QFN DESIGN CONSIDERATIONS TO IMPROVE CLEANING

Transcription

1 QFN DESIGN CONSIDERATIONS TO IMPROVE CLEANING Mike Bixenman, Kyzen Corporation Dale Lee, Plexus Corporation Bill Vuono, Raytheon Corporation Steve Stach, Austin American Technology ABSTRACT: Quad Flat non-leaded (QFN) devices are one of the fastest growing package types in the electronics industry. QFN acceptance in long-life devices exposed to harsh environments is currently limited due to reliability concerns. The large area heat sink associated with the QFN design, floods the perimeter of QFN in the soldering process with flux residue making the QFN one of today s toughest cleaning challenges. Un-cleaned flux residues trapped under the QFN have the potential to be hydroscopic and conductive. Concentrated ionic levels lower dielectric strength and can result in dendrite growth. From a cleaning perspective, many designers have poor insight into factors that assure a cleanable design. Variables such as solder paste, reflow conditions, component placement, component clearance (standoff), cleaning agent and cleaning equipment are important factors. Collaboration between process engineers, assembly designers, solder materials, cleaning agent and cleaning equipment experts can improve integration of the QFN design and assembly. Circuit board design plays an important role when cleaning is required. The purpose of this research is to study QFN ground pad and circuit board solder mask definition strategies. Past studies find that high levels of Z-Axis clearance result in less residue surrounding pads. Removal of solder mask opens flow channels that enable better cleaning. The designed experiment will study QFN cleaning effects and report data findings for current and improved design strategies. and can provide improved heat transfer to keep the IC cooler. The I/O and power and ground connections are typically arranged in one or two rows around the four edges of the device although the pattern can vary significantly from device to device as shown below. Figure 1: Diagram showing differing QFN pad patterns QFN s are more manufacturing friendly than other components because they are easier to handle and less prone to damage than alternative packages with leads or solder balls attached. The Dual Flat No-lead DFN is a cousin of the QFN having SMT leadless interconnects only on two sides of the package. There are two common QFN designs referred to as molded and open cavity. The most common type is the plastic molded package. The second type is an open cavity package and comes with a lid to seal the package. The molded package can be used with signal speeds up to 2-3 GHz. The open cavity package can be used with high frequencies up to 20-25GHz 1. KEY WORDS QFN, MLF, LPCC, QLP, HVQFN, LCC, Electronic Assembly Cleaning, Flux Residue, Electrochemical Migration THE CHALLENGE OF CLEANING QFN s A QFN is a Quad Flat No-lead electronic component package. It is also known in some circles as a Leadframe Chip Scale Package (L-CSP), MicroLeadFrame (MLF), or a MLP, LPCC, QLP, HVQFN and a LCC. The QFN is a poor man s BGA. It is becoming popular as an IC package because it is a small, near chip scale package size, Figure 2: Molded QFN vs Open Air QFN QFN PROBLEMS The two real down sides to using the QFN package is cleaning and rework. Rework requires a lot of heat to melt the solder connecting heat transfer pads to the board via structure. This increases the thermal stress to the board and can limit the rework yield and number of rework cycles.

2 The cleaning challenge is exacerbated by three points of the QFN design. One, there is a lot of flux to deal with compared to other packages. Two, the spacing under the bottom termination is very tight, just a couple of mils. And last, the fluid flow channels that normally form and facilitate rapid cleaning are blocked by the heat sink pad. Solder paste typically contains approximately 10% flux by weight, but by volume the flux comprises nearly 50%. When QFN s are reflowed most of the flux expelled from the molten solder from the heat sink pad(s) accumulates around the I/O pad structures in sufficient volume to seal gaps between component and board with solid flux. inconsistencies in bulk flux or random solder balls may change the RF properties on the surface of the circuit connection such as the dielectric strength, surface resistance, and Q-resonance 5. PROBLEM STATEMENT Traditional electronic interconnects had spacing distances between conductors that posed minimal reliability risks from soldering residues. In many applications, cleaning was not required. As the distance between conductors narrow, the Z- axis of the bottom termination reduces. Gap heights less than 2 mils from the bottom side of the component to the substrate create a vacuum effect during the soldering process. The capillary action of the flux residue flows under the component and in many instances bridges conductors. Figure 3: Flux residue fills open gaps in QFN structures Narrow gaps of this nature filled with flux have been shown in many papers to be very difficult to with conventional cleaning systems 2,3. It has also been reported that conventional batch and inline processes are not completely removing all flux residue from under QFN s especially when soldered with type RE and RO fluxes 4 WHY CLEAN QFN s There are two primary reasons for cleaning a QFN. Leaving flux residues under IC s is a common cause of electrochemical corrosion failure of a circuit. This is especially true with water soluble or other highly activated fluxes because the ionic compounds in the flux can be conductive and hydroscopic. Figure 5: Side View of Flux Residues Bridging Conductors To clean under the QFN bottom termination, long wash time combined with high pressure spray impingement is typically required to remove the residue surrounding the ground pad and leads. Long wash cycle times have the potential to slow production and in some cases create a production bottle neck. Another issue is that many of the batch and inline cleaning machines do not deliver the needed flow and pressure to clean under the QFN bottom termination. The problem is that partially cleaned flux residues remain under many QFN bottom terminations following cleaning, which is a reliability concern. Figure 4: Photo of electrochemical growth under device The second reason has to do with components with interconnect frequencies greater than 1GHz. In these applications, Figure 6: Partially Cleaned Flux Residue Post Cleaning

