Seeing the Interface: How Realistic is the Picture?
Nancy McGuire, Contributing Editor | TLT Feature Article April 2015
Close-up and real-time tribological testing help to produce relevant results.
www.canstockphoto.com
KEY CONCEPTS
•
Tribological testing is a tool kit, not a magic number. Relevance of the results depends on the questions asked and the methods used to answer them.
•
Wear behavior involves many interacting parameters and changing one parameter affects all the others.
•
Models based on very small effects during the early stages of wear can be preferable to those based on extreme temperatures and accelerated wear.
ONE APPROACH TO TRIBOLOGICAL TESTING GOES LIKE THIS: Take a sample, wear the heck out of it and see where it fails. Attach a number to this and compare that number to the ones from previous tests or the spec sheet for your machine part or lubricant.
This approach has the advantage of being relatively quick and inexpensive. The numbers let you screen materials for suitability, track performance over time or quantify the effects of changing one parameter in your system. What do they tell you about what’s going on out in the plant or on the road?
“Coefficient of friction—what is that, really?” asks Markus Grebe, lab manager and head of industrial research at the Tribology Competence Center, Mannheim University of Applied Sciences (the only German university to offer an undergraduate course in tribology). “The customer often wants one number they can use for everything.”
STLE-member Steven Shaffer of Bruker Nano Surfaces concurs: “Tribology is heavily empirical.” Lubricant designers often go with the tests they have always used, but how does a four-ball test correspond to a real-world system? Many of the standard lubricant tests and test geometries have a long history, but clients are less clear about how to use them to create realistic service conditions and produce relevant results. Sometimes even the act of running a test changes a contact surface in a way that influences the test result, Shaffer says.
True realism would require taking a fleet of trucks, for example, and attaching microscopes, video cameras, temperature sensors and chemical analyzers at every point where friction and wear occur. Each truck would be identical, except for one parameter, say a piston ring alloy or lubricant additive, and each would haul identical loads over the same stretches of road over the course of its lifetime, from factory to junkyard. The results would be extremely accurate but prohibitively expensive and not at all timely. And they might not tell you anything useful about different trucks hauling different loads on different roads in different weather conditions.
The real challenge is to find tribological test methods that keep the test time and expense within reason, provide useful answers to the questions being asked and provide some way to compare results with other similar studies. Even better would be a way to watch the wear process as it happens and see how the wear evolves under the most realistic conditions possible.
THE ANSWERS DEPEND ON THE QUESTIONS
Grebe describes his laboratory’s approach to selecting test methods and conditions: “What is the aim of your testing? To create and compare values for similar components and lubricants under defined conditions? To simulate a real system under real-world operating conditions? You need to know the limits of what your test shows,” he explains. “What questions are you asking? Where is the answer applicable?” Model tests are best for the early stages, and systems tests give you a better real-world picture, he says. (
See table titled Rating Your Test Method.)
RATING YOUR TEST METHOD
Table courtesy of Markus Grebe, Mannheim University of Applied Sciences.
“You have to combine various test methods to get a complete picture,” he adds. He and his colleagues are starting a new working group this year with German industry representatives to come up with a set of practical test procedures that answer the various types of questions.
Grebe describes two testing approaches. The first approach creates and compares numerical values and is most useful for quality control of materials and lubricants and screening them for suitability in practical applications. This approach uses common standardized tests to observe how changing one parameter affects a tribological value—varying additive concentration and observing the effects on extreme pressure behavior, for example.
Numerical results are useful for creating spec sheets for products that will be used in a variety of applications. This type of testing reveals if you are using the right base oil and additives, how an additive works at various concentrations or if your production process is working correctly.
Tests run according to standard protocols (ASTM or ISO, for example) facilitate meaningful comparisons of results from different laboratories, and specific parameters can be tracked over time. However, notes STLE-member Ali Erdemir, Argonne Distinguished Fellow and senior scientist for Argonne National Laboratory, many companies design their own test systems to provide information that is specific to their own products. Thus, parts manufacturers, lubricant manufacturers and tribometry instrument manufacturers might all be looking at the same system from different angles. Combining several types of probes may lead to a more informed judgment on the usefulness of a new product or technology, he says.
