Determination of Low Concentration of Multi-Target Gas Species Exhaled with the Breath

Abstract

Introduction We present a direct (on line) reading breath analysis method for determining concentration of multi-target gas species at ppm level or less. Recently a growing awareness of the portable analytical utilities for health care, sports and clinical medicine, has highlighted the need for determining such low concentration levels of exhaled multi-gas species. For example a portable acetone analyzer demonstrated in literature [1] comprises two semiconductor-based highly sensitive sensors. However, the small change in the acetone concentration at less than 1 ppm is difficult to be recognized. This may be partly attributed to the indeterminate concentration of interfering gas such as ethanol. This is the problem and the portable breath analyzer for multi-target gas is not used generally yet. We focus on determining low concentration level of two target gas species in breath, acetone and furthermore carbon monoxide (CO). It is known that the change in the concentration of acetone is associated with a lipid metabolism and CO with a chronic airway inflammation [2]. The multiplexed sensing is a promising way for more detailed inspection of health conditions, considering the progress of the commercially available and reliable gas sensors. The preliminary discrimination between acetone and interfering ethanol is investigated by using an artificial breath. The determining method used in this test is multiple regression analysis [3-5]. We evaluate the effectiveness of the sensors for determining such low level of concentration, and the applicability to CO, the 2’nd target gas. Experimental Two semiconductor-based gas sensors ‘S1’ and ‘S2’ were embedded inside the chamber and the artificial breath containing acetone and ethanol diluted in high humidity air (>80 %RH), was introduced at a constant flowing rate (1,200 ml/min.). Then the change in the electrical conductance of the sensors in accordance with the concentration of acetone (CA) and ethanol (CE) were monitored. The transient time (10 % - 90 % of the saturation) of the sensors’ response at the injection of acetone were observed to be 10 s. This rapid change, which is essential to realize a direct reading breath analyzer, was achieved by making the chamber as small as 40 ml in volume. Though both ‘S1’ and ‘S2’ react to acetone and ethanol, the sensitivity to acetone of the ‘S1’ is higher than that of the ‘S2’. Figure 1 shows the conductance change of the ‘S1’ in response to 0.1 – 10 ppm of acetone (circles), ethanol (squares) and the mixture (triangles). The x-axis represents CA+CE. As the mixture (triangles) has a slightly inclined slope, it is evident that the ‘S1’ was interfered by ethanol. The interference becomes a problem especially in the region of CA<1 ppm. Therefore in order to estimate the acetone concentration, the multiple regression analysis was applied to both ‘S1’ and ‘S2’. As a result of the preliminary discrimination test at the regions of interest, 0.6 ppm, 1.6 ppm and 3.7 ppm of acetone, the estimation error within 20 % were confirmed. This error deteriorates up to 50 % when neglecting the sensor ‘S2’. Conclusions We discriminated the concentration of the target acetone from that of the interfering ethanol in an artificial breath. The estimation error was within 20%. The multiple regression analysis was necessary in determining the concentration of acetone as low as less than 1 ppm. The analytical method might be applicable to the three species problem (acetone, ethanol and CO) in the same manner. We prepared the 3’rd sensor ‘S3’, which sensitivity to CO is higher than that of the ‘S1’ and the ‘S2’. The discrimination test for CO is in progress.