SMT 101: Adhesives, Epoxies and Dispensing

Estimated reading time: 19 minutes

By Barry Burns and Edward Fisher

The modern surface mount adhesive (SMA) must satisfy numerous performance criteria to ensure its suitability for a particular end user's production process, that it works in an effective manner and that it accounts for the lowest number of board defects, thereby increasing yield.

SMAs were developed in the 1980s specifically for bonding non-through-hole devices to printed circuit boards (PCB). Accordingly, SMAs were designed to provide qualities specific to SMT processing. For example, SMAs must have suitable thixotropic/rheological profiles to permit their application at speeds greater than 40,000 dots per hour (dph) via high-speed dispensing equipment leaving no pad contamination, material stringing or missed dots. Also, once dispensed on board surfaces, the adhesives must provide consistent dot profiles suitable for bonding a range of component sizes. Other requirements for SMAs:

• The adhesive dots suitable for securing very small components such as 0402s and 0603s require strict particle-size control created by specialized blending techniques that ensure good dispersion and prevent nozzle blockage.
• Uncured adhesive dots must have sufficient holding (green) strength to maintain the alignment of larger components to solder pads. Additionally, the uncured material must not absorb moisture from the atmosphere.

Cured SMAs must:

• Be strong enough to hold a range of component shapes and sizes to various substrates including low-stress molding compounds, various soldermasks and fluxes, and organic solder preservative (OSP).
• Maintain bond strength through the wavesolder process, which can subject the material to temperatures as high as 260°C.
• Exhibit excellent on-board electrical characteristics, such as high surface insulation resistance (SIR), and resistance in humid environments to prevent metal electromigration or corrosion, leading to device failure.

High-speed Dispensing RequirementsThe speeds of the newest generation of syringe dispensers, which approach 50,000 dph, together with the recent introduction of high-speed fluid jet dispensing technology, have placed new demands on the performance of SMAs. Many older materials lacked the correct rheology or material-flow characteristics, such as viscosity and thixotropy, required for high-speed dispensing. Hence, their performance fell short when used in the latest equipment.

Figure 1. The trend in product miniaturization naturally has been abetted by the growing demand for small-cased components such as 0402 and 0603.

An adhesive's rheological properties play a major role in both syringe-type dispensing and stencil printing. Thus, finding the correct combination of rheological properties can be the most important part of setting up a repeatable adhesive dispensing system. For their part, SMAs are formulated to permit rapid and controlled dispensing and to form dots of a defined shape. Hence, the materials are engineered to be thixotropic, i.e., viscosity decreases when a shear stress is applied during dispensing, promoting easier flow. At the PCB surface, however, shearing forces are eliminated and the adhesive quickly restructures and recovers its viscosity.

Small Component DispensingMiniaturization has been a major trend in SMT assembly (Figure 1), accompanied by a corresponding increase in the use of 0402 and 0603 devices. Table 1 illustrates the dimensions in mm of typical surface mount devices (SMD). As component size decreases, so too does the target adhesive dot diameter required to bond a component without contaminating solder pads.

The relationship between adhesive dot diameter and the nozzle's internal diameter (ID) typically is 1:2, the nozzle ID being approximately half that of the target dot diameter. Thus, for an 0603 component with a required dot diameter of 0.4 to 0.6 mm, a nozzle with an internal diameter of 0.2 to 0.3 mm is required.

To consistently dispense adhesive through such narrow bore nozzles, SMAs that will not cause nozzle blockage are required. Thus, as an approximate rule of thumb, adhesives dispensed through nozzles should not contain particles larger than one-third the ID of the dispensing nozzle. For example, to reliably dispense through a 25 gauge (260 µm) ID nozzle, the adhesive should have no particles greater than approximately 90 µm, and also should be processed under conditions designed to prevent aggregation of the primary particles. The latest SMA syringe grades can be dispensed through 0.4 and 0.33 mm ID nozzles without blockage problems.

Green/Wet StrengthGreen or wet strength refers to the uncured adhesive's ability to hold a device in alignment with the solder pads before the curing process. SMA rheology is very important in producing consistent dots with good green strength. An adhesive with poor rheological properties will exhibit lower green strength and tend to form misshapen dots that can string or slump and lose contact with components.

