Calibrated Procedure for Setting Pulse-Height Parameters in Wavelength-Dispersive Spectrometry

Abstract

Electron-probe microanalysis (EPMA) routinely makes use of wavelength-dispersive spectrometry (WDS) to measure characteristic X-ray photons over a wide range of concentration, so the WDS pulse processing system should separate counts from electronic noise over a wide range of count rate. WDS pulse processing hardware is simple compared to the sophisticated pulse processing hardware used in energy-dispersive spectrometry (EDS), as there is no monitoring of deadtime or active referencing to a zero strobe. WDS uses X-ray diffraction to select the desired X-ray wavelength, and all measured pulses are presumed to be for the element of interest. The pulse height analysis (PHA) system is set to discriminate between baseline noise and the measured X-ray pulses. Amplification of the detector signal is made using detector voltage referred to as bias (typically in the 1600-1800 volt range) and a gain setting in the electronic amplifier. On JEOL electron microprobes the gain setting is an integer multiplier (e.g., 16x, 32x, 64x, or 128x), and on Cameca microprobes the gain parameter is specified from a range of values. The bias and gain are set for a particular X-ray of interest, and a single-channel analyzer (SCA) scan is used to inspect the pulse distribution of the X-ray detector. For JEOL instruments this SCA scan is 0-10V with a pulse ideally located at 4V, and for Cameca instruments it is 0-5V with a pulse ideally located at 2V. Failure to set the PHA parameters to the correct values may result in a pulse not being registered at all, or some portion of baseline noise being registered as counts. Further, gain shift due to changing count rate requires attention to PHA settings to ensure processing over this range of count rate. It is therefore up to the analyst to understand and implement PHA settings. JEOL electron microprobes use flow-proportional detectors streaming P-10 (90% Ar-10% CH 4) for measurement of low energy X-rays, and sealed Xe detectors for the measurement of high energy X-rays. Cameca electron microprobes use either low or high-pressure P-10 detectors for low and high energy photons, respectively. Two different approaches are taken with setting of PHA parameters on these systems. JEOL PHA settings use an integral gain values coupled with a per-element bias, and Cameca PHA settings are made at constant bias and either fixed or per-element gain selected from a range of values. The JEOL SCA is set to use a constant baseline level (0.5 to 0.7 volt) with gain and bias set to place the pulse at a nominal 4V, and either an integral or differential window is used. The Cameca SCA ranges from 0-5 volts and baseline and window settings are defined by either the operating system or the user. Experience with measurement of WDS deadtime constants on JEOL microprobe detectors resulted in use of bias set on a per-element basis on each detector. This places the pulse at 4V and provides consistent clearance above the baseline with room for the escape peak, and headroom to permit inclusion of coincidence pulses at high count rates. The PHA calibration method described here is specific to the JEOL instrument, and was initially developed for Sandia-TASK [1]. First, X-ray energies are selected that span the energy range accessible by each spectrometer. For example, an instrument with several spectrometers each with PET and LiF crystals could use the K lines of Si, Ca, Ti, Fe, Cu, and Ge for calibration. Pure element standards or compounds are ideal for this use. All spectrometers with PET crystals are then peaked on the same X-ray line (e.g. Si K on Si metal). With all spectrometers at this peak position, the probe current is adjusted to obtain no higher than 20 kcps count rate. All spectrometers are then set to the same gain value for simplicity. Next, a bias scan is acquired by setting the SCA to a 4 volt baseline with 0.2 volt window in differential mode so that only 4-4.2V is counted. The bias values are then varied between 1500 and 1800V and the resulting peak on the bias scan, if any, is the bias value at which an SCA scan will produce a 4V pulse (Fig. 1A). An SCA scan can confirm the bias value does indeed produce a 4V pulse (Fig. 1B), but in practice the bias scan is all that is necessary. This procedure is then repeated for each set of gain values and then repeated for all the elements that are used to span the energy range as listed above. The data are plotted using the natural log of the X-ray energy vs. the bias value measured for each element. For Ti K the 420