What is the use and limitations of low-cost electrochemical sensors in air quality measurements?

Technical Note No. 066

Use and Limitations of Low-Cost Electrochemical Sensors in Air Quality 
Measurements

Date: 12 December 2024

Author: Andrew Turnipseed

Summary
This Technical Note provides guidance for using electrochemical sensors when applied to air 
quality monitoring as well as understanding their limitations. This Tech Note applies to the 
following 2B Tech instruments that use these sensors: AQLite (standard) Air Monitoring 
Packages, AQSync Air Quality Monitoring Stations that were custom ordered to use a sensor to 
measure carbon monoxide or other species, and Personal Air Monitors (PAM) previously 
offered by 2B Tech.

Tools/Materials Needed

none

Introduction
Over the past few years there has been an enormous surge in interest in the application of low-cost sensors to conduct measurements of air pollutants by educators, citizen scientists, and 
groups interested in air pollution levels within their own communities. Indeed, it is now 
recognized that sensors can fill important gaps that are virtually impossible to fill with 
conventional or even miniaturized instruments because of their low cost, small size, and ease of 
deployment. However, the limitations of sensors must be recognized and care must be taken to 
obtain meaningful results. 2B Tech incorporates low-cost electrochemical (EC) sensors in its 
AQLite and AQSync measurement platforms, and in its previously offered Personal Air Monitor 
(PAM). The EC sensors used by 2B Tech are manufactured by Alphasense – a leader in EC 
sensor technology. 

When using EC sensors for measuring ambient level pollutants one must keep in mind that 
these sensors were originally developed for industrial health and safety applications and were
primarily used to detect relatively high concentrations (> 1 to 100 ppm). Applying these sensors 
to the lower pollutant concentrations (~ 10’s of ppb) typical of ambient air can be challenging. 
At 2B Tech, we routinely incorporate EC sensors for measuring carbon monoxide (CO);
primarily because background CO concentrations tend to be ≥ 0.1 ppm, with urban levels 
extending upwards to several ppm. Other pollutants that can be measured by EC sensors (e.g, 
NO2, O3, SO2, H2S) tend to have ambient concentrations < 0.1 ppm. At these levels, effects of 
environmental changes, sensor noise and cross-sensitivities to other species are magnified. 
Therefore, we do not include these sensors in our measurement packages without prior 
discussion of a customer’s specific application.

In this Tech Note, we give some practical recommendations for using EC CO sensors for 
ambient air quality measurements. Although our recommendations are focused on CO sensors 
here, we will then mention how these ideas extend to other EC sensors. Our Tech Note 065
focuses on low-cost particulate matter (PM) sensors.

Recommendations for EC CO Sensors
(1) Field Intercomparisons with reference instruments are necessary for best accuracy. At 
2B Tech, we derive calibration coefficients (a zero offset and span) for EC CO sensors in 
the lab using gas standards. We have typically found that CO sensors calibrated in this 
way provide good relative measurements (i.e., they track CO concentrations well over 
time). However, significant zero offsets have been observed once deployed in the field 
(± 0.4 ppm or more). Comparing with nearby measurements from reference-grade 
instrumentation tends to be the most effective means of deriving correct calibration 
coefficients applicable to ambient conditions. This is supported by recent reports by the 
World Meteorological Organization (Lewis et al., 2018; Peltier, 2020). They conclude that
laboratory-based EC calibrations are not always valid under ambient conditions and that
intercomparisons in ambient air with established reference techniques provides the most 
reliable method of calibrating measurements from EC sensors. 

(2) Sensitivity degradation. CO sensor sensitivity slowly degrades over time – partially due 
contamination of the porous membrane that allows diffusion of pollutants into the EC 
sensor. Particulates are not filtered in the PAM and AQLite sample air, therefore, 
contamination is more rapid, and EC sensors should be either replaced or re-evaluated 
every 4-6 months. The air is filtered in the AQSync, thus, degradation is slower and EC 
sensors can be used for longer periods – up to a year of continuous use. CO sensors 
can be easily swapped (contact 2B for procedure); however, the original lab-derived 
calibration coefficients are likely no longer valid and should be re-evaluated.

