Improving sensor reliability
Delving into the science of failures pays dividends.
Jon Titus, Contributing Technical Editor -- Test & Measurement World, 3/1/2005
 |
| Read other articles from this issue: Table of contents, March 2005 AWARDS Test Engineer of the Year, Cover Story Test Product of the Year Test of Time OTHER FEATURES Reducing pin-count test Improving sensor reliability Bits battle noise Picture Perfect For more information on failure analysis, visit: www.tmworld.com/fa |
By applying coatings to sensors, Freescale has dramatically improved the sensors' lifetime in harsh environments. But it took lengthy failure analysis and experimentation to decide how best to apply the protective coatings and the thicknesses necessary to ensure reliable sensor operation without compromising sensor performance.
During analysis of failures, the Freescale engineering team focused on electrochemical techniques that help it determine coating permeability, corrosion rates, and adhesion. As the team made progress, it also developed some guidelines that can help engineers who face similar failure-analysis challenges.
Stop the corrosionTo protect its pressure sensors from acids, bases, or organic chemicals such as fuels, Freescale employs a conformal polymer, such as Parylene C, or a fluorocarbon or a fluorosilicon gel (Figure 1). Typically, failure mechanisms involve dissolution and swelling of the coatings caused by organic solvents and corrosion of silicon by aqueous solutions at the sensor-coating interface.
 |
| Figure 1. Applying (left) liquid silicone gel or (right) vapor deposited polymer film (Parylene C) on a sensor protects it from the effects of organic solvents or aqueous solutions. (Polymer film not to scale.) |
 |
| Figure 2. A typical organic-solvent test tank allows easy access to individual sensors. The tank purges open areas with nitrogen to reduce fire hazards. Courtesy of Freescale Semiconductor . |
Initial tests applied solvents such as gasoline, transmission fluid, and lubricating oil to unpowered devices. And the engineering team quickly realized the need to bias (power) the sensors during testing to duplicate realistic operating conditions.
 |
| Figure 3. A cut-away view of an organic-solvent test tank shows how the system exposes sensors to vapors and liquids. Pumps circulate liquids to keep immiscible materials suspended. |
To test sensors in aqueous solutions, the engineering team first tried a tank similar to the one used to test multiple sensors. Unfortunately, the electrochemical reactions in one sensor affected the electrical signals produced by other sensors in the shared conductive solution. The need to isolate sensors led the team's technician to build individual "tanks" made of off-the-shelf CPVC pipe and fittings (Figure 4).
 |
| Figure 4. An array of individual test cells subjects sensors to aqueous solutions. Courtesy of Freescale Semiconductor. |
During testing, the engineers monitor sensor outputs until a device fails, as indicated electrically by an open wire bond or a degraded device. At that point, the engineers can remove the failed sensor from its test environment and apply standard failure-analysis techniques to determine why and how it failed. (Figure 5 shows a new and a corroded sensor.)
Analysis results brought the team closer to understanding the sensors' failure mechanisms, but not close enough. They had to wait for over 100 days for a device to fail before they could examine it. During a test run, the sensors' electrical outputs didn't provide much useful information.
|
|
| Figure 5. A comparison of (top) a new gel-filled sensor and (bottom) a similar sensor exposed to transmission fluid shows gel erosion and corrosion. Courtesy of Freescale Semiconductor. |
To delve deeper into the failure processes, the team researched polymer-metal adhesion and corrosion. "When we understood that diffusion of a solution through a polymer caused the corrosion, we knew we needed to learn more about diffusion and corrosion processes," explained Dave Monk, the sensor development engineering manager in Freescale's Sensor Products Division. "After a while, we decided to hire an electrochemist for our team." The new team member applied electrochemical-analysis techniques such as linear-sweep voltammetry (polarization) and chronoamperometry to sensor testing.
Electrochem 101Electrochemical measurement techniques rely on apparatus that forms an electrochemical cell. The cell includes an ammeter, a voltmeter, a DC power supply, three electrodes, and a test solution (see a schematic diagram of the cell in Figure 6 ). By varying either the current or the voltage in the cell and then measuring the other quantity, electrochemists can deduce much about the chemical behavior and the composition of a solution or chemical activity at a surface. The setup typically includes a platinum auxiliary or counter electrode, a working electrode, and a reference electrode. A standard silver/silver chloride electrode can act as the reference. The Freescale team relies on commercial instruments from EG&G Ametek/Princeton Applied Research.
|
Figure 6. Electrochemical tests on a MEMS sensor use an arrangement of three electrodes. Tests change either the current or the voltage and monitor the other quantity. Courtesy of Freescale Semiconductor. |
Polarography, polarization, or linear-sweep voltammetry techniques all slowly increase the potential across the cell and monitor current changes. In polarography, current "plateaus" indicate the presence of a specific type of oxidation or reduction reaction, such as:
Zn2+ + 2e —> Zn
Polarography experiments use a single voltage sweep over a long period, while linear-sweep techniques may make many sweeps through positive and negative voltages using a triangle voltage wave to "probe" a test solution. Peaks in the sweep scan indicate chemical activity such as oxygen reduction or hydrogen generation.
 |
| Figure 7. Results from a linear-sweep voltammetry experiment show the current measured for a sensor coated with a 0.68 µm film of Parylene C. The two peaks represent chemical reactions on the sensor surface. Thicker films suppress these peaks. Courtesy of Freescale Semiconductor. |
The Freescale linear-sweep tests operated at 20 mV/s, and current flow had to overcome the resistance of the sensor coating. Thus, current flow provides a good indication of coating integrity. The Freescale team performed scans from 0 to 2 V and from 0 to –2 V with respect to the potential of the silver/silver chloride reference electrode (you can see a graph of the results in Figure 7 ).
