- THE MAGAZINE
- NEW PRODUCTS
Ceramic substrates are used as microelectronic interconnect devices in applications where the device will see high load, high temperature and high shock, and
require high reliability. High packaging density is a common advantage; the vias (the internal interconnects from one layer to the next) are buried under surface components, allowing the surface components to be placed closer to each other.
Ceramic substrates are commonly used in aerospace, military, medical, telecommunications, automotive and high-frequency circuitry applications. The interconnect substrate is made up of individual layers of unfired ceramic tape that is easily cut, punched, printed on, etc. Once all of the layers are complete, they are collated and aligned to each other using registration pins and then laminated and fired. The final product is a single ceramic substrate with all of the internal interconnect layers buried within a solid ceramic body.
As line widths and line spacing of printed circuits are reduced, both high-temperature co-fired ceramic (HTCC) and low-temperature co-fired ceramic (LTCC) manufacturing processes require an accurate screen printing process. All of the electrical connections are made once the substrate is collated, laminated and fired. Any misalignment of a circuit print on any layer can result in the fine interconnect lines misaligning to the vias, resulting in an “open” circuit and a scrapped substrate.
The circuit printing process is comprised of several elements:
• The screen frame
• The image to be printed
• The mounting of the screen frame in the printer
• The squeegee attack angle
• The squeegee speed
• The snap-off distance (i.e., the distance between the tape being printed and the underside of the screen)
• The method by which the tape layer is aligned on the printer stage
• The method by which the image is aligned to the tape layer
• Ongoing process verification
Early versions of the process had low print yields, excessive setup time and poor layer-to-layer alignment when all the layers were collated. Clearly, some well-thought-out changes were needed.
The first area that needed to be addressed was the setup time. The goal was to be printing good prints in five to 10 minutes from the end of the previous layer. This would greatly increase the printer uptime and contribute to greater daily throughput. As an added bonus, addressing the setup time would greatly increase yields, layer-to-layer alignment and screen life across all products. This article addresses innovations developed over a 14-year period at a medical company producing multilayer HTCC circuits at volumes of over 10,000 substrates per week.
The process engineer in charge of the print process decided to perform a complete design-of-experiment (DOE) on all of the significant printer setup variables. The intent was to define specific variables that could be “locked down,” thereby eliminating the time spent on adjusting those parameters. The parameters chosen included squeegee angle, squeegee speed, squeegee pressure and snap-off distance. Ink viscosity, screen mesh size/type, and emulsion type/thickness were not considered, as they were already determined and locked from a prior DOE.
As the experimentation progressed, the team quickly discovered that some mechanical features of the printer and screen frames played a significant role in elongating the setup time. First, the screen frames all varied in thickness. Since the screen frame is located in the printer on its uppermost surface, all of the screen frame thickness variation was translated into snap-off variation without even adjusting the snap-off. By changing the way the screen frame locates within the printer, this variation can be eliminated. The screen frame variation is eliminated from the equation when the mounting plate in the printer is changed to accept the screen frames on their bottom-most surface (see Figure 1).
In addition, the screen frames were mounted using four bolts, which required a wrench and took extra time. By incorporating some simple toggle clamps, this time could be mostly eliminated.
Significant time was also spent on aligning the printed image to the tape layer. Several prints were made, and an artwork overlay was used to determine the accuracy of the setup. By adding alignment bushings to each screen frame and alignment pins to the printer mount plate, the screen frame was able to be quickly and accurately located into the printer with only minor adjustment required.
Finally, the printed image on the screen was not consistently located to the screen frame, thereby requiring additional alignment once in the printer. This alignment was eliminated by accurately locating the artwork to the screen frame at the time of light exposure. This was accomplished by adding registration holes to the layer-specific artwork and using a simple see-through alignment fixture during the light exposure process (see Figure 2).
Once the DOE was complete, the squeegee angle, speed and pressure were all locked and no longer needed adjustment. The snap-off adjustment was eliminated when subsequent layer thickness remained unchanged. Having accomplished this, however, the setup time was still too long. The operator still required a lengthy time to achieve good alignment to the artwork, and a lot of debate ensued regarding the use of artwork as a “standard” for quality measurements.
A similar debate was brewing regarding the via-punch accuracy. By using the artwork to qualify the via-punching process, researchers found themselves modifying via-punch electronic files so that the via-pattern would better match the printed image. All of this led to the development of a custom measurement system.
Adding the measurement system eliminated the artwork-related inconsistencies and provided accurate information with which to accurately adjust the via-punches and the printers. However, the setup time was still too long; the operator required too much time to use the adjustment information from the measurement system. This pushed the research team into making a major modification to the printer that eliminated operator intervention in the alignment process. A servo-motor control was added to the X and Y axes, and a separate rotational axis was added so that the three adjustments were independent and individually controlled.
The key event that can trigger some significant paradigm shifts is a change from using the layer-specific artwork as the measurement “standard” to a vision
The key event that can trigger some significant paradigm shifts is a change from using the layer-specific artwork as the measurement “standard” to a vision measurement system that can be used for verifying several processes.
measurement system that can be used for verifying several processes. The first paradigm that must be shifted is the assumption that the screen printer must be aligned to the via-punch hole-pattern in the tape layer, or vice-versa. The second paradigm that must be shifted is that the via-punch via-pattern is naturally in perfect alignment to the via-punch registration hole pattern.
