Rowland circle geometry

Estimates of shifts of spectra in curved crystal geometries are often calculated for an ideal detector located on the Rowland circle. However, the detection surface is usually fiat and therefore cannot lie on the Rowland circle. Detectors located on a fixed length detector arm will additionally travel off the Rowland circle as the Bragg angle is scanned unless the crystal curvature is simultaneously scanned (which raises problems of stress hysteresis). Conventional shifts calculated for detection on the Rowland circle do not agree with shifts at a flat extended detector and this systematic error can be 100-200 ppm for any Johann curved crystal spectrometer. We have incorporated fiat surface detectors located off the Rowland circle into the general theory [18,17]. 

---