3 RESEARCH BACKGROUND From a cleaning perspective, many designers have poor insight into factors that assure a cleanable design. Variables such as solder paste, reflow profile, component placement, component clearance (standoff), solder mask definition, pad height, solder paste print thickness in mils, cleaning agent and cleaning equipment are important factors. Collaboration between process engineers, assembly designers, solder materials, cleaning agent and cleaning equipment experts can improve integration of the circuit design and assembly. QFN gap height is the critical factor when cleaning is required. Bixenman & Lee (2011 & 2012) researched the effects of solder mask definition on cleaning under bottom termination components. Solder Mask Defined (SMD) pads have a layer of solder mask printed over the pad termination. On components such as the QFN where the bottom termination is less than 2 mils, heavy flux deposits forms around the ground pad and component leads. Non Solder Mask Defined (NSMD) pads have a trough surrounding the lead terminations. During reflow, the trough around the pad allows air to penetrate under the component, which breaks the vacuum effect. The flux residues tend to be within the troughs around the pad. The bottom termination of the QFN has less of a tendency to be underfilled with flux residues, which makes for easier cleaning. No Solder Mask (NoSM) under the QFN increases the gap height, which also allows air to penetrate and exhaust under the component during reflow. The level of flux residue tends to flow onto the area where the solder mask was removed under the bottom termination. By removing the solder mask under the component, an additional 2-3 mils of standoff height is gained, which improves the ability to clean under the bottom termination. 1. Standard Figure 7: Standard Ground Pad Design 2. Square Pattern Figure 8: Square Pattern Design 3. Slot Pattern RESEARCH PURPOSE The purpose of this research is to study the effects of ground pad solder paste print design and the insertion of exposed via holes within the ground pad in an effort to correlate these design features with cleaning. Four ground pad designs were studied: Figure 9: Slot Pattern Design

4 4. Hexagon Pattern Three levels were studied Via Holes at (0.0135) diameter Figure 10: Hexagon Pattern Design Via hole strategy was designed within the test vehicle. Considering the current ground pad designs, the flux residue tends to push out into the streets surrounding the component pads. The thinking behind via holes was to provide an avenue for the flux residue to drain to the back side of the QFN ground pad. Figure 11 illustrates the design concept with the flux draining down into via holes and not into the streets surrounding the ground pad. The diameter of the drilled via hole is mils. After plating, the actual hole size is in the range of mils. Figure 12: 25 Via Holes Design 2. 9 Via Holes at (0.0135) diameter The via holes were plated after drilling. One of the questions that came out of this study is whether the via holes should be plated. Further discussion of this point can be found in the inferences of the data findings and conclusions. Figure 13: 9 Via Holes Design S t r e e t s 3. No Via Holes Figure 11: Flux Drains Through Via Holes Figure 14: No Vial Holes Design

5 Two Solder Mask defined levels were placed onto the test vehicle. 1. No Solder Mask Defined (NSMD). Figure 17: QFN Test Vehicle Two lead free solder pastes were evaluated. 1. Lead-Free SAC 305 No-Clean 2. Lead-Free SAC 305 Water Soluble The board finish was immersion silver. Figure 15: No Solder Mask Defined Pads 2. No Solder Mask (NoSM). An aqueous engineered electronic assembly cleaning agent designed to be processed in an inline cleaning machine was used to clean the boards. The inline cleaning machine was equipped with progressive energy coherent jets. METHODOLOGY The test vehicles were delivered to the Raytheon Circuit Card Assembly Center of Excellence (CCA-COE) in McKinney, TX. The components were delivered in feeder tubes, which were then transferred to a tape and reel format for compatibility with the placement equipment. Figure 16: No Solder Mask Defined Pads Single and double row QFN components were strategically placed onto the test vehicle. The Design of Experiment called for a combination of four solder stencils and two solder pastes, which would be used for the board fabrication. The solder paste stencils were fabricated from inch thick nano-coated laser cut stencil, with the exception of the Hexagonal Design stencil which was inch thick. The thinner stencil was recommended by the stencil supplier to prevent break-out of the delicate design of this pattern. The three other stencils designs (Standard, Slot, and Square patterns) were already fabricated and delivered, which will result in a slight variation in the final paste volume used. The hexagonal, slot and square stencil pattern designs were optimized to ensure no solder paste was printed into any via holes. The standard stencil design used a typical solder pattern without consideration of via hole locations. All stencil designs were designed for 55 to 60% coverage of the ground pad and 100% (1 to 1) for sign pin pads.

6 Pictures of the stencil designs are below: Figure 21: Hexagonal Stencil Pattern Figure 18: Standard Stencil Pattern Figure 19 Advanced Square Stencil Pattern The two solder pastes used were both a Type V solder paste mesh, for the fine stencil apertures. One paste was a common lead free paste, while the other was a water soluble paste. The solder paste application used an MPM Momentum screen printer using the solder pastes specified in the matrix. The vendor recommended solder paste print parameters (e.g. print speed, print pressure, etc.) were followed. The test vehicles were transferred to the Fuji NXT II M6II high speed pick and place machine for placement of the components. The test vehicles were generally fully populated per the DOE Matrix. In a few cases, the matrix defined unpopulated boards without components as control boards to determine the residues remaining from each stencil design and the overall print quality. The test vehicles were immediately reflowed using the Vitronics Soltec XPM3 convection reflow oven. This oven had twelve temperature zones for solder reflow, and the conveyor was set at 41 IPM. The reflow profile used was a ramp-tospike profile with a peak temperature target of 235 C. The reflow profile was identical for both pastes. Figure 20: Slot Stencil Pattern Figure 22: Ramp to Spike Reflow Profile

