Database for and statistical analysisof the interlaboratory determination of the conversion coefficient for the measurement of the interstitial oxygen content of silicon by infrared absorption

Abstract

The Institute's overall goal is to strengthen and advance the Nation's science and technology and facilitate their effective application for public benefit. To this end, the Institute conducts research to assure international competitiveness and leadership of U.S. industry, science and technology. NIST work involves development and transfer of measurements, standards and related science and technology, in support of continually improving U.S. productivity, product quality and reliability, innovation and underlying science and engineering.

Standard Reference Materials :

“…Calibration of the NIST FTIR instrument was based on measurements of a subset of eight reference specimens from a larger set of specimens previously used in an international study called the Grand Round Robin (GRR). [2,3] The bias of the eight-specimen subset with respect to the complete GRR specimen set is -0.72%, as established by the GRR data. The measurements made to certify the SRM have been adjusted to eliminate this bias.…”

“…Measurement of the oxygen number fraction is commonly made by infrared spectrophotometry in which the height of the strongest spectral oxygen peak, at 1107 cm"^is determined. The database and test method for this infrared measurement are well established [1][2][3][4][5][6][7]. It is a relative technique that relies on…”

“…There were significant discrepancies among the early values, but more recent work has produced conversion coefficients in much better agreement with the current internationally accepted value. That number is the result of the Grand Round Robin (GRR), a major worldwide experiment completed in 1988 [1,2]; it has been adopted by all of the major standards-writing organizations active in the semiconductor field.^More recently, a European (BCR) reference material for oxygen in silicon was completed in 1991 [3][4][5].…”

“…The uncertainties for these absolute techniques are not well characterized, but their uncertainty is much larger than the expanded SRM uncertainty, which is less than 0.2%. The spread of data among the different laboratories which made the absolute oxygen measurements was 10 to 30% [1,2J. In a subsequent BCR study, the intra-laboratory reproducibility of absolute measurements was +4%, and a propagation of errors estimate of absolute uncertainty was ±6% [3][4][5]. Since these absolute oxygen uncertainties, which are the best currently possible due to the limitations of the existing measurement techniques, are large, one must use these data (averaged over a number of specimens and laboratories) simply to arrive at a scaling factor for the measured optical parameter.…”

“…17% for three oxygen levels. These values are much lower than the uncertainty for absolute oxygen measurements in which case the interlaboratory variability can be as large as 10 to 30% [1][2][3][4][5]. Improvement in measurement consistency is more useful to users for quality control or comparison than is reliance on the uncertainty of absolute oxygen determination.…”