Smaller pitch and pad size

Yongmei Wang, Kazu Nakajima, Alington Lewis, Murad Kurwa, Ravi Bhat and Sammy Yi, Flextronics International Ltd., San Jose, CA

Many factors affect the printing process and its quality, such as the stencil, solder paste type, PCB, the machine and its capabilities. Within these categories there are many variables. The stencil, for instance, can vary in thickness, aperture opening, aperture shape, aspect ratio limitation, stencil fabrication and surface polish methods, stencil accuracy, etc.

In the following, the focus is on investigating solder paste printing capabilities and limitations for area-array packages, identifying the critical variables that affect the printing process. To measure the interaction among the variables, two separate designs of experiment (DOE) were conducted. We will discuss the method of the DOEs, as well as the relationships among the variables. Finally, we will describe the optimum conditions for solder paste printing and present design guidelines for manufacturing small area array packages.

Flextronics has implemented CSP on-board assembly for several years and has designed and processed two test vehicles, including most of the newer types of CSPs. On these test vehicles, 23 types of CSPs with fine pitch from 0.5 to 0.8mm, and I/O counts from 4 to 324 have been implemented. Several insights have been gained from this experience. First, paste powder grain size #4 is a better choice for fine-pitch and small pad-size printing. Second, in addition to paste selection, the most critical factors in the solder printing are process setting as well as design and technology of the stencil. In this study, we focused only on the printing process without assembling CSPs. Fine pitches with 0.4 to 0.6mm and small pad sizes with 6 to 16mil diameter were designed on the test vehicle. The objectives were:

• evaluation of solder paste printing limitations for fine pitch and small pad size, area-array components

• establishing optimum stencil design and printing process guidelines

To meet these objectives, the DOE was used to make experiment plans as well as to analyse the data. The test data to be used is solder paste height, solder volume before reflow, solder paste shape, Cpk for solder paste height, and number of defects.

Solder paste printing is influenced by many conditions, including material, process, environment and equipment, etc. To better understand the relationships among these factors, a fish-bone diagram (fig-ure 1) shows all of them, grouped into six categories:

ð Process setting

ð Environmentalcondition

ð Equipment capability

ð Stencil design

ð Solder paste type

ð PCB design

Within each category, factors are listed that affect the printing process. A total of42 factors are listed in all categories with each factor having more than two levels of variables. In this scenario, a full factorial, one blocked DOE would have more than 1020 combinations when taking only three levels each of a factor into consideration. Since this amount of complexity would make the experiment impossible, factors were limited to only those that would strongly influence the printing process. At the same time, two DOE plans were used: the goal of the first (I) was to identify the optimum printer setting parameters, the second one (II) wasto identify the relationships among thedifferent variables, thereby gaining a greater understanding of the CSP printing process.

As shown in the fish-bone diagram, 22 factors were selected which can affect the printing process setting (categories 1, 2 and 3). To simplify the DOE and obtain the optimized printing setting, only three factors from three categories were selected as follows:

ð Squeegee pressure

ð Printing speed

ð Separation speed

Other variables, such as solder temperature, mixing method, squeegee type and stencil cleaning method, remained constant. The stencil used in this DOE was 5mil thick and fabricated by laser cut with electro-polishing. This type of stencil has proven to be suitable for solder printing with aperture opening of 10mil from the CSP qualification. No-clean solder paste with particle size #4 was used.

This DOE was designed using the Box-Behnken design method with three factors by three levels. A total of 13 experiments were created using this method. The matrix of the DOE is presented in table 1. Thirteen sets of parameters for the machine setting were selected for the printing process. After proper printing experiments, paste heights were measured. The Cpk of solder paste heights for each setting was calculated and used as criteria for selecting optimum setting groups. As a result of the calculation, four process setting groups provided the highest Cpk numbers. To select the final optimum setting, a visual inspection of printed solder qualities from those four process settings was performed. Finally, one setting group with high Cpk and the best printing quality for small openings was selected as the optimum setting. As for DOE II, only the factors that were considered critical from categories 4, 5 and 6 (figure 2) were selected. These critical factors and their levels are listed in table 2.