As Tom DeMarco wrote in his book Controlling Software Projects: Management, Measurement, and Estimates, “You can’t control what you can’t measure.” All processes (including measurement processes) have inaccuracies within a measureable range. The better these process inaccuracies are understood and minimized, the more consistent the process will be. It is always best to reference each process from a single “master” standard in order to avoid additive tolerance errors. By incorporating a “glass alignment standard” within the measurement system, most of the positional errors within the measurement system can be eradicated, and the measurement system becomes the standard by which the via-punch and screen print processes are compared.
Incorporated within this glass alignment standard is a highly accurate frame containing the registration alignment pins that exactly match the registration hole-pattern of the individual tape layers. This creates the ability to measure the via-punched hole pattern created on the via-punch machine, as well as the printed circuit created on the screen printer. Both processes are periodically measured and validated to the glass alignment standard and are kept within tight specification ranges.
Most companies align the screen print process to the via-punched hole pattern or registration holes punched within the tape, which can cause additive error. The via-punch does not produce perfect via-patterns; aligning the screen printer to the via-punched pattern essentially adds the via-punch process tolerance to the screen print alignment tolerance. If this total potential error falls within acceptable limits, the process can function quite normally. Pushing the limits of the process to thinner lines and spaces or deciding to go to larger tape sizes could mean that the additive tolerance will no longer accomplish the necessary yield level because the total process variation is greater than the line or space width.
If the screen printer is aligned to the via-punch via-pattern, then the screen printer positional error is added to the via-punch positional error, thereby combining the positional errors that occur when the layers are collated. If all processes are independently aligned to the “standard,” the error range of each process falls directly onto the error range of the other process—keeping the accumulated error range no larger than the largest single process error.
The custom measurement system in Figure 3 consists of a camera and telecentric lens, an XY stage, a computer, a graphics screen, a report printer and a built-in glass alignment standard. This glass alignment standard consists of a hardened frame with registration pins to match the registration pattern. The tape layer goes on these registration pins and the system reads the via holes or fiducials printed on the layer, calculates X, Y and theta (rotational) corrections, and automatically uploads alignment offsets to the via-punch or screen printer.
Inside the frame is a highly accurate glass reticle that has a series of .004-in. diameter dots strategically and accurately placed on it using wafer process technology. The dots are used by the measurement system to self-correct its X and Y motions for orthogonal and positional errors. This keeps the measurement system calibrated to the same “standard” and makes it the best reference when it comes to measuring the via-punch and screen print process accuracies.
The measurement system has custom software, making it suitable for both validating the via-punch system and semi-automatically aligning the screen printer. The measurement system can recognize both vias of different sizes and the fiducials on the artwork. Once a printer is set up and the first layer printed, the printed tape is placed on the glass alignment standard of the system. The correct program corresponding to the specific layer being built is loaded, and the measurement system sequentially moves to and reads each fiducial and places a corresponding fiducial-shaped graphic on the graphics screen within a custom “bulls-eye” target to graphically show each fiducial’s location with respect to “true” position.
Once complete, the system calculates the “centroid,” total error, Cpk and any X, Y or theta error in the printed tape. Then the corrections to the respective process can be uploaded. This verification process should only need to be repeated periodically during long production runs to compensate for screen print image migration.
Before all of this work was completed, several steps needed to be performed when setting up the screen printer:
• Removal of the old screen frame with residual ink
• Squeegee cleanup
• Installation of the new screen frame
• Ink addition
• Screen flooding
• Loading the tape to be printed
• Setting the snap-off distance
• Making a test print
• Aligning the stage X, Y and rotation
• Additional prints/alignments
• Running the layers
After all of this work was complete, the team was able to consistently achieve a 5-10 minute setup time by eliminating some steps. In the new process, the new screen can be installed with ink already on it and the snap-off distance does not need to be set. Additional (and unexpected) benefits included:
• Layer yields increased significantly because the print accuracy was improved.
• Layer-to-layer alignment improved, which increased substrate yields.
• Screen life increased due to the elimination of catastrophic damage incidences from incorrect setup methods.
The ideal screen print machine will have the mount plate so the screen frame is located on the bottom (or epoxy) surface. This eliminates the snap-off adjustment within each setup, as long as the layer thickness remains unchanged. The screen print machine will have alignment pins on which the screen frame locates accurately. This minimizes any further alignment adjustment on manual printers, and, if the printer has auto-alignment, totally eliminates the need for further adjustment. Through careful engineering, many of the other adjustments within the screen print machine can be eliminated or at least specifically defined and locked. Figures 4 and 5 illustrate a printer transformation.
Individual process control can provide many benefits. Some of these accomplishments include:
• Tighter layer-to-layer collation alignment
• Elimination of all printer adjustments, including snap-off adjustment within a given tape thickness
• Setup times in the 5-10 minute range (vs. 30 minute minimum)
• Higher yields (this collection of projects took us from the high 70s to the mid 90s)
• Greater throughput (up to 25% increase)
• Reduced printer maintenance
In conclusion, it is possible to achieve world-class print quality and layer-to-layer alignment on HTCC or LTCC processes if individual process control is diligently pursued.
Listen to our podcast with the author!
For more information, call (480) 993-8446, email firstname.lastname@example.org or visit www.credenceeinc.com.