Scanning electron microscopy/energy-dispersive spectroscopy (SEM/EDS) can be an effective method of wear debris monitoring to help prevent equipment failures. Modern SEM/EDS has evolved to enable automated analysis and processes, which are key techniques in circumventing the technical challenges involved in selecting the right tools to perform a proper analysis in a timely and economical manner. The following three case studies will show how SEM/EDS technology can successfully resolve tough analytical problems using particle sizing and characterization, wear debris analysis and filter debris analysis.
The scanning electron microscope and energy-dispersive spectrometer (SEM/EDS) is a seasoned tool that is making its way into many oil analysis labs. This high-powered microscope with an emission spectrometer makes it possible to characterize particles from various contaminant sources for size, morphology and elemental composition. This sophisticated tool is currently referenced in many industry test methods to help troubleshoot potential sources of contamination1.
The main advantage of using SEM technology is that the technique can be employed across a variety of sample matrices and test mediums. While emission spectroscopy and automatic particle counting by laser technique requires a fluid sample to provide composition and particle distribution, SEM/EDS offers a one-step process to review fluids, solids and filter medium to obtain both particle size distribution and composition of particles and contaminants. Current industry methods have expanded to include SEM technology and its effectiveness into the lubricant and tribological industries across a variety of situations.
The use of both SEM and EDS technology for sizing and characterizing particle contamination was formally championed by the automotive industry’s test method for cleanliness, ISO 16232: Road vehicles - Cleanliness of components of fluid circuits. This standard was developed to report the cleanliness levels of parts and components fitted to automotive fluid circuits. The method is divided into many sections which address various aspects of sample collection and preparation. However, part 7 and 8 specifically address the use of an SEM/EDS tool to determine the size and number of contaminant particles, as well as the nature of the contaminants by identifying their elemental chemical composition, which can be used to classify the particles into likely material groups.
SEM/EDS technique is also recommended in ASTM D7919, Standard Guide for Filter Debris Analysis (FDA) Using Manual or Automated Processes, and ASTM D7898, Standard Practice for Lubrication and Hydraulic Filter Debris Analysis (FDA) for Condition Monitoring of Machinery. In these guides, it is recommended to evaluate the in-system filter elements for debris and solid contamination, which aids in the determination of machine condition and root cause analysis (RCA). The guides specifically address techniques to remove, collect and analyze debris from filters in support of machinery health condition monitoring.
As noted in ASTM D7919, filter debris analysis is an effective means of monitoring equipment wear because the wear history is efficiently captured in the filter matrix. The method states that correlating the filter contaminants to “normal” and “abnormal” lube system operation provides early indication of a contaminant- or component-wear-related problem. It goes on to suggest that analysis of the contaminant collected within the lube filter element provides a tool to identify the failure mode, its rate of progression and the source of the contamination. Typically, more than 95 percent of all released metal particles larger than the filter’s pore size are captured in the filter. In addition, other types of particulate contamination are captured, including seal wear material and environmental contaminants, which can also provide diagnostic information2.
These methods recognize that SEM/EDS analysis can be very time consuming to determine the composition of each particle in a filter. Unless an automated piece of equipment is available, elemental analysis of every individual particle on a test specimen is generally not practical. Modern SEMs are equipped with automated software to aid in the characterization of complex particle matrices. This technique employs a rule table to categorize components, which helps to further identify specific compounds. The classification rules may also allow for the trending of data from components of similar applications/systems. By utilizing high-resolution SEM and a computer-generated "rotating cord" algorithm, SEM determines the number, size and nature of contaminants and wear debris particles3.
An electron microscope generates high-energy electrons and focuses them on a specimen. The electrons interact with the specimen in many ways. Detectors measure the results of these interactions and provide signal data to software that converts the data to calibrated images and measurements. Two general designs of electron microscopes have been developed over the years: the scanning electron microscope (SEM) and the transmission electron microscope (TEM). The scanning electron microscope detects and measures three parameters that are the result of the beam electron interactions with the specimen: backscatter electrons, secondary electrons and X-ray photons.