Figure 2. Testing green strength of an adhesive, which must not permit a component's shift by more than 150 µm.

Before passing through the curing oven, PCBs are subjected to the movements of various chip placement machines that deposit SMDs on uncured adhesive dots on the board surface. Many such machines rely on X-Y table movement, subjecting the components and uncured adhesive to various acceleration/deceleration forces. If the adhesive does not have sufficient green strength, such impacts can cause misalignment of component leads with the solder pads.

An adhesive's green strength is tested using the Siemens standard SN59651 (Acceptance of SMD Adhesives), which involves sliding a populated PCB down a fixed incline to a stopper at the base of the ramp (Figure 2). The components' position on the board is noted before and after the test, and displacement is measured. The adhesive must not permit the component to shift by more than 150 µm (corresponding to half the track pitch on a standard 0.3 mm pitch component).

Viscosity Stability and Shelf LifeThe viscosity of an epoxy-based SMA increases over time as the adhesive undergoes very slow polymerization or cure. As this effect is accelerated at higher temperatures, the recommended storage temperatures are between 2° and 8°C to minimize polymerization, maintain product performance and guarantee maximum shelf life.

Figure 3. Bubbles and voids formed on an SMA's surface owing to excessive moisture can cause weak bonds or component loss.

The latest generation of SMAs have been formulated with a stabilizer to further reduce polymerization in refrigerated storage and to make the adhesives robust in viscosity stability. As these products have a refrigerated shelf life of 10 months and a room temperature (22° to 28°C) shelf life of one month, they can be left unrefrigerated on the factory floor (or in the dispensing equipment) for up to 30 days without compromising performance. Line operators also can expect dot-size consistency over that time at room temperature, allowing for more robust and reliable processing.

Moisture ResistanceSMAs must be exposed to as little moisture as possible or else the adhesive dot may boil and "explode" during cure. This phenomenon, called "popcorning," causes voids that weaken the joint and create paths for solder to penetrate under the device, which may cause circuit shorting because of solder bridging. Popcorning typically occurs in the cure oven when the moisture-laden adhesive is exposed to temperatures of 100°C or greater, or when a ramp rate of greater than 1° to 2°C is used during cure.

Figure 5. A lack of adhesion to the low-stress encapsulant on the IC causes the device to fall away from the PCB surface.

Popcorning can manifest itself in several ways: The numerous bubbles or voids that form on the surface of the cured adhesive cause it to take on a honeycomb appearance (Figure 3). These voids reduce the contact area between the adhesive and the underside of the component, which can result in weaker bond strengths or component loss during the manual insertion or wavesoldering steps. Also, moisture can produce voids large enough to form channels through which solder can flow during wave soldering, resulting in solder bridging and component/device failure. Finally, during the preheat cycle, solder flux may penetrate the voids in the cured adhesive dot (Figure 4). Because flux materials are commonly acid-based, they subsequently can cause corrosion or SIR failures.

Figure 4. An example of solder flux penetration into the voids of a cured SMA, caused by moisture uptake. Being acid-based, the flux can cause corrosion or insulation failure.

In a syringe, adhesives typically are exposed to very low moisture levels. However, when there is a long delay between dispensing and curing, the adhesive can be exposed to the prevailing temperature and humidity factors under normal operating conditions, which can vary considerably on a global basis. For example, moisture absorption can be a particular problem with stencil printing and pin transfer because of the large exposed surface area presented by open baths of adhesive. To minimize this problem, most SMAs are formulated with materials that have low moisture pick-up properties, permitting bath stability for up to one week.

Difficult-to-bond Surface AdhesionThe soldermask on some finished boards as well as SMDs can cause difficult bonding. Mask and component finishes can have a low surface tension or contain processing aids that act as releasing agents. Figure 5 illustrates a problem of bonding an integrated circuit (IC) component during manual insertion on the opposite side of the PCB. As shown, the IC device fell away because of a lack of adhesion to the low-stress encapsulant of the IC almost all the adhesive remains on the PCB surface with none on the device. Because the end-user has little or no control in specifying board and component finishes, the latest generation of SMAs exhibits a high affinity to low-stress encapsulants and processing aids used today.