7 No touch-up was performed on these assemblies. After reflow, the boards were inspected by X-Ray to determine voiding from the manufacturing process and the stencil/board combinations. Following assembly, a 5DX X-Ray Laminography system from Agilent Technologies (series 5000) was used, followed by a Nikon Metrology XTV160 2-D X-Ray inspection system. The boards were sent from Raytheon to Kyzen Corporation for cleaning. At Kyzen, boards were cleaned in an Inline cleaning machine with coherent jets. RESPONSE VARIABLES The response variables from this study include: 1. 5DX X-Ray Laminography 2. 2-D-X X-Ray Laminography 3. Microscopy of the Via Pattern on the Back Side of the Board 4. Side View Microscopy of QFNs Before and After Cleaning 5. Removal of the QFNs on Uncleaned and Cleaned Boards DATA FINDINGS 1. Agilent 5DX X-Ray Laminography The Agilent Medalist 5DX Series 5000 was used to inspect voiding of the ground pad. For each of the ground plane stencil designs evaluated in this study, a comparison of the standard ground pad on both the NSMD and NoSM components placements was completed to determine if the voiding was different on components with via holes. From the 5DX X-ray, a comparison of the voiding was done on the solid ground pad, 25 via holes and 9 via holes within the ground pad. The 5DX X-ray images indicate that void formations reduced when via holes were placed within the ground plain. A select group of Non-Solder Paste images illustrates a common pattern noted when reviewing the X-ray images. The data finds that the via holes placed within the ground pad reduced voiding (Figure 26). Additionally, the voids were smaller in diameter when placing via holes into the ground plane. Images 23, 24 & 25 show reduced voiding when via holes were placed within the ground pad. The images used came from the Hexagon ground plane pattern using the MLF88 component on NSMD pads. Figure 23: Hexagon Solid Ground Pattern Figure 24: Hexagon 25 Via Hole Ground Pattern Figure 25: Hexagon 9 Via Hole Ground Pattern Ground Plane voids were measured with 5D X-Ray to analyze the variance and to examine differences among level means for one or more of the factors. The MLF124 Double Row averaged roughly 8% fewer voids within the ground plane. The SLOT ground plane design had slightly fewer voids than did the HEXAGON, SQUARE and STANDARD ground plane designs. The NoSM and NSMD defined ground planes averaged 16% fewer voids than did the solid ground planes. The 25 via holes ground planes had 25% fewer voids than ground planes with no via holes and 20% fewer voids than ground planes with 9 via holes.

8 Main Effects Plot for 5D X-Ray Ground Pad Voids Data Means 0.54 Component Ground Plane Pattern 0.48 % Ground Pad Voids Mean MLF124 Double Row MLF88 Singe Row Solder Mask Definition HEXAGON SLOT SQUARE STANDARD Via Holes NoSM NSMD Solid Figure 26: 5D X-ray Ground Plane Analysis of Variance Figure 28: Hexagon 25 Via Hole Ground Pattern 2. 2-D-X X-Ray Laminography Nikon Metrology 2-D-X X-ray and Computed Tomography (CT) inspection system images the internal structure. The 2-D imaging was used to detect and compare voiding. Similar to the 5DX X-ray, the data findings indicate that void formations reduced when via holes were placed within the ground plane. Images 27, 28 & 29 show reduced voiding when via holes were placed within the ground pad. The images used came from the Hexagon ground plane pattern using the MLF88 component on NSMD pads. Figure 29: Hexagon 9 Via Hole Ground Pattern Figure 27: Hexagon Solid Ground Pattern Ground Plane voids were measured with 2D X-Ray to analyze the variance and to examine differences among level means for one or more of the factors. The MLF124 Double Row averaged roughly 8% fewer voids than did the MLF88 Single Row within the ground plane. The SQUARE ground plane design had slightly fewer voids than did the SLOT and STANDARD ground plane designs. The HEXAGON ground plane design average 5-8% more voids than the other ground plane designs. The NoSM and NSMD defined ground planes averaged 16% fewer voids than did the solid ground planes. The 25 via holes ground planes had 20% fewer voids than ground planes with no via holes and 16% fewer voids than ground planes with 9 via holes.

9 Main Effects Plot for 2D Xray Ground Pad Voids Data Means % Ground Pad Voids Mean Component MLF124 Double Row MLF88 Singe Row Solder Mask Definition NoSM NSMD Solid Ground Plane Pattern HEXAGON SLOT SQUARE STANDARD Via Holes Figure 30: 2D X-ray Ground Plane Analysis of Variance 3. Microscopy of the Via Pattern on the Back Side of the Board The via holes placed onto the ground pads were approximately 11 to 12 mil in diameter. The center of the ground plane was copper plated. The purpose of the via holes was for the flux residue to drain to the opposite side of the board. The thinking behind the via holes are as follows: (1) The air supplied through the vent during reflow repels the capillary forces that can form under the component. When capillary forces are attractive, flux residue tends to fluid heavy flux deposits into the streets and around the leads along the ground pad. This underfilling effect creates a very challenging cleaning requirement. (2) Flux residues from the large ground pad drain to an area of the board that is easier to clean. The via holes were inspected to determine: Solder bumps formed on the back side of the board Flux residue drained to the back side of the board Review of the four stencil patterns to determine which pattern provided the best result Figures 31 & 32 provide an example of via holes as inspected on the back side of the board. Appendix A provides an overview of Via Hole images for the four ground patterns. Figure 31: 25 Via Hole Pattern Figure 32: 9 Hole Via Pattern The component placements were number by rows using the following matrix. NSMD NSMD NoSM NoSM NSMD NSMD NoSM NoSM 25 vias vias vias vias vias 5-1 Solid vias vias vias vias vias vias 25 vias 25 vias 25 vias 25 vias vias 25 vias 9 vias 25 vias 9 vias \ vias 9 vias 9 vias 9 vias 9 vias vias 9 vias 9 vias 9 vias 9 vias Solid Solid Solid Solid Solid Figure 33: Component Labeling Matrix 25 vias vias vias vias vias 5-7 Solid vias vias vias vias vias 5-8 Solids 6-8

