Figure 1. Laser-machined stencil (unpolished) showing vertical walls of opening, which tend to be rough.

Transcription

1 Subtleties of 1 Stencil PrintingLr2F Solder W Though applying consistent volumes of paste to component pads is vital for reliable solder joints, there are process limitations. by Carl Missele, Motorola, Schaumburg, Illinois SMT is essentially a packaging philosophy, and packaging has driven the printed circuit board industry. This has resulted in improvements in all related technologies, such as soldering equipment, solder paste, solder stencils, pick-and-place machines, board materials, and vision systems, to name a few. In some cases, such improvements have leapfrogged packaging technology, leading the way toward its further improvement. One area of the technology that has lagged (but not for lack of trying) is solder paste application for surface mount devices (SMDs). Part of the reason stems from the process' inherent limitations; another is inadequate understanding. Process Control: Methods of Measurement Regardless of the method of application, the key to process control is measurement and rapid feedback. Though there are many measurement methods used (most are noncontact), these are the most common: 1. Depth-measuring microscope depends on the shallow depth of field of a microscope at high magnification. The method employs the technique of focusing on the pad metallization, zeroing the micrometer attached to the microscope stage, and refocusing on the top of the printed paste. The micrometer reading is the height of the paste. There are several problems with this method: Figure 1. Laser-machined stencil (unpolished) showing vertical walls of opening, which tend to be rough. The thickness is influenced by the solder resist on the board on which the stencil rests during the printing process, rather than on the circuit metal. Thus, the operator using the metallization as the reference will find the paste print thicker than expected. The operator does not know if the printed pattern is flat on top and will not always select the optimum location for the measurement. Local camber in the substrate can affect the reading since the reference is not flat to the lens. When viewing the top of the print at high magnification (necessary for shallow depth of field) the solder spheres are relatively large. Does one measure at the top of the sphere, the center, or somewhere between? A limited number of printed pads are viewed and may not represent the average height of the paste over the entire surface of the board. 2. Weighing method. The board to be used is weighed on a three- or four-decimal place balance before the application of paste, and then again after the application. The difference is the weight of paste deposited. Some preestablished range is used to accept or reject the print. The problem with this method is that one or a few pads may be missing and still fall into the acceptable range. This method is not sensitive enough for good process control. 3. Light section scope. A sliver of light is projected onto the printed pad at a 45" angle. The light is viewed through a video camera that projects the image on a CRT. Also projected on the CRT are horizontal and vertical lines which are controlled by a calibrated electronic measuring system. One pair of lines is moved to the reference of the projected sliver of light on the pad and zeroed. The other is moved to the offset that occurs due to the 45" projection. The measuring system compensates for the measuring angle and gives a height reading on the screen. The problem: The sliver of light is wide when magnified, limiting the resolution of the measurement. It is also difficult to distinguish where the top of the print is owing to the irregularities of the solder spheres. Surface Mount Technology April

2 SUBTLETIES OF STENCIL PRINTING SOLDER, 4. Laser light section is similar to the regular light section system except that the light sliver is thinner. The additional problem with this system is absorption of the laser energy by the resist or the board material, leaving insufficient contrast to accurately establish a reference with the light. 5. Contact method. A micrometer with vertical travel is used for this measurement. The foot of the micrometer is placed on the board and zeroed, then moved up to make contact with the top of the print. The micrometer reading is the height of the print. The problem here is obvious: The method is not very accurate since there is no feedback on when the probe touches the solder print D-laser imaging probably gives i the most needed information of any of these mentioned. A laser scans the printed paste and a computer processes the reflected light to produce a three-dimensional image of the pad. Much can be determined from the information, such as volume, height, area, and print irregularities. However, that is also one of the problems: There is too much information. The other problem is that in some systems, the measurements take too long to feed back to the process. It should be noted that a new laser system does gather the information but also can process it in a time short enough to be useful. For example, a 25 cm2 board could be scanned in 1.5 to 3 min and the data can be presented on a process control chart displayed on a CRT to show whether the process is still in control. Height references for the operation are established from conductors under the solder resist in the vicinity of the printed pad to be measured. This eliminates the error introduced by camber or warp in the board, but does not compensate for the thickness of the solder resist on which the stencil rests during the printing process. This is not a closed-loop system but approaches that capability. Systems of this type cost nearly $200,000 and many small businesses would find the cost difficult to justify. It is apparent that all of the measurement techniques fall short in supplying useful process control data partly due to inherent characteristics of the solder paste and partly due to the techniques employed to make the measurements. It should also be apparent that solder paste height, and more importantly, volume, is difficult to measure accurately (see sidebar). B. Solder spheres flattened by force squeegee on the stencil (arrow), whic adversely affects print definition. ifference is 70pm. The solder spheres themelves are 45 to 75pm in diameter and it must whether to measure at the top or in -supm) previously esiadiisnea wouia DF) ex- *.,, l.yl _._I_LL_..!_._ ceeaea. worse, IT me rounn, nnn, or sixin prim are measured without wiping the underside of the stencil, another!5opm could be added to the paste height. This occurs because the paste, on release from the stencil openings, splashes on the underside of the stencil and remains * *....I., (They are slightly flattened by the force of the squeegee on the stencil; see photo B.) These also adversely affect the print definition. Process Control at the Printing Machine 1. Solder Paste Characteristics. If one cannot measure the results of the process effectively, how is process control established? The most effective way is to perform the operation correctly at the printing equipment. In operations where statistical process control is used, this is called precontrol. The first step is to se lect a solder paste with characteristics matched to the application. For fine-pitch (0.6 and lower), it is important to use the finer particle materials. With pads 0.3 wide and solder spheres in the range of 30pm in diameter, it takes only 10 spheres to span the pad and only five or six to meet the height requirement. If using standard solder paste with an average diameter nearly twice that of the finer particles, it would take only half as many spheres to do the same-obviously impractical. Printing solder paste with a viscosity at the lower limit, approximately 800 Kcps, appears to yield better results than with higher viscosity materials. At the same time, one must not compromise on room temperature slump or on heating slump. Materials can and should be evaluated for these characteristics before using in a production environment. This is at a time when many companies are converting to no-clean solder pastes. While actually labeled noclean, they are not no residue. In fact, many leave large quantities of residue. If one considers that a solder paste with 90 percent solids by weight is approximately 22 April 1993 Surface Mount Technology i

