XRF is a spectrometric technique that can be applied to many materials for composition analysis, as well as coating and multilayer thin-film thickness measurements. As with all analytical instrument purchase decisions, the driving factor should be the specifics of the application.

The following will present important factors to consider when choosing XRF configurations to measure thin-film coatings. We will be focusing on Energy Dispersive XRF (EDXRF) systems that are now offered in a number of benchtop configurations and more recently in portable, handheld configurations¹.

EDXRF Spectrometers (major components important to selection):

Atmosphere:

XRF instruments have analysis sensitivity to a large portion of the Periodic Table. In very general terms, from Sodium (atomic number, Z=11) to Uranium (Z=92). However, air starts absorbing the characteristic emissions of lower Z elements, below Ti (Z=22). The fluorescent yield from lower atomic number elements is lower and other elements in the sample matrix are more likely to absorb the lower energy emission of lower Z elements. Common methods of improving low Z sensitivity are to evacuate the instrument sample chamber or to purge the sample chamber with Helium (He flush).

Detectors:

Newer detector technology, Silicon Drift Detectors (SDDs) offer improved low energy sensitivity enabling analysis of some low Z elements, even in air. For example the Phosphorus (Z=15) content of Electroless Ni (EN) coatings¹. However, many low Z analyses will still require getting rid of the air.

Silicon detectors have become very prevalent in EDXRF. Today they are either the newer SDDs noted in the previous paragraph or Si-PIN detectors. The third type of detector prevalent in EDXRF instruments is a sealed, gas filled proportional counter (Prop Counter). Table 1. presents the relative characteristics of these detector types and the general application significance.

Expanding upon the significance of the EDXRF detector choices with regards to various applications—qualitative analysis requires a silicon detector. The large FWHM of Prop Counter results in heavy overlaps of adjacent elements to the level where peak search algorithms and/or visual spectrum observation may not detect the presence of one or more component. This also holds true for some incoming inspection of manufactured materials that require the identification of elemental components that may or may not be present. Although peak overlaps do occur with Si detectors they can, for the most part, be easily separated and identified making Si detector systems ideal for qualitative and incoming material inspection.

The electronics that make up EDXRF spectrometers are generally very stable and do not affect analysis repeatability (precision). Random counting error typically has the greatest impact on measurement precision. Counting error follows Poisson statistical distribution—the more counts acquired per measurement the better the precision. The very high count throughput of the SSD detector is an important consideration when measurement applications require high precision with high sample volumes, but to take advantage of this attribute requires high fluoresced intensity from the sample.

The amount of fluoresced intensity will be sample dependent—what the material is, but importantly how much is there. In the case of very thin films or small samples, small sample features, this can be tiny. An SDD may not offer a great advantage over a Si-PIN if the counts are not available.

When samples or sample areas are small (10 tens of microns in diameter) the solid angle of the detector can play an important role. This can often be the case with the measurement of electronic component functional coating thickness measurement and is why the Prop Counter is a viable and popular choice. The large capture angle enables the use of smaller collimators. When the sample spectrum is fairly simple, two or three elements and the analysis area is small ~100-200 µm in diameter, a Prop Counter based instrument can be an ideal configuration.

The noise characteristics of the detection system will play a large role in conjunction with the resolving capabilities. The result is the peak-to-background response of the system, and therefore, the detections limits whether these are expressed in terms of low concentrations or very thin coatings – RoHS analysis or electronic contact ENEPIG measurement. For these applications a Silicon detector is essential and the SDD the preferred choice.

X-Ray Source (Tube, Power Supply, Primary Filtration, Beam Size)

Here I am lumping a number of components into what I am calling the source, which is usually considered the tube and power supply. The perspective taken here is what the sample “sees.” The tube and power supply determine the energy distribution and intensity that the sample sees. The majority of analytical EDXRF tubes used in commercial spectrometers are 50 kV and 1 mA (50 W). The 50 kV potential provides good excitation efficiency for K emissions to Sb (Z=51). The L emissions of higher Z elements are then used to Uranium. The source tube flux is controlled with the filament current settings. The source will generate a continuum (also known as Bremsstrahlung – breaking radiation) with characteristic lines of the anode material superimposed on top. Common anode materials are Tungsten (W), Rhodium (Rh), Molybdenum (Mo), and Chromium (Cr). The most common are W and Rh. W generates higher Bremsstrahlung and therefore better high energy (17-30 keV) excitation efficiency. Rh is often selected for low Z element excitation efficiency.

Primary filters are inserted between the tube window and the sample to filter specific energies emitted by the tube. They serve two functions: filtering (removal) of tube characteristic lines when they may interfere with the emissions of analyte elements from the sample and to remove backscattered continuum from the tube, which is the primary source of spectrum background. The removal of background can enhance peak-to-background response and thereby, detection limits.