10 The data findings were analyzed statistically. The main effect plots analyze the variance to examine differences among level means for one or more of the factors. % of Solder Bumps on the Bottom Side of Board % Flux Residue on the Bottom Side of Board M LF 124 Dual Row Main Effects Plot for Formed Solder Bumps Data Means Component Placement MLF88 Singe Row HEXA GO N Figure 34: Solder Bumps Formed in Via Holes Component Placement STA NDA RD Figure 35: Flux Residues Present in the Via Patterns The two factor interaction plots analyze the pairs of factors present in the design. Flux Residue Mean Values per Component Placement Figure 36: % Flux Residues as a Function of the # of Via Holes NoSM Ground Plane Pattern S LO T SQ U A RE Solder Mask Definition Main Effects Plot for Flux Residue Present Data Means Ground Plane Pattern Solder Mask Definition NSM D M LF 124 Dual Row MLF88 Singe Row HEXA GON S LO T SQ U A RE NoSM STA NDA RD NS MD Interaction Plot for Flux Residue Present Data Means Placement Via Holes 9 25 Flux Residue Mean Values Flux Residue Mean Values Interaction Plot for Flux Residue Present Data Means Placement Figure 37: % Flux Residues as a Function of Solder Mask Definition Figure 38: Flux Residue on Component Back Side of Board Solder Bumps Formed Under Component Placement Figure 39: # of Solder Bumps formed on Via Holes as a Function of Ground Pattern Design Interaction Plot for Flux Residue Present Data Means Interaction Plot for Formed Solder Bumps Data Means Placement Solder Mask Definition NoSM NSMD Ground Plane Pattern HEXAGON SLOT SQUARE STANDARD Ground Plane Pattern HEXAGON SLOT SQUARE STANDARD

11 # of Solder Bumps Formed on Via Holes Solder Bumps as a Function of Via Holes Interaction Plot for Formed Solder Bumps Data Means Placement Figure 40: # of Solder Bumps formed on Via Holes as a Function of Solder Mask Definition Figure 41: Solder Bumps formed as a function of Via Holes Present When removing the QFN to inspect for flux residue and ground plane pattern formation, the data finds that 1. Flux residues in the streets was reduced 2. The solder on the ground plane tended to coalesce and fill many of the via holes 3. By draining the flux to the opposite side of the board, the levels of flux residues was reduced in the streets 4. Side View Microscopy of QFNs Before and After Cleaning The level of flux residue under the bottom termination is a function of attractive and repulsive capillary forces. When the Z-Axis (the distance from the board to the bottom of the component) is less than 2 mils, the capillary forces of flux residues attract during reflow. Heavy flux residue deposits accumulate in the streets around the ground pad and next to the components. This attractive force renders a significant level of flux residue that blocks flow channels under the component Interaction Plot for Formed Solder Bumps Data Means Placement Solder Mask Definition NoSM NSMD Via Holes 9 25 during cleaning. To clean, fluid flow, wash pressure and longer wash time is needed to remove all flux residues. Bixenman and Lee (2011 & 2012) studied the effects of solder mask definition on the capillary forces of flux attraction and repulsion. The data found that NSMD (non-solder mask defined) and NoSM (no solder mask) increased component standoff height and repelled capillary forces. The data found that flux residues using these two solder mask definitions was less and easier to clean. For NSMD pads, the flux residues tended to accumulate in the troughs next to the component. On NoSM under the bottom termination, the flux residue flowed away from the soldered pads but did not form heavy deposits. Both NSMD and NoSM rendered less flux residues under the bottom termination and created an easier to clean condition. Based on the solder mask research, both NoSM and NSMD strategies were built into the test vehicle. The QFN component design has a ground pad under the center portion of the component. This large ground pad can pull the component down during reflow. To overcome this design limitation, the test vehicle designed for this research studied four different ground pad designs. Additionally, via holes were placed into some of the ground pads. The purpose of the via holes was to break the flux attractive forces during reflow and provide a path for the flux residues to drain to the bottom side of the board. The gap heights were measured to analyze the variance and to examine differences among level means for one or more of the factors. The data found that the MLF124 Double Row QFN averaged 15-20µm higher standoff gap height than did the MLF88 Single Row QFN. The no solder mask defined pads (NoSM) averaged 15-20µm higher standoff gap height than did the non-solder mask defined pads (NSMD). The SQUARE patterned ground plane had the highest standoff height. No via holes solid ground plane components had higher standoff height than did the ground planes with via holes. Gap Height (µm) Mean MLF124 Double Row MLF88 Singe Row Solder Mask Definition NoSM Main Effects Plot for Gap Height (µm) Data Means Component NSMD HEXAGON 0 Ground Plane Pattern SLOT SQUARE STANDARD Via Holes 6 25

12 Figure 42: Standoff Height Variances The stencil maker recommended a 3 mil solder paste thickness for the Hexagon ground pattern as compared to 4 mil solder paste thickness for the Slot, Square and Standard ground patterns. The data found that solder paste thickness factors into the components standoff height. The data also found that ground pads with 25 pads reduced gap height. component before cleaning. The component gap is greater than 4 mils and no visible flux residue can be seen next to component pads. Interaction Plot for Gap Height (µm) Data Means Gap Height (µm) Mean Ground Plane Design 25 Via Holes 9 Via Holes Solid Figure 45: Standoff Gap Greater than 4 mils Clearance Figure 46 provides a side view illustration of a cleaned component on a part with higher standoff gaps. No flux residues were present after cleaning. 70 HEXAGON SLOT SQUARE Ground Plane Pattern STANDARD Figure 43: Standoff Gap as a Function of Solder Paste Thickness Figure 44 illustrates a side view of one of the QFN components. The standoff gap is estimated to be less than 2 mils. Note how the flux residue underfilled the visible component pads. Figure 46: Illustration of a Cleaned Part with a 4 mil Standoff 5. Removal of the QFNs on Uncleaned and Cleaned Boards The test vehicle contained 48 QFN patterns that were strategically placed onto the test vehicle. Rows 1 & 2 were populated with MLF88 and MLF124 Dual Row components. Each component had 25 via 11 to 12 mil in diameter located within the ground. Half of the components were NSMD and the other half was No-SM defined. Row 3 components had both 25 & 9 via holes within the ground pads. Rows 4 & 5 had 9 via holes within the ground pads. Row 6 had a solid ground pad. The via holes protruded to the back side of the board. Figure 44: Flux Residues Underfill the Bottom Termination The data finds that when the component has a standoff height greater than 2 mils, the flux residues during reflow tend to accumulate next to the soldered pads. The level of residue is far less under the components bottom termination. During cleaning, the residue is easier to reach since the flow channels are not restricted. Figure 44 illustrates a side view of a