3 50 percent liquid by volume, it is clear that dealing with that liquid in residue form can be a problem. If the liquid is comprised of low-boiling solvents, drying and poor tack might result. Unfortunately, the organics that promote tack and prevent rapid drying are also difficult to volatilize during the reflow process and thus leave a residue. Retaining tack in a solder paste for from two to six hours cannot be compromised if high yields are expected. 2. Solder Stencil Characteristics. Stencils are the tools used for mass application of solder paste. Stencils can be perimeter-bonded to a polyester mesh, to a stainless steel mesh, or directly to the stencil frames. Each has advantages. The polyester mesh makes the stencil more compliant and able to conform to long wave camber in the substrates. The stencil bonded directly to the frame works well when stenciling solder paste on the second side of a PCB that has minimum support from the tool plate. Stencils that are chemically milled have smooth walls in the openings but show signs of undercutting. Laser-machined stencils have vertical walls in the Figure 4. Stepped stencil cleared by Delrin squeegee. Other openings will not be affected by hard-edge squeegees. openings, but tend to be rough (figure 1). Milling the opening with a numerically controlled router gives smooth, accurate openings but leaves rounded corners and can take hours to complete. If choosing to use a stepped stencil, chemical milling is the simplest method for producing the stencil. 3. The Printing Machine. The stenciling process requires a printing machine to deposit the solder paste on the substrate. The machine should be capable Figure 2. Dual print heads of a printing machine are capable of performance in both contact and off-contact mode. Figure 3. Polyurethane (tor and Delrin squeegees for dual-head screen printer. Hard edges prevent paste scoop-out. of printing in both the contact and offcontact mode. A distinct down stop and squeegee pressure (force) control are both necessary. Dual squeegee heads are a distinct advantage and vision is an asset (figure 2). Often neglected in printing equipment is proper initial setup and preventative maintenance. These steps ensure that the stencil will always be parallel to the toolplate or workholder and that the squeegee will always travel parallel to the work surface. 4. Squeegees. Ideally, a hard bladetype squeegee would be used in both squeegee heads. These could be made of steel, Delrin, nylon, or polyethylene (figure 3). These hard blades prevent scooping of solder paste from large stencil openings, but will permit printing a fine-pitch pattern in a stepped stencil. They will also conform to long wave camber or to thickness variations in the PCB. 5. The Setup. The blade squeegees are placed into the holders and checked to see if they are in contact with the work surface over their entire length, and then to see if they travel parallel to the work surface. Next, the squeegees are raised and the stencil placed in the machine.?he stencil is then lowered to the top of the substrate on the workholder and checked for parallelism. (The stencil must be parallel to the work surface but may not be parallel to the top of the substrate.) The stencil is then raised 0.1 to 1.5mm, depending on the size of the stencil and whether it is perimeter-bonded to polyester mesh. For a small stencil directly bonded to the frame, 0.15 is acceptable. For a 60cm2 frame with a polyester perimeter, 1.5mm can be used. In the setup position, the squeegees are lowered to the top of the stencil until Surface Mount Technology April

