Share via email
This article investigates the effect of temperature on near-infrared (NIR) absorption spectroscopy and why it is crucial to control it – especially when analyzing liquid samples. These insights will help you to understand how to improve the accuracy and repeatability of NIRS measurements.
Basic introduction to near-infrared spectroscopy
Near-infrared spectroscopy is an analytical method based on the interaction of light and matter. NIR spectrometers measure the absorption of light from the sample in the NIR region at wavelengths between 780 to 2500 nm. Chemical as well as physical and rheological parameters can be detected in both liquids and solids. Results are fast (< 1 minute) and no sample preparation or chemicals are required. Since NIRS is a secondary method, a primary method such as titration must be used to create a prediction model.
Learn more about the basics of NIRS in our blog post
Theory linking vibrational transitions and temperature dependence
The most fundamental model explaining the vibrational behavior of molecules is the harmonic oscillator model (Figure 1) [1,2].
This theory, developed within the framework of quantum mechanics, explains the vibrational energy of molecules or functional groups using the following formula:
E = Energy
n = quantum level
h = Planck constant
ν = frequency
As depicted in Figure 1 and described by the equation above, the harmonic oscillator asserts that only certain discrete energy levels (quantum states n) are allowed. Therefore, the transition between different vibrational states (e.g., n = 0 to n = 1) occurs only when a specific amount of energy (∆E) is available.
∆E = hν
The energy difference ∆E depends on the Planck constant h and the frequency ν, with ν being influenced by the binding strength of the atoms within the molecule or functional group. Since the calculated energy differences fall within the range of infrared (IR) light and near-infrared light, IR and NIR light can induce vibrational transition. Furthermore, the model explains why the resulting absorbance bands can be associated with different functional groups.
Although temperature is not explicitly mentioned in the harmonic oscillator formula, temperature plays an important role because it defines in which energy state the molecules are. The probability of molecules being in a certain energy state level is described by the Boltzmann distribution [3]:
Pn = probability of population of quantum level n
En = energy
kb = Boltzmann constant
T = temperature
Z = partition function
At very low temperatures, molecules predominantly populate the lowest energy state (n = 0). As temperature increases, the probability to occupy higher states (n = 1, 2, 3, ...) increases.
The temperature also influences the movement of the molecules, which in turn affects the width of the spectral bands. Higher temperatures cause a broadening of peaks due to the Doppler effect and increased molecular collision because of a higher mobility of the molecules. The impact of these factors is more pronounced in gases than in liquids, and is least pronounced in solids [4].
Effect of temperature changes on NIR predictions
To investigate the effect of temperature on NIR results, we selected various liquid applications and monitored the change in prediction results for specific temperatures. The analysis was conducted over a temperature range of 26–38 °C.
The sample was measured three times at each temperature to determine the repeatability error of the NIR predictions. All measurements were performed using the OMNIS NIR Analyzer Liquid and OMNIS Software. Standard glass vials with an optical pathlength of 8 mm and a total fill volume of 1 mL were used as sample containers. Temperature control was managed using the built-in functionalities of the OMNIS NIR Analyzer. A representative measurement series is shown in Table 1.
Table 1. Measurement series for a polyol sample. The sample was initially cooled to 25 °C using the OMNIS NIR Analyzer and held at this temperature for 300 seconds. The sample was then heated to the target temperature (e.g., 26 °C) and the measurement was initiated. This procedure was repeated two additional times to obtain three measurements per target temperature.
Qualitatively, the high repeatability of measurements performed at the same temperature is clear as shown by the excellent overlap of spectra (Figure 2a). This is further confirmed by the quantitative analysis of the reproducibility shown in Figure 2b, which indicates a low repeatability error (absolute error = 0.05 mg KOH/g, relative error = 0.20%) as calculated from the repetition measurements.
When comparing the NIR spectra from measurements taken at different temperatures, differences in spectral shape are directly observable (Figure 3a). This change affects NIR prediction results as displayed in Figure 3b, which shows a clear trend toward decreasing values at higher sample temperatures.
The investigation of other applications confirmed the observation that temperature affects the predicted results. Figure 4 illustrates the impact of temperature on the predicted values for hydroxyl value in polyols, moisture content in methoxypropanol, and cetane index and viscosity in diesel. A comparison across all applications reveals that the predicted results change linearly with variations in temperature. This constant absolute change in prediction results per degree of temperature change for each parameter reflects a consistent alteration in spectral shape with changes in sample temperature.