13 Figure 49: QFN Removal Method Figure 47: MLF88 Component Placements After removal the component was inspected and imaged. The level of flux residue for each component was graded and placed into Minitab to analyze factor interactions. Not Cleaned Boards For each of the ground pattern designs, the components were removed on uncleaned boards to inspect the level of flux residues around component pads and within the streets surrounding the ground plane. Main effects and interaction plots of the data analyze the variance differences among level means for one or more of the factors. Flux Residues Bridging Pads Figure 48: MLF124 Dual Row Component Placements The QFN components were removed after cleaning on both uncleaned and cleaned boards within the designed experiment. An X-acto knife was used to shear and remove each component (Figure 49). The components with via holes were highly difficult to remove. Some of the via holes were soldered to the back side of the board, which resulted in added shear resistance strength. In many cases, it was not possible to cleanly remove the QFN ground pad on many of the components removed to see the resulting pattern. Figure 50: Example of Flux Residues Bridging Pads

14 Flux Residue Means Main Effects Plot for Flux Residue Bridging Pads Data Means MLF124 Dual Row MLF88 Singe Row HEXAGON Solder Mask Definition NoSM Component NSMD 0 Ground Plane Pattern SLOT SQUARE STANDARD Via Holes 9 25 Regarding the ground plane patterns, the stencil fabricator recommended a 3mil aperture thickness for the Hexagon design and 4 mil thickness for the Slot, Square and Standard ground plane designs. The data findings strongly infer solder paste thickness is one of the factors that correlate to the Z-Axis gap height. The data finds that the 3 mil Hexagon pattern left more flux residues next to the pads and within the streets than did the ground patterns printed with 4 mils of solder paste Interaction Plot for Flux Residue Bridging Pads Data Means Solder Paste Thickness (mils) 3 4 Figure 51: Flux Residue Bridging Pads on Uncleaned Boards Flux Residues in the Streets Flux Residue Mean MLF124 Dual Row MLF88 Singe Row Component Figure 54: Flux Residue Bridging Pads as a Function of Solder Paste Thickness Interaction Plot for Flux Residue in Streets Data Means Figure 52: Example of Flux Residues in Streets Flux Residue Mean Solder Paste Thickness (mils) 3 4 Flux Residue Means Main Effects Plot for Flux Residue in Streets Data Means MLF124 Dual Row MLF88 Singe Row HEXAGON Solder Mask Definition NoSM Component NSMD SLOT SQUARE STANDARD Via Holes Figure 53: Flux Residues in Streets on Uncleaned Boards 0 Ground Plane Pattern Figure 55: Flux Residue in Streets as a Function of Solder Paste Thickness Cleaned Boards MLF124 Dual Row MLF88 Singe Row Component The cleaned boards were processed in an inline cleaning machine using Progressive Energy Dynamics. The coherent spray jets provide high deflective energy for removing flux residues from under bottom termination components. The wash chemistry contact time was 5 minutes in the wash using an engineered aqueous cleaning agent processed at 20% concentration at 65 C. The main effect and interaction plots report the data for the lead-free no-clean flux residue.

15 Main Effects Plot for Flux Residue Next to Pads - Cleaned Boards Data Means Component Ground Plane Pattern Flux Residue Mean MLF124 Dual Row MLF88 Singe Row Solder Mask Definition HEXAGON SLOT SQUARE Via Holes STANDARD NoSM NSMD Figure 56: No-Clean Flux Residue Present Next to Pads on Cleaned Boards Flux Residue Mean Main Effects Plot for Flux Residue in Streets - Cleaned Boards Data Means MLF124 Dual Row NoSM Component Solder Mask Definition MLF88 Singe Row NSMD HEXAGON SQUARE STANDARD Figure 57: No-Clean Flux Residue in Streets on Cleaned Boards The data finds that the level of no-clean flux residues under the bottom termination was very low. Most all of the components were clean. On components where residue was found, it was very minor. On boards soldered with water soluble flux, the data found no evidence of wet residue. The data finds evidence of white ghosting in areas where the flux residue propagated. The higher lead-free temperatures may contribute to the white ghosting effect. The organic acid activators may also be penetrating solder mask finish. Follow on testing is needed to better understand the root causes and reliability effects of white ghosting. 0 Ground Plane Pattern SLOT Via Holes 9 25 Figure 58: White Ghosting on Solder Mask on a Board Soldered with Water Soluble Solder Paste INFERENCES FROM DATA FINDINGS 5DX and 2D-X X-Ray Laminography The addition of via holes within the ground plane resulted in smaller and less voiding. There was not a significant difference found based on the ground plane design. The data findings on the 5-DX and 2-DX X-Rays were relatively consistent. Ground planes with 25 via holes leave behind smaller and fewer voids. Ground planes with 9 via holes leave behind slightly fewer voids as compared with the solid ground pad but the size of the voids were not smaller. Microscopy of the Via Pattern on the Back Side of the Board Placing vias within the ground plane contributes to: 1. Level of flux residues within the QFN streets. On the ground planes with 25 via holes a range of 40-60% of the pad area on the back side of the ground pad had flux residues present. On the ground planes with 9 via holes a range of 15-30% of the back side of the ground pad had flux residues present. The via holes displace a significant level of flux to the back side of the board. 2. The level of flux residue draining to the back side of the boards is slightly greater for No-SM component placements. 3. On the 25 via ground plan, nine of the vias at the center had a tendency to form solder bumps. 4. Solder bumps were formed at a wide range of 10-80% of the vias on the back side of the board. 5. The number of vias formed into solder bumps is scattered based on the ground pad design. 6. The shear strength of the ground pad is strengthened with the addition of via holes.