SIR RequirementsSMA's role remains incomplete even after the PCB has been through wave soldering. Since the cured adhesive remains on the board surface for the duration of the device's lifetime, it must not cause corrosion or electrochemical migration when subjected to a high-temperature/high-humidity environment under electrical loading. SIR is just one of many tests that the new generation of SMAs must pass. The test involves coating the adhesive across an interlinking comb pattern of bare copper tracks. The track width, pitch, applied bias voltage, test voltage and adhesive thickness requirements vary by customer as do the environmental test conditions and duration. Typical test conditions are illustrated in Table 2.

ConclusionSome say that SMA chemistry is mature and there will be no more significant advances. But with the introduction of jet dispensers and the increased market demand for high-speed stencil printable SMAs, there always will be a call for faster, "high tech" adhesives. The industry already is calling for dual-cure systems (UV/heat) for simultaneous double-sided reflow. The anticipated arrival of lead-free soldering technology also will impose new performance demands on next generation SMAs. In fact, SMA formulators are waiting for end-users to set future performance criteria. Once the bar has been raised, the technology will be adapted and advanced to meet the newer requirements.

BARRY BURNS, senior chemist at Loctite Ireland, may be contacted at 353-1-404-6551; Fax: 353-1-451-9073; E-mail: barry.burns@henkel.de. EDWARD FISHER, electronics application chemist, may be reached at Loctite Corp., 1001 Trout Brook Crossing, Rocky Hill, CT 06067; (860) 517-5359; Fax: (860) 571-2511; E-mail: ed.fisher@loctite.com; Web site: www.loctite.com.

______________________________

ABCs of SMTAdhesion: The attractive force between materials of dissimilar molecules.

Anisotropic: A quality of a conductive adhesive that permits the flow of electrons in the Z-axis only.

Popcorning: A condition in which micro-fissures occur in the epoxy body of a component or bubbles form on an adhesive layer, caused by excessive moisture within the materials and activated by heat.

Rheology: Describes the flow of a fluid (adhesive) and, hence, its viscosity and surface tension properties.

Slump: The spreading of an adhesive after stencil printing but before reflow, curing or drying.

Surface insulation resistance (SIR): A test to measure, in ohms, an insulating material's electrical resistance between conductors.

Syringe dispensing: Adhesive application via pneumatic or hydraulic pressure to deposit a specific amount of material through a needle to a target site.

Thixotropic: A characteristic of an adhesive that renders it viscous when static, and fluid when stressed.

Viscosity: The measure of an adhesive's resistance to flow.

_____________________________________

Adhesives/Epoxies & DispensingBy Bob Hoffman

Automated dispensing systems are used today in a wide variety of electronics assembly processes. Specific requirements for placement and volume accuracy can be used to describe and define any electronics dispensing application. The bottom line is that application-specific placement and volume accuracy requirements must be consistently achieved within a control process environment to produce acceptable products. Process alternatives and application requirements of epoxy dispensing are provided, enabling an educated decision when choosing between dispensing speed and accuracy.

Automated dispensing systems are used today in a wide variety of electronics assembly processes, from relatively low-tech "industrial" gluing operations through increasingly demanding applications, such as solder paste, SMAs, encapsulants and underfills. Specific requirements for placement accuracy (where the fluid is dispensed) and volume accuracy (how much is dispensed) can be used to describe and define any electronics dispensing application. Whereas accuracy requirements for low-tech dispensing operations might be in the 5 to 20 percent range, more precise applications often can require placement or volumetric accuracy of 1 to 3 percent or better.

Although it is a critical requirement, dispensing accuracy cannot be viewed without regard to other considerations. Generally, dispensing speed and total production throughput also are critical factors, and must be addressed to have a truly viable and sustainable manufacturing process. After first quantifying the basic accuracy capabilities of a dispenser, the user can move on to the other parts of the decision matrix.

By focusing on SMA dispensing, this article explores some specific methods for quantifying the performance of a dispenser's placement and volume accuracy, while balancing overall speed/throughput considerations.

Defining SMA Application RequirementsSMA constitutes a very common dispensing task in microelectronics assembly, used to bond surface mount components to the bottom of a PCB and hold the component in place during the wavesolder operation. The typical definition for a good dispense shot is one that gets enough SMA in place to wet the component's bottom side, sufficient to hold it in place without covering it or the PCB bond pads.