A second approach involves simulating a real tribological system. This is a more complex approach and involves several interacting factors. Simulations are useful for determining wear-related influences on a machine’s overall function and operational capabilities, diagnosing operating conditions or testing original parts to create information about a particular tribosystem.
Although this type of testing gives you a more realistic look at entire systems, the results are representative of specific systems under specific conditions. Extrapolations can be made to similar systems, but it’s dangerous to forecast behaviors in a completely new system, says Grebe.
You can use your past experience on similar situations as long as you are open to differences that you weren’t aware of previously, Grebe adds. Manufacturing processes matter. Things that didn’t matter before (additives, manufacturing methods, packaging materials) can become important in a new situation. Grebe says that customers often claim they haven’t changed anything, but when they look more closely, they discover changes that they didn’t attach any significance to or anticipate the resulting differences in performance. “But that’s what makes the research interesting,” he adds.
Combining the one-parameter test approach with the systems simulation approach is useful for optimizing components and tribological systems to achieve a predetermined goal. Such goals include setting manufacturer’s guarantees for service lifetimes and creating data for maintenance scheduling.
SYSTEMS APPROACH
Grebe’s laboratory takes a systems approach, selecting test methods and using combinations of tests depending on what they hope to find out as a result, he explains. About two-thirds of the research at his laboratory involves automotive parts. “The Tribology Handbook is our bible,” he says (
1).
Before this book came out, he continues, everybody developed their own tests and instruments, but they didn’t integrate the information. “They didn’t publish the test parameters or the characteristics of the instrument. It makes a difference whether your instrument has polymer parts or metal because the heat transfer characteristics, for example, are different.”
This approach also emphasizes running tests for a sufficient amount of time to produce realistic results. “Often tests are much too short to be realistic,” Grebe says. He notes that most of the older model testing methods use extreme wear conditions, including high loads and accelerated wear that don’t adequately simulate real operating conditions.
The results of these tests are often more representative of the running-in period—a period of accelerated wear at the beginning of the wear process where rough spots produced during manufacturing are worn away. The running-in period is typically followed by a long period of steady-state wear, more typical of normal operations, followed by an accelerated wear period as the part starts to fail. Extrapolating results from the running-in period can lead to service-lifetime predictions that are much shorter than typical under moderate loading parameters, start-stop cycles and changes in loads and speeds.
Grebe’s group is most interested in the periods of accelerated wear, but they don’t assume that these periods are representative of what happens most of the time. “Most systems don’t operate at peak performance all the time,” he says. Running your tests “flat out” risks premature failure caused by putting too much energy into the system, in the form of heat or electrical charge transfer. Testing real parts or specimens cut from real parts lets you observe the effects of specific manufacturing processes, realistic surface topographies and real-world variations in quality. Lab specimens are good for observing systematic changes in one or a few parameters, but they tend to be more homogeneous and have smoother surfaces than real-part specimens.
Daimler AG, the German automotive company, is using a novel real-parts method, partly developed at Mannheim University of Applied Sciences, to test motor oil additives and develop lubricant specifications for a new silicon–aluminum cylinder liner material. Daimler used a high-frequency oscillation sliding test rig (SRV) and 10 x 15-mm liner samples to determine that some of the oils drew silicon particles out of the liners and others did not. (
See Figure 1.)
Figure 1. Piston-ring/liner test results showing the effects of various oil additives on liner friction and wear. (Chart courtesy of Markus Grebe.)
Tribology testing by varying the oils and using liner samples was a new approach for Daimler, and it was especially cost-effective. Tests that used liner samples ran 250 € each, whereas a complete combustion engine test would run 250,000 € and destroy the motor, Grebe says. “This test was very effective in showing how one important thing is relevant to the complete system.”
A CLOSE-UP LOOK
Laboratory studies give another type of insight into friction and wear processes. By closely observing processes under controlled conditions, researchers can gain understanding of the fundamental processes at work and use this knowledge to make sense of observations in the field. Technological developments have made it easier to place instruments close to the site of friction and wear and record ongoing processes—movies instead of snapshots.