16 7. The air flow from the via holes during reflow appears to reduce the attractive capillary forces, which results in less flux residue under the bottom termination. 8. The via holes appear to slightly draw the QFN down, which results in a low gap height as compared to components without vias. 9. The via holes appear to improve the cleaning under the bottom termination due to less flux residue present under the component. Side View Microscopy of QFNs Before and After Cleaning When the gap of a component is filled with liquid flux at the moment of reflow, the capillary forces pull the component and substrate together according to the following equation. 8 Equation 1: Interfacial pressure differential (for planes) Δp = 2γ / R Where γ = surface tension; R = radius meniscus Capillary forces can attract or repel depending on the resultant forces of adhesion of, cohesion, and surface tension of molten solder and flux fluids in contact with the part and substrate surfaces. The resultant interfacial pressure is referred to as the capillary force. Capillary forces can work for you and against you. If the radius is negative, as is the case with molten solder, the force is negative or repulsive. If the radius is positive, as is the case with liquid flux the force is attractive. These two forces interact to set the gap under the component. When the flux is not allowed to fill the gap, it s capillary force is 0 and the solder alone sets the gap. These forces are illustrated in Figure 59 & 60 below. channel, which allows cleaning fluids to rapidly penetrate, wet and remove flux residues under the QFN. Figure 60: Attractive Flux Capillary Forces The absence of air flow and exhaust under a components bottom termination creates an attractive flux capillary force. When this condition exists, heavy flux deposits in the streets and next to the component pads. These heavy flux deposits underfill and dam the bottom termination. To clean under the bottom termination, high deflective forces and long wash times are needed to clean under the bottom termination. Solder pad heights less than 2 mils result in this condition. Removal of the QFNs on Uncleaned and Cleaned Boards The level of flux residues in the QFNs bottom termination streets for this study was lower than normal. The average flux residue level in the streets ranged from 22-28%. The factors that contributed to this lower amount are standoff gaps greater than 2mils, solder paste print thickness, the insertion of via holes and air flow and exhaust during reflow. These factors resulted in an easier than normal cleaning condition. The MLF124 double row QFN was slightly easier to clean than was the MLF88 single row QFN. The ground pad under the double row QFN was smaller and the addition of the second row of interconnects resulted in a slightly higher gap height. Of the four ground plane designs, the slot pattern resulted in fewer residues in the streets. Twenty five via holes in the ground pad resulted in fewer flux residues next to the pads. Figure 50: Repulsive Flux Capillary Forces On QFNs with a standoff gap greater than 2 mils, the level of flux residues in the streets and next to the component pads is significantly reduced. Solder pads greater than 2 mils and the addition of via holes allow air flow to penetrate and exhaust during reflow. The air flow under the bottom termination repels flux capillary forces resulting in less flux residue. Less flux residue under the bottom termination creates an open flow The cleanliness of the bottom termination was near perfect. This resulted from the higher gap height, lower levels of flux residues in the streets and pads. The white ghosting issue on boards soldered with lead-free water soluble solder paste appears to be an interaction of the organic acids with the solder mask. Additional research is needed to determine root cause and reliability effects of white ghosting. CONCLUSIONS QFN components are extremely challenging to clean. This is due to the low standoff gap, which creates attractive capillary forces under the bottom termination during reflow. During reflow, the tight gap prevents air flow and exhaust currents.

17 When this condition exists, the capillary forces of attraction allow flux residues to pool in the streets and next to the solder pad connections. Prior research found that NSMD and NoSM solder mask strategies improve the ability for air flow and exhaustion during reflow. On NSMD pads, the flux residue tends to pool in the trough next to the soldered pads. The removal of the solder mask under the QFN bottom termination (NoSM) increases the gap height by roughly 1 mil. Flux residues tend to spread under the component but not in thick masses. This additional gap height improves the ability for air to penetrate and exhaust during reflow. As a result, fewer residues are formed in the streets and next to the soldered pads. This research studied the effects of different solder paste and printed circuit board ground pad designs and the insertion of via holes within the ground pad. The data found that the via holes within the ground pad provide a path for air to penetrate and exhaust during reflow. This air path repels the flux capillary forces, which results in less flux in the streets and next to the solder pads. The slot ground pad pattern design resulted in fewer residues in the streets and next to the solder pads. Additional work is needed to study optimal via hole size and the use of non-plated via holes. The Hexagon ground pattern was printed with 3 mils of solder paste as compared to 4 mils for other three ground pad designs. This resulted in a lower gap height for the Hexagon ground plane. The data findings for the Hexagon design found higher levels of flux residue in the streets. This data point provides strong evidence that the gap height is a critical factor to reducing the level of flux residue under the bottom termination. This study reported inline cleaning using coherent jets. Boards were also processed in other cleaning machines but not reported from this study. The data findings and inferences from those other cleaning machine designs will be reported within follow on research papers. The white residue ghosting issue from water soluble solder paste residues needs future study to determine root cause and reliability effects. ACKNOWLEDGEMENTS Several people worked behind the scenes to make this research possible. First, Dale Lee and his Plexus team helped to frame the cleaning problem and design options that may improve the ability to clean QFN components. Plexus personnel designed the board and provided micro-section analysis. Raytheon McKinney under the leadership of Bill Vuono worked with stencil designer, provided the stencils, assembled the boards and X-rayed the bottom terminations. The people behind the scenes at Kyzen Corporation worked with the board fabricator and component suppliers to obtain the test vehicles. The boards were cleaned at Kyzen s Application Testing lab. Removal of components, imaging and grading was completed by Kyzen technicians. Steve Stach of Austin American Technology provided both insight and images that helped explain the capillary forces or repulsion and attraction. This data finding is a critical factor that explains why heavy flux deposits form under bottom termination components with gap heights lower than 2 mils. The gap heights on test QFN boards used in this study were higher than normal. The via hole patterns may have contributed to this effect. The air currents provided by the via holes appear to allow expansion of the solder as opposed to contraction during reflow. Additional research is needed to conclusively answer this hypothesis. If this hypothesis is valid, the use of via holes can have a positive cleaning effect in respect to gap height. Outside of cleaning, the ground pad designs and use of via holes may render other effects, which may be good or bad from a design perspective. The data findings from this research will be reviewed to gain a better understanding of the positive and/or negative. These findings and follow on research will be reported within future studies. REFERENCES 1 Wikipedia, Definition of a QFN (2013) 2 S. Stach, M Bixenman (2008, Sep) Optimizing Batch Cleaning Process Parameters for Removing Lead-Free Flux Residues on Populated Circuit Assemblies SMTAI, Rosemont, IL 3 M. Bixenman, D. Lee, S. Stach (2012, Sep) High Speed Cleaning in a Reduced Manufacturing Footprint SMTAI, Orlando, FL 4 U. Tosun, R. Ravindran, M. Mccutchen; (2013) Determining Critical Cleaning Process Parameters for QFN s part one, Printed Circuit Design and Fab; p29-33

