Printed Wiring Board Industry: Part 2 - 2 Clusters For Rigid Multilayer PWB Manufacturing
For purposes of this use cluster profile, the fabrication of rigid multilayer PWBs has been broken down into nine process steps (see Figure 2-4). These steps form a generic process flow, with many processes and potential alternative processes within each function. Each process step and its associated use cluster is described below, identifying the most common processes, alternatives (where appropriate), and the general technology trends.
A. Data Acquisition and Computer-aided Design (CAD)
The method by which PWB shops receive circuit images and other data has changed dramatically over the past 15 years. As late as 1980, manual methods of circuit layout were not uncommon. These included the familiar hand-tape method, whereby a layout artist would apply strips or doughnuts of black tape onto a mylar sheet. The resulting circuit artwork, usually on a 2: or 4:1 scale, would be submitted either to a photographic service bureau or the PWB shop for photographic reduction. Photo-tools were created from contact prints of the reduction. Paper plots from early computer-aided design (CAD) systems were also photographically reduced. Drilling data was often a diagram or plot of the circuit image, with letter codes placed beside hole locations to signify hole size.
The wide use of vector photo plotters had a significant, but not revolutionary, effect on the front-end of PWB manufacturing. These plotters, for the most part now obsolete, consisted of a light source, aperture wheel, and moving table. Film was exposed trace by trace. Rather than creating artworks, 1-1 film plots were generated and the reduction step eliminated. Vector plotters were generally too slow to be employed for imaging large, step-and-repeated, ready-to-use photo-tools. For example, an 18 x 24-inch image could have taken several hours per layer to plot on a vector plotter. Therefore, PWB shops were still required to manually create photo-tools by making contact prints of the photo-plotted original and registering them to the hole pattern of a drilled panel.
Personal computing equipment, inexpensive circuit layout software, and laser photo plotters have eliminated manual methods of circuit layout. Currently, virtually all circuit layout is done on computers and the output image files are sent to laser photo-plotting service bureaus or PWB shops with laser photo plotters. Since these machines are much faster than their vector predecessors, it became practical to manipulate the image data (and the associated drill and route data) prior to photo-plotting. CAD software became available and performed simple editing of image files, such as step-and-repeating, adding borders, vents, or text even adding, changing, or deleting circuit features. The image file sent to the photo-plotter would return as a ready-to-use photo-tool.
In the modern PWB shop, data is received by modem or magnetic media. The circuit image files (usually "Gerber" files, so named for the company that created the format for its vector photo-plotters), along with drill and route programs, are displayed for sales, quoting, and overall manufacturability purposes. Thereafter, the image files are manipulated and edited to produce the image files for a customized photo-tool. The department responsible for this data processing, usually referred to as the CAD department, may also perform design checks and other engineering functions. Higher-end software packages include design-rule check modules to automatically search the image and drill files for violations such as power-to-ground shorts, trace and space widths or annular ring sizes.
B. Inner Layer Image Transfer
The function of this large use cluster is to transfer an image of a circuit layer from a film photo-tool (or directly from data, in the case of direct imaging) to the copper foil of PWB base laminate (Figure 2-5). Two basic strategies, subtractive and additive, exist. Subtractive inner layer image transfer is accomplished through a series of processes that together are referred to as "print-and-etch," and is by far the predominant method. Additive methods of multilayer circuit manufacturing are covered in Section I.C.6 therefore, only one broad technology exists in this cluster. Several secondary clusters exist within this use cluster one is associated with imaging. Direct imaging looms as a major advancement to conventional photo-tool imaging, but has failed to make significant inroads.
Etching is another secondary cluster with two major chemistries as alternatives. Inner layer base material selections can be made that eliminate some of the wet processing in this cluster. We have included in this cluster steps not generally thought of as part of image transfer. These include the material preparation processes of shearing and surface cleaning, photo-plotting, and oxide treatment. In each case it will be seen that specific print-and-etch regimes can be employed to eliminate one or all of these steps making their inclusion logically desirable.
2. Conventional Print-and-Etch
Print-and-etch is a series of process steps that accomplish the goal of image transfer for inner layers from photo-tool to copper (print-and-etch may also be used for outer layers, although complicated somewhat by the need for through-hole plating). During the "print" step, photoresist is applied to the surface copper of PWB material. Photoresist is a light-sensitive organic coating that can be imaged with a photo-tool and a light source. When exposed and developed, the circuit image is transferred to the photoresist (Figure 2-6). During the etch step, copper that is not prot now becomes the etch-resist), is etched away and only the image of the circuit remains. The photoresist is then stripped, revealing the copper circuit remaining beneath it. Print-and-etch is a subtractive process there is much less copper on the layer after the process than before.
a. Photo-tool Creation
Image transfer begins with photo-tool preparation. The photo-tool functions like a negative in photographic printing the image of the circuit is created on the photo-tool. A photo-tool is usually a film image (the film may be mounted on glass or the image transferred to glass) of the circuit layer. It is placed between the light source and the resist-coated panel during the exposing (or printing) process. The line drawn between data acquisition and photo-tool creation is blurred by the fact that photo-tools are "created" virtually at computer workstations, then physically by laser photo-plotters. Prior to the widespread use of laser photo-plotters, photo-tool creation was largely a conventional photographic process handled by the graphics arts (or "camera") department. This department, now obsolete, was responsible for photo-tool creation. Now, photo-tools are laid out by the CAD department, then photo-plotted.
Presently, all PWB shops have in-house laser photo-plotting or have access to laser plotting service bureaus. Furthermore, concurrent with the advent of laser photo-plotting has been the development of inexpensive CAD software. Complex manipulation of image files, such as rotation, mirroring, or the addition of vents, identification marks, tooling marks or targets, and so on, is done on computer workstations in the CAD department. The resulting photo-plots may be directly used as photo-tools for initial production runs. Film developing is usually done in a three- or four-chambered conveyorized developer that includes developer, stop, fix, and drying steps. Silver is present in the small amount of waste water generated. The presence of silver necessitates a method of silver removal (since this waste stream is regulated), although most shops do not approach the limit in their combined waste stream since the total mass of silver discharged from this one process is very small.
Film photo-tools are usually precision punched to match holes punched in the interior layers of the PWB or to match tooling pins preset in exposing frames. Tooling regimes vary from shop to shop but all are designed to preserve layer-to-layer registration. Film photo-tools are inspected for defects by eye or by automated optical inspection devices. These machines are capable of flagging defects in the plotted image due to handling, dust, or other causes.
b. Material Preparation
The PWB base material, or "laminate," consists of a dielectric material sandwiched by copper foils. Both the copper foil and the dielectric core are manufactured in various sizes. Foil thickness is stated in ounces/ft2. Core thicknesses range from 3 to 47 mils (1 mil = 0.001 inches) for inner layer manufacture. Most PWBs are manufactured on glass-woven epoxy-based materials, with the most common resin system being FR-4 (flame retardant-4). Other materials are selected for specific electrical, physical, performance, or cost characteristics. Two examples of alternative materials include: high-temperature stable/low-dielectric constant material is sheared to panel size, then cleaned mechanically, chemically, or by a combination of both. The purpose of this cleaning step, referred to as "pre-clean" or "chem-clean" is to remove surface contamination, including any anti-tarnish coating present and to condition the surface copper topography so as to promote the subsequent adhesion of photoresist.
Mechanical scrubbing methods include abrasive brush scrubbing and pumice scrubbing. Brush scrubbing removes a thin layer of surface copper, thus ensuring a clean surface, but tends to impart stress to thin core material by deforming it during the scrub. Brush scrubbing can also produce a surface not compatible with fine-line circuit designs. Pumice or aluminum oxide scrubbing imparts less or no stress to the material and produces a favorable surface for photoresist lamination, but is known to be ineffective at removing anti-tarnish coatings applied by laminate manufacturers. Thus, pumice scrubbing is often accompanied by at least some chemical cleaning components.
Chemical cleaning is usually accomplished in a conveyorized spray chamber. Two chemistries are sprayed onto the surface of the panel. The first is usually a proprietary product designed to remove anti-tarnish coatings. The second is a micro-etchant such as potassium persulfate, which is applied to further clean the surface and leave a desirable surface finish. A third chamber may include amild anti-oxidizer.