Based on the requirements, a FR4 board was designed. The dimensions of this board were 8-in in length, 7-in in width and 0.062-in in diameter. All the pads had Au/Ni surface finish. Ninety test cells were designed with various pad dimensions and pitches on the board. A cell is a 20 x 20 grid array with round shape pads. Zone A, B and C have pad pitches of 0.4, 0.5 or 0.6mm, respectively. Each zone has six columns with one column consisting of five cells designed with five different pad sizes. Other kinds of footprints, such as fine-pitch QFPs, 0201 discretes and µSMD components were designed on the board as references. Since this board was created only for testing the solder printing process, no via or circuit traces were designed.

The stencil design is one of the great challenges of the solder printing process. To better understand the relationship between a design and its performance during solder printing, nine different stencils will be prepared for this DOE. As listed in table 4, different stencil technologies, such as chemical etch and laser cut with electro-polish, were tested. Two kinds of stencil materials will be tested: stainless steel (s.s.) and Invar (Fe-Ni alloy). The tested stencils are laser cut with electro-polish, stainless steel, with 4, 5 and 6mil thickness.

All three stencils shared the same design of aperture openings. For each test cell, six types of aperture opening were used in the stencil design. The aperture shapes were round, square and oval. The apertures were either the same size as the pads or 2mil larger.

Since the stencil opening can be as small as 6 to 10mil, powder grain becomes the most important factor in solder paste characterization. Therefore, the paste particle size was the only factor considered in this study. Solder paste powder types #3 and #4 were selected. Size #3 was used because it has become an industry standard among assembly manufacturers for the components that exist with the CSP. The objective in testing size #3 was to identify the limitations with this type of solder print process and providing guidelines for printing. Here, in this context we will focus on powder size #4.

For each process setting from the matrix of the DOE I with 13 sets of parameters,10 boards were printed with the solder paste height measurements on six locations. An auto printing machine was used in this process. From the data, the Cpk for each setting was calculated. The best four process settings that had high Cpk were repeatedly tested visually. The results are presented in figures 1 through 3.

These figures represent the relationship between printing speed and squeegee pressure when holding the separation speed at low, medium and high. In figure 2, when the separation speed was low, the low printing speed provided relatively high Cpk, while the squeegee pressure exerted less influence. However, comparing with the Cpk contour in figure 3 and 4, the highest Cpk range is not shown in figure 2.

In figure 3 and 4, with median and high separation speed, low printing speed and high squeegee pressure, the resulting Cpk was as high as 1.80. Comparing figures 3 and 4, it is clear that high Cpk can be achieved with low printing speed and median separation speed (figure 2) regardless of the setting in squeegee pressure.

Three boards were printed using the bestfour-process settings A to D. The visual evaluation found some bridging failures that are listed in table 5. Also, missing solder paste on a cell was ranked up to five grades. Higher grades represented more missing solder paste on a cell. From these results, setting D was selected as an optimum print setting for the DOE II.

In this experiment, the same printing machine was utilized and the same solder paste was tested as in DOE I. The stencils had a thickness of 4, 5 and 6mil. DOE II was processed on three groups of printing with the optimum setting that was determined in the DOE I. Three boards for each stencil were printed. One of the three boards was inspected and recorded. The inspection includes solder paste height, visual solder paste volume, solder paste shape, and bridging count.

The solder paste height versus area ratio (AR) of each aperture opening was calculated for each stencil thickness. The AR is defined as the area of a stencil opening over the wall area of the aperture opening. Paste height versus AR for each stencil is presented here as figures 5, 6 and 7. Through these figures, the area ratios (AR) above 0.5 seem to result in stable solder printing.

