Monitoring of randomly varying chemical processes by correlation chromatography M. Kaljurand a,* , H.C. Smit b a Institute of Chemistry, Estonian Academy of Sciences, Akadeemia Road 15, Tallinn EE0026, Estonia b Laboratory for Analytical Chemistry, University of Amsterdam, Nieuwe Achtergracht 166, 1018 WV Amsterdam, The Netherlands Received 8 May 1997; received in revised form 15 May 1998; accepted 26 May 1998 Abstract Correlation chromatography (CC) is a well known method for reducing detection limits in chromatographic analysis. Its usefulness for the monitoring of varying substance ¯ows is however more ambiguous. Although generally believed as a useful tool for this purpose, very little work has been done to prove it. In this paper CC and common, single injection chromatography (SIC) were compared with respect to their restoration capabilities of a stochastic ®rst order process, which is assumed to be the varying concentration input function of a chromatograph. It is demonstrated that CC can restore the input function with a smaller error than SIC when the process rate and input sequence length are related by a certain condition. The effect was found to be more profound when measurement noise is present due to the multiplex advantage of the CC method. # 1998 Elsevier Science B.V. All rights reserved. Keywords: Process monitoring; Multiple input/correlation chromatography; Measurement noise 1. Introduction Monitoring can be de®ned as the measurement of the changing properties of an object in order to detect too large deviations from a pre-set value [1]. The need for such an operation frequently appears in the che- mical industry, where the input/output ¯ow of materi- als have to be controlled in order to manufacture a product of a desired quality. Chromatography, being a simple and straightforward analytical method, is a popular monitoring tool. In common chromatography the sample is introduced by a single injection and after that a chromatogram is recorded. The variations in time of the process components are reconstructed by using the values of the chromatographic peak areas obtained after a certain time interval. We call this interval the time resolution of the process. The chro- matographic method, however, appears to be slow when the separation time of the components of the injected sample is comparable or larger than the characteristic time of the monitored process. In this case the process course cannot be reconstructed (in the sense of the sampling theorem) using the concentra- tion values obtained at discrete time moments. In most cases the separation time in contemporary gas chro- matography (GC) is (depending on the type of the column) hardly less than 1 min. Processes that could Analytica Chimica Acta 373 (1998) 175±187 *Corresponding author. Tel.: +372-2-536438; fax: +372-2- 536371; e-mail: vmihkel@argus.chemnet.ee 0003-2670/98/$19.00 # 1998 Elsevier Science B.V. All rights reserved. PII S0003-2670(98)00391-2