Comment on "Analysis of high-resolution x-ray diffraction in semiconductor strained layers" [J. Appl. Phys. 86, 782 (1999)]

The multilinear regression analysis proposed for optimizing the measurements and minimization of errors in the x-ray peak separation method is examined. It is shown that the interpretation of the residuals as the errors of the physical variables leads to paradoxical conclusions, and prevents the definition of a unique numerical value for the probable error. It is also shown that if the method is to yield a permissible set of lattice parameters, the residuals of the best fit must fall within the range of experimental uncertainties estimated in advance. Thus, the residuals are only useful as a measure of the internal consistency among the data, but they cannot be used to quantify or minimize the errors within that set. (C) 2000 American Institute of Physics. [S0021- 8979(00)03511-8].