A minority of shops have eliminated the cleaning step by purchasing pre-treated material, often referred to as "double-treat" material. The laminate is purchased with an oxide coating already present and the sheared panels (the laminate may be purchased already sheared to panel size to reduce handling) are immediately ready for resist application. While this laminate is generally more expensive, two wet processing steps are eliminated (cleaning and oxide). Lower discharge and fewer process lines make this alternative attractive, especially to smaller shops. Although economies of scale are realized when material manufacturers treat the surface copper, the overall efficiency of this alternative is clouded by the fact that most of the surface copper on an inner layer is etched away. Thus, most of the resources spent treating the surface copper prior to etching is wasted. Ultimately, the solution to this problem could focus on reducing the amount of materials put onto the laminate temporarily. For example, direct imaging of copper could reduce the need for copper etching. Double-treat material is not used on outer layers because the oxide coating on the copper lines prevents the deposition of nickel, gold and solder onto the copper circuitry. These metals are required for assembly.
The term "imaging" usually refers to three process steps: photoresist coating (sometimes referred to as "lamination"), exposing (or "printing"), and developing. When completed, the panel is ready for etching, with the imaged photoresist now ready to act as the etch-resist. The most common photoresist in use today is dry film. Prior to the rapid conversion to dry film, which occurred in the early 1980s, etch-resists were screened onto panels. The performance benefit and ease of use of dry film ensured the virtual revolution. Other photoresist schemes exist liquid resists can be sprayed or applied by dipping, or even electrostatically deposited. The continuing improvement of dry film performance (0.003-inch traces and spaces are routinely achieved with various dry films) has, however, left little room for the alternatives. Furthermore, modern dry film resists for common print-and-etch functions are fully aqueous and are developed in a simple carbonate solution.
Dry film photoresists can be purchased in a variety of thicknesses, ranging from less than 1 to 3 mils, and a range of products with subtle performance differences are available. Dry film resist is typically packaged in rolls and is applied with heat and pressure to the surface of the panel, usually with a hot roll or cut-sheet laminator. The photo-polymer film layer is sandwiched between a separator sheet that is automatically peeled away by the lamination equipment, and a mylar cover sheet, which may remain atop the resist layer until development when it is removed by hand. Dry film used by the PWB industry is negative-acting film, i.e., that portion of the resist layer which is exposed to an appropriate light source will polymerize and will not develop, whereas the remaining areas of resist will remain soluble and develop away in a suitable developing solution. Thus, the photo-tool for inner layer print-and-etch should be a negative film; the circuit area should be transparent to light, while the non-circuit areas should be opaque. This will result in a resist image of the circuit remaining atop the copper layer after development.
Exposure (printing) is accomplished by placing the panel and its photo-tools in a vacuum frame and directing a light source for the appropriate time at the panel. Exposing equipment consists of one or two light sources and a drawer that holds the vacuum frame. Numerous photo-tool registration strategies exist. The most common is to mount the top and bottom photo-tools onto a glass frame that can be adjusted to bring the two layers into registration. Panels are placed into the frame, which is then closed, a vacuum is drawn, the light source is switched on, and both sides of the laminate are exposed simultaneously.
The development of fully aqueous film resists is performed in a warm sodium or potassium carbonate solution. Concentrations of less then 2% are common. Developing is done in conveyorized spray chambers that have a developer module along with one or more rinse modules. Both sides of the laminate are developed together. The use of semi-aqueous or solvent-developing dry films is no longer common for any application and is very rare for inner layer parts. The spent developing solution, which contains the carbonate and dissolved resist, is one of the largest (in terms of volume) spent process fluid waste streams in the PWB shop.
Panels approaching the etcher have exposed copper and a resist image of the circuit pattern. Etchant, sprayed onto the surfaces of the panel, removes the exposed copper, but cannot significantly dissolve the copper residing under the resist. In this way, a copper circuit is formed. Etching is performed with conveyorized equipment that typically includes a main spray chamber, an etchant flood rinse, and several cascading water rinses. Long conveyorized units that include developing, etching, and film stripping are common only in large production shops. Acidic cupric chloride and alkaline ammoniacal are the most common etchants (sulfuric-peroxide, a common microetchant, is also acid and ferric chloride, dominant in the past, are now rarely found. A complex array of issues surround the choice between the remaining chemistries, and the decision is tied to downstream process material choices as well as economic considerations.
For example, cupric chloride is generally incompatible with the metallic resists (tin or tin-lead) which are commonly applied to outer performance (for fine-line etching), pollution prevention (ease of on-site regeneration and copper recovery), or other issues. In this example, two etching systems are required, cupric chloride for inner layers and ammoniacal for outer layers. Etching, a one-step conveyorized process, is not often a difficult to justify. For this reason, many small shops employ the more versatile ammoniacal etchant for both inner and outer layers. Cupric chloride etchants consist of cupric chloride (CuCl2) and hydrochloric acid. The simple etch reaction is driven by copper's two oxidation states:
Since no metal complexer is present and the reaction is reversible chemically by chlorination, oxidation with peroxide or other oxidizer, or electrolytically, cupric chloride is far easier to regenerate on-site than ammoniacal etchants. Several closed-loop regeneration systems have been developed that reoxidize cuprous chloride and maintain total copper content at desirable levels (usually in the 15 to 20 ounce/gallon range). Chlorination, the most common method, is performed in a closed-loop arrangement in which spent etchant is circulated through the chlorinator and back to the etcher sump. Copper oxide waste is produced. Furthermore, since this etchant is acidic, no attack on the alkaline-sensitive dry film resists occur. Cupric chloride has a similar etch rate to ammoniacal but is not, as mentioned above, compatible with many metal resists.
Ammoniacal etchant is popular due to ease of use and general compatibility with most resists. Ammoniacal etchant systems are comprised mainly of ammonium hydroxide and ammonium chloride. Other ingredients are present to a lesser degree and serve a variety of functions. As with cupric chloride etchant, the etching reaction is driven by the cupric (Cu++) ion:
Ammoniacal etchants are maintained for continuous operation with a feed-and-bleed arrangement based on baume or specific gravity measurements. In this arrangement, a pump is connected to a baume-activated switch. When the baume of the etchant in the sump rises due to the increasing copper concentration, the pump is switched on. Copper rich etchant is removed from the sump while fresh etchant is introduced. In this way, a steady concentration of copper (critical in maintaining a steady etching rate) is maintained. Spent ammoniacal etchant is not commonly regenerated or recycled on-site, rather, it is usually pumped into barrels or tanks and shipped off-site for recycle and metal recovery. on solvent extraction, is commercially available for on-site regeneration. The system incorporates electrolytic copper recovery in addition to etchant regeneration. The spent ammoniacal etchant stream is the largest stream shipped off-site by PWB shops.
e. Resist Stripping
After etching, film stripping is performed. Stripping may be done in-line with etching as simply another spray chamber at the end of the conveyorized etch line or in a tank. For dry film resists, a wide array of strippers exist. The simplest is hot potassium hydroxide (KOH). Simple aqueous strippers tend to remove resist in strips that do not fully dissolve, making them generally inappropriate for spray operations. Monoethanolamine (20% by volume in an alcohol solvent) is frequently the chemistry of choice for many applications. Other proprietary formulations abound. Methylene chloride is declining in use, and is unnecessary for fully aqueous films.
Oxide treatment, common in the plating industry as a paint base, is used in PWB manufacture to promote copper-to-epoxy adhesion in multilayer manufacture. The oxide process line usually contains four or five process tanks and three or four rinse systems (Figure 2-7). The process tanks consist The micro-etch may be persulfate- or peroxide-based. Oxide chemistries are usually proprietary a common oxidizer is sodium chlorite with sodium hydroxide. Other ingredients vary from vendor to vendor. The oxide bath must be quite hot, usually 140 to 150°F or hotter. The process takes 15 to 30 minutes to complete.
3. Image Transfer Options
a. Direct Imaging
There are two approaches to direct imaging, an option that eliminates the need for photo-tooling. In the first approach, the inner lay imaging, but the pattern is generated by direct exposure of the resist with a laser rather than through a patterned film. In the second approach, no photoresist is applied. The dielectric layer is coated with a bound plating catalyst material that is photo-sensitive (either positive- or negative-working). For a positive-working acti presence of an aromatic diazo compound and cupric acetate is used to reduce the Cu+2 ion to its metallic state. Unexposed areas remain non-conductive. (This process is described in further detail in Section I.A.7, Additive Metallizations.) Further plating can be done using some form of electroless plating to build adequate copper thickness. Because the sidewalls of the resulting metal lines are unconstrained in this process, there are design limitations to this option. This process has been successfully demonstrated on ceramic materials.
