By John R. Morris and Jemes H. Conway A T&T has recently made significant progress toward eliminating the need for PCB cleaning through the use of low-solids flux for wave soldering in an inert atmosphere. For reflow processes, cleaning is performed after reflow for testability or for cosmetic reasons, even though the solder pastes we use produce noncorrosive, nonconductive residues. To eliminate the need for cleaning, residue must be significantly reduced. This study identified several candidate materials for scaleup and factory trial. The following sections dcscribe the development of a manufacturable no-clean reflow soldering process. The previously mentioned low-residue solder pastes were initially reflowed using a commercially available infrared reflow oven with atmosphere-control capability. Ovens from several vendors were subsequently evaluated. Prior to optimizing conditions for a particular solder paste, a standard set of conditions was established to reduce the number of variables. The two major variables considered were the thermal profile and the oxygen content of the reflow environment. In general, a standard thermal profile (Figure 1) was maintained similar to that used with RMA pastes in AT&T s factory ovens, \ with the lowest possible 0, content. An example of the soldering results obtained using this process is illustrated in Figure 2. diffirences in thermal mass and IR shadowing effects. Because low levels of 0, must be maintained in inertatmosphere IR ovens, an inert gas such as N, must be actively introduced into the heated region of the oven. This tends to increase convective heating and reduce the shadowing effect, which occurs more frequently in naturally convective IR ovens. The soak period also allows for continued volatilization of solvents and other inactive components of the solder paste flux while activating those components that chemically remove from pads and leads the surface oxides that were present before reflow. This step may be the most critical portion of the noclean LRSP process because it may determine the amount of residue left on the board after reflow. If the soak period is too short or conducted at too low a temperature, sufficient activation will not occur. On the other hand, if the soak period is too long or conducted at too high a temperature, the activators may volatilize before they can begin to work. In either case, poor wetting or solder ball formation can occur. Reflow. Reflow requirements include the three preceding factors. The reflow temperature must exceed the eutectic temperature of the metal alloy being used. It is almost impossible to achieve a uniform temperature across the PCB due to differences in heating rates and heat dissipation; therefore, a minimum and maximum temperature must be specified. The profile should be adjusted to minimize this difference (AT). Board discoloration due to oxidation at elevated temperatures typically occurs in ambient-air atmosphere reflow. The low 0, concentrations of inert atmospheres allow for slightly higher maximum board temperatures. A specific duration above the liquidous temperature is 28 CIRCUITS ASSEMBLY AUGUST 1991
plementat ion most promising LRSP candidates excant performance degradation under ing an assembled PCB e board to be soldered. be done on each product type unless the types are negligible. Next, several ons of 230 AWG, usually K-type, ly placed and soldered ture alloy (typically e are three general lo- mponent land, some al-mass components. ipated temperature d of a chip carrier ture here is usualmaximum and minimum. h thermal mass-under a lead that is adjacent large, thermally massive dehis is the coolest connected to a recording derecorder or a digital data log- ty into their equipment. Next, the 0, window, or range of oxygen concentration that yielded favorable results, was determined. We attempted to allow the natural leaks in the system to establish varying 0, levels by adjusting the flow of N, into the oven and exhaust venturis drawing gas out. This was a very difficult and inconsistent process. We then introduced a metered air leak into the N, source line that fed the oven s heated zones. This measure was effective for oxygen concentrations from 510 to 1,000 ppm. Figure 3 illustrates a reproducible, consistent 0, level map of the oven that sets the reflow region monitoring position to 510, 100, 500, 800, and 1,000 ppm. Levels above 1,000 ppm were extrapolated from a calibration curve. The entrance, exit and cooling zones have considerably lower 0, concentrations than the heated regions because air was not introduced into the inlet N, in these regions. Comparison of Oven Types A general comparison of several IR reflow oven vendors systems follows. Methods of maihtaining inert-atmosphere control. For an inert reflow environment, the IR oven must be sealed as tightly as possible in order to maintain a steady-state inert status. Two methods are used to achieve this goal. One is to enclose the working length of the oven in an airtight casing. The other is to place sealing materials in gaps that need not be accessible and gasketing between opposing edges of hinged areas. The latter method can be ineffective and difficult to implement and test, but it provides easier repair access than the former alternative. Gasketing may also be suitable for retrofitting existing ambient-air oven designs.. clean, low-residue reflow soldering e. Due to the volatile nature of the ganic solids in low-residue solder in and cooling zones and installing re able to monitor each section of FIGURE 1: Typical thermal profile.
