Proper Calibration of Gas Analyzers: "Cold" and "Hot" Models, Monogas and Complex Components

The accuracy of a gas analyzer's readings is not just a matter of having a good instrument. It is a matter of correct and timely calibration. Even the most expensive and precision analyzer will drift over time: the zero point shifts, sensor sensitivity changes, and eventually the display shows figures that cannot be trusted. Calibration brings the instrument back to its certified accuracy.

However, there is no universal recipe here. Calibrating a flue gas analyzer for a boiler house is fundamentally different from setting up a precision NDIR instrument for a scientific laboratory. In this article, we will examine the differences between calibrating "cold" and "hot" analyzers, which gas to use for establishing the zero point, how to correctly select a calibration span gas mixture, what linearization is, and why monogas should be used whenever possible.

"Hot" and "Cold": The Fundamental Difference

The first and most important question to ask before starting any work is: how does your analyzer handle the sample?

"Cold" analyzers operate with a dried sample. Before entering the measurement cell, the gas passes through a sample conditioning system — it is cooled and dehumidified. There is no condensate, no water vapor to interfere with the measurement. The humidity and temperature of the sample do not affect the optical signal, so for such instruments a pure zero gas is critically important — most often high-purity nitrogen. Using ambient air to calibrate the zero point of a "cold" analyzer is strongly discouraged, as even minor impurities or air humidity can introduce unacceptable error into the "zero" from which all subsequent measurements are calculated.

"Hot" analyzers, on the other hand, operate with a wet, hot sample. The entire sampling line, including the measurement cell, is heated to prevent condensate from forming. For them, zero calibration with dried nitrogen is dangerous: a sharp difference in humidity and temperature between the calibration gas and the actual sample will cause significant zero point drift. Therefore, for "hot" analyzers, the zero is calibrated with first-class purity ambient air — not the outdoor air of a metropolis, but specially prepared air that is practically free of dust, oil, moisture, and measurable impurities.

Zero Point Calibration: Why You Can't Just Use Any Air

Calibrating the zero point is the foundation. An error at zero automatically transfers across the entire measurement range. That's why choosing a zero gas is not a mere technical trifle, but a deliberate decision.

• High-purity nitrogen (N₂) — the universal zero gas for most "cold" analyzers (NDIR, chemiluminescent, electrochemical for oxygen). It guarantees the absence of the measured component and minimal matrix influence.

• First-class purity ambient air — used for "hot" analyzers and thermocatalytic sensors. The latter require oxygen to function, so supplying pure nitrogen would cause a deep negative drift. The critical requirement here is that the air must be precisely of first purity class; otherwise, background impurities will distort the zero point.

• Synthetic air — an acceptable alternative in laboratory conditions when first-class ambient air is unavailable, but it is inferior to the latter in terms of temperature and humidity stability.

The key rule: the temperature and humidity of the zero gas must be as close as possible to the conditions under which the analyzer operates with a real sample. Supplying dry cylinder gas during calibration of a sensor that works in a humid environment guarantees zero drift.

Span Calibration: The Secret Lies in the Upper Limit

If the "zero" sets the starting point, then span calibration determines the slope of the entire measurement line. And here there is an iron rule that many people violate.

To level the characteristic across the entire scale, the span point is calibrated using a calibration gas mixture with a concentration of 80–100% of the upper measurement limit (UML). For example, if an analyzer is designed for a range of 0–1000 ppm, the span gas should contain 800–1000 ppm of the target component. Calibrating at the midpoint of the scale (50% of UML) is a common mistake. It ensures accuracy only at that single point, leaving the upper and lower sections of the range unpredictable.

Why does this work? The optical or electrochemical characteristic of a sensor may not be perfectly linear. By setting the slope near the upper limit, we minimize the accumulation of error across the entire scale.

What Is Linearization and Why You Won't Do It in Your Garage

Linearization is the procedure of constructing and correcting a calibration curve using multiple points across the entire measurement range. Unlike a two-point calibration (zero and span), which only corrects offset and slope, linearization corrects the nonlinearity of the optical path.

This procedure is often confused with ordinary calibration, but there is a world of difference between them. Linearization:

• is performed using 6–9 points evenly distributed across the scale (e.g., 0%, 20%, 40%, 60%, 80%, 100% of UML);

• requires precision equipment and the highest accuracy reference gas mixtures;

• creates a table of raw detector values and their corresponding reference concentrations, which is then stored in the instrument's memory.