In a chronoamperometry experiment, apparatus applies a voltage step to a solution for a few milliseconds and then monitors the change in current over time. Experiments can use one or two voltage steps. This technique provides information about diffusion of chemicals into sensor coatings and about the rate of a chemical reaction, such as corrosion, that takes place on a sensor's surface.
During a test, these electrochemical techniques let the team determine how much water and oxygen have permeated the protective coatings. The tests also furnished information about the adhesion of the coatings to the sensor's surface.
Chemical tests run faster"Electrochemical tests saved us a lot of time," explained Freescale's Monk. By running a polarization experiment, for example, the team obtained results about nine times faster than if it had to wait for the completion of exposure tests alone. "We tested experimental parts using electrochemical techniques and we then used the same types of parts in exposure tests. It took over 125 days [3000 hr] for the exposure-test parts to fail, but in only 2 weeks [330+ hr], the electrochemical experiment detected the start of corrosion." The shorter test periods helped shorten product-development cycles.
A catastrophic corrosion failure would still take the same time—about 125 days—to appear, but the electrochemical tests let the engineers detect the start of the corrosion or delamination that would eventually lead to a complete failure. In addition, electrochemical techniques let the team monitor corrosion processes from start to finish, something they couldn't do with exposure tests alone.
"We now understand the failure mechanisms in qualitative terms," said Monk. He explained that an aqueous solution will diffuse through the polymer coatings down to the metal-polymer surface, "but we don't know yet which process starts first, delamination of the polymer from the metal, which leads to corrosion, or corrosion that leads to delamination."
As more corrosion occurs, it produces chemicals that cause more delamination, which propagates the process. More delamination leaves more room for liquid reactants on the sensor's surface, which accelerates corrosion. No matter how the process starts, by carefully applying electrochemical techniques and through statistical analysis of the data, the engineers can predict the life of a specific type of coating on a sensor under a variety of operating conditions.
Results of the electrochemical tests also let Freescale's engineers model coatings for a variety of temperature, voltage, thickness, and chemical conditions, and they use the results to produce sensor coatings to meet customers' needs. "We didn't create one low-cost media-compatible pressure sensor that meets every requirement," noted Monk. "We offer a low-pressure sensor for appliance manufacturers that works with soapy water, but if a manufacturer wants to measure tire pressure, they need a different part."
Guidelines for other projectsIn addition to learning how to apply a new failure-analysis technique, the team also developed guidelines that other engineers and analysts can apply in similar situations.
- Investigate the sciences that apply. "We investigated things that didn't seem obvious at first," said Monk. "Over time, the automotive and paint industries have improved the adhesion of paint to metal because you don't see as much corrosion on cars as you did 20 years ago." Learning more about the science of polymer adhesions helped the team determine how to apply thin coatings so MEMS sensors will survive in harsh environments.
- Explore the literature in new areas. Often, engineers narrow their focus and stay confined within an area, say circuit design. "We had to think differently about reliability problems and educate ourselves," explained Monk. "We read about what smart people in the paint and polymer industries had done until we had a thorough exposure to those fields."
- Work with experts. Consultants and academic researchers in other fields can help break barriers between disciplines. An electrochemist on the Freescale team provided insight into the use of standard techniques most chemists know about. Many engineers, however, might have had no exposure to electrochemistry or to electrochemical instrumentation.
- When people from other fields gather, innovation happens quickly. As Monk explained, "You'll develop new approaches to problems when you bridge the 'space' between professions such as electrical engineering and electrochemistry. Adding an electrochemist to our engineering group led us to different failure-analysis techniques that quickly homed in on problems."
- Apply design-of-experiment (DOE) techniques. Although increasing the temperature during testing accelerated corrosion somewhat, the engineers found temperature wasn't the most important accelerant of sensor failure. "We applied DOE techniques and found that bias voltage accelerated corrosion more than temperature," said Monk. "We could apply a higher voltage and 'tune' an experiment. From an electrochemist's perspective, that makes a lot of sense, but it might not be obvious to traditional engineers."
The use of DOE techniques also led the team to examine concentration as another key acceleration factor. If you expose a sensor to a nitric acid solution with a pH of 5 to 6, the sensor will chemically react slower than it would if exposed to nitric acid with a pH of 0.5.
The Freescale engineering team has tested sensors for many years under a variety of harsh conditions found in vehicles and appliances. Inevitably, though, a customer will arrive with a plan for a product that exposes sensors to other challenging environments. So, the engineers expect to continue testing a variety of sensor coatings for some time to come.