Backscatter electrons are the electrons from the electron gun that scatter the electrons in the target atom orbitals and come out of the specimen to be detected. They are mostly high in energy level and are useful in identifying the atomic number of the target atoms with which they scatter because their backscattering coefficient is proportional to the target’s atomic number. Secondary electrons are mainly those electrons ejected from orbitals of the target atoms as a result of interaction with the electrons from the gun. They are much lower in energy and, because of this, are absorbed in the specimen very easily before they can be detected. Consequently, only those electrons ejected very close to the surface of the specimen are available to be detected. This makes them very useful for defining the surface topography of the specimen.
When secondary electrons are generated, they leave an unoccupied position in the orbital arrangement of the atom in the specimen. An electron from a higher orbital immediately fills this vacancy. The electron gives up some energy in the form of a photon as it “relaxes” into the lower orbital. The exact energy level of the emitted photon is unique to the specific atom and orbital relationship involved. The photons in the useful range for elemental identification are within the X-ray energy range and typically fall between 1 kiloelectron volt (keV) and the maximum beam energy, up to 20 keV. They come from the inner orbitals (K, L and M).
The instrument obtains an image by directing the beam electrons in a scanning action, scanning rows of pixels from top to bottom in a selected square area. The scanning stops at each pixel for a controlled short period of time and measures the current in the detector at that point. The image processing system then converts the varying measurements to a grayscale image. The scanning action is controlled by a separate set of coils in the objective lens. Magnification is accomplished by adjusting the size of the area on the specimen that is scanned. The image processing system creates a screen image that has the same dimensions, regardless of magnification. As magnification is increased, a smaller area of the specimen is scanned and expanded to the screen image size. Electrons have mass and electric charge, so they interact with matter very easily. The interior of the instrument’s electron beam column must be kept at a very high vacuum so the electrons can reach the specimen. The compartment where the specimen is located, called the stage, may also be under a vacuum so the scattered and emitted electrons can travel from the specimen to the detectors.
The beam electron’s path of travel is an electric circuit, concentrating electron current on a point on the specimen. The electron current must find a path off the specimen and back to the power supply or an electric charge will accumulate somewhere, probably on the specimen, resulting in the distortion of the emitted electrons. The method for mounting the specimen (slide, tape, filter, etc.) must be either capable of easily conducting electric current under high vacuum conditions or discharging any built-up charge to a limited atmosphere in the sample stage under variable pressure conditions.
The images produced by the instrument are the result of digital image processing. Electron interactions occur at the speed of light. Charges accumulate in receptors at rates of thousands of pulses per second. The X-ray photon detector and its electronics must be kept at a very low temperature to keep the detector stable and to minimize electronic noise.
Examining a specimen manually in great detail under the SEM is a very time-consuming and tedious process. Many lubricant users want a quantified analysis of their sample. The analysis is often triggered by abnormal results from a different analysis, such as a particle count, inductively coupled plasma (ICP) analysis or analytical ferrogram. Users of the lubricant want to know what the material is, the quantity of each major constituent, and what shapes and sizes are involved. The automated feature analysis (AFA) capability of modern systems provides the means to have the software operate the system and perform the repetitive task of searching and measuring an entire specimen. This frees up the analyst to evaluate the raw data, establish criteria to separate out the major constituents and sort the data for each of those constituents into useful information for the user.
