Peak-width determination of step-scan profiles: the floating-baseline method

A new method of measuring peak widths for Bragg reflections, collected by the step-scan technique on single-crystal X-ray diffractometers that utilize crystal monochromators, is reported. The method is based on the assumption that the baseline of each peak profile is nearly linear, an assumption that can be tested by the runs test for randomness. A series of linear regression fits is calculated employing the profile steps and their intensities and the resulting baselines are tested for linearity. The strategy is to eliminate from the peak profile, beginning near the peak center, scan steps that do not contribute to a linear baseline. After each scan-step elimination, the linear regression fit is recalculated and the resulting baseline retested for linearity. If the resulting baseline is linear, then the process exits and the innermost steps are used to determine the peak width, otherwise a profile step near the peak center is removed and the process is repeated. The method reported does not employ a windowing scheme and is not dependent on the initial starting location for the calculations. The results indicate that the method reported determines wider peak widths than those calculated by other peak-width-determination methods. The metric results for data sets of several crystal structures illustrate that data analyzed by the reported method are similar in quality to data analyzed by the learnt-profile method, and examination of the weighted residuals for all data analysis methods employed indicates that the standard deviations of the intensities are generally underestimated. For data sets that contain poorly defined peak profiles, the reported method can generate lower residual values. The method has been named the floating-baseline technique because of the illusion of the successive regression lines floating on the peak profile until a linear baseline is detected.