Direct imaging plotters have been available for a decade or more, but few shops currently employ this method. High capital costs for direct imaging equipment are a barrier for medium to small shops. Furthermore, exposure times are 2 or 3 times longer than with conventional imaging. Ironically, it is the prototype manufacturers, least able to make purchases of this magnitude, that could benefit most from this technology. Film usage in prototype shops is much higher per unit of manufactured PWBs than in large production houses (where a single photo-tool may be used to expose hundreds of panels), and in some cases, prototype manufactures may be able to better tolerate the longer print times.
In addition to the potential cost-savings in a high-mix, low volume shop, elimination of photo-tools may also increase yields in all shops by improving alignment between layers. Film photo-tools suffer from dimensional instability and of the plotting and print area. Film stretch (or shrinkage) may reach several mils in the corners of 24-inch photo-tools and in many cases it is the largest contributor to the overall registration tolerance of the manufacturing process. With direct imaging, only the positional tolerance of the imager (usually claimed to be under 1 mil over 24 inches) contributes to misregistration.
b. Pre-treated Materials
Post-metal etch oxide treatment can be eliminated by purchasing material that is already oxidized or otherwise treated (such as double-treat copper). An oxide line consumes considerable energy and water and can include as many as 12 to 15 process and rinse tanks (if all of the rinse systems are two-stage counterflowing). Obviation of this process step is particularly attractive to small or quick-turn prototype shops. On the other hand, pre-oxidized material may be eliminated as an option in some shops that use automated optical inspection (AOI) to search for defects. The oxidized copper surface, black or dark brown in color, is unacceptable for many AOI units that depend on reflected light.
Lamination is the process of putting together the various layers. While some variety of materials (such as release sheets) exists, this process is quite uniform between manufacturers. Some production shops have used off-site lamination of 4- and 6- layer boards, enabling them to produce multilayer circuits without investing in lamination and other multilayer equipment. As higher layer counts have become more common, most shops have been forced to create in-house multilayer lamination departments, including lamination presses, oxide and desmear pressed lines, and other multilayer-lamination departments.
Lamination may be divided into several steps. First, the multilayer inner layers and ancillary materials are assembled in a stack in preparation for lamination. At this point, the circuit patterns have been generated on each layer of copper/dielectric laminate. The stack is held in registration by tooling pins. A typical circuit stack will include copper foil, B-stage (substrate material with semi-cure epoxy, also referred to as prepreg) and the etched inner layers. B-stage styles are selected to provide the multilayer circuit with the specified overall thickness, and several sheets may be placed between successive inner layers. Several circuit boards may be pressed in one stack or "book" and they are separated by coated aluminum sheets or other release materials. Cleanliness is essential during the lay-up operation; however, not all manufacturers perform this process in a cleanroom environment. Dust and other contaminants can degrade the bondline between copper and epoxy.
During the press cycle, heat and pressure are applied. Modern presses have platens that are enclosed in a vacuum chamber, because vacuum greatly enhances overall performance by removing air and volatiles from the panels as the B-stage cures. Press cycles are usually computer-controlled. The specific cycle is dictated by the substrate employed; for FR-4, temperatures of 350°F (176°C) and pressures of 150 to 350 psi are common, and the cycle requires more than 1 hour to complete. Press platens may be heated with stream or electricity.
Vias (holes drilled through a PWB for the purposes of layer-to-layer interconnection) are drilled into the PWB to connect the inner and outer layers (Figure 2-8). The etched copper pattern of the inner lay After drilling, the inner layer foil extends to the barrel of the hole and is available for interconnection when the hole barrel is metallized. This drilling of PWBs is performed almost universally with computer numerical control (CNC) drilling equipment and tungsten-carbide drills. Most variation between manufacturers occurs in the selection of entry and back-up materials, where a number of options exist these are discussed below. Laser ablation, an alternative drilling technology, has generated considerable interest. Laser ablation is particularly good at removing organic materials but removes copper at a much slower rate.
2. Conventional CNC Drilling
Common CNC drilling equipment consists of a table, which moves along two axes, and one or more drill spindles that move in the z-axis. The machine is controlled by a computer that reads a list of commands referred to as the drill program. These programs are created, nearly ready to use, by circuit layout software and are then step-and-repeated along with the image files for optimum panel usage. Basic commands found in all drill programs direct the machine to move the table to a specified location, cause the spindle assembly to drill a hole, or effect a tool change. Other commands set the "feeds and speeds" or the vertical and rotational speed of the drill as it enters and exits the PWB material. In operation, the drill machine will drill holes at a rate of a few per second, depending on the distance between the consecutive drilled holes and the entry, and withdrawal speeds of the drill. High-end CNC drilling equipment include features such as large tool trays that hold 100 or more tools and facilitate automated tool-use management, broken tool detection, automatic stacking, loading, and unloading.
b. Entry Material
The top layer that the drill enters before PWBs are drilled is called "entry material." The PWBs are drilled in stacks that consist of a sheet of entry material, one or more circuit panels, and back-up or exit material. Entry material is required to reduce or eliminate exit burring and to reduce drill wander, which is the tendency of the drill to briefly skate on the surface before penetrating. A variety of entry materials exist. The most common are paper-phenolic (10 to 24 mils thick), paper-melamine (10 to 24 mils thick), aluminum (7 to 15 mils thick), and an aluminum-clad material consisting of a phenolic, melamine, simple paper, cellulose, or other core. Entry material is discarded for recycle or disposal after use.
c. Back-up Material
The drill bit terminates its downward stroke with the point penetrating the back-up material in order to complete the drilling of the bottom panel in the stack. Back-up materials are generally 0.062 or 0.093 inches thick. Common materials are pressed wood products (pulp or fiber), paper-phenolic-clad with a wood product core, or aluminum-clad with a wood product core. Since the drill penetrates only halfway through back-up material, it is generally flipped over and used a second time before being discarded.
d. Drill Bits and Process
Drills used by the PWB industry range in size from 0.005 to 0.250 inches in diameter. As circuit densities have increased, hole sizes have decreased in an effort to conserve space. Common via sizes today are in the 0.015 to 0.025 inch range, while the leading technology is demanding considerably smaller holes. The basic improvement in conventional drilling equipment to accommodate the small hole demand has been the air-bearing spindle, which is capable of the high rotational speeds (greater than 100,000 rpm) demanded by the delicate small tools. New drills are capable of 1000 to 5000 hits before becoming unacceptably dull. Drills are usually resharpened 2 to 4 times before being discarded.
3. Laser Ablation
Lasers can be used as an alternative to mechanical drilling to form holes (vias). They are especially well-suited to blind via formation when a via is drilled from the surface of a PWB that terminates within the substrate to provide interconnection for some, but not all, of the layers of a multilayer PWB but can do a very good job at through-hole and buried via formation. Buried vias are drilled through the inner layers prior to lamination and provide interconnection for a pair of inner layers (such as layers 2 and 3 of a 4-layer PWB) when the PWB is laminated, the via is buried by the surface layers. Research has been done in blind via formation by groups researching MCMs at IBM and MCCi as well as by equipment vendors such as Litel and ESI. Table 2-1 shows some of the different laser drilling options and their respective advantages and disadvantages.
|CO2 laser||Low equipment cost||Minimum geometry limited
by wavelength to
>30 m, thermal ablation requires cleanup,
serial drilling process (one hole at a time).
|IR/Argon Ion/Nd:YAG laser||Low equipment cost,
smaller features than CO2
|Serial drilling process
(one hole at a time), requires cleanup.
|Frequency quadrupled YAG
laser (UV wavelength)
|Modest equipment cost,
small features possible
not robust, serial drilling process
(one hole at a time).
|Excimer shadow mask||Small features possible||Low throughput (most energy
wasted), need for in situ
masking, laser sources expensive.
|Small features possible,
high throughput, simple optics
|Expensive laser, large
features difficult, potential via density limitations.
Because of the intriguing cost-performance potential of the Litel projection technique relative to other options, MCC's efforts were focused on demonstration of projection excimer (short wavelength/high energy) machining in MCM applications. The Litel technique involves proprietary tooling that serves to concentrate a flood illumination excimer beam source to the areas requiring machining. Shadow mask techniques reflect back laser energy from all areas but the features (vias) to be machined. Since for most applications the net via area is small compared to the part, the majority of the excimer energy is lost and throughput is very low. In contrast, the Litel technique collects energy spatially and concentrates it at the areas to be machined, vastly improving energy density at the surface and improving throughput dramatically. In addition, non-contact tooling is used with the Litel approa masking characteristic of the shadow mask approach. Because of the proprietary nature of Litel's technique, further details of the approach should be acquired from Litel.j
MCC and Litel did a series of experiments to explore the use of lasers for drilling laminate materials. In the first experiment, an array of greater than 100,000 vias having diameters less than 1 mil (<25 m) and spread over an area less than 10 sq. inches were drilled simultaneously in a spin-coated high-temperature thermoset plastic dielectric material. The pitch was as small as 2 mils. Researchers found that:
- Drilling of all holes was finished in less than one minute.