These figures show a trend that, when the area ratio is equal or greater than 0.5, the paste height will become stable around the stencil thickness. In other words, at an area ratio of 0.5 or greater, a stable process of solder paste printing can be achieved. This result is in line with experience obtained from CSP qualification.

During inspection, attention has been drawn to the shape changes of solder paste with different aperture opening. Generally, three types of solder shape were observed. Type #1 is a cone-cylinder shape, which has a flat-top surface. This shape indicates that a sufficient solder printing is obtained. Type #2 is cone shape with a sharp top, but still sufficient solder printing is presented. This kind of shape conceals that the aperture opening and aperture shape prevent a good solder release. Type #3 observed is a small pile of solder particle, which in-dicates insufficient paste was printed. Figure 8 shows each type of solder printing. The inspection result was provided in table 6 in which the term sufficient refers to printing with type #1 shape; inconsistent volume refers to solder of type #2 and type #3 shapes; insufficient refers to the solder printing with type #3 shape.

The data in table 6 tells that with 4mil stencil, when an aperture size is larger than 10mil, sufficient solder printing will be obtained. However, when an aperture size is about 8mil, different solder shapes will start to be observed. When an aperture size is smaller than 6mil, insufficient solder printing will be expected. In one sentence: aperture size equal to 8mil is a printing limitation for laser-cut, polishing stencil with 4mil thickness. Aperture size below 8mil is not suggested.

Similar phenomena are found on 5 and 6mil thick stencils. The only difference is the limitation for those stencils moving toward larger aperture size than that for 4mil stencil. For CSP stencil design, aperture sizes larger than 10mil on 5mil stencil or aperture sizes larger than 12mil on 6mil stencil are recommended.

The recommendation list above reflects that for round-shape and square-shape apertures, when AR is 0.5, the apertures are at their limitation. Comparing with the knowledge obtained from data analysis of solder paste height, a unanimous result is achieved. The summarized results are presented in figure 9.

Within the solder shape/volume inspection, on locations that AR is around 0.5, numbers of different solder paste shape were recorded as well. These locations include the cells printed with: 1) on 4mil stencil, 8mil round, square apertures, or 8mil by 10mil oval apertures; 2) on 5mil stencil, 10mil round, square apertures, or 10mil by 12mil oval shape apertures; 3) on 6mil stencil 12mil round, square apertures, or 12mil by 14mil oval shape apertures.

When solder paste has a shape of type #1, it means a good printing definition has been achieved. Solder paste with the shape of type #2 indicates that a poor printing definition is presented. For 4, 5 and 6mil stencils, most of good printing definition percentages on round-shape apertures, square-shape apertures or oval-shape apertures are higher than 90% of printing sites. These high percentage values proved that at apertures with AR about 0.5, good printing definition could still be achieved. The test also reveals that with the same AR opening, square-shape aperture has better printing definition that with round shape.

On locations that AR is less than 0.5, numbers of different solder paste shape were also recorded and analyzed. The locations include: 1) on 4mil stencil, 6mil round, square aperture, or 6mil by 8mil oval aperture; 2) on 5mil stencil, 8mil round, square aperture, or 8mil by 10mil oval shape aperture; 3) on 6mil stencil 10mil round, square aperture, or 10mil by 12mil oval shape aperture. The aperture shapes observed consist of solder shape types #1, #2 and #3. Both solder shapes with types #1 and #2 have supplied sufficient solder on sites, but solder paste with type #3 shape has insufficient solder paste volume after printing.

The recorded data percentage is plotted in figure 10 in which site percentage of sufficient solder is always lower than 50% regardless of aperture shape. These low site percentage values provide data evidence that, when the apertures with AR less than 0.5, insufficient solder paste printing is definitely expected. Therefore, apertures with AR less than 0.5 are not recommended. It becomes also clear that with the same AR value, apertures with square-shape provide more solder paste than that with round shape. Therefore, from paste-volume side, square-shape aperture is recommended.

