www.Herguth.com

800-645-5227

Particle Characterization and Sizing

Scanning Electron Microscopy (SEM)

Utilizing Automated Electron Beam and Automated Feature Analysis (AFA) Software

For Particle Counting and Particle Characterization

 REGISTER NOW FOR A FREE GIFT!

Introduction:

In recent decades oil analysis laboratories have used automatic particle counters to determine the number and size of particles in many oil wetted systems. 

Once determined, there has been the question of; what are they?  The logical next step was to look at the metals analysis or extract them onto a filter membrane

and view them under magnification.  Characterizing the make-up of the particles is very useful in determining the source.  Once the type of particles was

determined a cost effective action plan could be undertaken to rectify the problem.  The problem with these two approaches is; the metals analysis is often

unreliable in identify larger particles and; under optical magnification it is not possible to tell for sure the make up of the particles.

In some cases SEM analysis was undertaken to further identify the particles.  This proved to be time consuming for the SEM operator and it was left to the

operator to judge which areas of the sample to investigate.

This report illustrates the performance and innovative approach to resolving the aforementioned limitations on sizing and identifying particles using an “Automated

Electron Beam Particle Analyzer” equipped with “Automated Feature Analysis” (AFA) software to characterize complex matrix of particles extracted onto a

membrane filter from oil samples. 

Conclusion:  The SEM/AFA analysis offers the analyst a superior method of determining the size and make up of particles in lubricating oil, grease, filters

and process materials.  This knowledge and appropriate actions can save a plant thousands of dollars in needless speculation and inappropriate expenditure.

Methods:

In the case studies Automated Feature Analysis (AFA) was performed using the Rotating Chord Analysis algorithm to determine Size, Shape, Number and

Type of particles > 4 um that were on the surface of a filter sample.  Most often, prior to analysis, a liquid sample was filtered on 0.45 µm pore filter membrane.

Once analyzed, Tabular Reporter is used to mine the data, and determine the best way to interpret and visualize the data. Microsoft Excel® is used

to do graphical representations of tabular data. Each filter when analyzed is divided into stage movements and electronic fields. The electron beam analyzers

do not take an image and then processes/analyzes the image; rather it is a dynamic scanning of the filter. When a particle is located based on a defined

signal change, the system stops the scan and acquires an Energy Dispersive Spectroscopy (EDS) spectrum.

The spectrum is analyzed in real-time to provide the elemental composition of the individual particle.  Table 2 provides “Guaranteed Performance Characteristics”.

Definitions:

Dmax: The maximum detected diameter of a particle. See figure below.

Dperp: The diameter perpendicular to Dmax. See figure below.

Aspect Ratio: The Aspect Ratio is defined by dividing the Dmax by Dperp.

Dperp

Dmax

Detected Particle

Rotating Chord Algorithm (RCA)

Description: Rotating chord is a computer algorithm which uses a two-step “Detect-and-Measure” mode. The Detect mode allows the instrument to search

and detect when a feature/article is present, or not, on the surface of the membrane patch based on atomic number contrasting using backscatter

electrons. Once the feature/particle is detected the instrument switches automatically to the Measure mode. In RCA, the electron beam of the instrument

rasters across the particle in lines that are called “chords”. These chords determine the center of the particle and the size and shape of the particle, by

effectively drawing a series of chords which encompass the particle (See Figure 1). RCA provides different characterization parameters which commonly

are used to determine the types of particles in a sample.

                                                                                      Figure 1.  Rotating Chord Analysis.

                                                                           16 Chord illustration of how RCA determines the size

                                                                                       and shape of a particle ( DMAX 80 µm)

                                                                   Table 1 Lists possible parameters that are determined by RCA.

    Table 1 Definitions (41 Categories are available)

Table 2 - Guaranteed Performance Characteristics**

** Performance as measured using Performance Grading Software TM

for features 1 to 100 µm

                                                                      IMAGE #1                                                               IMAGE #2

                                                                        IMAGE #3                                                                       IMAGE #4

 Rules for Particle Characterization:

Initially the area of interest on the membrane filter is defined.  This can range from 100% of the effective area or as little as 1%.  The selection is usually based

on the level of detail required for a thorough analysis vs. time and budget considerations.  The system is then set up to determine what size particles are of interest

and in what ranges (bins).  After these decisions have been made the analysis can begin.  The system scans (rasters) the filter in as little as 0.5 micron if desired. 

Once complete the results of the analysis can be viewed by the operator.  At this point the operator decides what particles are on interest.  The analysis rules can now be written and the data obtained and presented to the client.

In Case Study #1 the Rules are shown in Table #3.  After scanning the filter is was determine that the majority of the particles had the elements described in the Rule.  