- Vias of less than 1 mil diameter were fabricated on a 2-mil pitch (one via of 1 mil diameter every 2 mils of linear spacing). This is good for high density interconnect.
- Via placement was accurate over the whole drilling area.
- Stray energy from the laser created surface damage in the areas adjacent to the via areas.
A second experiment was done to investigate the applicability of laser drilling for high-end PWBs. In this experiment, several thousand 2- to 3-mil diameter holes were drilled simultaneously in a 1- to 2-mil thickness of polyimide film (Upil two layers of circuitry in the x-y plane and two via layers (connections normal to the z-axis). The vias were larger and more widely spaced than in the previous experiment.
Both wafer and laminate samples showed excellent results. Holes were accurately machined on-center (locally) and corner to corner took less than 30 seconds per site in the second experiment. According to the MCC technical report, standard drilling costs are approximately 0.5 to 1.0¢/hole. Laser drilling has the potential to lower drilling costs substantially (capital + labor + materials).k In addition to the potential cost benefits, laser drilling could have environmental advantages over mechanical drilling. Since vias can be accurately placed and their dimensions controlled, the routing density per layer can be higher. In many cases, this would allow manufacturers to achieve the same level of connectivity with fewer layers and the associated environmental costs.
E. Hole Cleaning
Hole cleaning generally refers to a process called desmear and/or the closely related process of etchback (Figure 2-9). Desmearing holes refers to the removal of a small amount of epoxy-resin from the hole barrel including any that may have been smeared across the copper interface during drilling. The smear on the copper surface, if not removed, would prevent interconnection between it and the electroless copper which is to be plated in the hole barrel. Etchback, performed less frequently on standard materials due to the advances in the performance of desmear chemistries and the subsequent relaxation of most specifications, is the removal of a significant amount of epoxy-resin and glass fibers (as much as 1 to 3 mils) that leaves the copper interface protruding into the hole. The protruding copper surface allows a large surface area for the interconnection with the subsequent copper plating and the surfaces exposed by the removal of epoxy cannot have been smeared during drilling. Most of the demand for etchback stems from military specifications. The necessity, even the advisability, of a 3-mil etchback has been frequently called into question and, for the most part, specifications covering this process have been relaxed.
Deburring and scrubbing are processes performed immediately prior or after desmear or etchback. During drilling, copper burrs may be raised on both sides of the panel by the action of the drill entering and exiting the material. The burrs are sanded smooth on a deburring machine, which consists of a sanding wheel and a conveyor. In wet deburrers, copper dust is carried off in a waste stream. Dry machines usually are outfitted with vacuum units. Deburring is more correctly considered a surface preparation step rather than hole-cleaning. Scrubbing is performed as a surface preparation step prior to electroless copper (and during other stages, such as before solder mask). Scrubbing may be performed similarly to deburring except a much less aggressive surface abrasion occurs. Pumice or aluminum oxide scrubbers, which direct a high-pressure spray of abrasive particles at the PWB are also used for surface preparation.
This cluster has seen considerable change over recent years. Wet chemical methods have been refined and permanganate seems to have been settled on as the desmear oxidizer of choice. Electrolytic regeneration systems are now commonly applied to permanganate baths to greatly extend their life and make overall control of the process quite simple. Plasma desmear/etchback has become a common alternative. Originally, high capital costs proved to be a barrier for wide acceptance of this technology. Used equipment and the availability of somewhat less expensive models have increased overall use. Also driving the popularity of plasma desmear/etchback is the elimination of the wet process line otherwise required, and the fact that it can produce both epoxy-resin and glass etchback.
2. Wet Chemical Desmear and Etchback
a. Wet Chemical Methods
Currently, the most widely used chemistry is sodium or potassium permanganate when significant etchback is not required or specified. Permanganate-based systems remove a thin layer of epoxy-resin (typically less than 1 mil) and smear and are quite adequate for desmear-only applications.
The permanganate desmear line consists of three process baths (Figure 2-10). The first is a solvent conditioner subsequent removal of the epoxy-resin smear. The constituents of this bath are usually proprietary, some include n-methyl pyrrolidone (NMP). The second bath is usually the permanganate itself. The make-up is 8 to 10 ounces/gallon of permanganate in a sodium hydroxide solution. The bath is generally heated to 160°F or higher. Concentration, temperature, and dwell time are all varied to arrive at an optimum epoxy removal. Dwell times are quite long 20 minutes or more is not uncommon. By-products, consisting of the reduced manganate ion (MnO4-2) and other manganese compounds, develop rapidly in the heated bath and are usually removed by filtration. Frequent analysis and permanganate adds are necessary. Electrolytic regeneration units, consisting of a ceramic porous pot, cathode, anode, and rectifier, have been developed and are designed to anodically re-oxidize the manganate ion back to permanganate (MnO4-). These units can be calibrated to effect a steady concentration of the permanganate ion over long periods of time, eliminating the need for frequent analysis while reducing sludge formation and extending overall bath life. The final bath is a neutralization step, designed to remove permanganate from the holes and surface of the panel. Sulfuric acid based chemistry is common.
Other systems have been largely discarded for various reasons. Originally, chromic acid was the chemistry of choice, but health and waste disposal issues have eliminated it as an option.
Concentrated sulfuric acid (usually 93%) is still in use, but generally requires a permanganate step for final hole cleaning, making it a significantly longer and more expensive process than permanganate alone. It still finds application, however, when etchback is required because it is more aggressive than permanganate and it remains as the only viable wet chemistry option. Handling the concentrated acid and operation of the line has proved to be a problem in many shops. The amount of epoxy-resin removal is controlled by the dwell time in the sulfuric bath, which must be precisely monitored. Sulfuric acid does not etch glass, however, and a second step is required to perform the glass fiber etch. Glass etchants include hydrofluoric acid (rarely used), hydrochloric acid, and ammonium bifluoride.
b. Plasma Etchback
Plasma etchback can be used to remove both epoxy resin and glass. By varying the parameters of the etch chamber, responses such as etch rate, throughput, and selectivity of glass fiber to epoxy etch can be controlled. The cost of plasma etch systems appropriate for PWB applications could be below $100,000 for certain applications. Very little process gases are used in the plasma etchback process [typically 100 standard cubic centimeters/minute (sccm) or ~400 gallons/day running at 100%]. In the integrated circuit fabrication industry, it is common practice to vent these unregulated inert gases to the atmosphere, although in some cases they are scrubbed with other gases. Inside the etch chamber, plasma etching is a six-stage process including the following steps:
- Forming reactive species,
- Transferring reactive species to PWB surface,
- Transferring reacted species away from PWB surface, and
- Removal of reacted gases from process chamber.
Critical parameters for the etch process include: etch gases, chamber power, pressure, and gas flow rate. Table 2-2 shows where some of these parameters might be set:
|Parameter||Epoxy Etch Settings||Glass Etch Settings|
|Etch Gas||O2, O2 + CF4, O2 + SF6||CF4, SF6, O2 + CF4, O2 + SF6|
|Power||200 to 2000 Watts||200 to 2000 Watts|
|Pressure||200 to 1500 mTorr||200 to 1500 mTorr|
|Flow rate||10 to 200 sccm||10 to 200 sccm|
Oxygen is required to etch organic materials like epoxy or polyimide. It also increases the etch rate of glass when combined with fluorine-containing gases. Plasma etching can be done on multiple substrates either as a batch process or on one substrate at a time. Batch processing can usually be completed in a few minutes; however, etch uniformity is usually somewhat better on substrates that are processed one at a time. Large substrate size is not a fundamental problem, but increasing the chamber size also increases equipment cost. Etch rates between 0.1 m to 1.0 m/minute are typical. CF4 with O2 are commonly used to simultaneously etch glass and epoxy. Possible by-products from this process include some fluorocarbon molecules.