One way of determining what’s going on, especially in the very early stages of wear, is to home in on the events happening on the surface in a very local sense. “What is the actual temperature at the contact interface?” Shaffer asks. “High temperatures, even those that are extremely local and short-lived, can produce chemical changes in a lubricant. Elevated temperatures also can affect microstructures below the base material’s surface and in the friction tribo-layer.” The lubrication regime influences the frictional forces imposed at the surface. These forces, in turn, influence the formation of microstructures in the surface and sub-surface layers, he explains.
Boundary friction has the greatest effect on wear lifetime, Shaffer says. Temperature can change the nature of boundary film constituents in a lubricant additive package. These constituents affect the evolution of subsurface microstructures, which influences the resulting wear rate. Measuring temperatures at the contact interface requires getting as close to the interface as possible without affecting the contact itself, which can be especially challenging.
New additives and other lubricant constituents change the nature of chemical and microstructural interactions at the boundary, in the lubricant and in the surface itself. Evolving government regulations and changes in pricing and supply can create a need for effective substitutes when the old standbys become illegal or scarce. A thorough understanding of the mechanisms of each constituents’ effectiveness and how these surface interactions occur is paramount in identifying suitable substitutes, Shaffer says.
The accumulation and transfer of electrical charge associated with solid surfaces in contact with each other has not been studied widely. Erdemir and his group recently published a study that measured tribocurrents and macroscopic friction forces for a metal ball in contact with a rotating polytetrafluoroethylene (PTFE) disc. They observed significant differences in electrical behavior, depending on the atmosphere surrounding the test setup (
2). (
See sidebar titled The Body Electric.)
THE BODY ELECTRIC
Surface melting, plasticization, strain and abrasion are commonly understood as direct effects of friction. However, electrical charge transfer is also present and can have significant effects on macroscopic friction behavior. A recent study by Thiago Burgo and Ali Erdemir at Argonne National Laboratory demonstrated charge transfer effects between a metal ball and a rotating polytetrafluoroethylene (PTFE) disc. (
See Figure 2.)
The authors cite a strong correlation between friction and triboelectrification, noting that “macroscopic friction originates from and scales to the intrinsic electronic interactions and modifications of the electronic properties of surfaces (e.g., anodic oxidation), stimulating changes in the macroscopic friction force.” Tribocharges can exceed all other factors for mechanical energy dissipation, they note.
They demonstrated this by running their ball-on-disc tests under high vacuum, nitrogen, hydrogen and open-air atmospheres. Tribocurrent amplitude was greatest in the vacuum and significant in the relatively inert nitrogen atmosphere. The tribocurrent was greatly reduced under the reactive hydrogen atmosphere and almost completely absent under air. Under the more reactive atmospheres, mechanochemical reactions, triboluminescence, corona charging reactions, phonons and heating propagation consumed the triboplasma generated by mechanical stress at the interface. In contrast, less-reactive atmospheres caused a large part of the high-energy species formed by mechanical stress to convert into electrical current.
The authors observed macroscopic flakes of PTFE strongly adhering to the tail surface of the metal ball, indicating that the ball had picked up wear particles after it had slipped on the surface. They proposed a mechanism by which the PTFE picks up negative charges as a result of mechanochemical reactions during contact electrification, and then retransfers these negative charges to the metal ball by material transfer (polymer surface flaking and adhesion to the ball). The authors note that friction force and tribocharges have a common macroscopic origin: They vary concurrently and with similar rates depending on the test conditions (
2).
Figure 2. Schematic of a metal ball and a rotating polytetrafluoroethylene (PTFE) disc. (Illustration courtesy of Thiago L. Burgo and Ali Erdemir and Argonne National Laboratory.)
Close-up study shows wear in its early stages: the atomic dislocations that grow into cracks or the particle releases that eventually cause a part to fail. Microscopic images can show whether a particle is a large flake, indicative of a serious wear problem or an agglomeration of small particles released during normal operations. Electron microscopy and atomic force microscopy reveal the nanoscale topography in a worn region, giving insight into the causes and effects of the wear.
