Available online at www.sciencedirect.com ScienceDirect Procedia Engineering 168 (2016) 197 – 200 30th Eurosensors Conference, EUROSENSORS 2016 Gas composition sensor for natural gas and biogas Arjen Boersmaa*, Jörgen Sweelsena, Huib Bloklandb a TNO, De Rondom 1, 5612AP Eindhoven, The Netherlands TNO, Leeghwaterstraat 44, 2628 CA Delft, The Netherlands b Abstract The calorific value of energetic gasses is an important parameter in the quality assessment of gas steams, and can be calculated from the chemical composition of the gas An array of capacitive sensor electrodes was developed, each functionalized with a gas responsive coating to measure the concentration of methane, ethane, propane, carbon dioxide, water and nitrogen in various gas mixtures Using the data from six functionalized chips, the calorific value could be measured with 5% accuracy Part of this error is caused by the noise in the electronics, another part by the non-linearity of the responsive coatings Future developments aim at improving the response of the coatings and a higher integration of electrodes and will lead to a higher sensitivity and an expected accuracy of 0.5 % The sensor should be implemented in flow equipment, restricting the size to several cubic centimeters © 2016 The Authors Published by Elsevier Ltd This is an open access article under the CC BY-NC-ND license © 2016 The Authors Published by Elsevier Ltd (http://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under responsibility of the organizing committee of the 30th Eurosensors Conference Peer-review under responsibility of the organizing committee of the 30th Eurosensors Conference Keywords: Gas composition sensor; Calorific value sensor; Capacitive comb electrodes Introduction The need for transition towards a sustainable energy supply will result in an increase of the production of biogas A significant part of the biogas will be fed into the distribution grids for natural gas Furthermore gas is imported from different sources all over the world, through pipelines and as Liquid Natural Gas (LNG) Both biogas and LNG have a deviating composition related to the traditional sources in The Netherlands, such as the Dutch Slochteren gas and North Sea gas The current infrastructure dealing with the processing and use of natural gas in The Netherlands is tuned to the gas obtained from the Groningen fields, having a very constant composition and energy content Due to the change in supply, it is expected that the composition will change in the future, thus leading to different energy contents (i.e calorific values) Gas producers, gas network companies and end-users ask for low-cost calorific value * Corresponding author Tel.: 0031 888665713 E-mail address: Arjen.boersma@tno.nl 1877-7058 © 2016 The Authors Published by Elsevier Ltd This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/) Peer-review under responsibility of the organizing committee of the 30th Eurosensors Conference doi:10.1016/j.proeng.2016.11.216 198 Arjen Boersma et al / Procedia Engineering 168 (2016) 197 – 200 sensors, which are currently not available The currently available gas quality measuring systems (e.g GC, Wobbe Index analyzer, etc.) cannot fulfill the need for a cost-effective inline measuring method Alternatives are based on a combination of different physical gas characteristics An example is proposed in the EDGaR project, in which the gas composition is measured using a combination of NDIR, photoacoustics, thermal conductivity and viscometry to obtain a value for the gas composition[1][2] Although this combined approach can measure the composition very accurate, the size and costs of such a system is too high for widespread implementation This paper presents the development of a low cost calorific value sensor for natural gas and biogas, based on the measurement of the composition of the individual components methane, ethane, propane, carbon dioxide and nitrogen This gas sensor is based on gas sensitive coatings on an electronic platform New coating formulations were developed that selectively absorb the target gas and consequently will give rise to a change in material behavior (i.e dielectric constant) This change in material properties is monitored using capacitive comb electrodes Combining the response of multiple sensor chips makes it possible to simultaneously obtain the concentrations of the individual components of the target gas Subsequently the calorific value of the gas mixture can be calculated Approach For the development of a low cost, small sensor, the most logical and promising approach is the use of responsive coatings on a sensing platform The responsive coatings can be based on polymers, metal oxides or semiconductors that change a physical property upon absorption of the target gasses The sensing platform can measure the change in dielectric constant, electrical or thermal resistance, mechanical resonation, optical reflectivity or absorption, etc One approach that is widely used is based on the change in resistivity of a nanoparticle based metal oxide layer at elevated temperatures However, this approach cannot be used for in-line measurements in calorific gas, because there is no oxygen present for surface oxidation, and the related safety issues limit the power and temperature of the working electrodes Subsequently, a capacitive platform was chosen, made from an array of comb electrodes, each of which was coated with a polymer based coating, specifically tuned to one of the target gasses[3][4] For the first prototype six capacitive comb electrodes were used, manufactured by NXP (Fig 1) Each of these electrodes was coated with a thin layer of the responsive coating, by means of drop casting Fig 1: Capacitive chip with a responsive coating (left, courtesy of NXP), SEM image of capacitive comb electrodes (right) These six chips were inserted in a home build gas exposure chamber in which gas composition, temperature and pressure can be controlled The capacitances of the six chips are simultaneously measured using an IviumStat (Ivium Technologies) and a HiMUX XR Multiplexer in order to switch between the six chips The capacitance measurements were done at two frequencies (700 and 8000 Hz) in order to increase the number of data points and include some frequency dependent behavior The noise levels of the capacitive measurements were approx fF for a capacitive change during exposure between 50 and 300 fF This gives a typical signal to noise ratio of 20 The chips were exposed to pure gasses, simple mixtures of the hydrocarbons with nitrogen and more complex mixtures of three or four gasses Each exposure experiment was started using a nitrogen baseline, after which the gas mixture was led into the chamber This resulted in values for each chip for the capacitance at two frequencies before and during exposure So, we obtain an array of 12 measured values (=6x2) For each gas mixture, the actual values for the gas partial pressures can be calculated from the response matrix ([R]) and the chip array ([C]): Arjen Boersma et al / Procedia Engineering 168 (2016) 197 – 200 ܲுସ ܲۍ ې ܴଵǡଵ ǥ ܴଵǡଵଶ ܥଵ ێଶு ۑ ܲ ڭ ڰ ڭ ڭ൩ ൌ (1) ێଷு଼ ۑ ǥ ܴ ܴ ܥ ܲ ێேଶ ۑ ହǡଵ ହǡଵଶ ଵଶ ܲ ۏைଶ ے The response matrix [R] results from the calibration experiments of the sensor array to a series of gas mixtures Results The six responsive coatings that were applied on the capacitive comb electrodes were based on fluoro, silicone and imide polymers, some having porous additives for the capture of the gas molecules The cavity size and porosity was tuned to the chemistry and molecular size of the individual gasses When gas molecules are captured inside a cavity, the dielectric constant increases, giving rise to an increasing capacitance On the other hand some responsive polymers showed a decrease in capacitance upon exposure An example is given in Fig of two coated chips exposed to propane: chip shows a decrease and chip an increase when the propane pressure is increased Fig Example of two chip responses to the exposure to propane Baseline is 100% nitrogen Besides absolute response and accuracy of the calorific gas sensor, a second important parameter is response time When the sensor is used for burner or engine management, the response should be fast The chips that are shown in Fig respond within minute However, the response time of other chips depends on the nature and amount of additives used An example is given in Fig Two chips were coated with methane responsive coatings One had a times higher concentration of the methane capturing cage molecule The absolute response was indeed times higher, but the response time increased from below minutes to 50 minutes Fig Response time of two chips, depending on the concentration of methane capturing additives: the concentration of the red curve is times higher than the concentration of the blue curve 199 200 Arjen Boersma et al / Procedia Engineering 168 (2016) 197 – 200 Sensor performance and conclusions The sensor performance was assessed for a series of 15 mixtures of methane, ethane, propane and nitrogen For each mixture, the gas concentrations were calculated from the measured sensor values and converted to the calorific values It was found that the accuracies in the calculated gas concentrations showed a wide range: standard deviations ranged from mbar for propane to 50 mbar for methane A plot of the measured partial pressure of methane versus the real partial pressure is shown in Fig 4A It can be seen that the error at high methane pressures is significant The reason for this high value is a combination of a large noise in electronic hardware, lower sensitivity of the responsive coating for methane at high pressures and linearization of the response for the matrix calculations Especially the latter has a large influence at higher partial pressures The accuracy of the propane partial pressures is markedly better (Fig 4B) A B Fig Measured methane (A) and propane (B) partial pressure (PP) versus the real values The partial pressures of the gasses in the mixtures are converted to the calorific value (CV) by making use of the energy content of methane (35.7 MJ/m3), ethane (64.0 MJ/m3) and propane (91.0 MJ/m3) The gas mixtures are prepared in a gas exposure box and from the mixing ratios, the real CV’s (@ STP) are calculated and plotted versus the measured values in Fig The values measured by the sensor chip array correspond well with the real values of the gas mixtures that were fed into the gas exposure system The accuracy of the measurements for these 15 gas mixtures is 5% over the whole CV range Since the CV of natural or biogas will be between 20 and 50 MJ/m3, a higher accuracy can be obtain if the sensor array is only calibrated for that CV range Fig Measured calorific value of 15 gas mixtures versus the real values References [1] [2] [3] [4] EDGaR, www.edgar-program.com G de Graaf, F Bakker and R.F Wolffenbuttel, Sensor platform for gas composition measurement, Proc Engin 25 (2011), 1157-1160 D Snelders, A Boersma, A-J de Jong, Gas Sensor Array and Method, 2014, WO2016003272 https://www.tno.nl/en/about-tno/news/2016/4/brightlands-materials-center-gas-sensor-technology/ ... sensor for natural gas and biogas, based on the measurement of the composition of the individual components methane, ethane, propane, carbon dioxide and nitrogen This gas sensor is based on gas. .. (2016) 197 – 200 Sensor performance and conclusions The sensor performance was assessed for a series of 15 mixtures of methane, ethane, propane and nitrogen For each mixture, the gas concentrations... de Graaf, F Bakker and R.F Wolffenbuttel, Sensor platform for gas composition measurement, Proc Engin 25 (2011), 1157-1160 D Snelders, A Boersma, A-J de Jong, Gas Sensor Array and Method, 2014,