Therefore, neglecting sample temperature control during measurements will affect both the accuracy and reproducibility of NIR predictions. Table 2 displays the changes associated with each degree of temperature variation. Due to the absolute change per degree of temperature, the induced relative error is more significant for samples with lower concentrations.
Table 2. Overview of the absolute and relative change of NIR predictions with each degree of change in sample temperature for different applications. Relative errors induced by temperature changes can be very significant for lower concentrations of the parameter of interest.
Table 3 summarizes the total error of the polyol example with the measured parameter of hydroxyl value, including the repeatability error as well as the temperature-induced error for a deviation of 1 °C or 2 °C. As shown, a deviation of two degrees in temperature already causes a significant error of more than 1%.
Table 3. Overview of the total error (repeatability error and temperature variation error) for a polyol sample with a predicted value of 24.91 mg KOH/g at 26 °C.
How to improve the accuracy and reproducibility of NIR results
Based on these findings, it is highly recommended to use a reliable method for heating and/or cooling samples to their respective target temperatures. ASTM D6122, which provides general implementation guidelines for NIR applications, underscores this need:
- A1.5 Sample Temperature
Sample temperature greatly impacts the reproducibility of spectral measurements due to density changes and intermolecular interactions and may consequently affect predicted values.
One common solution for this when using NIR analyzers is to heat the sample holder to the target temperature and use a defined waiting time after inserting the sample to ensure thermal equilibrium. A challenge with this approach is determining the ideal waiting time to ensure the sample reaches the target temperature while leveraging the speed of NIR analysis. This is particularly challenging because the starting temperature of the sample can be influenced by variations in the laboratory due to seasonal effects (e.g., winter/summer). In many cases, waiting times of 30–60 seconds are used, but experiments show that such short periods are insufficient (Figure 5).
Therefore, a more sophisticated approach is to monitor the sample temperature itself. The OMNIS NIR Analyzer used for these experiments allows for such a procedure thanks to the combination of multiple temperature sensors and a sophisticated algorithm. With the OMNIS NIR Analyzer, temperature-controlled measurements can be defined to evaluate and regulate sample temperature automatically before the measurement begins. This offers multiple advantages:
- No arbitrary waiting time is needed, ensuring the target temperature is reached while maintaining a high analysis speed.
- Temperature fluctuations are minimized in measurements due to seasonal temperature changes in the laboratory environment.
Conclusion
The effect of temperature variation on NIR measurements and its influence on accuracy and repeatability are not always immediately apparent. This is because temperature fluctuations typically occur over extended periods (e.g., seasonal temperature changes in the laboratory) and are not as noticeable during the initial state of developing applications and creating NIR prediction models or libraries.
However, as demonstrated in this measurement series, such fluctuations can significantly affect the accuracy and repeatability of NIR predictions by more than 1% per degree of change in sample temperature and should, therefore, be controlled. Ideally, this should be done with functionalities that allow monitoring of the sample temperature, not just the temperature of the sample holder.
References
[1] Heisenberg, W. Über quantentheoretische Umdeutung kinematischer und mechanischer Beziehungen. Z. Für Phys. 1925, 33 (1), 879–893. DOI:10.1007/BF01328377
[2] Landsberg, Gr. Molekulare Lichtzerstreuung in festen Körpern. I: Lichtzerstreuung im kristallinischen Quarz und ihre Temperaturabhängigkeit. Z. Für Phys. 1927, 43 (9–10), 773–778. DOI:10.1007/BF01397337
[3] Boltzmann, L. Weitere Studien Über Das Wärmegleichgewicht Unter Gasmolekülen. Sitzungsberichte Akad. Wiss. Zu Wien 76, 373–435.
[4] Herzberg, G.; Herzberg, G. Infrared and Raman Spectra of Polyatomic Molecules, 22. print.; Molecular spectra and molecular structure / by Gerhard Herzberg; van Nostrand: New York, 1987.
OMNIS NIRS: An efficiency boost for your laboratory
This White Paper presents the basics and benefits of NIR spectroscopy and discusses applications from the petrochemical, food and beverage, semiconductor, and pharmaceutical industries to demonstrate the unique functionalities of OMNIS NIRS in different situations.