heaters while others use an additional set of heaters. In all cases direct IR heating is enhanced by convective gas flow. Each also contains an entrance and exit curtain or baffle areas purged with N, to prevent the influx of air from the open, though restricted, ends of the oven. Retrofitting existing ambient-air systems is a third alternative to purchasing an IR oven with built-in inertatmosphere control. There are several commercially available retrofit kits on the market today. The means of attaining an inert atmosphere are generally the same as for infrared reflow applications except that no major alterations are made to existing hardware. Specialized plumbing, gas preheating and sealing are performed as additions. Oxygen mapping was performed on one oven of each type. While stabilized at setpoint temperature, the ovens were loaded with PCBs spaced approximately one board length apart. The results are shown in Figure 4. lnerting level comparisons. A major factor in oxygen level is N, consumption rate. The gas consumption rate of the ovens ranged from a low of about 1,100 SCFH to a high of about 2,600 SCFH. The average consumption rate was approximately 1,600 SCFH. A significant reduction is not expected unless a major redesign of the primary leak areas, such as the oven entrance and exit, is performed. The better enclosed the oven and the higher the gas flow rate, the lower the overall 0, content. The optimum value depends on the oven design.
FIGURE 3 Oxygen level map at various inlets. Two other sources of oxygen concentration increase are exit cooling fan operation and venturi exhaust flow rates. These were a factor in all the ovens evaluated. Since the cooling fans are outside the controlled inert atmosphere, they introduce air that can be reflected off of PCBs exiting the oven. Potential solutions would be to redirect the air circulation or to enclose the cooling fan area and supply it with N,. Most vendors were typically able to maintain an inert Stencil Cleaning the Clean Way! The Atlantis Closed Loop Aqueous Stencil CleanerA that is environmentallv cr endy 1. Closed Loop Wash and Rinse Tanks 2. Computer Controlled Cycles 3. Continuous ph Monitoring Systems 4. DualFiltering 5. Optional Waste Water Treatment 32 CIRCUITS ASSEMBLY AUGUST 1991..-. Exeter Corporate Park Watson Brook Road Exeter, NH 03833 Telephone: (603) 778-8500 FAX (603) 778-8689 CIRCLE 21
CIRCLE 24. -.--...--. mal profile. We suspect, however, that insufficient $&eating of the incoming gas could have deleterious effects. These issues will be studied further in a number of ctory trials. A final issue, not specific to processing LRSPs in an ert environment, is differential board temperature (AT). This value should be kept to a minimum to reduce thermal damage to the PCB and facilitate good solder reflow without overheating. These benefits are achievusing a forced-convection/ir reflow oven. Because process requires introduction of inerting gas, this of reflow oven is an automatic candidate. One of ore difecult-to-reflow PCBs exhibited a 50% ree is, however, a down side to forcedtion/ir reflow using an inert atmosphere. The e of the gas flows must be more stringently mainotherwise, a net gas flow in one direction can be ed and air will be drawn into that end of the oven. flow of LSRPs studying the effects of oxygen concentrations on mance of LRSPs, we attempted to in a standard-air and -temperature nment. The results ranged from highly inadequate. In one case the solder did not rex dried and hardened, leaving a of solder balls instead of joints. At the other end board after reflow. In some cases a level of 800 ppm did not severely impair joint quality, but an increase in solder ball formation occurred. These results indicate that the low-residue solder pastes are probably more forgiving in terms of 0, concentration than originally expected. However, the optimum conditions can be determined only after defect data from this and subsequent soldering operations are analyzed. The most promising no modified process a nology assembly. The best available materials are ade- S quate, and the rapid evolution these materials have experienced indicates that better ones will probably C be available in the near future. 2 We have defined a process using commercially available $ equipment for implementing these materials into produc- 2 tion. Further optimization will be performed as the process ' is studied in a manufacturing environment. E2 John Morris is a member of the technical stafl and James Conway is a senior technical associate with AT& T Bell Laboratories, Princeton, NJ. less postreflow residue than with ion pastes. The remaining air reeen these extremes, the majority iny, dull joints with solder balls. flow rates to the heated zones flow rates to the entrance, exit and cooling zones flow rates to the system's highly sensitive venturi Concentration Effects initial uncertainty regarding the order-ofconcentration of 0, that would be necessary good solder joints with LRSPs. A level of d to contain 54 ppm 0,. Some of the e low-residue solder pastes were eliminated of this information. The use of the standard file was continued. candidate LRSPs were evaluated in 800, 1,000, and 1,500 ppm 0, concentration es. The results indicate that a level of 5500 1 BEX supplies many of the major P.C.B. equipment I] manufacturers. We have direct replacements for P.C.B. equipment, E]... :- available in Polypropylene, PVDF, Brass, 303SS, 316SS, PVC, CPVC, andmore. k:j k].... If you would like to improve your ChemicaVCleaning Operations, call for a FREE sample of our new models specifically designed for your needs. ;<;!. 2: ' 0. :. BEXm... :.. a. :..... :..... :.. SPRAY NOZZLES BEX, Inc. 37709 Schoolcraji Road Livonia, MI 48150-1009 (313) 464-8282 * : :.:..:. -..: : :. *.*: 12: