Oxygen transport at cathodes of polymer electrolyte fuel cells (PEFCs) is one of the key factors for determining their power generation performance. To maintain their stable operations, it’s necessary to establish a technique for real-time monitoring of oxygen concentration inside cathode channels of operational cells. In the previous works, the measurements of oxygen distribution within gas channels of operating PEFCs have been conducted using gas chromatograph (GC) [1], oxygen-quenching of fluorescence and phosphorescence [2], and tunable diode laser absorption spectroscopy (TDLAS) [3]. GC techniques provide better accuracy for gas detection in flow channels, however, it takes a long time (at least a few minutes) to collect a gas sample and identify a species. On the other hand, TDLAS enables the non-invasive in-situ monitoring of chemical species inside cells with a high time resolution of 100 milliseconds. Fujii et al. [3] applied the single-ended TDLAS sensor to measure the oxygen concentration in a PEFC channel. In their experiments, since the laser beam is injected perpendicular to the flow direction, the optical path length is extremely short and the sensitivity is insufficient. The detection sensitivity of oxygen is relatively low due to the weak infrared absorption. Therefore, the optical path needs to be lengthened to measure its concentration with high accuracy. In this study, to solve the above-mentioned problem, we developed the fiber-optic transmission TDLAS system and applied it to the quantitative measurement of oxygen concentration in the narrow channel of a simulated cell.Fig. 1 shows the experimental apparatus for measuring the oxygen concentration in a simulated flow cell using the fiber-optic transmission TDLAS system. TDLAS is one of the high sensitive absorption spectroscopy techniques which measure the absorbance of a chemical species and identify its concentration using infrared diode laser. The triangular modulated beam emitted from a DFB laser diode (wavelength: 764 nm) is directly injected into the straight narrow channel (width and depth: 1.0 or 1.5 mm, length: 50 mm) through the single-mode optical fiber. To collimate the laser beam at the channel inlet, a GRIN lens is attached at the top of the pitch fiber. The injected beam passes one or two channels along the flow direction. In the case of passing two channels, the beam transmitted through the first channel is turned 180 degrees by a triangular prism at the bend section (length: 5 mm). The returned beam is passed through the second channel and detected at a photodiode. The output signal is transmitted to a lock-in amplifier, and then the second harmonic (2f) absorption spectrum can be obtained by phase-sensitive detection. In the experiments, the mixed gas of oxygen and nitrogen was supplied to the channel at 100 ml/min. The oxygen concentration in the channels was adjusted to 0 to 100% by controlling each gas flow rate. This measurement was performed at 25 oC and 1 atm.Fig. 2 presents the 2f absorption signals of oxygen in the narrow channel of the simulated flow cell. The incident laser beam passed two straight channels (channel width and depth: 1.5 mm, optical path: 105 mm). The oxygen mole fraction was varied from 0 to 100%. Although the TDLAS measurements were performed inside the narrow channels, the optical fringe noise was hardly observed and the 2f spectra were clearly obtained. It was also noted that the peak-valley height of 2f spectrum increased with an increase in oxygen mole fraction. The oxygen concentration in fuel cell channels can be quantitatively estimated from only the 2f spectrum data using the fiber-optic transmission TDLAS technique.