These observations are by definition very localized, observing phenomena over areas that can be less than a millimeter across. Some techniques, including ion bombardment methods, require stopping the process to take samples or destroying machine parts by cutting samples from them. Some analysis techniques are nondestructive and can be conducted online in the field. Others require vacuum chambers and other special setups and can be used only on model systems in the lab. Some real-system analyses require special handling and may cause changes in the material properties of the parts being studied.
Are nanometer-sized changes important in everyday operations? “Yes!” says Grebe. “Piston rings wear in nanometers per hour, and it all adds up.” Higher resolution lets you interpret results earlier in the wear process, he explains. Measuring small effects lets you use more realistic conditions—you don’t need to place a sample under artificially high temperature or accelerated-wear conditions to create an effect that is large enough to observe.
Examining small changes also sheds light on why something
did work: a chemical structure in a thin film, for example. That way you can keep the best features and expand on them rather than relying on trial and error. Screening for beneficial features also helps in designing and planning subsequent tests, Grebe adds.
REAL-TIME TECHNIQUES
Some commercial systems already include real-time wear monitoring and analysis capabilities. They incorporate temperature sensors, moisture sensors and particle counters that monitor lubricant streams continuously and send alerts to a crew member’s cell phone when conditions are out of spec (
3). These sensors provide information on the degree of wear over time, provide an early warning when a part is about to fail and narrow down the possible locations where excessive wear is occurring.
Other types of online testing require setting up a model system in a laboratory in order to provide access to specialized instruments and controlled environments. Embedded thermistors or thermocouples can record time-averaged temperatures for specific regions of a mechanical system. Temperature-sensitive phosphorescent coatings and phosphor-containing optical fibers track rapid, localized temperature fluctuations by monitoring light decay as a known function of temperature. Lubricant film thicknesses can be measured in real time by monitoring the capacitance, inductance or resistance of the lubricant film. These techniques require sensors and data transmission devices at the main friction points, and the components may need to be adapted to keep the areas of interest from moving out of the viewing range of the sensors (
4).
Chemical probes, including infrared and Raman spectroscopy and x-ray photoelectron spectroscopy (XPS), can monitor real-time chemical changes in lubricants and mechanical part surfaces. X-ray reflectivity and fluorescence studies monitor nanoscale structural changes as they evolve.
Some techniques that are typically used in post-mortem studies can be adapted for real-time observations. Mass spectrometry has been used to study volatile wear products as they are released. This requires installing a tribometric setup in a vacuum chamber, with an outlet port leading to the mass spectrometer (
5).
Tracer particles can provide details about wear rates and localize the sources of wear particles. Implanting radioactive particles in metal or polymer machine parts, or irradiating steel parts, is a well-known materials analysis technique that Grebe’s lab is adapting for use in tribological studies. (
See sidebar titled Tagging the Wear Source.)
TAGGING THE WEAR SOURCE
Radioactive tracers are useful for measuring wear as it occurs in a real system under normal operating conditions. Sensors placed at various locations downstream of friction points can detect wear particles released by machine parts during operations. Tagging specific parts by implanting radioactive species is one means of establishing a specific source of wear particles in the lubricant stream.
Steel parts like piston rings and cylinder walls can be activated by placing them in a cyclotron. Some of the iron atoms on the steel’s surface pick up extra neutrons in the cyclotron and become cobalt-57. As the cobalt decays back to iron, it releases neutrons that can be detected and counted.
Grebe notes that cyclotron exposure times have to be limited, because producing too much cobalt alters the mechanical properties of the surface. Polymers and nonferrous metals can be ion-implanted with radioactive species, as long as the amounts are kept low enough to not affect the material’s strength, elasticity or other properties significantly.
Radiation detectors measure the amount of radioactive wear debris that accumulates in the lubricant stream during operations, providing a continuous real-time readout of the wear process. The tracer particles indicate the source of the wear, pointing to areas that bear closer inspection afterward.
Because radioactive materials are involved, special precautions are needed for handling and disposing of the irradiated parts, lubricant and oil filters. Other components that come into contact with the irradiated parts or the lubricant, as well as any gaseous emissions, must be monitored for radiation levels.
