Using Online Near-Infrared Spectroscopy for Real-Time Analysis of Industrial Gases

Near-infrared (NIR) spectroscopy is a rapid, non-destructive method for analysis of products measured routinely that can eliminate lost time and product with real time quality control. This technique is fast and nondestructive and enables real-time analysis of industrial gases. In addition to this, NIR spectroscopy can be utilized for tracking different types of samples such as liquids, solids, slurries, and pastes. It can be used in those applications where real-time analysis or high throughput is required. NIR provides real-time results, which make it a perfect analytical tool for process control. It also offers several benefits such as high accuracy, high precision, high throughput, and is capable of tracking a range of physical and chemical properties in a short period of time.

NIR spectroscopy is used at different stages of production such as online or inline analysis of multiple process streams, feedstock qualification, and qualification of final products. This article explores the use of NIR spectroscopy for real-time and online analysis of industrial gases through a high pressure gas cell. While the emphasis is given more to gases that are studied during the production of ethylene, NIR spectroscopy can also prove suitable for a range of industrial processes. The aim of this study is to assess the viability of NIR spectroscopy for online, real-time analysis of industrial gases, and the results that were obtained showed that this technique is ideal for tracking the gas concentration. The detection limits and precision for gas concentration determination were measured to be about 1% at 1atm. With higher pressures it may be possible to measure concentrations to a higher degree of accuracy within a narrow range.

Background: NIR in Ethylene Production

Ethylene is an organic material that is produced in large volumes at an industrial level. Roughly 150 million tons of ethylene is produced every year. Feedstock such as naptha or light gases like propane and ethane is initially heated to increased temperatures so that the feed is decomposed into tiny hydrocarbon molecules. Once the products have been cooled, they are sent through a range of separation processes with a single product stream among many products of high purity ethylene. One way to optimize the process is to improve the output of ethylene and other similar commercial products. Process analytics like NIR spectroscopy play a major role in such process optimizations providing real-time gas concentrations at different points in the process. Fiber optics enables real-time sequential measurements from up to nine positions such as feed, product streams, recycle streams, and so on using a single instrument, thereby allowing quick adjustments to be made to the process to account for differences in temperature, feed, etc. The overall benefits of integrating NIR into the process is enhanced process reproducibility, better capacity, better product quality, enhanced plant safety, and reduced in-process testing.

Analyzed Gases

The following materials were used in this analysis:

  • Ethylene (C2H4)
  • Carbon dioxide (CO2),
  • Propane (CH3CH2CH3)
  • Acetylene (HCCH)
  • Butane (CH3CH2CH2CH3)

High pressure gas cell and Metrohm NIRS XDS process analyzer with single fiber instruments were used in this study. Figure 1 shows the experimental setup, where the Metrohm NIR process analyzer relays light to the gas cell through a 50m of a single fiber optic.

Figure 1. Metrohm NIRS XDS Process Analyzer (left) with high pressure gas cell ( right).

Gas concentrations were controlled by varying gas flow rates which were measured with flow meters. The light sent to the gas cell is transmitted via the 25cm path length cell filled with ~ 1atm of gas. Light that is not absorbed is resent to the NIR analyzer, which in turn, records the NIR absorption spectrum. Also available are in-line transmission cells with path lengths measuring up to 10cm and rated to 5000psi and 300°C. For on-line analysis, the gas may directly flow within the flow cell. In some cases, adjustment of the temperature, flow rate, and/or pressure would need to be made.


The aim of this experiment is to obtain NIR spectraNIR spectra for the various gases listed in the Analyzed Gases section. It would be possible to monitor the concentrations of each individual gas only if the absorption spectra of each gas is different.

The next goal fo this experiment is to determine the approximate accuracy of the NIR gas concentration prediction and probable detection limits. It should be noted that the NIR absorbance value is directly proportional to the gas concentration. In this analysis, the concentration of specified gases was different at different increments and studied by NIR. This data helped in measuring the lowest specified increment that would provide a clear change in the absorption of gases and ultimately helped in approximating the precision of the concentration prediction and possible limits of detection.

Results and Discussion

Figure 2 shows the NIR spectrum of 1% and 10% CO2 when compared to the empty cell. A perceptible change can be seen in the spectra near 1960nm and 2010nm. In Figure 3, an expansion of this part demonstrates that the difference is visible even at 1% CO2. NIR spectra are typically applied with industry-standard math treatments to minimize the scattering effect and to highlight the target absorption features. The spectra were then applied with the 2nd derivative math treatment. Figure 4 shows the 1900-2100nm region for different concentrations of CO2, indicating that there is a major reproducible change in the NIR spectra even with a difference of 1% CO2. In order to associate the 2nd derivative spectra value to CO2 concentration, a regression technique can be developed. NIR predictions are likely to have a precision and detection limit of below 1% CO2.