(3) Temperature affects both the sensitivity (or span) and the zero offset (signal at CO = 0). 
Specification Sheets provided by Alphasense indicate the sensor sensitivity roughly 
changes about 1% per °C (from 0 and 30 °C) and is moderately consistent across sensors (Figure 1). Temperature impacts on sensor offsets are generally nonlinear, exhibit more sensor-to-sensor 
variability, and have a larger impact attemperatures > 25-30 °C (see Figure 2). In general, the 
CO offset becomes increasingly more negative at high temperatures. Note that with a typical sensor sensitivity (~350 nA/ppm), an offset of -100 nA is equivalent to -0.35 ppm. The large
sensor-to-sensor variability makes it difficult to apply any simple temperature correction for the 
offset. However, temperature changes below about 25 °C tend to have small impacts on measured CO concentrations.

(4) Humidity effects on EC CO sensors are often transient (see Alphasense Tech Note AAN 110) and observed during rapid humidity changes. This effect is very difficult to both elucidate or correct. However, ambient humidity typically changes slowly enough that the effects on CO measurements are fairly small (< ± 0.2 ppm), such that 2B Tech does not currently recommend any method for correcting humidity effects.

Electrochemical Sensors for Other Pollutants
EC sensors are also available that target many other pollutants such as NO2, SO2, and H2S. In 
general, the same recommendations given above apply to these sensors as well; however, 
these sensors should be used with more caution when applying them to measurements in 
ambient air. The EC sensors that target these species have similar absolute sensitivities as the 
CO sensor, and the impacts by temperature and humidity are of similar magnitude, but these 
pollutants are generally observed at much lower concentrations than CO. This essentially 
magnifies the effects of temperature and humidity. For example, a temperature change 
between 20 to 30 °C can easily alter the offset output of an EC sensor by ± 10 nA (see Figure 
2). Since both CO and NO2 sensors have approximately similar absolute sensitivities (~ 350 
nA/ppm), a 10 nA signal change is equivalent to ~ 30 ppb (10 nA/350 nA/ppm = 0.029 ppm = 29 ppb). This is a relatively small amount compared to typical CO concentrations, but not for NO2.
In fact, ambient levels of NO2 are often only about 30 ppb and rarely extend above 100 ppb. 
Thus, for EC NO2 measurements, it becomes critical to compensate even for moderate changes 
in environmental variables. These corrections can be substantial and difficult to apply uniformly 
across multiple sensors.

A further concern for some EC sensors is that they respond to other pollutants in addition to the 
intended target species (referred to as a “cross sensitivity”). For example, the SO2 sensors exhibit significant negative responses to both NO2 and O3. Therefore, measurements of NO2 and O3 (each with its own associated measurement uncertainties) are typically required to correct the observed SO2 sensor response. These corrections can be relatively large, especially if the interfering species is at a much higher concentration than the target pollutant. In any case, these corrections always increase the uncertainty of the target pollutant measurement. 

Overall, when using any EC sensors for ambient air quality measurements, it is always critically 
important to understand the typical concentration levels of the target pollutant one wants to 
measure compared to the limitations of the specific sensor. Limitations in EC sensors arise 
from the sensor responses to temperature and humidity changes as well as cross sensitivities to 
other species that could be present for a given application. A general rule of thumb would be 
that concentration variations below about ± 20 are difficult to accurately discern when using EC 
sensors. Furthermore, it appears that laboratory-derived calibrations and correction algorithms 
(i.e., for temperature or cross sensitivities) are not always applicable once the EC sensor is 
deployed measuring ambient air. The reasons for this are not fully understood, but it seems 
clear that measurement comparisons to more established instrumental techniques in ambient 
air are required to reduce EC sensor uncertainties to levels where meaningful pollutant 
concentrations can be obtained.

References
Lewis, Alastair, Peltier, W. Richard and von Schneidemesser, Erika (2018) Low-cost sensors for 
the measurement of atmospheric composition: overview of topic and future applications. 
Research Report No.1215, World Meteorological Organization (WMO), Geneva, Switzerland.
https://eprints.whiterose.ac.uk/135994/

Peltier, Richard W. (editor) (2020), An update on low-cost sensors for the measurement of 
atmospheric composition, Research Report No. 1215, World Meteorological Organization 
(WMO), Geneva, Switzerland. https://library.wmo.int/records/item/37465-an-update-on-low-costsensors-for-the-measurement-of-atmospheric-composition#.YL3zF0w8y7