Continuous online Fourier transform infrared (FT-IR) spectrometry analysis of hydrogen chloride (HCl), carbon dioxide (CO2), and water (H2O) in nitrogen-rich and ethylene-rich streams

• Published in: Applied spectroscopy Vol. 67; no. 9; p. 1019
• Main Authors: Stephenson, Serena, Pollard, Maria, Boit, Kipchirchir
• Format: Journal Article
• Language: English
• Published: United States 01.09.2013
• ISSN: 1943-3530
• DOI: 10.1366/12-06706

The prevalence of optical spectroscopy techniques being applied to the online analysis of continuous processes has increased in the past couple of decades. The ability to continuously "watch" changing stream compositions as operating conditions change has proven invaluable to pilot and world-scale manufacturing in the chemical and petrochemical industries. Presented here is an application requiring continuous monitoring of parts per million (ppm) by weight levels of hydrogen chloride (HCl), water (H2O), and carbon dioxide (CO2) in two gas-phase streams, one nitrogen-rich and one ethylene-rich. Because ethylene has strong mid-infrared (IR) absorption, building an IR method capable of quantifying HCl, H2O, and CO2 posed some challenges. A long-path (5.11m) Fourier transform infrared (FT-IR) spectrometer was used in the mid-infrared region between 1800 and 5000 cm(-1), with a 1 cm(-1) resolution and a 10 s spectral update time. Sample cell temperature and pressure were controlled and measured to minimize measurement variability. Models using a modified classical least squares method were developed and validated first in the laboratory and then using the process stream. Analytical models and process sampling conditions were adjusted to minimize interference of ethylene in the ethylene-rich stream. The predictive capabilities of the measurements were ±0.5 ppm for CO2 in either stream; ±1.1 and ±1.3 ppm for H2O in the nitrogen-rich and ethylene-rich streams, respectively; and ±1.0 and ±2.4 ppm for HCl in the nitrogen-rich and ethylene-rich streams, respectively. Continuous operation of the instrument in the process stream was demonstrated using an automated stream switching sample system set to 10 min intervals. Response time for all components of interest was sufficient to acquire representative stream composition data. This setup provides useful insight into the process for troubleshooting and optimizing plant operating conditions.