Machine wear material identification by FingerPrint

This application note shows that the Epsilon 4 is fully capable of identifying small differences in lubrication oils.

The FingerPrint technique can be used for identifying the presence of machine wear in lubricating oils but is applicable for the identification of all kinds of materials including liquids, powders and solids.

Introduction

This application note shows that the Epsilon 4 – a high- performance benchtop energy dispersive X-ray fluorescence spectrometer – is fully capable of identifying small differences in lubrication oils. The FingerPrint technique can be used for identifying the presence of machine wear in lubricating oils but is applicable for identification of all kinds of materials including liquids, powders and solids. FingerPrint software is available for the Epsilon 4, the Epsilon 3X and Epsilon 3 ranges of XRF spectrometers.

Application background

When analyzing lubrication oils it is common practice to relate wear element composition and concentration ratios to critical machine parts. During its use, lubrication oil can undergo several changes, such as oxidation and reduction in hydrocarbon chain lengths. These changes are evident in EDXRF spectra, but typically ignored when applying traditional XRF techniques (for example quantification). The advanced statistical approach of FingerPrint can readily differentiate between all these oil properties. Furthermore,  as a simultaneous technique, detecting all elements (from Na to Am), it is well suited for rapid elemental analysis.

Instrumentation

The instrument used for this study was equipped with a rhodium anode X-ray tube, a 10W, 50 kV and 2 mA generator, 6 filters, a helium purge facility for elements lighter than titianium, a high- resolution silicon drift detector, a spinner and a 10-position removable sample changer.

Sample preparation

Approximately 5 g of each oil sample was weighed into disposable P1 sample cups. Each P1 cup was constructed with a 3.6 micron Mylar® X-ray support film. Total sample preparation time was less than 2 minutes per sample.

Method setup

The optimal FingerPrint conditions were determined using the standard feature: Optimize Conditions. Measurement times were 3 minutes per analysis. The FingerPrint application was chosen by toggling the FingerPrint button and using the default settings.

Create a library – measure typical samples

The FingerPrint option permits custom library development. In this study 7 certified lubrication oil standards were used to set up the library. Each standard was given a name to mimic a suitable machine problem, for example: contaminated oil or engine problem. No elemental concentrations were required and no spectral interpretation or deconvolution was necessary. All materials were simply loaded and measured.

Measurements

Samples were simply placed on the sample tray and measurements were started after the corresponding sample positions were identified.

Technique robustness

To investigate the robustness of this method, several parameters were varied including measurement time, sample composition and sample weight.

Instant results – unknown sample identification

The identification of likely candidates is instantly reported where the Chi2 values are below 3. In this example the standard named contaminated oil, was measured as an unknown sample. The FingerPrint module identified only one lubricating oil from the library, as it was the only material with a Chi2 value less than 3, Table 1. For illustration purposes an expanded list is shown below. The result was a correct match demonstrating the distinguishing power of the FingerPrint software with Epsilon 4.

Table 1. Expanded view candidate list and relating Chi2 scores

table1.PNG

Lubricating oil spectrum library

Using the spectrum viewing function it is possible to view sample and standard material spectra. This step is not necessary but illustrates the resolving power of the Epsilon 4. A selection of oil spectra is shown in Figure 1.

Figure 1. Selection of oil spectra

figure1.PNG

Application time variations

Although the method was initially set up for 3 minute measurements, changing the application time had negligible effect on the FingerPrint identification capability. This was illustrated by the large difference between the Chi2 scoring of the confirmed and nearest candidate, Table 2.

Table 2. The effect of application time on Chi2 scores

table2.PNG

Composition variations

To demonstrate the influence of compositional variation on identification power, the standard named engine problem was diluted with white oil and the Chi2 of the first and second  candidates are reported. In all cases the standard named engine problem, was the first candidate in Table 3.

Table 3. Chi2 scores for different dilutions

table3.PNG

Sample weight variations

The mass of standard named suspect engine oil was varied from 5 to 1 g. The Chi2 and the next candidate are listed in Table 4.

Table 4. Chi2 scores for different sample weights

table4.PNG

The results demonstrate that the built-in scaling feature of the FingerPrint software can accommodate significant reduction in sample weight without compromising on performance.

Repeatability

The standard named engine problem was measured 254 times as unknown with default settings.

Table 5. Repeatability of Chi2 scores

table5.PNG

In each instance this technique correctly identified the unknown as engine problem in Table 5.

Conclusion

This work clearly demonstrates that the Epsilon 4 can be used for rapid and reproducible identification of used and unused lubrication oils. The results demonstrate that the excellent distinguishing capabilities are relatively insensitive to variations

of measurement time, dilution and sample weight. FingerPrint software is an effective tool for preventive maintenance programs and is easily operated.

Einloggen

Noch nicht registriert? Konto erstellen