Solder bridging is defined before inspection. The criteria are when the space between two solder piles is shorter than two diameters of a solder grain, a bridging is present. Bridging is counted when there is a path that links two solder piles together. Gap refers to the distance between two adjacent apertures. In this inspection, the number of bridging was recorded.

Apertures with square-shape opening have higher a possibility of solder bridging. Because a square aperture has one edge which equals to gap, a round aperture has only one point which equals to gap. From data collected for solder bridging, no clear trend versus stencil thickness was found at this moment. However, it is clear that round-shape has less bridging than square-shape aperture.

In figure 11, percentage of bridging versus different aperture shape on locations of gaps equal to 2 and 4mil are provided. No solder paste bridging was found when gaps are larger than 4mil regardless of aperture shape. The results from bridging inspection present that with the same closest gap distance between two adjacent apertures, square-shape leads to more bridging potential than round and oval-shape. The reason could be explained as a different situation for each kind of shape. When a printing site has four other sites surrounded, a square-shape aperture has four edges with closest distance, while a round-shape aperture has four points with closest distance, and an oval-shape aperture has only 2 points with closest distance (the other 2 points have 2mil larger gaps). Therefore, although observed from analysis on visual solder volume/shape inspection that square-shape apertures supply more solder paste than round-shape with the same AR value, when a gap between two apertures goes down to 4mil, square-shape apertures will have more chance to cause bridging. Consequently, when designing a stencil, if a small gap is required, comparing to round or oval-shape apertures, square is not recommended. When a gap larger than 4mil is allowed, a square-shape aperture is suggested.

On printer settings, the squeegee pressure and printing speed clearly influence the Cpk of the solder paste height. In general, a higher squeegee pressure and low printing speed will result in better Cpk.

Area Ratio is an important parameter for obtaining good printing results. AR of 0.5 for square and round openings can serve as a guideline for stencil design. An AR below 0.5 is not recommended. This result is suitable for stencil thickness of 4, 5 and 6 mil. Square aperture openings provide better solder paste release than round openings as AR # 0.5. For the solder paste used in this project, particle size #4, when gap between apertures is smaller than 4mil, solder paste bridging will be expected. Also, as the gap narrows, square-shape openings on the stencil may cause bridging more easily than round-shape apertures. When solder paste powder size #4 is used, a square-shape aperture is recommended for small openings as AR # 0.5, but not for the locations having gaps smaller than 4mil. To complete this project, especially DOE II solder paste with powder size #3 needs to be tested as well as chem-etch, laser-cut stencils with Invar material.

Stencil Recommended Acceptable

thickness aperture size aperture size

Zusammenfassung

In Experimenten wurde untersucht, worauf beim Lotpastendruck von CSP-Anschlußstrukturen auf Leiterplatten zu achten ist. Bekanntlich stellen diese höhere Anforderungen an Genauigkeit und Zuverlässigkeit, weil sie geringere Padabmessungen sowie Pitch-Raster dazwischen aufweisen. Bei Beachtung der hier aufgestellten Guidlines ist in der hochvolumigen Fertigung die CSP-Verarbeitung sichergestellt.

Des expériences ont été faites pour déterminer les paramètres à prendre en compte lors de la sérigraphie de pâte à braser de structures CSP sur des cartes imprimées. Comme on le sait, celles-ci sont très exigeantes quant à la précision et à la fiabilité car elles présentent des pistes de dimensions inférieures et des écartements réduits. L’observation des lignes de conduite indiquées assure une parfaite production de CSP en grands volumes.

Durante degli esperimenti è stato analizzato a cosa bisogna fare attenzione durante lo stampaggio di pasta per saldatura a dolce di strutture di collegamento CSP su circuiti stampati. Come si sa essi pongono dei requisiti più severi in fatto di precisione e di affidabilità in quanto essi posseggono delle dimensioni di pad o di reticolo pitch minori. Osservando le direttive qui riportate è possibile garantire una regolare lavorazione CSP anche in produzioni ad alte quantità.

www.flextronicscom

Fax +44-1628-501-510

EPP 152