18 5 EMPF Tech Tips, (2007, January). 6 Lee, D., Helvestine, R & Bixenman, M. (2011, Sep). Design for Cleaning Medical Electronic Circuit Devices. Meptec/SMTA Medical Conference. Phoenix AZ. 7 Bixenman, M. & Lee, D. (2012, Sep). Cleaning Medical Electronics. Meptec/SMTA Medical Conference. Pheonix, AZ. 8 Stach, S. & Bixenman, M. (2004, Sep.). Optimizing Cleaning Energy in Batch and Inline Spray Systems. SMTAI, Rosemont, IL

19 QFN Design Considerations to Improve Cleaning Mike Bixenman, Kyzen Corporation Dale Lee, Plexus Corporation Bill Vuono, Raytheon Corporation Steve Stach, AAT Corporation The Science of Cleaning Green

20 Introduction

21 QFNs Floods the perimeter/streets with flux residue One of today s toughest cleaning challenges Un-cleaned flux residues trapped under the QFN Potential to be hydroscopic and conductive Can lower dielectric strength and result in dendrite growth May change the RF properties Dielectric strength Surface resistance Q-resonance

22 Problem Statement

23 Low Standoff Gap As the distance between conductors narrow Z-axis of the bottom termination reduces BTC gap heights less than 2 mils Create a vacuum effect during the soldering process Capillary action of the flux residue flows and bridges conductors Flux residue fills open gaps in QFN structures

24 Flux Residues Bridging Conductors

25 Long wash time Slow cycle times Slow production May create a bottle neck Cleaning Requires Many batch and inline cleaning machines do not deliver flow and pressure needed to clean under the QFN bottom termination.

26 Partially Cleaned Flux Residues

27 Design for Cleaning

28 Research Purpose The purpose of this research is to Study QFN cleaning forces as a function of Ground pad designs Solder mask definition strategies Via Holes within the ground pad Gap height Capillary forces

29 Standard Ground Pad Design

30 Square Ground Pad Design

31 Slot Ground Pad Design

32 Hexagon Ground Pad Design

33 Via Holes Provide an avenue for Flux residue to drain to the back side of the QFN ground pad Air flow and exhaustion S t r e e t s Via Hole diameter mils

34 25 Via Holes

35 9 Via Holes

36 No Via Holes

37 Non-Solder Mask Defined

38 No-Solder Mask Defined

39 Methodology

40 Test Vehicle

41 Assembled at Test Vehicles Raytheon Circuit Card Assembly Center of Excellence (CCA-COE) McKinney, TX Components delivered in Feeder tubes Transferred to a tape and reel format Compatibility with the placement equipment

42 Four solder stencils Design of Experiment 4 mil thick Nano-Coated Stencil for Square, Standard and Slot designs 3 mil thick Nano-Coated for Hexagon Prevent break-out of Hexagon pattern design

43 Stencil Pattern Designs Optimized to ensure No solder paste printed into via holes Standard stencil design used a Typical solder pattern without consideration of via hole locations All stencil designs were designed for 55 to 60% coverage of the ground pad 100% (1 to 1) for sign pin pads

44 Standard Stencil Design

45 Advanced Square Design

46 Hexagon Stencil Design

47 Slot Stencil Pattern

48 Type 5 Solder Paste Fine Stencil Apertures No-Clean Lead-Free Solder Paste Water Soluble Lead-Free Solder Paste MPM Momentum screen printer Recommended solder paste print parameters followed Print speed Print pressure Fuji NXT II M6II high speed pick and place machine Placement of the components

49 Test matrix Fully populated Cleaned Not Cleaned Test Vehicles Unpopulated boards without components View residues remaining from each stencil design Overall print quality

50 Reflow Profile Vitronics Soltec XPM3 convection reflow oven Twelve temperature zones for solder reflow Conveyor was set at 41 IPM Ramp-to-spike profile Peak temperature target of 235 C Identical for both pastes

51 Inspection Inspected by X-Ray to determine Voiding from the manufacturing process Stencil/board combinations Following assembly Agilent Technologies 5DX X-Ray Laminography system from Agilent Technologies Nikon Metrology XTV160 2-D X-Ray inspection system

52 Cleaning Boards were sent from Raytheon to Kyzen Corporation for cleaning Boards were cleaned in an Inline cleaning machine Progressive Energy coherent jets

53 Response Variables 1) 5DX X-Ray Laminography 2) 2-D-X X-Ray Laminography 3) Microscopy of the Via Pattern on the Back Side of the Board 4) Side View Microscopy of QFNs Before and After Cleaning 5) Removal of the QFNs on Uncleaned and Cleaned Boards

54 Data Findings

55 5D X-Ray Laminography

56 Agilent 5D X-Ray Laminography Inspect voiding of the ground pad A comparison of component placements Solder Mask Definition NSMD NoSM Via Holes

57 5D X-Ray No-Clean Data Findings Void formations reduced when Via holes were placed within the ground plain Voids were smaller in diameter when placing via holes into the ground plane Solid 9 Via Holes 25 Via Holes

58 Analysis of Variance Main Effects Plot for 5D X-Ray Ground Pad Voids Data Means Component Ground Plane Pattern % Ground Pad Voids Mean MLF124 Double Row Solder Mask Definition MLF88 Singe Row HEXAGON SLOT Via Holes SQUARE STANDARD NoSM NSMD Solid No Clean Data

59 2D X-Ray Laminography

60 2D X-Ray No-Clean Laminography Nikon Metrology 2-D-X X-ray Images the internal structure Detect and compare voiding Data findings Via holes reduced void patterns Solid 9 Via Holes 25 Via Holes

