Objective. This study evaluates a method for calibrating mainstream CO2 analysers in which CO2 partial pressure (PCO2) is calculated as a function of the outputs of CO2 and O2 analysers. Methods. Three mass flow controllers were used to generate 25 different reference mixtures of O2, N2 and CO2. Reference gas mixtures were combinations of PCO2 = 2, 4, 6, 8, 10 kPa and O2 partial pressure (PO2) = 10, 20, 40, 60, 80 kPa (balance N2). CO2 and O2 analyser data were fitted by a calibration equation which took into account the effects of oxygen partial pressure and nonlinearity of the CO2 analyser. The calibration coefficients were tested in a separate validation data set with a variety of combinations of CO2 and O2. Results. Our new calibration method yields a standard deviation of CO2 measurement error that is significantly lower than a CO2-only calibration method in the validation data set (0.54% versus 2.72%, P < 0.05). PCO2 measurement errors produced by the single gas calibration equation are significantly correlated with PO2 in both the calibration (R = - 0.9906, P < 0.05) and validation data sets (R = -0.9642, P < 0.05), but the errors given by our new calibration equation are independent of PO2 (R = -0.0364, NS, and R = -0.0305, NS, for calibration and validation data sets respectively). Calibration with only CO2 cannot eliminate the error related to the collision broadening effect of O2, which in our CO2 analyser is approximately a 1% underestimation of PCO2 for every 10 kPa (75 mmHg) increase in PO2. Conclusions. This study shows that non-dispersive infrared CO2 analyser readings can be substantially affected by background oxygen. This effect can be corrected for by calibrating the CO2 analyser with gases containing known proportions of both CO2 and O2.
- CO analyser
- Collision broadening