F. Making Holes Conductive
For holes to serve their intended purpose of creating layer-to-layer interconnection, they must be coated, or plated through with a conductive substance (Figure 2-11). Since PWB substrate material is not condu the substrate is not possible. First, a seed layer of copper or other conductive material must be plated or coated onto the hole walls, or barrels. After the seed layer is applied, electroplating of a relatively thick (0.001") layer of copper is possible.
The next step in PWB fabrication is to provide electrical connection of the various layers of the PWB through the holes created during drilling or laser ablation. We have called this part of the process "making holes conductive." This is accomplished by depositing metal along the sidewalls of the holes (vias). This is usually done with electroless copper plating. Other methods that may offer improved cost and environmental performance are emerging to challenge the existing practice. Until the latter half of the previous decade, all shops were using electroless copper to metallize hole barrels and create the interconnect between circuit layers. The general strategy has been to plate a thin layer of electroless copper to make the conductive surface that is required for electrolytic copper plating. This mature technology has produced reliable interconnects for decades. However, the industry has sought alternatives to electroless copper, and this search intensified during the 1980s. While none of the chemistries present in the electroless copper line are particularly expensive (with the possible exception of the palladium-based catalyst), the typical line is long (17 or more tanks, depending on rinse configurations) and may have 8 or more process baths. The electroless copper line is also a major source of chelated, or complexed, copper. Chelaters, such as EDTA, are common not only in the electroless bath itself, but also in the cleaner found at the beginning of the line. The discharge from this line often must be treated separately for this reason. Falling permit discharge limits and new formaldehyde regulations provided the impetus for the search for non-copper, formaldehyde-free alternatives.
During the first part of this decade, three basic alternatives were emerging, and still others remain to be tested. The first alternative, a carbon-based proprietary product called Blackhole®, has gained respectability and provides several definitive advantages over the conventional electroless copper process. The advantages include no chelated metal sources (in fact, aside from the conventional micro-etch chemistry, no copper sources exist at all), no formaldehyde, and a much simpler process. The same advantages are offered by the second product, a graphite-based proprietary product named Shadow®, which coats the hole barrels with conductive graphite rather than carbon. The Shadow® process contains only 4 process tanks (including a conventional micro-etch and anti-tarnish). The third alternative is a group of products that avoid the use of electroless copper by heavily seeding the holes with a palladium activator. These processes remain relatively complex (one product has nine process tanks), but are generally formaldehyde-free.
Other systems are in development. DuPont has produced an excellent technical bulletinl detailing 17 different approaches to direct metallization. An electroless nickel system, which has the considerable advantage of being a mature technology, is being tested, and it probably contains neither formaldehyde nor chelaters. The "Lomerson" system, described several years ago, in which the retracting drill bit smears the hole barrel with a conductive substance, continues to generate interest due to its obvious efficiencies.
Despite the commercial availability of the aforementioned three alternatives, electroless copper remains entrenched as the dominant process. Many shops have waited for second or third generation products and do not feel compelled to change. Nevertheless, the industry appears poised to move away from electroless copper during the remainder of this decade.
2. Electroless Copper
The electroless copper process consists of four basic segments: cleaning, activation, acceleration, and deposition (Figure 2-12; see full-size graphic). An anti-tarnish bath is common after deposition. Virtually all shops purchase a series of proprietary chemistries from a single vendor that are used as the ingredients for the several process baths in the electroless copper process line. Only the micro-etch, its associated sulfuric dip, and the anti-tarnish baths are likely to be non-proprietary chemistries.
The cleaning segment begins with a cleaner-conditioner designed to remove organics and condition (in this case swell) the hole barrels for the subsequent uptake of catalyst, followed by a micro-etch step. The cleaner-conditioners are typically proprietary formulations, and mostly comprised of common alkaline solutions. Micro-etching, found on many PWB process lines, is itself a cluster of three chemistry alternatives: sulfuric acid-hydrogen peroxide (consisting of 5% sulfuric acid and 1% to 3% peroxide) is most common, followed by sulfuric acid-potassium (or sodium) persulfate (5% sulfuric, 8 to 16 ounces/gallon persulfate), and ammonium persulfate. In each case, the micro-etch bath is followed by a sulfuric acid dip, whi sulfuric-peroxide bath has some advantages, including high copper capacity (3 to 4 ounces/gallon) and easy waste treatment (chilling to <40°F causes copper sulfate crystallization). Ammonium persulfate is uncommon due to high waste treatment costs.
b. Activation and Acceleration
Activation, through use of a catalyst, consists of two process tanks. A pre-dip, for the drag-in protection of the expensive activation (also called catalyst) bath, usually contains hydrochloric acid and possibly tin or sodium chloride. The activation bath itself consists of hydrochloric acid, tin chloride, and palladium chloride. The Sn+2 ion reduces the Pd+2 to Pd, which is deposited on the panel. The remaining Sn+2 and Sn+4 are selectively removed from the hole barrels by the accelerator (also called the post-activator). Fluoboric acid is a common accelerator, as is sulfuric acid with hydrazine.
c. Copper Deposition
Electroless copper baths can be divided into two types: heavy deposition baths (designed to produce 75 to 125 micro-inches of copper) and light deposition baths (20 to 40 micro-inches). Light deposition must be followed immed more common heavy deposition can survive the outer layer imaging process and copper electroplating occurs thereafter. The main constituents of the electroless copper chemistry are sodium hydroxide, formaldehyde, EDTA (or other chelater), and a copper salt. In the complex reaction, catalyzed by palladium, formaldehyde reduces the copper ion to metallic copper. Formaldehyde (which is oxidized), sodium hydroxide (which is broken down), and copper (which is deposited) must be replenished frequently.
Most heavy deposition baths have automatic replenishment schemes based on in-tank colorimeters. Light deposition formulations may be controlled by analysis. Formaldehyde is present in light deposition baths in a concentration of 3 to 5 grams/liter, and as high as 10 grams/liter in heavy deposition baths.
When light deposition is applied, the next process step must be electrolytic copper plate. This is either a full panel plate (the typical 1 mil is plated in the holes and on the surface) or a "flash" panel plate, designed only to add enough copper to the hole barrels to survive the imaging process. Flash-plated panels return to copper electroplating after imaging to be plated up to the required thickness. This double plating step has made heavy deposition the more common electroless copper process.
d. Process Waste Streams
The electroless copper line typically contributes a significant portion of a PWB shop's overall waste. Water use is high due to the critical rinsing required between nearly all of the process steps. Copper is introduced into the waste water stream due to drag-out from the cleaner-conditioner, micro-etch, sulfuric, accelerat this copper is complexed with EDTA and requires special waste treatment consideration. Furthermore, waste process fluid generation is high. Micro-etch baths are exhausted when 2 to 4 ounces/gallon of copper is dissolved and this bath life is usually measured in days. While the electroless copper bath is relatively long-lived (usually several weeks or months), a considerable bail-out stream (including formaldehyde) is generated (several gallons of concentrated bath chemistry per day in production shops).
3. Carbon-based Alternatives
The only carbon-black dispersion process (Blackhole® by MacDermid) became commercially available in 1989. The advantages of this type of process versus conventional electroless plating are numerous, if one assumes that overall performance is similar (performance parity has not been established). Production rates are higher for Blackhole® since it is applied in about half the time required for electroless copper, formaldehyde is not a constituent of any of the process formulations, and copper is dragged into the wastewater stream from only the micro-etch bath. Furthermore, overall water use is reduced. Although it has become more common in larger shops, this process is still an unusual choice for small shops with sales of less than $5,000,000/year, which account for the majority of PWB facilities in the country. Capital costs for the Blackhole® conveyorized process line are much higher than for an electroless copper tank line, and represent a barrier of entry for small shops. Payback from production time savings and waste reduction are likely to be quite long for small manufacturers.
Although originally designed for either batch immersion or continuous conveyorized spray application, refinements in the carbon-black dispersion process have eliminated the immersion option. The conveyor consists of 11 chambers, 5 process baths, 4 rinses, and 2 dryers. The cleaning and conditioning processes are plumbed to share rinse water fresh water is used for the conditioner rinse and then re-used for the cleaning rinse before being discharged from the system. The carbon-black dispersion bath itself follows. The carbon deposition layer is dried and then removed selectively from the surface copper foil by a conventional persulfate micro-etchant. A conventional anti-tarnish is the final step in the process. The anti-tarnish and micro-etch processes are also plumbed to share rinse water.
Processed panels proceed to either imaging or panel plating. No special considerations or downstream process changes are required. The carbon layer is rapidly coated with copper during electrolytic copper panel or pattern plating and the resulting through-hole interconnection has passed industry standard performance testing (such as MIL-P-55110D).
