Chemical Sensor Accuracy Limits

Chemical sensors detect specific substances and report a concentration, but their accuracy and sensitivity are fundamentally limited by noise, drift, cross-sensitivity, and the physical laws governing the sensing material. This article explains those limits and what you can do to improve a sensor’s real-world performance.

Formula

The sensitivity factor S is a dimensionless correction that accounts for temperature and humidity deviations from a reference condition (25 °C, 50 % RH):

S = 1 / (1 + 0.02·(T − 25) + 0.005·(RH − 50))

where:

• T = ambient temperature in °C
• RH = relative humidity in %
• 0.02 = temperature coefficient (per °C) for a typical metal-oxide gas sensor
• 0.005 = humidity coefficient (per % RH) for the same sensor

The corrected concentration is then C_corrected = C_raw × (1 + 0.02·(T − 25) + 0.005·(RH − 50)).

Worked Example

Using the default values: T = 25 °C, RH = 50 %, C_raw = 100 ppm.

1. Temperature difference: 25 − 25 = 0 °C
2. Humidity difference: 50 − 50 = 0 %
3. Correction term: 0.02×0 + 0.005×0 = 0
4. Sensitivity factor: 1 / (1 + 0) = 1.000
5. Corrected reading: 100 × (1 + 0) = 100 ppm

At the reference conditions, the correction factor is 1 and the reading is unchanged. If the temperature were 35 °C and RH 60 %, the factor would be 1 / (1 + 0.02×10 + 0.005×10) = 1 / 1.25 = 0.800, and the corrected reading would be 100 × 1.25 = 125 ppm.

What Do “Accuracy” and “Sensitivity” Actually Mean?

Accuracy is how close a measured value is to the true concentration. Sensitivity is the change in output per unit change in concentration (e.g., mV per ppm). A sensor can be very sensitive (detects tiny changes) but inaccurate (reads 120 ppm when the true value is 100 ppm). Conversely, a sensor can be accurate at high concentrations but insensitive at low ones. Both are limited by the signal-to-noise ratio, calibration quality, and the sensor’s intrinsic response.

Key Factors That Limit Performance

Noise and Detection Limit

Every sensor has electronic noise (from the readout circuit) and chemical noise (random fluctuations in the sensing material). The limit of detection (LOD) is typically defined as the concentration that produces a signal three times the standard deviation of the noise. Below that, you cannot reliably distinguish the signal from noise. For example, a cheap electrochemical CO sensor might have an LOD of 1 ppm, while a laboratory mass spectrometer can reach parts-per-trillion.

Cross-Sensitivity and Interference

Most sensors respond to more than one gas. A metal-oxide sensor for carbon monoxide will also respond to hydrogen, methane, and ethanol. This cross-sensitivity causes false positives and overestimation. For instance, a breathalyzer using a fuel cell sensor can mistake acetone (from a ketogenic diet) for ethanol, leading to a falsely high reading.

Drift and Aging

Sensor output changes over time due to poisoning (irreversible binding of contaminants), catalyst degradation, or changes in the sensing film. Drift can be as high as 5–10 % per month for some non-dispersive infrared (NDIR) sensors. Regular calibration with known standards is required to compensate.

Temperature and Humidity Effects

Many sensors are temperature-dependent. For example, a pellistor (catalytic bead) sensor’s output changes with ambient temperature. Humidity can also affect readings by condensing on the sensor or altering the chemical reaction rate. The formula above shows how a typical metal-oxide sensor’s sensitivity changes by 2 % per °C and 0.5 % per % RH.

Response Time and Flow Rate

Sensors don’t respond instantly. The time constant (time to reach 63 % of the final value) can be seconds to minutes. If the concentration changes faster than the response time, the reading will lag and be inaccurate. Also, for many sensors, the sample must be delivered at a constant flow rate; variations cause errors.

Calibration and Reference Standards

A sensor can only be as accurate as its calibration. If the calibration gas is off by 5 %, the sensor will be off by 5 %. Calibration drift over time, improper calibration intervals, and using a single-point calibration instead of a multi-point one all limit accuracy.

