| by Ron Abramshe, Ph.D.
Any material made by humans is destined to have some flaw. Although we strive for perfection, because that’s what our customers expect, errors can still occur. Whether they’re made in engineering, medicine, or day-to-day business transactions, mistakes are something that many companies don’t like to acknowledge, much less correct. Ironically, by not openly admitting mistakes and correcting them, organizations lose a valuable opportunity to strengthen the relationship with their customers.
This article tells the story of how a little plastic particle could have become a huge problem for the supplier of micronized diamond powders and its customer, a well-respected Japanese manufacturer. Rather than shuffle blame or deny the problem, the supplier admitted to several key mistakes and, by working with its customer, sought effective solutions to correct the problem.
Warren/Amplex Superabrasives, a manufacturing unit of Saint-Gobain Ceramics & Plastics Inc., worked hard at securing the supply of Resin Bond, a species of diamond, to Disco Corp., a respected slicing-blade manufacturer in Japan that serves the Japanese electronics industry.
A slicing blade is an ultra-thin metal blank—from 0.004-in. to 0.008-in. thick—upon which a micron-sized diamond is deposited in an electrolytic bath. When it comes to the slicing blade’s thickness, stability, and particle size, stringent terms and measurements are demanded of the applied micron diamond.
These blades are used in industrial applications such as row-slicing of alumina titanium carbide (AlTiC) read/write heads, silicon wafers for memory chips, and alpha silicon carbide for LED applications. Among the stringent blade requirements is particle-size distribution (PSD). These are Gaussian distributions of a number of discrete particles that are graded using a special technique.
Controlling the PSD is critical to the success of increasing yield after a cut is made. If the distribution is too fine, productivity decreases and becomes apparent in low wheel-life and burning, which can result in subsurface damage to the part.
If there are large particles in the distribution, then chipping occurs during the cut. If the chip is too large — i.e., greater than 10.0 µm — the piece will be useless because it’s impossible to lap out such a large chip during later stages of processing. (“Lapping” is a method of reducing the number of peaks produced after a coarse particle has gouged its way across a surface.) Both conditions lead to yield loss and low productivity, which in turn create bottlenecks downstream in the manufacturing process.
Warren/Amplex (where I work), an industry leader and established micronizer, worked hard at perfecting the PSD with Disco and had established a good supplier relationship with that company during the past few years.
During the course of this relationship, Disco had made sporadic complaints about particle-size analyzers and correlation coefficients. The company was using a different type of particle analyzer to measure Warren/Amplex’s products and, as new people were assigned to this product line, the same questions arose as to the validation of our correlation studies. Countless hours of statistical discussions in Japan were spent proving that the micron size that we supplied was stable and — more importantly—repeatable.
The problem gained momentum when Disco, routinely testing the product, found a 30-µm diamond particle. We were dumbfounded. With all the quality control work and procedures implemented in our micronizing process, we were confident that an oversize particle couldn’t slip through.
Trying to avoid “who, me?” and related self-denial paradigms, we asked Disco for tangible evidence to support the existence of the whopper diamond particle. We knew that Disco was as capable of detecting particles as we were —and that we were as serious and dedicated to product quality as our customer.
Meanwhile, we took an introspective look at our own practices to check for any oversights. We hadn’t experienced a complaint like this in several years, and never one of this magnitude. We formed a team of experienced members from quality control, manufacturing and management, and embarked on a journey that, when all was said and done, surprised all of us.
We started at packaging and reviewed our manufacturing procedures to ensure that we were keeping within the process steps. We checked off the main steps as follows:
• Packaging hood wipe down before and after packaging
• Antistatic spray used before and after packaging
• Packaging hood HEPA filter maintenance done in accordance with the schedule
• Packaging containers stored in a sealed bin
• Packaging fine-to-coarse product
• Morning and afternoon general area cleanup
• No cardboard or other material, including lead pencils, in the area
• No spills close to the time of the complaint
• Ventilation filters in accordance with type and replacement schedule
• No disgruntled or new employees at the time of the complaint
Having satisfied ourselves that our packing department wasn’t the cause, we moved backwards to our micronizing department, headed by a seasoned micronizer, Charles “Chuck” Washburn, who had more than 18 years of experience in the industry. Chuck was well aware of the complaint and had completed a review of his department. The ultimate perfectionist who knows his business inside out and expects his team to aspire to the same professionalism and product quality, Chuck was dispirited to find his department under scrutiny. Nevertheless, no area was excused from our internal auditing.
This wasn’t a “go through the motions” audit. We visually inspected each of the area’s manufacturing steps—a real-time, hands-on review. Here’s a summary of the audit checks that we performed (except a few that are proprietary to the Warren/Amplex process) as we hunted for the root cause of the contamination:
• Product separation upon receipt of feedstock
• Full area wet-down prior to feedstock loading
• Ventilation and filtration completed
• Deionizer water cartridges in accordance with scheduled changes
• Water-system filter cartridges changed in accordance with the schedule
• Lids placed firmly on all containers
• Area wipe down before and after feedstock loading
• Water-system pressure and temperature controls functional, no abnormal conditions
• No leaks or abnormal conditions reported
• No unusual employee absences or problems around the time of this complaint
While we were still proceeding through the audit, the Disco sample arrived, along with some microscopy photographs and Raman infrared scans, the latter being a method of determining crystal structures by vibrating their crystal lattices with infrared light and reading the resonances.
