Factors Affecting Sensitivity and Stability
Depending on the point of measurements, meteorological factors such as wind speed, pressure,
humidity, and temperature will affect the sensitivity
of a sensor, more precisely the baseline and sensor
gain. This can be corrected quite precisely using
generic data from the manufacturer describing how
the temperature, humidity, etc. affect the sensor
signal. For low-cost sensors, such calibrations can
be more complex as these kinds of sensors are
often used in mobile applications. Given that the
aim is to monitor ambient air concentrations, a lowcost sensor must be able to operate in both cold and
hot weather posing a challenge for manufacturers
of these sensors. Many applications therefore
require continuous calibration, since a given calibration may only work for one location, for a
limited time [46]. Some low-cost sensors must be
re-calibrated every month or even every week. This
is also due to their low stability leading to sensor
aging [47].
Especially when ppb sensitivity is needed and
for conditions with large temperature variations, a
better correction than what the manufacturer
offers may be required. Such a situation can be a
challenge for low-cost sensors, as data processing
may be highly resource demanding. If not done
correctly, monitoring the other variables and
correcting onboard can create a large uncertainty.
The calibration procedure may cause false results,
also as many low-cost sensors are made by
smaller companies where the sensors are not regulated properly [48]. Many users find that manufacturer calibrations will never be sufficient for
ambient air measurements, since there is a substantial difference between laboratory measurements and ambient air measurements with
interference from other gasses.
Improvements to S.S.S
Sensitivity, stability, and selectivity are the key
parameters for sensor performance and for lowcost sensors to become more prevalent, improvements must be made in these areas.
For instance, when metal oxide sensors are
used for gas measurements, filter materials can
be incorporated into the sensors in a number of
ways. This may be done to improve the sensitivity
of gas sensors as for when pre-concentrators are
used. These may take the form of metal-organic
framework (MOF) based filters that can reversibly
adsorb the target gas over a period of time, before
releasing it as a concentrated burst when heated,
enabling the sensor to effectively operate at lower
limits of detection and with increased sensitivity
[49]. As mentioned above, however, there is often
a trade-off; one sensor parameter is improved at
the cost of another. Pre-concentration is an example of improving sensitivity at the cost of a lower
sampling rate. Depending on the application, this
approach is still valuable since loss of performance may not necessarily be detrimental. For
instance, low selectivity is not an issue for sensors
in highly controlled industrial environments, but a
high sampling rate may be key for safety, whereas
for atmospheric measurements of well-mixed
gases, temporal resolution is a lower priority and
the need for sensitivity is high.
As mentioned above, metal oxide sensor
response can degrade over time, for instance,
due to poisoning of the metal oxide or physical
degradation causing a drift in the signal [50, 51]. If
trained personnel are needed to frequently
recalibrate sensors, then they are no longer a
low-cost option. However, the loss of stability
may in some cases be compensated by including
a high efficiency filter in the sensor system. If the
sample flow is periodically passed through a filter,
which removes the target gas, then the zero offset
can be reset, and the sensor effectively
recalibrated. Automatic calibration, without the
need for trained operators, is one key to low-cost
sensors becoming viable as an accurate, costeffective source of data.
To improve the selectivity of metal oxide sensors, an array of sensor nodes can be operated in
combination. As described, metal oxide sensors
operate by measuring the change in resistance of a
metal oxide semiconductor when free-electrons
are removed by oxidizing gases. This means that
oxidizing gasses like O 3 and NO 2 can be difficult
to distinguish. If a filter that selectively removes
one of these gases is included in the sample
stream, then the sensor node’s response to individual pollutants may be distinguished [52]. Alternatively, metal oxide sensors synthesized with
Low-Cost Sensors for Indoor and Outdoor Pollution
431
Depending on the point of measurements, meteorological factors such as wind speed, pressure,
humidity, and temperature will affect the sensitivity
of a sensor, more precisely the baseline and sensor
gain. This can be corrected quite precisely using
generic data from the manufacturer describing how
the temperature, humidity, etc. affect the sensor
signal. For low-cost sensors, such calibrations can
be more complex as these kinds of sensors are
often used in mobile applications. Given that the
aim is to monitor ambient air concentrations, a lowcost sensor must be able to operate in both cold and
hot weather posing a challenge for manufacturers
of these sensors. Many applications therefore
require continuous calibration, since a given calibration may only work for one location, for a
limited time [46]. Some low-cost sensors must be
re-calibrated every month or even every week. This
is also due to their low stability leading to sensor
aging [47].
Especially when ppb sensitivity is needed and
for conditions with large temperature variations, a
better correction than what the manufacturer
offers may be required. Such a situation can be a
challenge for low-cost sensors, as data processing
may be highly resource demanding. If not done
correctly, monitoring the other variables and
correcting onboard can create a large uncertainty.
The calibration procedure may cause false results,
also as many low-cost sensors are made by
smaller companies where the sensors are not regulated properly [48]. Many users find that manufacturer calibrations will never be sufficient for
ambient air measurements, since there is a substantial difference between laboratory measurements and ambient air measurements with
interference from other gasses.
Improvements to S.S.S
Sensitivity, stability, and selectivity are the key
parameters for sensor performance and for lowcost sensors to become more prevalent, improvements must be made in these areas.
For instance, when metal oxide sensors are
used for gas measurements, filter materials can
be incorporated into the sensors in a number of
ways. This may be done to improve the sensitivity
of gas sensors as for when pre-concentrators are
used. These may take the form of metal-organic
framework (MOF) based filters that can reversibly
adsorb the target gas over a period of time, before
releasing it as a concentrated burst when heated,
enabling the sensor to effectively operate at lower
limits of detection and with increased sensitivity
[49]. As mentioned above, however, there is often
a trade-off; one sensor parameter is improved at
the cost of another. Pre-concentration is an example of improving sensitivity at the cost of a lower
sampling rate. Depending on the application, this
approach is still valuable since loss of performance may not necessarily be detrimental. For
instance, low selectivity is not an issue for sensors
in highly controlled industrial environments, but a
high sampling rate may be key for safety, whereas
for atmospheric measurements of well-mixed
gases, temporal resolution is a lower priority and
the need for sensitivity is high.
As mentioned above, metal oxide sensor
response can degrade over time, for instance,
due to poisoning of the metal oxide or physical
degradation causing a drift in the signal [50, 51]. If
trained personnel are needed to frequently
recalibrate sensors, then they are no longer a
low-cost option. However, the loss of stability
may in some cases be compensated by including
a high efficiency filter in the sensor system. If the
sample flow is periodically passed through a filter,
which removes the target gas, then the zero offset
can be reset, and the sensor effectively
recalibrated. Automatic calibration, without the
need for trained operators, is one key to low-cost
sensors becoming viable as an accurate, costeffective source of data.
To improve the selectivity of metal oxide sensors, an array of sensor nodes can be operated in
combination. As described, metal oxide sensors
operate by measuring the change in resistance of a
metal oxide semiconductor when free-electrons
are removed by oxidizing gases. This means that
oxidizing gasses like O 3 and NO 2 can be difficult
to distinguish. If a filter that selectively removes
one of these gases is included in the sample
stream, then the sensor node’s response to individual pollutants may be distinguished [52]. Alternatively, metal oxide sensors synthesized with
Low-Cost Sensors for Indoor and Outdoor Pollution
431