W. HUENERS, vice president of Marketing, may be contacted at Palomar Technologies, 2230 Oak Ridge Way, Vista, CA 92081; 760/931-3600; e-mail: firstname.lastname@example.org.