Figure 2. NIR spectra of empty gas cell, cell with 1% CO2, and cell with 10% CO2.

Figure 3. Expanded portion of Figure 2 showing absorbance of CO2.

Figure 4. 2nd derivative of NIR spectra for varied amounts of CO2 in gas cell.

Absorption of more gases was also analyzed. Figure 5 shows a similar absorption change for different concentrations of propane, with the strongest absorption taking place at 1695nm.

Figure 5. Select region of 2nd derivative of NIR spectra of varied amounts of propane in gas cell.

Both the detection limit and the precision are analogous to CO2 and anticipated to be roughly 1% at 1atm. It was observed that acetylene displayed a strong absorption near 1528nm. A broader range of gas concentrations were studied in this analysis, differing from 0% to 81.8%. Figure 6 shows the 2nd derivative spectra of various gas concentrations.

Figure 6. Select region of 2nd derivative of NIR spectra of varied amounts of acetylene in gas cell.

In order to associate the acetylene concentration to the spectral changes, a linear regression model was developed. The standard error of calibration (SEC) was found to be 1.8%. The NIR values in contrast to the accurate value for the calibration set are shown in Figure 7.

Figure 7. NIR value (y-axis) compared to tank value (x-axis).

Next, spectra were recorded for more gases including butane and ethylene. The wavelengths showing highly intense absorption for individual gases are listed in Table 1. The spectra of other industrial gases are shown in Figure 8.

Table 1. Wavelengths showing spectral absorption features in NIR region for various industrial gases

Gas 1600-1650nm 1650-1700nm 1700-1750nm
Ethylene 1624 1680 1748
Propane - 1696 1748
Butane - 1698 1752
Acetylene - 1678 1732

Figure 8. NIR spectrum of ethane, ethylene, and propylene.

For individual gases, regression models can be developed and the same can be used to monitor the concentration of individual gases. Selection of appropriate wavelength, or the usage of partial least squares regression, or a combination of both may help in determining precise concentration of gases for mixtures as well.

Factors Affecting Gas Measurements

A number of factors should be taken into account when improving the NIR measurements of gases. Firstly, the absorbance is directly relative to the path length as incited by beers law:


Where ‘A’ stands for absorbance, ‘e’ for molar absorptivity that remains constant for a specified gas, ‘L’ stands for the path length, and ‘c’ refers to the concentration. As the path length increases, the absorbance also increases. The effect of path length on the absorbance of ethylene is shown in Figure 9.

Figure 9. NIR spectrum of ethylene with 0, 1, and 3cm path length.

To ensure optimum results, it is critical to optimize the path length at the proper pressure and temperature, so that the target absorbance is within 2 AU.

Both pressure and temperature are other major parameters that need to be taken into account. Any difference in pressure and temperature can considerably impact the quantified absorbance. According to the deal gas law, the gas concentration is related to pressure and temperature by:

Where ‘c’ stands for concentration, n/V stands for the number of moles/unit volume, ‘R’ refers to the gas constant, ‘P’ is pressure, and ‘T’ is temperature. As the temperature decreases or increases, the absorption and concentration of gas also increases. Likewise, as the pressure decreases or increases, the absorption and concentration also increases or decreases. The effect of pressure on propylene’s NIR absorbance spectra is shown in Figure 10.

Figure 10. NIR spectra of propylene at various pressures.

Furthemore, differences in pressure and temperature may impact the band profile and peak position, promoting a bias in NIR predictions. Hence, under suitable conditions, the temperature and pressure of the gas will continue to be constant throughout the analysis.


This article has shown how NIR spectra were obtained for a wide range of industrial gases at different concentrations. The overall results demonstrate that NIR spectroscopy is suitable for monitoring gas concentrations. It was shown that the NIR spectra of individual gases have less overlap in comparison to the spectra for solids and liquids, with varied spectra for each type of gas. Featuring better specificity, the NIR models are less susceptible to matrix effects. The limits of detection and precision for gas concentration determination were predicted to be approximately 1% at 1atm. At higher pressures, gas concentrations can be determined accurately within a narrow range. For best results, the pressure within the gas cell should be stable and optimized. Using NIR for industrial gas monitoring can help enhance process reproducibility and reduce cycle time. NIR allows for precise and real-time results that can be utilized for process control and optimization, thus enabling enhanced product quality and improved plant safety.

This information has been sourced, reviewed and adapted from materials provided by Metrohm AG.

For more information on this source, please visit Metrohm AG.

Tell Us What You Think

Do you have a review, update or anything you would like to add to this article?

Leave your feedback