The data indicated that a particle about 30 µm in diameter was evident. The Raman scan did show a strong diamond peak. The team was baffled. How did this happen, with all our quality assurance systems, procedures, and checkpoints? How could a 30-µm particle get by us?
Meanwhile, Disco was pressuring us for an explanation and replacement material. Now we were being asked to explain this quality incursion while simultaneously researching and implementing immediate preventive actions.
The replacement material was inspected using scanning electron microscope (SEM) analysis, which is an electron-beam-based microscope used to examine, in a three-dimensional screen image, the surface structure of prepared specimens. Literally hundreds of SEM scans were done to ensure that no oversize particles were found. After we were satisfied that the replacement material was fine, we sent it to Disco.
We then prepped the returned sample for our own analysis. As we were completing our analysis of it, our SEM technician, Nick Tumavitch, Ph.D., noted that under high magnification, the particle in question looked more like an agglomerate or foreign particle. Chuck was beside himself because he performs a conductivity wash on all Disco-ordered diamonds. “Impossible,” he stated, despite the evidence under the microscope.
Nick looked at the particle again and decided to perform an energy dispersive X-ray (EDX) analysis, which bombards a sample’s surface with X-rays to read the signature. The EDX provided us with a qualitative element analysis.
After some adjustment for bandwidth, Nick determined that the particle was mostly plastic. There was a strong carbon peak but also high peaks from nitrogen, chlorine, and oxygen, which aren’t indicative of diamond.
At this point we believed we were on to something —perhaps a round plastic bead had somehow become imbedded in the diamond. But where were these beads coming from?
Chuck announced that he was going to “sieve some water,” at which point the team unanimously decided that he’d finally gone around the bend. What Chuck really did was take samples of deionized water before and after the filtration process and pass those through a 1.0-µm electroformed sieve. What he screened out were numerous plastic deionizer resin beads during the prefilter and perhaps a dozen after filtration.
As evidenced by both light microscopy and SEM, the filtering system we use employs a number of carbon prefilters, and a succession of filters up to 1.0 µm. Our maintenance person took the system apart piece by piece, carefully analyzing each filter. He found that after several years of filter replacement, the plastic housings that seal the gasket of each filter had worn enough to permit 10-µm or 20-µm particles to eventually work their way through the gasket.
During a follow-up team meeting, we asked ourselves how the beads got into the system, and, more important, how did they get by the resin tank guard? Knowing that resin beads decrease in size after a certain amount of regeneration, we decided to call the filtration company and launch an in-depth inquiry. The filter company assured us that a bead of that size wouldn’t be permitted to be regenerated.
Still no answers, but the evidence was before us. We decided to ask the technician from the water filter company how he replaces the deionizer cartridges. Accordingly, at the next scheduled replacement, Chuck asked the technician questions from a prepared list. One key question was, “Did you ever see anything out of the ordinary during the past year?”
The technician recalled that, about a year before, one of the retainers holding the resin-retaining grid was loose. When questioned further, the technician said it was possible that some of the resin beads could have escaped at that time.
We had our answer. A tiny fraction of these beads had escaped — passed through our filters because of a worn housing — and found their way into our main deionization tank. Randomly, these beads would get into our grading system. Unfortunately, they also found their way into our products.
We now had to find a solution and corrective action to satisfy our Japanese customer as well as other customers who could experience the same mishap. We considered sieving all the water through a 1.0-µm sieve (as Chuck had done to find the problem), or emptying the 1,500-gallon central storage tank, scrubbing it, then refilling it and checking the water with a 1.0-µm sieve.
Both of these solutions were cumbersome and time-consuming. The team came up with a different idea: Why not dia-filtrate the water before each grading station, like we did in our submicron grading? We agreed that this would solve the problem because these filters wouldn’t pass anything, including beads, over 500 molecular weight.
We purchased the required number of filters and presented this solution to Disco as a preventive action. They weren’t completely convinced and countered by requiring us to take samples and sieve-check the water before we used it for their products. We agreed to most of their requests, and the corrective action has proved successful.
The lessons to this story can be summarized as follows:
• The customer is always right.
• Don’t second-guess yourself.
• Look for facts, and question everything when you’re not sure.
• Trust your co-workers.
• Be honest with yourself and your customers.
• Communicate, communicate, communicate.
• Remember that nothing touched by human hands will be perfect.
It’s never easy to admit a mistake, especially when you believe that you have everything engineered perfectly. Senior management must create a corrective action system that permits the free flow of ideas and risk taking. The actual structure isn’t as important as how efficiently it functions.
Senior, as well as mid- to lower-level managers, must all be committed to the same good manufacturing practices that define the company, as well as to the company’s ideals and the products it produces.
Here are some suggestions to keep in mind to establish and maintain principles behind good manufacturing practices (GMP):
• Talk with your customers frequently; ask them to comment on your products.
• Ask your customers what they want in terms of features and benefits.
• Pay close attention to customer complaints. Measure them, correct them, and ensure that closure is made to the customer’s satisfaction. The ISO 9001 registration process is an excellent way to establish a reliable quality management system and GMP adjunct.
• Train and reward employees who are outstanding customer advocates.
Ron Abramshe, Ph.D., is the product manager and technical sales manager for Warren/Amplex Superabrasives. He earned his doctorate in engineering management from Kennedy-Western University, a master’s degree in engineering from Polytechnic University of New York, and a bachelor’s degree in industrial engineering from the University of Dayton.
|