61 Analysis of Variance Main Effects Plot for 2D Xray Ground Pad Voids Data Means Component Ground Plane Pattern % Ground Pad Voids Mean MLF124 Double Row MLF88 Singe Row Solder Mask Definition HEXAGON SLOT SQUARE Via Holes STANDARD NoSM NSMD Solid No Clean Data

62 Via Holes on Back Side of Board

63 Via Hole Patterns on Backside 11 to 12 mils in diameter Via holes copper plated The purpose of the via holes Flux residue to drain to the opposite side of the board Air supplied through the via holes during reflow Repels the capillary forces that can form under the component

64 Via Hole Patterns on Backside 25 Via Holes 9 Via Holes

65 Component Identification

66 Formed Solder Bumps

67 Flux Residue % Flux Residue on the Bottom Side of Board M LF 124 Dual Row Main Effects Plot for Flux Residue Present Data Means Component Placement MLF88 S inge Row HEXA GON NoSM Ground Plane Pattern S LO T SQ U A RE Solder Mask Definition NS MD STA NDA RD

68 Flux Residues as a Function of Via Holes

69 Flux Residues as a Function of Solder Mask Definition Interaction Plot for Flux Residue Present Data Means Flux Residue Mean Values Placement Solder Mask Definition NoSM NSMD

70 Flux Residues as a Function of Ground Pad Design Flux Residue Mean Values Interaction Plot for Flux Residue Present Data Means Ground Plane Pattern HEXAGON SLOT SQUARE STANDARD Placement

71 Solder Bumps as a Function of Ground Plane Interaction Plot for Formed Solder Bumps Data Means Solder Bumps Formed Under Component Ground Plane Pattern HEXAGON SLOT SQUARE STANDARD Placement

72 Solder Bumps as a Function of Solder Mask Definition Interaction Plot for Formed Solder Bumps Data Means # of Solder Bumps Formed on Via Holes Placement Solder Mask Definition NoSM NSMD

73 Solder Bumps as a Function of Via Holes Interaction Plot for Formed Solder Bumps Data Means Solder Bumps as a Function of Via Holes Placement Via Holes 9 25

74 Via Hole Data Finds 1. Flux residues in the streets was reduced 2. The solder on the ground plane tended to coalesce and fill many of the via holes. 3. By draining the flux to the opposite side of the board, the levels of flux residues was reduced in the streets.

75 QFN Gap Height

76 QFN Gap Height Main Effects Plot for Gap Height (µm) Data Means 110 Component Ground Plane Pattern 100 Gap Height (µm) Mean MLF124 Double Row MLF88 Singe Row Solder Mask Definition HEXAGON SLOT SQUARE Via Holes STANDARD NoSM NSMD

77 Standoff Gap as a Function of Solder Paste Thickness Interaction Plot for Gap Height (µm) Data Means Gap Height (µm) Mean Ground Plane Design 25 Via Holes 9 Via Holes Solid HEXAGON SLOT SQUARE Ground Plane Pattern STANDARD

78 Less than 2 Mils Standoff Gap

79 Greater Than 4 mil Standoff Gap

80 QFN Gap Height Data Findings Flux residue under the bottom termination is a function of Attractive and repulsive capillary forces When the Z-Axis is less than 2 mils Flux residue capillary forces attract during reflow Heavy flux residue deposits accumulate in the Streets Interconnecting pads Attractive force renders Significant level of flux residue Underfills component with flux residue Flow channels closed Repelling force render Low levels of flux residue Flow channels open

81 Removal of Components to Inspect for Residues

82 Flux Residue Under QFN Test Board 48 QFN patterns Strategically placed MLF 88 Component Placements MLF 124 Component Placements

83 Component Removal

84 Not Cleaned Boards MLF 88 dirty component example

85 Not Cleaned Boards MLF 124 Dual Row dirty component example

86 Flux Residues Bridging Pads Main Effects Plot for Flux Residue Bridging Pads Data Means Component Ground Plane Pattern Flux Residue Means MLF124 Dual Row MLF88 Singe Row Solder Mask Definition HEXAGON SLOT SQUARE Via Holes STANDARD NoSM NSMD

87 Flux Residues in Streets Main Effects Plot for Flux Residue in Streets Data Means Component Ground Plane Pattern Flux Residue Means MLF124 Dual Row MLF88 Singe Row Solder Mask Definition HEXAGON SLOT SQUARE Via Holes STANDARD NoSM NSMD

88 Flux Residue Next to Pads as Function of Solder Paste Thickness Interaction Plot for Flux Residue Bridging Pads Data Means Solder Paste Thick ness (mils) 3 4 Flux Residue Mean MLF124 Dual Row MLF88 Singe Row Component

89 Flux Residue in Streets as a Function of Solder Paste Thickness Interaction Plot for Flux Residue in Streets Data Means 0.30 Solder Paste Thick ness (mils) 3 4 Flux Residue Mean MLF124 Dual Row MLF88 Singe Row Component

90 Cleaned Boards Cleaning Parameters Inline Cleaning Machine Progressive Energy Dynamics using Coherent Jets 5 minute wash cycle Aqueous Engineered Cleaning Agent 20% Concentration 65 C Wash Temperature The data finds Minor levels of no-clean flux residues No wet water soluble flux residues Some white ghosting on solder

91 No-Clean Flux Residue Next to Pads Main Effects Plot for Flux Residue Next to Pads - Cleaned Boards Data Means Component Ground Plane Pattern Flux Residue Mean MLF124 Dual Row MLF88 Singe Row Solder Mask Definition HEXAGON SLOT SQUARE Via Holes STANDARD NoSM NSMD

92 No-Clean Flux Residue in Streets Main Effects Plot for Flux Residue in Streets - Cleaned Boards Data Means Component Ground Plane Pattern Flux Residue Mean MLF124 Dual Row MLF88 Singe Row Solder Mask Definition HEXAGON SLOT SQUARE Via Holes STANDARD NoSM NSMD

93 Water Soluble Example

94 Inferences from Data Findings

95 Voiding is reduced when Voiding Via holes are placed into ground plane Results in smaller and less voiding