Fuel cell sensors have now been in use worldwide for about 25 years for testing the breath of suspected drunk-drivers.

For most of that time the application was simply for the "screening" test, usually roadside. Those failing this test would then receive the "evidential" test at a police station, using a more established analytical procedure: originally this was blood analysis by gas chromatography. The cost and delay inherent in calling a doctor to administer the blood test led to the gradual introduction of evidential breath testing, and the leading technology was infrared absorption. Fuel cell sensors, although showing far better specificity towards alcohol among the constituents of breath, failed originally to reach the required standards of speed and accuracy. Continuing development has addressed these problems, and courts of law are increasingly relying on a fuel cell to provide the evidence to convict a drunk-driver. The world's leading manufacturer of breath alcohol instrumentation, Intoximeters Inc. of St. Louis, Missouri, have a wide product range in which the fuel cell plays the analytical role.

Breath alcohol instrumentation using the fuel cell has almost always taken a snap sample of about 1cm³ breath near to the end of an exhalation. Originally, to save time, the peak current developed was used as a measure of alcohol content, this measurement taking typically a minute. The residual signal was burned off afterwards by short-circuiting the sensor. This method suffered from the great disadvantage that over a sequence of such measurements the response profile would change, leading to a steady reduction in sensitivity. The original sensitivity would be regained after "resting" the sensor for up to 24 hours.

Attention to the fuel cell electrocatalyst has steadily improved performance characteristics such as speed, accuracy and calibration stability both short and long term. The course of development can be illustrated by looking at the time taken for an ethanol fuel cell to discharge 95% of a typical sample.

	Year	Time
	1978	5 minutes
	1983	2 minutes
	1988	1 minute
	1993	15 seconds
	1998	6 seconds*

[*The latest evolution would actually take about 3 seconds, but it has a membrane overlay to damp the response spike]

This gradual improvement in sensor speed brought an additional benefit: the original technique of measuring the peak current as a quicker but approximate measurement could now be abandoned and the electrical output integrated, rendering the analysis largely immune to residual changes in the response profile, giving exceptional stability of calibration. The Intoximeter EC/IR instrument in UK police installations must maintain its calibration between six monthly service intervals, which it does without difficulty: in fact 2 years is typical.

As was pointed out earlier, the fuel cell sensor is highly specific to ethanol among the possible constituents of breath. Very few other substances likely to appear in human breath will be detectable, but those which do will alter the shape of the response profile. If the response profile is found to differ from that of an uncontaminated ethanol sample, the test is aborted, thus reducing the chance of a miscarriage of justice.

The effectiveness of the fuel cell sensor is now well established in the field of evidential breath alcohol analysis, and in recent years the principle has been applied to measuring many other gases such as carbon monoxide for various monitoring purposes such as personal exposure, emissions, and environmental monitoring.