The required SMA volume depends on the physical dimensions of the surface mount component, the proximity of the PCB bond pads and the SMA's adhesion strength. Assuming the component dimensions and placement are repeatable, an upper and lower control limit on the fluid volume that will meet the bonding requirements within worst-case process parameters can be specified.

Figure 1. Example of a DUT sample.

If the original dot is a hemispherical cross section, it has a volume that is easy to calculate based on the diameter of the dot. If the dot is shaped more like a "chocolate kiss," as it typically is when needle dispensed, then the volume often is modeled as a cone with the diameter and height of the dot. Regardless of the dispensed dot shape, it must be tall enough and of sufficient volume to wet the bottom side of the component. As the component is placed, the SMA is squished out to a final shape of something more like a cylinder. Too much SMA will be a problem because the fluid will be pushed onto the bond pads and too little will compromise the adhesion between the component and the substrate. Even if one could dispense the perfect volume dot every time, placement accuracy is key in the successful bonding of the surface mount component.

Measuring Placement AccuracyFor analysis purposes, placement accuracy can be isolated from the application. A straightforward approach is to dispense dots on a glass master or other stable substrate and then use a downward looking camera on a coordinate measurement machine (CMM) to determine dot locations. The substrate should be dimensionally stable and provide a good contrast to the dot material so that the edge of the dot can be determined. Additionally, it is important to take into account the fluid's potential to flow after dispense, especially if the test surfaces have a "grain," such as brushed aluminum or glass reinforced epoxy. The best practice is to use a highly polished surface with a high surface tension, which encourages the dot to stay put rather than run or wick.

Figure 2. 3-D surface profile image of needle-dispensed SMA ? high magnification (a). 3-D surface profile image of needle-dispensed SMA ? low magnification (b). Photo courtesy of Optimet.

The CMM resolution and its attendant optics generally should be an order of magnitude higher than that of the dispensing system. For today's high-end dispensing systems, it is common to find both positioning control and optical resolution in the single digits to the tens of microns. Thus, to measure dispensing accuracy, the coordinate measurement system should have optical and position resolution of approximately 1 to 5 µm.

Basically, the measurement process consists of finding the center of the dot, determining its roundness and comparing the actual center location to the intended location to find the placement error (e.g., X actual - X intended = X error). CMMs with advanced vision algorithms such as edge detection are well suited for acquiring these values and determining placement error quickly and accurately.

The dispensed part, or device under test (DUT), should be fixed to the CMM to ensure that it does not move even slightly during the measurements (Figure 1). Once the DUT coordinate system is established, the operator proceeds to target the individual dots. A table of data points consisting of the X and Y center point location for each dot is developed (Table 1). Spreadsheet comparisons of actual dot locations vs. intended dot locations are analyzed to create a statistical portrait of dispense placement accuracy. While it is common to analyze the error in X and Y separately, it also is possible to use a root mean square calculation to determine the error distance as a single value function of both X and Y.

Test result data interpretation depends on the circumstances of the tests and the interests of the interpreter. For example, the above data suggest that both the X- and Y-axis have approximately the same deviation about the mean. One observer might be interested in reducing the magnitude of the standard deviation of both axes by investigating what it is that contributes to the variation. Another observer may be more interested in eliminating the apparent offset in the Y-axis to improve overall dot placement accuracy.

Typically, a non-zero average placement error is representative of a systemic condition such as imperfect calibration of the needle-to-camera offset, which can be isolated and eliminated.

Measuring Volume AccuracyGenerally, there is a correlation between dot volume and dot diameter, but it is quite possible that two dots with very different volumes could appear to have the same diameter. Therefore a two-dimensional (2-D) inspection system is not adequate for quantifying the volume accuracy of a dispense system.

A conventional method to calculate volume is to measure the weight of the dispensed dot samples and then use the known density of the fluid.

Dispensing a large number of dots on a pre-weighed substrate can provide a reality check by providing a comparison for the dispensed sample, with the difference between the two weights representing the weight of the dots. The average weight per dot can be calculated, but the variability of

the dots within the sample is indeterminate from this test. The technique's ultimate accuracy depends on the precision of the scale and the process used to determine the weight.