Linearization is the primary factory adjustment of the optical line. It is at the manufacturer's facility that the calibration curve for each specific instrument is recorded and corrected under ideal laboratory conditions. Re-linearization in field conditions is an exceptionally rare and complex procedure required only after serious intervention in the optical path: replacement of the radiation source, detector, or measurement cell. In all other cases, regular two-point calibration (zero + span) is sufficient; it corrects drift without touching the shape of the curve.

Why Monogas Is Preferable: The Danger of Cross-Sensitivity

One of the most underestimated rules of accurate calibration is the use of monocomponent gas mixtures (monogas). Monogas is a mixture containing only one measurable component in an inert diluent gas (nitrogen or air). Why is this so important? The reason lies in cross-sensitivity — a phenomenon where a gas analyzer reacts not only to the target gas, but also to other components present in the sample.

Cross-sensitivity is particularly characteristic of:

• electrochemical sensors (SO₂ and NO₂, CO and H₂, H₂S and SO₂);

• semiconductor detectors;

• certain optical methods if absorption bands overlap.

If a multicomponent calibration gas mixture containing several gases at once is used for calibration, it is impossible to say for certain which component caused the signal. The result is a systematic error that propagates to all subsequent measurements. This is precisely why in metrologically critical tasks only monogas is applied — one for each channel.

This also explains why "hot" systems are calibrated with first-class ambient air rather than a synthetic multicomponent mixture. Clean ambient air does not contain measured components and introduces no distortions related to cross-sensitivity. Its main constituents (nitrogen, oxygen, argon) form an inert background that does not affect the readings. No extra peaks, no cross-interference — the zero stays true.

Complex Gases: Why NO Requires Patience

Some gases behave capriciously during calibration, and NO is one of the most prominent representatives. Nitric oxide is chemically active, easily oxidizes to NO₂, and tends to adsorb onto the walls of gas lines.

When calibrating an NO channel, several features must be taken into account:

• The material of the sampling tubes must be inert (PTFE, electropolished stainless steel) — otherwise, some NO will simply "stick" to the walls.

• The purging time with the span gas must be significantly longer than for inert gases. While 2–3 minutes may be sufficient for CO or CO₂, for NO the minimum recommended purging time is at least 5 minutes, and ideally up to 10–15 minutes, to achieve full saturation of the gas path.

• After introducing the span gas, a pause must be observed until the readings stabilize. The signal from an NO sensor can "creep" for several minutes due to the slow equilibration between the gas phase and the system walls. Stabilization usually takes around 10–15 minutes, and only after that can the span point be recorded.

An attempt to speed up the process and record the span point without waiting for complete stabilization guarantees a deviation in readings. The instrument will systematically understate the NO concentration in the sample.

Practical Checklist: How Not to Ruin a Calibration

Let's conclude with a set of rules to keep in front of your eyes during every calibration:

1. Identify the type of analyzer — "cold" or "hot". This determines the choice of zero gas and the overall methodology.

2. For "cold" analyzers, use high-purity nitrogen as the zero gas. Do not use ambient air.

3. For "hot" analyzers and thermocatalytic sensors, use first-class purity ambient air. Nitrogen without oxygen will cause drift.

4. The zero gas must be temperature- and humidity-stabilized to match the measurement conditions.

5. For span calibration, use a calibration gas mixture with a concentration of 80–100% of the upper measurement limit.

6. Linearization is not calibration. It is performed at the manufacturer's facility and should not be repeated in operation unless absolutely necessary.

7. Always prefer monogas. Multicomponent mixtures are only acceptable where the absence of cross-sensitivity between components has been precisely proven.

8. When calibrating NO and other complex gases, ensure prolonged purging of the path (up to 10–15 minutes) and wait for complete stabilization of readings.

9. Keep a calibration log — record dates, values before and after, gases used, and cylinder numbers of calibration mixtures.

Proper calibration is not a one-time action but a regular process that ensures the reliability of all subsequent measurements. Neglecting these rules comes at a high cost: fines for exceeding emissions, erroneous technological decisions, and loss of control over production processes.