4. Graphite-based Alternative
The only available graphite-based process (Shadow® by Electrochemicals) is very similar to the carbon-based alternative discussed above. In this case, graphite particles suspended in a colloid are dispensed onto the surface and act as conductive pathway for electroplating. The graphite solution is proprietary. Four specific process steps are required: cleaning/conditioning, graphite (Shadow®) application, micro-etch, anti-tarnish. Graphite application can be done on conveyorized equipment or with an immersion (vertical) process. One pass through the graphite application step is sufficient to prepare holes for copper deposition. The vendor claims to be able to run this process from cleaning to lamination in less than 15 minutes.m Graphite particles adhere well to the laminate surfaces and can tolerate mechanical scrubbing. The micro-etch solution is usually based on persulfate. Some facilities do not use anti-tarnish coatings on the copper surface if the dry film lamination process is done in-line immediately following graphite application.
a. Cleaning and Conditioning
Panels can be cleaned using conventional methods or in the conveyor line. Cleaning solutions used are similar to systems employed for electroless copper. As noted above, the cleaner-conditioner step is designed to remove organics and condition hole barrels for the subsequent uptake of catalyst or graphite in this case. Wet chemical and dry etch methods for cleaning and conditioning are discussed in Section II.E.2.
The graphite is applied in the form of colloidal graphite. The particles are suspended in a slightly caustic colloidal solution (pH = 9). In the conveyorized process, up to 200 18" x 24" boards come out of the cleaning step and are dipped through a liquid colloid-containing graphite. Surface activation is rapid and solution that runs off the boards can be reused. In an immersion (vertical) setup, conventional process tanks are used to apply graphite. In this case, some efficiency is lost but existing equipment can be used. Following the graphite application, the panels must be dried to fix the graphite on dielectric surfaces. This step is critical for obtaining good copper adhesion during subsequent copper plating. Drying is done with air knives followed by a short bake (170°C). The air knives used are run at close to room temperature; their principal function is to knock excess solution off the panels. As such, these air knives consume far less energy than ones used for hot air solder leveling (HASL). Very little waste is generated during graphite deposition because panels are dipped through a graphite-containing solution. No rinse is required&
Micro-etch follows graphite application. The drying step does a good job at removing colloid material from the field area of the panel. However, simply drying the solution does not remove graphite particles from exposed copper. The vendor recommends spraying a persulfate solution onto the panel surface. A filter below the panels prevents graphite from entering waste streams. Also, the graphite captured at the micro-etch step is returned to the process at the graphite step.
Following micro-etch, the exposed copper is subject to oxidation. Because of this, an optional "anti-tarnish" step can be done. The system vendor recommends using a benzothiadzole-containing solution to protect copper surfaces. Some PWB manufacturers skip this step, in which case it is recommended that image transfer films be laminated onto the panels directly following micro-etch. Skipping the anti-tarnish step should reduce chemical use, cost, waste, and cycle time. Effects on reliability were not reported but some commercial vendors use this approach.n
e. Process Waste Streams
The quantity of waste water produced by a typical horizontal conveyorized graphite application system is less than 5 gallons/minute. Also, wastewater is only produced when the system is running, rather than constantly, as in the rinse tanks used for electroless copper plating. In short, the graphite system seems to reduce both waste streams from PWB manufacturing and PWB manufacturing costs. There are five principal environmental benefits from using the graphite process instead of electroless copper. These include:
- Reduction in chelated copper and metal waste,
- Elimination of formaldehyde,
- Reduced water use,
- Reduced treatment chemical use, and
- Reduced sludge disposal.
5. Palladium-based Alternatives
Another alternative to electroless copper is palladium-based, which has the benefits of completely removing formaldehyde from the process, reducing the amount of water consumed to make drilled holes conductive, and reducing the amount of wastewater generated from such a process. It can also be run with conveyor equipment to increase throughput and lower cost. Like the graphite process discussed above, panels can be sent directly from a palladium-based process to a dry film laminator.
The first stage of cleaning/conditioning is the "sensitizer" step. Two things must be accomplished: first, through-holes must be cleaned, and second, a charged polymeric material must be applied to the inside dielectric surface of holes. This charged material can then receive a catalyst during subsequent processing. Cleaning is accomplished with the desmear chemicals discussed in Section II.E.2. Chemical use and waste water produced in the cleaning step of the palladium-based process is similar to volumes used for the carbon- and graphite-based processes. However, each step must be followed by a rigorous rinse to prevent contamination of the following steps.
- Pre-dip: During pre-dip, a base salt version of the catalyst (which does not contain catalyst metal) is applied to the panels.
- Activator: After pre-dip, the catalyst (activator) is applied. In this case the catalyst is palladium/tin in a colloidal solution. The catalyst adheres well to glass/epoxy laminates.
- Accelerator (Enhancement): After the activator step, panels are rinsed with a caustic soda (accelerator or enhancer). The accelerator removes stannous tin (Sn+2) or reduces it to a metallic form (Sn). Also during this step, metallic palladium is converted to palladium sulfide (PdS).
- Alkaline Rinse
Micro-etch is done with hydrogen peroxide (H2O2) and sulfuric acid (H2SO4). This step removes excess palladium sulfide from exposed copper surfaces without oxidizing the copper. This step enhances adhesion between exposed copper and electroplated copper added later without degrading adhesion between laminate materials and plated copper.
After micro-etch, the panels are dried to prepare them for lamination.
d. Process Waste Streams
There are two principal process waste stream considerations: first, formaldehyde is completely eliminated from the process of making holes conductive; and second, the amount of water used (and waste water produced) is greatly reduced when compared to conventional electroless plating. According to Shipley, the maker of one palladium-based process, the cost of this process is competitive with electroless copper plating.
G. Outer Layer Image Transfer
This large cluster includes outer layer imaging, copper plating, etch-resist plating or application, etching, and etch-resist stripping (Figure 2-13). The cluster includes copper electroplating& with image-transfer copper-plating. Although arguably an independent function, copper electroplating is performed in a sequence determined by the overall image transfer strategy. Furthermore, with copper sulfate being the overwhelming choice of PWB shops, a cluster of alternatives does not actually exist (pyrophosphate baths, the other chemistry, have vanished). With the exception of process material improvements such as dry film photoresist, plating chemistries (copper sulfate vs. pyrophosphate, in particular), or the substitution of tin-only etch-resist for tin-lead, this cluster has remained essentially unchanged for many years. It also forms the core of double- and single-sided processing, thus many of the processes described here predate multilayer manufacturing.
Two major subtractive options, starting with copper clad laminant, are available to the PWB manufacturer. The first sequence, print, pattern-plate, and etch, is the most common. While copper pattern-plating is a uniform process shop-to-shop, the etch-resist metal plated over the copper is not and forms an interesting and important second level cluster. Etching is theoretically a cluster of two chemistry options, but with most metallic etch-resists, only ammoniacal is possible. In the other sequence, panel plate, print, and etch is less common. Here, copper is plated to full thickness over the entire panel prior to imaging. The photoresist imaged during the "print" process serves as the etch-resist precisely as in inner layer image transfer.
It can be seen that the panel plate, print, and etch process (also referred to as "tent-and-etch" because the drilled holes are tented over and protected from the etchant by dry-film photoresist) eliminates process steps required in the more common pattern-plate process. There is no etch-resist plating (the photoresist serves as the etch-resist) or metal stripping Therefore, the tin or tin-lead problem is obviated. Furthermore, cupric chloride etching is an option when dry-film photoresist is the etch-resist. Unfortunately, it is the inefficiency of panel plating, along with certain technical limitations of this process, that prevent its widespread use. Most of the copper on a typical circuit panel is etched away, thus most of the plated-on copper of this process is promptly removed during etching, unlike the copper added during pattern plating. Furthermore, the panel-plated copper can cause difficulty in etching, particularly fine-line etching. The thicker the copper to be etched, the greater the undercut. This problem has dampened the enthusiasm for tent-and-etch. On the other hand, layout design changes can easily rehabilitate tent-and-etch. One method, referred to as "pads-only outer layers" eliminates difficulty of fine-line etching on outer layers (along with eliminating the need for solder mask) at the expense of requiring two extra layers of inner layer circuitry.