How to Choose a Sensor for Your Application

Start by defining the required range, detection limit, and acceptable cross-sensitivity. Then match the sensor technology to the target gas:

• Electrochemical – good for low concentrations (ppm) of toxic gases like CO, H₂S, O₂; low power; limited temperature range.
• Non-dispersive infrared (NDIR) – excellent for CO₂ and hydrocarbons; high accuracy, long life, but expensive and requires a clear optical path.
• Metal-oxide (MOX) – cheap, sensitive to many gases, but high cross-sensitivity and high power consumption.
• Catalytic bead (pellistor) – for combustible gases in % LEL; requires oxygen to operate.
• Photoionization detector (PID) – for VOCs; uses UV light; not specific, but very sensitive.

Also consider the environment: temperature, humidity, pressure, and the presence of poisons (e.g., silicones can poison pellistors).

Common Mistakes and Misconceptions

• Ignoring temperature compensation – assuming the sensor reads the same at 0 °C and 40 °C can cause errors of 20–30 %.
• Using the wrong calibration gas – calibrating with methane when you’re measuring propane will give wrong readings.
• Assuming linearity – many sensors are linear only over a limited range. Outside that range, the response is nonlinear and the reading is unreliable.
• Forgetting that “sensitivity” is not “accuracy” – a sensitive sensor can still be inaccurate if not calibrated.
• Not accounting for humidity – in outdoor applications, humidity changes can cause drift that is mistaken for real concentration changes.

Improving Accuracy and Sensitivity

You can improve performance by:

• Using a temperature and humidity sensor to apply corrections (as in the calculator above).
• Calibrating regularly with certified standards, and using multi-point calibration.
• Adding a filter or selective membrane to reduce cross-sensitivity.
• Using a more sensitive measurement technique (e.g., a PID instead of a MOX for VOCs).
• Ensuring a stable flow rate and proper sample conditioning.
• Employing signal averaging and advanced algorithms to reduce noise.

Frequently Asked Questions

What is the difference between accuracy and precision?

Accuracy is how close a measurement is to the true value. Precision is how repeatable the measurement is. A sensor can be precise (same reading every time) but inaccurate (always 10 % high). Both are needed for a trustworthy result.

Why does my sensor read differently on a humid day?

Water vapor can interfere with the chemical reaction on the sensor surface, or change the thermal conductivity of the sample gas. For metal-oxide sensors, humidity typically increases the apparent concentration. Use a humidity correction or a sensor that is less humidity-sensitive, such as NDIR.

How often should I calibrate my sensor?

It depends on the sensor type and the application. For safety sensors, monthly or quarterly calibration is common. For research instruments, calibrate before each experiment. Follow the manufacturer’s recommendation, but also monitor drift – if readings drift more than 10 % between calibrations, calibrate more often.

Can a sensor detect multiple gases at once?

Some sensors are inherently cross-sensitive and will respond to several gases, but they cannot distinguish between them. To measure multiple gases simultaneously, you need an array of sensors with different selectivities (an “electronic nose”) or a gas chromatograph, which separates gases before detection.

What is the lowest concentration a chemical sensor can detect?

It varies widely. Electrochemical sensors can detect down to parts-per-billion (ppb) for some gases. NDIR sensors for CO₂ typically detect down to a few ppm. PIDs can detect VOCs in the ppb range. The theoretical limit is set by the noise level and the sensor’s sensitivity.

Why does my sensor take so long to respond?

Response time is limited by the diffusion of gas into the sensing element and the speed of the chemical reaction. For thick sensing layers or low operating temperatures, response times can be minutes. To speed up response, you can increase the operating temperature (for MOX sensors), use a thinner sensing layer, or increase the flow rate.

Related

• Hide Your Number When Calling: A Simple Guide
• System Partition: Definition and How It Works
• Absolute Relative Mixed Cell References Explained

in t | Temperature | °C | 25 | 0
in rh | Relative humidity | % | 50 | 0
in c | Target gas concentration | ppm | 100 | 0
out sens | Sensitivity factor | - | 1 / (1 + 0.02 * (t - 25) + 0.005 * (rh - 50)) | 3
out corrected | Corrected reading | ppm | c * (1 + 0.02 * (t - 25) + 0.005 * (rh - 50)) | 1