Beam size is usually controlled by circular collimators of varying diameters and are sometimes rectangular. Typical EDXRF collimation will range from ¼ inch (for bulk analysis) down to 50 microns for small part or feature analysis. The collimator size in conjunction with collimator to sample dimension determines the analysis area on the sample. This is often referred to as “spot size” and determines what is being measured. This can be critical when performing truly nondestructive analysis of materials that are not homogenous, i.e., a populated printed circuit board. It is important that the fluoresced X-rays are coming from the Au pad and not an adjacent solder joint.

Applications and Instrument Configuration:

As indicated in the introduction of this paper, we will focus on instrument selection for thin-film coating applications. All X-ray spectrometers will have some combination of the components discussed thus far. This is true of benchtop and portable, handheld instruments.

However, handheld instruments do place some constraints on these components, because they must be small and relatively light. Most handheld XRF instruments weigh around 1.3 to 1.8 kg (3 to 4 lbs). This is significant when trying to hold an analyzer steady for any period of time. This and dimensional size, like a ray gun, means that the detection system of choice is Si based (Si-PIN or SDD). It is noted above that most EDXRF instruments use 50 Watt tubes. This is true of benchtop units, but cannot be implemented in a handheld package due to the power draw that requires a larger tube and power supply and is not sustainable by a light battery. Handheld X-ray tubes are mini tubes that operate at ≤ 10W. This means that tube flux (excitation intensity) is very limited. The lack of flux is largely overcome by keeping the working distance between the sample, mini-tube and detector very small and the beam collimators relatively large, 1 to 3 mm in diameter. This is large for many thin-film applications, such as electronic industry applications.

All EDXRF systems are capable of bulk sample analysis—benchtop units and handheld units. This means that the samples are infinitely thick (thicker than the highest energy analyte line escape depth) and homogeneous. Thin film measurement capability is enabled by software algorithms that can deal with layers that are less than infinitely thick. This means that the algorithms correct for interlayer matrix effects (the same as bulk samples), but also intra-layer matrix effects, but there are also geometry considerations. Bulk analysis compares relative intensities, but thin-film analysis deals with absolute intensities. If the sample is off the spectrometer’s analysis plane the thickness reading will be off.

Benchtop units can maintain the focus plane by moving the source and detector with the aid of a camera and precise motor drives or focusing measuring distance algorithms to a fixed focus distance. Portable handheld units define the focus plane at the face of the “gun.” Additionally, the escape angle of the analyte emissions must be defined for accurate thin-film coating measurement. This dictates that the snout of the gun be flush with the sample surface (no tilting allowed), and so, handheld units that offer thin-film software are usually fixed in a stand to maintain geometry, but then they are no longer truly portable.

Conclusions – Practical Thin-film Applications for Benchtop and Handheld EDXRF Systems:

Many benchtop EDXRF systems have been specifically designed to measure thin films; they have the geometry defined and adjustable focus to maintain it, as well as highly collimated beams to analyze small samples and/or small sample features. Most handheld held instruments on the market were designed for bulk analysis, largely alloy material identification.

There is application for handheld systems that can offer the required software algorithms and can maintain the geometry that the software is “looking” for—geometry parameters that are fixed by the handheld design. The need for portability in thin-film analysis can be binned as follows:

1. Proximity to the coating process—taking the analyzer to the process in order to better control the process. In this case the sample area needs to be larger than the beam size, i.e., > 3x3 mm with a 1 mm collimator.

2. Large production parts where they are too large to fit in a sample chamber and too valuable to cut up. In this case sample area is typically not a problem.

An ideal application for Handheld EDXRF is corrosion coatings in the aerospace industry. Such an example is ZnNi coatings that are gaining popularity as a replacement for Cd coating. One that we have demonstrated is the ZnNi corrosion coating on aircraft landing gear. Of note is the design to hold the spectrometer window flat against the sample analysis surface and thereby maintain the geometry necessary for accurate coating measurements (Figure 1). Recent testing yielded the same measurement precision as a benchtop system (Figure 2).

An area where handheld instruments are not appropriate is with many electronic product applications, such as SnPb analysis for RoHS or Hi-rel solder analysis on PCBs. The beam size and multiple geometries of a populated board make any kind of quantitative analysis impossible, many of which should be handled as thin film analyses. Even as a screening tool, handheld XRF systems can fail where the total mass of the SnPb is so small (thin finish or small component) relative to the spectrum intensities from a larger collimator determined area that may include scatter from the epoxy board and emissions from adjacent components that the Pb emissions are either not detected or incorrectly quantified to pass levels for RoHS or fail levels for Hi-rel. Only benchtop units should be considered for this application.

 1 “X-Ray Fluorescence for Today’s Quality Assurance”, J. Bogert, Quality Magazine, January 31, 2014