It should be pointed out that several of the process options listed here and in Section II.H are not true options, but rather, alternative methods required in some cases by customer specification and end-product performance. Although tin and tin-lead perform identically as etch-resists, shops that are otherwise anxious to remove lead plating are unable to do so. A small number of parts continue to require reflow (a finish necessitating tin-lead plating) rather than hot-air-solder-level (HASL) (a finish for which tin can be substituted for tin-lead).
2. Image, Pattern Plate, and Etch
Outer layer imaging is quite similar to inner layer imaging. The panel is thicker and has drilled holes, but it is essentially processed the same. Small shops may use the identical photoresist product for both inner and outer layers for convenience. Photoresist thickness, not critical for an etch-resist, takes on significance as a plating resist. Most shops use a thinner resist for inner layers (0.001 inches) and are forced to use thicker resist for outer layers (typically 0.0015 or 0.002 inches). Generally, the resist thickness should equal or exceed the thickness of the metals to be plated onto the pattern to avoid copper or tin-lead "mushrooming" over the top of the resist. Resists other than dry film are extremely uncommon for outer layer imaging.
Exposing may be done with first-generation photo-plotted photo-tools or with diazo, a reddish transparent film that allows for manual registration. With a diazo photo-tool, an operator can see through the dark areas of the film (the circuit pattern) and can align the photo-tool to the hole pattern, eliminating the need for tooling regimes. Although not practical for production shops, manual registration with diazo photo-tools is not uncommon in prototype shops. When pattern plating is to follow, outer layer photo-tools are positive images of the circuit. When developed, the circuit image is developed away, exposing the underlying copper. The photoresist remaining on the panel is the plating resist for the pattern plate process. Developing is done in a sodium carbonate solution (1% to 2%).
b. Pattern Plating Copper
Pattern plating, so named because only the circuit pattern and hole barrels are plated, is a two-step process (Figure 2-14; see full-size graphic). First, copper is plated. Only the thin electroless copper layer has been deposited in the hole barrels up to this point in the process and it is far short of the typical 0.001-inch specification for copper thickness. Copper is also plated onto the circuit pattern. None of the copper plated during this process is etched away, but rather, remains on the circuit and is part of the finished product. Second, immediately after copper electroplating, a metallic etch-resist is plated over the copper, usually tin, tin-lead or nickel-gold.
Although a few copper electroplating chemistries exist, nearly all modern PWB shops use simple copper sulfate. The bath has an extremely long life (measured in years) and is generally easy to maintain and control. The bath is typically made with 10 ounces/gallon of copper sulfate, 25 to 40 ounces/gallon of sulfuric acid, and a small amount of hydrochloric acid to provide a chloride concentration of 30 to 90 ppm. Proprietary organic additives, usually referred to as brighteners, distinguish one vendor's bath from another. The pre-plate line consists of an acid cleaner (dilute phosphoric acid is a common constituent), a micro-etch, and a sulfuric pre-dip.
High-performance copper plating is reliably performed at a current density range of 20 to 35 amperes/ft2. Manufacturers generally plate from 0.0013 to 0.0017 inches of copper to ensure that all hole barrels meet the minimum of 0.001 inches in all areas of the panel. Dwell times depend on current density and target thickness, but generally range from 30 minutes to somewhat more than one hour. Although a source of copper in the waste water stream, the copper sulfate plating process lends itself to several recovery schemes. A drag-out tank immediately following the plating bath can be electrowinned (an electrolytic recovery process) while in service to maintain a low copper concentration therein (reducing copper dragout to the flowing rinses), while recovering dragged-out copper in metallic form. Some have employed ion exchange to produce a closed-loop. In this arrangement, waste water is purified and reused as rinse water and the cation regenerant is returned to the copper sulfate bath.
c. Pattern Plating Etch-Resist
Immediately after copper pattern plating, an etch-resist metal is plated over the copper. Tin-lead has been the most common etch-resist fo reflowed solder, tin-lead efficiently serves as both etch-resist and finish. With the rise of the solder-mask-over-bare-copper (SMOBC) surface finish method, tin-lead plating has become unnecessary for most panels. Shops facing prohibitive lead discharge limits have rapidly embraced tin-only plating, which has become the primary alternative. While the regulatory status of tin varies from locality to locality, it is safe to say that tin concentration receives considerably less scrutiny than lead. Many shops have, however, been reluctant to eliminate tin-lead plating due to the fact that a certain percentage of their work still requires a reflowed finish. Some military specifications continue to call for tin-lead reflow so tin-lead plating is quite common in shops serving military clients. Still other shops have been forced to maintain both tin and tin-lead plating baths to satisfy their customer base. As an etch-resist, there is no difference in the performance of tin or tin-lead.
Two tin-lead plating chemistries currently exist. The fluoborate bath is most common. This bath consists of fluoboric acid, tin and lead fluoborate, and proprietary organic additives. Stannous tin (Sn+2) is maintained at 2 to 4 ounces/gallon and lead at 1 to 2 ounces/gallon. The other chemistry, which is entirely proprietary, consists of methane sulfonic acid (MSA) and is not in wide use. Tin-lead is plated to a thickness of 0.0002 to 0.0005 inches.
Tin fluoborate can be used as an alternative to tin-lead plating. Some shops have phased out tin-lead by simply replacing the tin-lead anodes concentration to gradually fall. Tin sulfate, however, has become the tin bath of choice. Tin is plated for etch-resist purposes only a 0.0002 inch thickness is adequate. The sulfate bath consists of 20% sulfuric acid and enough stannous sulfate to provide 2 to 3 ounces/gallon of stannous tin.
Nickel-gold is also pattern plated electrolytically as an etch-resist and surface finish. Electrolytic soft gold is the surface finish of choice for certain performance considerations, including high corrosion resistance, low contact resistance (although wear resistance is poor), and long shelf-life. Electrolytic pure gold may be called plating line consists of a nickel pre-dip, the nickel plating bath, a gold pre-dip, and the gold plating bath. The nickel plating chemistry of choice is nickel sulfamate or nickel sulfate. Nickel is quite concentrated in either bath and reaches 17 ounces/gallon in typical nickel sulfate formulations. Unlike many other process baths, many shops use decades-old nickel plating formulations and do not purchase proprietary chemicals. Nickel is plated to any specified thickness, usually in the range of 50 to 500 micro-inches. Gold is then immediately plated over the nickel. Acid gold cyanide formulations are most common and are similar, but not identical, to the hard gold baths designed for edge connector plating. Sulfite-based alkaline baths are also in use. Gold is generally plated to a thickness under 100 micro-inches, and for many applications 10 to 30 micro-inches will suffice. Both the nickel and gold plating baths are quite long-lived barring unusual events, each may
Other etch-resists include tin-nickel and tin-lead over tin-nickel, well-known to have certain performance advantages over conventionally processed boards, but are rarely found. Metals such as rhodium or gold may be plated selectively over certain areas of a circuit, then masked off and a conventional etch-resist is plated over the remainder. Rhodium is the metal of choice for maximum wear resistance.
d. Photoresist Stripping
Photoresist is stripped from the outer layers in the same chemistry (often in the same bath or spray chamber) as the inner layers. When tin-lead is the etch-resist, small amounts of lead can be removed as metal or dissolved in the stripper solution, which complicates waste treatment. It should be noted that, although resist stripping is a simple, one-tank operation, the majority of a shop's total resist consumption is dissolved in this bath (the remainder is dissolved in the developer). Both the developer and the stripper have very short bath lives (often measured in hours) compared to most plating solutions, and these operations generate a large volume of waste process fluid.
e. Outer Layer Etching
Ammoniacal etchant is the only etchant used for outer layer panels plated with a metallic etch-resist (Figure 2-15). The etchant is generally inert to both tin and lead, although some manufacturers do report low concentrations of lead present in their spent ammoniacal etchant. Etching of outer layers is otherwise identical to that of inner layers described in Section II.B.2.e. Spent ammoniacal etchant is a major waste stream and is usually transported to recycling plants.
f. Tin and Tin-Lead Stripping
Etch-resist is stripped off the panel immediately after etching for the solder mask on bare copper (SMOBC) process. Tin-lead may remain on the board if reflowed tin-lead is the specified finish; in these cases the strip process is bypassed. Previously, reflow was a common finish. It is now quite rare for multilayer circuits, but still is found in "low" technology single- and double-sided circuits. Tin or tin-lead stripping chemistry is proprietary. Nitric acid, ammonium bifluoride, and peroxide-based systems are available. The stripper bath-life is moderate. If lead is present on the s major source of waste lead.