Electron microscopes define features by detecting contrast in signal intensity from one pixel to the next as they scan the specimen. In automated analysis, particles are usually defined based on their being of different chemical composition than their surroundings. The backscatter electron detector is used to accomplish this definition. The operator sets a general brightness level, and then a range of intensities, called contrast, near this brightness. Automated feature analysis (AFA) makes use of the difference in signal intensity caused by changes in backscatter efficiency as the scan passes over particles of differing chemical composition. A threshold signal intensity is set above which the instrument defines a particle as opposed to the lower (carbon) background intensity of the test medium. The instrument scans the specimen at a defined rapid rate, and when it detects an intensity at this threshold, it immediately shifts to its highest resolution and measures the particle. When it completes the measurement, it will obtain X-ray data from the particle if so desired. The instrument can also save an image of the particle. It then goes back to its more rapid search mode and looks for the next particle, and on it goes until it finishes. It will search and measure the entire specimen or randomly analyze the number of chosen fields. It can X-ray every particle or only those particles over a certain dimension, aspect ratio, area, video level or other operator-defined criteria.
Compiling contrast data occurs at a much more rapid rate than compiling X-ray data. It is often useful to divide the analysis into two parts, the dimensional measurements and the X-ray measurements. Accurate dimensional measurements can be taken rapidly over a large portion of the sample. The sample can then be scanned randomly for X-ray data, which takes 4 to 6 seconds per particle. The system works most efficiently and accurately when the specimen is not too heavily loaded with particulate, so generally a small volume of sample is used, diluted in solvent and drawn through a low-contrast filter for analysis.
Specimen preparation for lubricating oils and filters basically involves diluting a portion of the sample in a solvent and drawing the solution through a filter. Polycarbonate filters are the choice medium because the filter texture presents a low signal contrast. The specimen filters are then affixed to aluminum stubs using carbon conductive tape and placed in the microscope stage (see Figure 1).
Figure 1. Four sample specimens installed in SEM stage
The filter locations are defined to the instrument by designating their locations, shape and focus settings. When the instrument starts its process, it divides the sample into as many square fields as it can fit within the defined sample shape. These are called stage fields. Their size depends on the magnification used during the automated run and the number of sections into which the stage field is divided. Each stage field (a square within the circle) is then divided into 25 smaller squares called e-fields.
The instrument runs through its process by centering the stage in a stage field and then moving from e-field to e-field within the stage field using the scan generator. This is much faster because it involves less mechanical movement of the stage. Because the instrument is faster at measuring the particles than it is at collecting compositional data by X-ray, the process typically is divided into two runs, one for morphological measurements and the other for X-ray data. When the system finishes the automated run, it has collected a large file of data that can be processed like a spreadsheet.
The software has various tools to assist in analysis. An image can be saved for each particle and called up with its X-ray spectrum and critical data. The system remembers where the particle is located, so it can drive directly to it if desired. The rules applied while collecting the data initially can be called up and changed. The data set can also be recalculated to the new rules. The same is true for the X-ray vector file. Elements can be added or subtracted, and the data recalculated.
A sample of gearbox oil was submitted to determine the source of large, visible ferrous particles in the oil. The SEM/EDS results shown in the images at the top demonstrated that the large particles were actually small iron and sulfur-rich platelets that had agglomerated into larger visible particles. This proved to be quite a different problem than if they had been large wear particles. These types of 1- to 2-micron iron-sulfur particles are typical in gearbox oil and are considered to be mild corrosive wear versus severe wear1.
Figure 2. SEM/EDS of sulfur and iron-rich particles
Process contamination can be of interest, especially for those in the manufacturing industry. It is expected that debris will find its way onto finished products during the manufacture of parts and components. In order to evaluate the cleanliness of the manufacturing process, components can be analyzed to review remaining particle contamination.
A roller bearing assembly was submitted to evaluate the size distribution of particles remaining on the components after manufacture. The set of bearing rings and rolling elements, which contained one outer ring, one inner ring, two seal rings (same steel material as outer and inner rings), and rolling elements, was washed in laboratory-filtered solvent and evaluated for bearing cleanliness per Military Standard 1246C, product cleanliness levels and contamination control program requirements. The size distribution of the particles is summarized in Table 1. The bearing cleanliness level per Military Standard 1246C is 200. This assembly meets at 15 microns and above.