4 i 50 percent liquid by volume, it is clear that dealing with that liquid in residue form can be a problem. If the liquid is comprised of low-boiling solvents, drying and poor tack might result. Unfortunately, the organics that promote tack and prevent rapid drying are also difficult to volatilize during the reflow process and thus leave a residue. Retaining tack in a solder paste for from two to six hours cannot be compromised if high yields are expected. Figure 4. Stepped stencil cleared by Delrin squeegee. Other openings will not be affected by hard-edge squeegees. to a stainless steel mesh, or dire more compliant and able to conform to long wave camber in the substrates. The stencil bonded directly to the frame works well when stenciling solder paste on the second side of a PCB that has minimum support from the tool plate. Stencils that are chemically milled have smooth walls in the openings but show signs of undercutting. Laser-machined stencils have vertical walls in the openings, but tend to be rough (figure of printing in both the / ntact and off- 1). Milling the opening with a numerical- contact mode. A dlct down stop and squeegee pr9ss ure (force) control are both nec5hat-y. Dual squeegee heads ke hours to complete. If / are a igfinct advantage and vision is an a stepped stencil, chem- (figure 2). simplest method for pro- Often neglected in printing equipment is proper initial setup and preventative maintenance. These steps ensure that the stencil will always be parallel to the toolplate or workholder and that the squeegee will always travel parallel to the work surface. 4. Squeegees. Ideally, a hard bladetype squeegee would be used in both squeegee heads. These could be made are capable of \ of steel, Delrin, nylon, or polyethylene performance in both contact and off-contact mode. Figure 3. Polyurethane (top) and Delrin squeegees for dual-head screen printer. Hard edges prevent paste scoop-out. and then to see if they travel parallel to the work surface. Next, the squeegees are raised and the stencil placed in the machine. The stencil is then lowered to the top of the substrate on the workholder and checked for parallelism. (The stencil must be parallel to the work surface but may not be parallel to the top of the substrate.) The stencil is then raised 0.1 to 1.5mm, depending on the size of the stencil and whether it is perimeter-bonded to polyester mesh. For a small stencil directly bonded to the frame, 0.15 is acceptable. For a 60cm2 frame with a polyester perimeter, 1.5 can be used. In the setup position, the squeegees are lowered to the top of the stencil until Surface Mount Technology April

5 t ~, SUBTLETIES OF STENCIL PRINTiNG SOLDER they force it to contact the substrate along their entire length. In the run position, the stencil is inked with a large quantity of solder paste in front of the first squeegee. Running the printing in alternate directions, the following are observed: Squeegees are traveling at about 25 to 40mm/sec. Squeegees are traveling past the substrate at 12 to 25. Squeegees are clearing the top of the stencil of paste over substrate. Squeegee pressure (force) should be used to control cleaning function. If the squeegees easily clean the solder paste on the first pass, squeegee pressure is decreased until the squeegees will not clear, and then enough pressure is returned to clear the paste (figure 4). If a stepped stencil is used, greater squeegee pressure will be required to 24 April 1993 Surface Mount Technology clear the stepped portion of the stencil. The other openings in the stencil will be minimally affected by the additional force as long as the hard squeegees are used. Polyurethane squeegees, even at 90 durometer, will scoop solder paste from the stencil openings. The scooping grows progressively worse with increased pressure. Polyurethane squeegees used with stencils introduce a variable to the process that is very difficult to control. Once the setup is satisfactory, based only on the requirement that the surface of the stencil is being cleared by the squeegee, it is time to examine the print. For best definition and consistent thickness, the stencil must be wipedon the underside after every few prints. How Patterns Will Appear The printed patterns will appear slightly convex for the smaller pads. The larger Figure 5. Crown pattern at periphery of a large solder pad. A good, nonslumping paste will retain this shape (including a depression in the center). Figure 6. Characteristic convex shape of a small solder pad, which Is produced because more of the paste is retained by the walls of the opening than by the pad. pads will be concave for two reasons: Even though the squeegees do not scoop solder from the stencil openings, the paste pushed forward to the squeegee will tend to perform in that manner, and when the stencil is raised off the substrate after the print is completed, the paste at the periphery of the stencil will separate from the paste remaining on the pad. Some paste will stay with the walls of the openings, but in the separation process, the entire perimeter of the pad is lifted before separation. After separation, a good, nonslumping paste will retain the peaks around the perimeter and an apparent depression in the center (figure 5). As the pad widths decrease, the ratio of the width to the thickness of the stencil approaches unity. The surface tension of the solder paste is such that more of the paste is retained by walls of the openings than by the pad. Therefore, only a small amount of the solder paste in the center of the opening is deposited on the pad, producing a convex shape (figure 6). This pattern is also the rationale for choosing thinner stencils to print smaller openings. The need is to increase the ratio of the width to the thickness of the stencil. Since a mix of stencil openings is generally found on a stencil, using stepped sections for finepitch parts is reasonable because it satisfies the ratio requirement. The negative aspect is that the step must exceed the pattern area by at least 3mm around the pattern, an area that cannot be used for other components. Toward Defect-Free Soldering The solder stenciling process has limitations and process control deficiencies. However, following the procedures outlined above will help eliminate the need to measure the thickness, volume, or weight of the printed pattern. As the industry strives toward defectfree reflow soldering, stenciling solder paste will not be the surviving process. (Perhaps solder plating circuits will replace solder stenciling.) Meanwhile, paste dispensing may be the answer, or perhaps, conducfive adhesives will replace solder. Whatever the means, SMT will be with us for many years to come and this and many other problems will be solved. smt Contact author at Motorola, Land Mobile Products Sector, 1301 E. Algonquin Rd., Schaumburg, ; telephone:

6