3. Panel-Plate, Print, and Etch ("Tent-and-Etch")
With the exception of the additional panel plating (so named because the entire panel is plated, not just the circuit image) step, this process precisely tracks the inner layer print-and-etch method described in Section II. fewer process steps, requires less time, and creates somewhat less waste. Although copper is inefficiently plated, then etched, etching can be performed in cupric chloride or ammoniacal systems.
One major drawback of tent-and-etch is the difficulty of etching through the plated-on and base-laminate copper layers. With circuit densities steadily increasing, circuit features have become smaller and smaller. Most shops are required to produce patterns of trace widths that measure less than 10 mils across. Fine-line etching is quite difficult if the sum of the copper residing on a panel is 2 to 3 mils thick, which is the typical sum of 0.5 ounce/ft2 base laminate copper and the panel plating. A second problem arises when SMOBC is not the surface finish required. Electrolytic gold and tin-lead must be pattern plated in most cases. Thus, a shop faced with a variety of requirements cannot use tent-and-etch exclusively.
H. Surface Finish
For most parts, the functions of the surface finish are to prevent copper oxidation, facilitate solderability, and prevent defects during the assembly process. Other surface finishes are dictated by the environment in which the part will reside or by specific performance criteria. It is in outer layer finishing that the greatest variety of manufacturing options and processes exist (Figure 2-16). Furthermore, t may vary from shop to shop. For example, gold edge-connector plating may occur either before or after solder mask application.
Solder-mask-over-bare-copper (SMOBC) with hot-air-solder-leveling (HASL) has become the industry-wide standard finish, replacing tin-lead reflow, which is now an unusual finish for multilayer parts. Nickel-gold is a significant alternative finish. Nickel-gold coatings may be electrolytically plated as an etch-resist (a direct substitute for tin or tin-lead) or they may be applied electrolessly after solder mask application (a direct substitute for HASL). Organic, "pre-flux" coatings, which prevent oxidation and facilitate subsequent soldering, are gaining in popularity. Beyond these finishes are a wide variety of combinations that vary from shop to shop selectively plated combinations of gold, tin-lead, and even rhodium. Shops have produced SMOBC, tin-lead reflow panels by plating tin-lead as an etch-resist, then after etching, selectively str After solder masking, the panels are reflowed.
2. Solder Mask Over Bare Copper (SMOBC), Hot Air Solder Level (HASL)
This method predominates for several reasons. Copper is a surface that lends itself to rigorous cleaning, which is essential for solder mask adhesion. Tin-lead under solder mask will liquefy during soldering and may cause the mask to blister and peel. The hot air solder leveling process generally produces less waste water and introduces less lead into the waste water stream than tin-lead plating and reflow. Despite these advantages, well-known disadvantages also exist. The shelf-life of hot air solder leveled circuits is short and solder thicknesses on pads and hole barrels is notoriously difficult to control. For these reasons, a small minority of specifications continue to call for tin-lead plate and reflow or other alternati air solder leveling, nomenclature screening, and finally, gold edge plating if necessary.
a. Solder Mask
The purpose of solder mask is to mask off and insulate physically and electrically those portions of the circuit to which no solder or soldering is required. Increasing density and surface mount technology have increased the need for solder mask to the point that, with the exception of "pads only" designs, nearly all parts require it. Manufacturers have had some autonomy in selecting masks. Many specifications do not call out a specific product or product type and this has allowed the manufacturer to choose masks based on processing as well as performance issues.
Three basic type of masks are commonly applied: thermally cured screen printed masks, dry film, and liquid photoimagible (LPI). Thermal masks have predominated for decades but are gradually being replaced by LPI, despite being the lowest cost alternative. Dry film has some specific advantages, such as ease of application, but its use seems poised to decline as well in the face of improving LPI formulations.
b. Hot Air Solder Level (HASL)
The HASL process consists of a pre-clean, fluxing, hot air leveling, and a post-clean. Pre-cleaning is usually done with a micro-e peroxide micro-etch is not common in the process. Dilute ferric chloride or a hydrochloric-based chemistry is favored for compatibility with the fluxes that are applied in the next step.
Hot air level machines consist of a panel transport mechanism that carries the panel into a reservoir of molten solder, then rapidly past jets of hot air. All areas of exposed copper are coated with solder and masked areas remain solder-free. Boards are then cleaned in hot water, the only step in the SMOBC process where lead may enter the waste water stream, albeit in very small quantities. Once cleaned, the panels may again enter the screening area for optional nomenclature screening, or proceed directly to the routing process.
3. Reflowed Tin-Lead
The reflow process, once the predominate method, uses the tin-lead etch-resist to create the final surface finish, thereby creating some efficiency when compared to the plate-etch-strip sequence of SMOBC. It is not uncommon in lower-technology designs. Some military specifications require reflow and explicitly exclude the SMOBC/HASL method. Tin-lead is required, however, and this has contributed some to the demise of reflow. Distinct performance issues have also played a role tin-lead is difficult to clean, is a poor surface for solder mask, and will liquefy during wave soldering, causing defects downstream. The basic function of the reflow process step is to fully encapsulate the copper on the panel (after etching, copper is exposed along the vertical flanks of circuit features) and to create a more durable surface.
Two types of reflowing, or fusing, methods exist. The oldest is hot oil reflow. In this process, the board is fluxed, then immersed in a pot of hot oil long enough to heat the tin-lead plate to its melting point. Infra-red reflow, a somewhat more modern approach, is done in a conveyorized machine that includes a fluxing station, an infra-red oven chamber, and a cool-down chamber.
Nickel-gold finishes may cover an entire circuit or be selectively plated onto certain areas of a circuit. Furthermore, nickel-gold formulations produce hard (cobalt or other metal is co-deposited in small amounts for hardness) or nickel-gold can be either electrolytically or electrolessly deposited.
a. Hard Gold
Hard gold is electrolytically plated. The most common application of hard gold is edge connectors, but hard gold may also be plated over circuit areas as well. Automated edge plating machines are common since manual plating is quite labor-intensive. Typically, a plater's tape is applied to the board, masking off all of the circuit above the edge connector area. The panel is then processed through a nickel-gold plating line, with just the edge connectors immersed in the plating fluid. Nickel is plated first and Watts or sulfamate nickel is common. Cyanide gold is the most common gold electroplating chemistry.
b. Soft Electrolytic Gold
Soft gold is a pure gold over nickel deposit. It may be plated over the entire circuit or selectively over certain portions of a circuit (excluding edge connectors, which require hard gold). Selective electroplating requires a combination of masking and bussing (providing current to the portion of the circuit being electroplated). Selective gold applications include contact points (which may require hard gold), press pads, wire bond sites, or portions of a board that may reside in a corrosive environment. Selective gold plating can be labor-intensive and is not frequently specified for production lots (all gold plating is often substituted; the labor savings offset the extra gold).
c. Electroless Nickel/Immersion Gold
This method of applying soft gold has received considerable attention. Electroless plating can be conveniently performed after etching because no bussing is required. Therefore, these all-gold boards can be processed with a standard tin etch-resist and processed identically as SMOBC, except the gold plating step replaces the HASL step. This process has advantages over SMOBC/HASL and electrolytic gold plating. When compared to SMOBC/HASL, electroless all-gold circuits have a much longer shelf life. The flat surface profile of the electrolessly plated surface-mount pad and overall excellent solderability make electroless nickel/gold ideal for surface-mount technology. When compared to electrolytic gold, electroless has the advantage of full copper encapsulation because plating is performed after etching, not before, as with electrolytic gold plating. Selective gold plating is made somewhat easier by the electroless plating method since no electrical bussing is required. Cost is the main disadvantage. Immersion gold and electroless nickel process baths are short-lived compared to electrolytic formulations and maintenance and control of these baths is more difficult. The main application of electroless nickel-gold coatings is chip-on-board technology, where component leads are ultrasonically or thermosonically bonded to gold pads rather than soldered.
I. Final Fabrication
During the final fabrication process, non-plated through-holes and other tooling features may be added to the circuit, and the circuit itself is completely or partially depanelized. Depanelization is accomplished with numeric controlled routers that are quite similar to drilling machines. With complete depanelization, the circuit is routed out of the panel. With partial depanelization, common with production lots, most of the circuit profile is routed, but it remains tabbed to the panel during testing and assembly. In this way, several circuits can be tested and assembled at once. After assembly, the individual circuits of a panel can be conveniently snapped, or broken, out of the panel. Such panels are often referred to as "breakaways" or "snaps." Tool-and-die depanelization methods are not common with multilayer circuits.