| Abstract
| - A wavelet−neural network signal processing method hasdemonstrated ∼10-fold improvement over traditionalsignal processing methods for the detection limit ofvarious nitrogen and phosphorus compounds from theoutput of a thermionic detector attached to a gas chromatograph. A blind test was conducted to validate thelower detection limit. All 14 of the compound spikes weredetected when above the estimated threshold, includingall 3 within a factor of 2 above the threshold. In addition,two of six spikes were detected at levels of half theconcentration of the nominal threshold. Another two ofthe six would have been detected correctly if we hadallowed human intervention to examine the processeddata. One apparent false positive in five nulls was tracedto a solvent impurity, whose presence was subsequentlyidentified by analyzing a solvent aliquot evaporated to 1%residual volume, while the other four nulls were properlyclassified. We view this signal processing method asbroadly applicable in analytical chemistry, and we advocate that advanced signal processing methods should beapplied as directly as possible to the raw detector outputso that less discriminating preprocessing and postprocessing does not throw away valuable signal.
|