Tin and iron being the most dominate, but certainly other particles and alloys were present.  So, the Rule was made requesting number and size of particle containing the elemental proportions in the Rule.

Table# 3

CASE STUDY #1

SEM/AFA

January 9, 2007

Customer:  Nuclear Power Plant

Description of Sample:  Containment Spray #1 - #1 Motor Inboard ID: 3485 6:31:23 AM MRSR# DB-2612

Background:  Due to the critical nature of the application the client samples the Containment Spray Motors about every month.  The standard test package

includes metals analysis, particle count and physical properties testing.  The particle count for this motor bearing oil had been consistently higher than all of the

other motor of the same make and model (ISO Code 4406 = 21/18/14 vs. 16/14/10).  There was no indication of silicon or wear metals by the emission spectroscopy. 

It was very important for the root cause investigation to determine what these particles were and their sizes.  It was decided to employ the SEM/AFA to resolve

any doubt in the analysis.

Sample Preparation:  5 ml of thoroughly mixed oil was diluted in heptane solvent and drawn through a 0.45 um filter. 

SEM Observing conditions:  The specimen was observed under reduced vacuum at 15keV of beam energy to perform automated feature analysis on a non-conductive substrate.  Working distance was approximately 16 mm to facilitate x-ray spectroscopy.

General observations:  Approximately 9.5% of the surface area of the filter was analyzed with the automated process.  5618 particles were identified and classified

by chemical composition and size.  This extrapolates to 1,182,740 total particles per 100 ml, which is approximately 6.67% lower than indicated by HL-1185 particle count. 

The difference is mainly due to the way the two processes account for particles of less than 4 um in size. 

Most debris in the sample is dominant in tin or iron.  The tin dominant material displays as a range from nearly pure tin to various brass-like mixtures with copper,

zinc and lead. The iron-dominant particles are mainly a low chromium alloy or oxides, however about 5% of the iron particles contain greater than 5% chromium.

Graph 1- Shows the predominance of particle composition found in the sample

The particles were classified by chemical composition to sort out the most frequently appearing particle types. 

TABLE 1

Table 1- Lists the classifications, their definitions and the relative percentages of each in the specimen.

TABLE 2

Table 2- Lists the particles sorted by size in the same ranges as HL-1185.

TABLE 3

Table 3- Shows the differences in the relative percentages of the various size groups between the SEM/AFA and the HL-1185 particle count.

                                                                                                                    Case Study #2

December 8, 2006

Customer:  Coal Fired Power Generation

Sample Description:  Pulverizer Gear Box, ISO VG 150

Sample Preparation:  20 ml of thoroughly mixed oil was diluted in heptane and drawn through a 0.45 um nitrocellulose filter, followed by thorough rinsing with heptane.

Analysis Parameters:  100% of the exposed filter area was searched with a criteria to find all particles of 4 um and greater in dimension.  15 keV beam energy and

reduced vacuum were used with a backscatter electron detector. 

Summary of Findings:  A total of 2083 particles were identified on the filter.  Eight general categories of particles were identified based on their chemical composition. 

The particles are identified and separated out of the group in the top to bottom order listed in the below table according to the listed criteria.  The criteria were selected

based on the dominant patterns of chemical distribution in the 2083 particles.

This is a summary of the particles found, sorted by classification:

The table below shows the particles sorted by classification and ISO 4406 particle size criteria.

The chart below shows the top % elemental composition by average for each of the listed elements in the 10.67% miscellaneous classification.

Mining the Data:

The table below shows key parameters for a typical analysis.  There is far too much data to present it all here 285 Particles counted with 41 different categories of particle characteristics.  However, this table illustrates that you can easily sort by parameters of interest.  In this case we sorted on DMAX (descending). However, DAVE , the average length of all 16 Chords is the definition of the “particle size”.

It is clear these data show the particles are:

a. On average the particles are oblong due to their ASPECT ration of ~2.

b. The particles are primarily iron and silicates.

c. The average size is 59 µm (DAVE).

ASPECT Ratio:  The aspect ratio gives information about the relationship of the length of a particle to the width.  Illustrated below are seven different aspect ratios ranging from 1 to 7.  If each square below represents a particle with an aspect ratio of 1 (DMAX/DPERP) you can see, the higher the aspect ratio the longer and narrower a particle is in size.  Examples of particles with large aspect ratios are cutting wear particles or fibers.  However, the aspect ratio is not always indicative of cutting wear.  We have demonstrated that different alloys produce cutting wear particles with aspect ratio nearing 1.  The alloy of 310 Stainless is an example of a alloy that forms low aspect ratio cutting were particles

                                       ******************************************************************************************************************************