These methods provide good detail on what is happening at a contact surface and perhaps a few atomic layers below. The difficulty lies in finding out what goes on in buried or confined interfaces, Erdemir says. Even surface observations can be difficult because of the small size of the contact area. Actual contact may occur over areas ranging from a few nanometers to 100 microns or so. The scale of the measurement is also a factor, he continues. Even polished, optically smooth surfaces are rough at the nanoscale. High spots, where wear first occurs, can represent less than 0.001% of the total contact area.
The speed of events during the wear process also can complicate observations, Erdemir says. Complex interfaces have many interacting factors, all acting at once: chemical, thermal, mechanical and electrical. “The best system would be one that lets you collect all the information at once.” Does such a system exist? “I’m not aware of one,” he says.
IN-SILICO TRIBOLOGY
Computer modeling and simulation of tribological processes are in the very early stages of development. The complexity and sheer number of possible conditions involved require a lot of computing power. Fortunately, high-performance computing has evolved to the stage where it can begin to tackle this type of problem. “We can now simulate systems containing hundreds of millions of atoms,” says Erdemir. (
See Figure 3.)
Figure 3. Computer simulation of wear particles adhering to an atomic force microscope probe. (Illustration courtesy of Sandia National Laboratories.)
Coupling models with experimental data allows you to observe things that happen too fast or that are too entangled with other phenomena to observe otherwise. As modeling capabilities develop, they will be useful for designing new materials and they may help to predict wear mechanisms and effects in new systems.
Ideally, modeling could help you design parts that last for the lifetime of the system, Erdemir says. This would be a significant contribution to developing sustainable modes of transportation. “I hope to see this in my lifetime,” he adds.
Computer modeling for tribological studies has yet to catch on at Mannheim University. Tribological testing is “more complex than weather forecasting,” says Grebe. It’s very difficult to model because it involves so many parameters, he says. Real tribology requires a lot of contact points: the base body, counter body, surrounding medium and intermediates. Friction and wear processes transfer energy from one component to the surrounding components, causing changes in their elemental composition and structural features, which in turn affect the parts in contact with them and so forth. “If you set up a simple standardized system, you can calculate that, but it’s only just that one system. You have expended a lot of effort on just that one part,” Grebe says.
Often chemists don’t know exactly how certain additives work and this makes constructing realistic computer models impractical, Grebe notes. The chemistry of lubricants is very complex. Computer models can be used for the hydrodynamic situation, he says, referring to operating conditions where the mechanical parts are perfectly separated by a lubricant film. His group is more interested in areas of mixed friction and boundary friction, typical of what happens when a motor stops and starts.
FROM REACTIVE TO PROACTIVE
Real-time and close-up testing have joined standard tribometric techniques in the tribologist’s tool kit. Erdemir predicts a further evolution to online tribo-sensors that interact with mechanical systems during operations to adapt to changing conditions and prevent catastrophic failure. “You could design sensors coupled with computers that could tune your [car] engine to adapt to driving conditions,” he says. Cars already on the market have engines that shut themselves off when the driver is at a long stoplight and turn on again when the driver presses the accelerator pedal. “The engine companies are getting very smart,” he says.
REFERENCES
1.
Czichos, H. and Habig, K.H. (2010),
Tribologie-Handbuch: Tribometrie, Tribomaterialien, Tribotechnik, Third Edition, Springer Vieweg.
2.
Burgo, T.L. and Erdemir, A. (2014), “Bipolar tribocharging signal during friction force fluctuations at metal–insulator interfaces,”
Angewandte Chemie International Edition,
53, pp. 12101-12105. DOI: 10.1002/anie.201406541.
3.
McGuire, N. (2014), “Oil analysis: Keeping military systems running smoothly,” TLT,
70 (7), pp. 34-43.
4.
Sherrington, I. (2010), “Chapter 11: Measurement techniques for piston ring tribology,”
Tribology and Dynamics of Engine and Powertrain, Woodhead Publishing, Ltd., pp. 387-425.
5.
Stachowiak, G. and Batchelor, A.W. (2005),
Engineering Tribology, Third Edition, Butterworth-Heinemann, pp. 659-660.
Nancy McGuire is a free-lance writer based in Silver Spring, Md. You can contact her at nmcguire@wordchemist.com.