The goal of this portion of the project is to develop a balloon-based instrument to measure low concentrations (~ 10 ppb) of CO based on the R series CO sensor (RCO1000RH/F) developed by Transducer Technology Inc.

It is an attractive choice because it consumes very small amounts of power and is rugged and portable. The sensor is an electrochemical cell that measures CO concentration by measuring the oxidation current of CO. The challenge is to extend the range of sensitivity from 1 ppm stated in the product literature to the desired 10 ppb and to characterize the sensitivity to temperature and humidity. This will require us to measure current with nA resolution. We must correct for the effect of stray induced currents by proper shielding, mechanical effects from temperature changes, and from interference from SO₂.

The sensor consists of three porous platinum electrodes printed on porous Teflon membranes that sandwich the solid state acid electrolyte. The chemistry of interest (oxidation of CO) occurs at the working electrode. The counter electrode completes the circuit, allowing charge to flow across the cell and the reference electrode is used to maintain a constant interfacial potential difference regardless of the current flowing in the cell. All potential measurements are made with respect of the reference electrode. The output current of the cell is measured at the working electrode.
The reaction occurring in the working electrode is CO + H2O → CO2 + 2H+ + 2e- and the reaction at the counter electrode is O2 + 4H+ + 4e- → 2H2O. We assume that both of these reactions are diffusion controlled, but that the limiting current density for the CO oxidation is more than an order of magnitude smaller than that for the oxygen reduction because of the large differences in concentration. 
 
Current Work

We have performed straightforward calibrations of current vs. CO using a static 100 mV potential and a gas mixer box with zero air and 1000 ppm CO. The measured response was 0.164 μA/ppm CO from (Figure 6). The error in the CO concentration is attributed to the limitations of the mixer box. The source of error in the current is under active investigation.
Figure 7 shows the dependence of current vs. time for zero air and 50 ppm CO. The low frequency drift has amplitude of ~100 nA and a period on the order of 10-100 seconds. We believe that some of the low frequency noise is from stray currents induced by external fields. This can be solved by better shielding. We suspect another source is thermal drift.
Mechanical noise and thermal expansion noise are difficult to quantify and control. The primary mechanism is that the spacing of the electrodes changes thereby changing the already large capacitance of the cell. The mechanical resonant frequency of the sensor is probably well over 10 Hz and can probably be ignored. However, the large thermal expansion coefficients of all the polymer components in the cell can cause significant changes in the spacing of the electrodes. The thermal time constants of the cell are on the order of the period of the low frequency noise. This could explain the presence of the low frequency noise on the zero air curves.

Future Work

We will use electrochemical impedance spectroscopy to determine the appropriate electric circuit analog for this cell. This will enable us to separate out the electrochemical signal from the capacitance and solution resistance. We can then evaluate the sensitivity of the output current to all of the sources mentioned above. This will be invaluable to determine the ultimate resolution of this cell. In addition, we can use this circuit analog to evaluate the effectiveness of pulsed voltage methods to separate the electrochemical component from the extrinsic current fluctuations.
In parallel, we will try to induce in the cell by controlled application of the possible noise sources and control the noise using standard methods for controlling that noise. For instance, we will simulate the effect of electric fields by waving a strong magnet around the leads and around the sensor. Then, we will repeat our measurements using extensive shielding to determine if we can eliminate the currents induced from stray fields. We will also change the temperature of the cell by cooling and heating the gas flowing through the chamber. It may be necessary to dissect the cell and redesign it so that it is less sensitive to temperature fluctuations. In any case, we will determine the dominant noise sources and determine whether the output of this sensor can be processed and controlled to make CO concentration measurements with 10 ppb resolution and ~30 second time constants