Today, advanced metrology systems using scanning lasers and sophisticated surface profilom-etry algorithms rapidly capture and analyze extensive three-dimensional (3-D) data for each dot. These systems offer more accurate methods for tracking volume accuracy across large numbers of dispensed dots, while also providing the granularity to assess dot variability within the sample. This can be especially helpful in identifying and analyzing inherent process inconsistencies or potential volume accuracy changes that result over time from fluid aging or machine variations (Figure 2a-b).

Balancing Speed and Accuracy ConsiderationsIn real-world production environments, it is vitally important to understand the factors that can cause accuracy to be sacrificed at the expense of greater speed, and to strive for a balanced process that delivers sustained throughput levels without undermining placement and volumetric accuracy. By knowing the application requirements and understanding the available process alternatives, the process engineer can make an educated decision regarding the tradeoffs between speed and accuracy (Figure 3).

Figure 3. Balancing process speed and accuracy is key, and can be achieved by understanding application requirements and available process alternatives.

One such decision lies between using contact-based needle dispensing vs. noncontact jet dispensing methods. Maintaining accuracy in contact-based needle dispensing depends critically on consistent board flatness and accurate needle Z-height positioning in relation to the PCB. This is because the dot formation relies on the fluid's surface energy or wetting ability to the substrate, which provides a motive force for the dot to leave the needle tip and attach to the substrate. To produce consistent results, contact-based systems must carefully manage variables such as the degree of warp in the PCB, the state of wear in the dispenser's mechanical standoff, and whether the standoff lands on a trace or pad.

If the needle is too far above the substrate, the bottom of the dot will not touch the substrate, causing the dot to cling to the tip of the needle. If the needle is too close to the substrate, however, the dots are dispensed with an inconsistent, squashed appearance as material builds on the outside of the needle, potentially plugging the tip.

In contact-based dispensing, each dot requires an X-Y move to the dot location, plus a Z-move down to the dot height. A typical dispenser repeats these steps approximately four to six times a second. In contrast, noncontact dispensing uses a jetting technique that fires dots directly onto the substrate as it flies over the PCB at a uniform height of 1 to 3.5 mm above the board, resulting in highly consistent dome-shaped dots. Because the jet head can deliver a per-shot cycle time of as little as 12 milliseconds and no Z-axis movement is required to physically position the dispense head against the PCB, noncontact dispensing can run at nearly the maximum speed of the X-Y positioning systems. By eliminating all wasted vertical motion, repeated height-sensing steps and wetting dwell times that are required in traditional systems, these new noncontact dispensing platforms can achieve sustained throughput rates as high as 40 to 50 thousand dph without compromising accuracy.

ConclusionFor the process engineer, the bottom line is that application-specific requirements for dispensing accuracy cannot be ignored or compromised if product quality and production yield objectives are to be achieved. At the same time, throughput rates must be optimized to achieve required production volumes and overall business objectives. To deliver on these goals, the process engineer must possess a thorough understanding of how to measure dispensing placement and volumetric accuracy, as well as an awareness of the process alternatives available for sustaining both speed and accuracy. SMT

REFERENCES

1. "Inspection Strategies for Process Control," Bob Ries, Circuits Assembly, March 2000.
2. "Adhesives/Epoxies and Dispensing," Dr. Ann Van Den Bosch and Tony DeBarros, SMT Magazine, May 1997.
3. "Determining the Repeatability of Dispensing Samples," Al Lewis, Adhesive & Sealants Industry, September 1996.
4. "Managing the SMT Adhesive Dispensing Process," Lisa Collins and James Klocke, Electronic Packaging & Production, December 1992.
5. "Process Control in Adhesive Dispensing," Alec Babiarz, SMT Magazine, June 1992.
6. "Adhesive Dispensing for Surface Mount Assembly," Alec Babiarz, Circuits Assembly, July 1989.

BOB HOFFMAN, Win3 business development manager, may be contacted at Asymtek, a Nordson Co., 2762 Loker Ave. West, Carlsbad, CA 92008; (760) 431-1919; E-mail: bhoffman@asymtek.com.