Table 1. Particle count per MIL STD 1246C
X-ray data was also collected for the measured particles. The criteria used to define and separate each particle class is listed in Table 2. The particle classifications were determined by identifying trends in the chemical composition of the particulate. The rules were applied with the data normalized without including carbon, as carbon usually comprises a major component of all the particle groups. Carbon is not included when defining classifications because it skews the other elemental concentrations too low. Some of the carbon represented in the measurement is from the carbon filter substrate. Figure 3 shows that approximately 65 percent of the particulate greater than 5 microns are comprised of various steels (carbon steel, stainless steel and iron oxides/corrosion), copper, zinc and tin.
Table 2. Classification rules
Figure 3. Particles greater than 5 microns by classification
Figure 4 shows the particle distribution data sorted by DMax (the longest chord in each particle). Notice that the distribution indicates the size range for each of the particles that were classified. The distribution shifts when the particulate is sorted for size using equivalent diameter as the criteria. Equivalent diameter is the diameter of a circle with the same area as the area of the particle measured (see Figure 5).
Figure 4. Particle size and distribution, particle classifications by DMax
Figure 5. Particle size and distribution, particle classifications by equivalent diameter
The average elemental concentrations for each classification can also be reviewed. Table 3 offers assistance in determining the general alloy of the component producing the metallic wear debris. It can also help determine the source of inorganic mineral dirt and other contaminants. When reading the table, find the classification of particle in the top horizontal row. Looking down that column, the percentage of other elements associated with that class of particle is displayed. This is the average concentration of all the particles in the class. For example, in trying to determine the source of the steel, it is useful to review other elements that might define the alloy, such as cobalt, chromium, manganese and nickel.
Table 3. Elemental concentrations per particle classification
Filter debris analysis (FDA) is a condition-based monitoring tool used by various industries to identify wear mechanisms in critical oil-lubricated machinery5. Filters are often discarded during regular maintenance, yet the filter traps potentially valuable wear data that may not be collected or analyzed consistently using standard oil analysis. FDA is not currently a standard condition monitoring tool used across any industry, mainly due to the barriers surrounding the filter’s larger size and costs of logistics and analysis. The benefit of filter collection and classification over standard oil analysis is that the filter can provide a more consistent collection of particles over time between periodic maintenance intervals5.
Recent studies in the wind industry proved that integrating the SEM/EDS tool with filter debris analysis on a routine evaluation is a great way to capture trendable data. The SEM/EDS characterized the particulate for both size and composition. The ability to identify all captured particulate, including non-ferrous particles and contaminants, allowed for an in-depth evaluation of potential wear and failure modes6. The process was also productionized, suggesting that customized SEM/EDS techniques can become a more affordable and trendable testing option.
The data in Figure 6shows good correlation of SEM/EDS data obtained from one test turbine between the large main gear oil filter and a small prototype side-stream filter with respect to particle size and classification5. The data is an average of data collected on the same turbine over the course of a year and a half, including similar and consistent preparation processes. The data is a clear indication that as many filters as possible should be analyzed for a comprehensive component profiling of failure modes6. SEM/EDS provides the consistent source for analyzing wear and contamination particles.
Figure 6. Particles by classification: large main filter (top) and small side-stream filter (bottom)
The SEM/EDS process and instrumentation are the ultimate tools for evaluating deposits and wear debris, particle sizing and characterization, failure analysis, filter debris analysis, contaminant analysis, and metallurgical studies. The sophisticated and automated software found on modern instruments offers the analyst a superior method of determining the size and makeup of particles in lubricating oil, grease, filters and process materials in one evaluation technique. Using the correct tools is critical to the time sensitivity of lubrication problems, and SEM/EDS analysis provides the means for evaluating a variety of complex situations effectively and timely.
This article was previously published in the Reliable Plant 2019 Conference Proceedings.