Gasoline properties determination with phononic crystal cavity sensor

Gasoline properties determination with phononic crystal cavity sensor

Sensors and Actuators B 189 (2013) 208–212 Contents lists available at ScienceDirect Sensors and Actuators B: Chemical journal homepage: www.elsevie...

882KB Sizes 5 Downloads 383 Views

Sensors and Actuators B 189 (2013) 208–212

Contents lists available at ScienceDirect

Sensors and Actuators B: Chemical journal homepage:

Gasoline properties determination with phononic crystal cavity sensor A. Oseev, M. Zubtsov, R. Lucklum ∗ Institute of Micro and Sensor Systems (IMOS), Otto-von-Guericke-University Magdeburg, Germany

a r t i c l e

i n f o

Article history: Available online 8 April 2013 Keywords: Phononic crystal sensor Gasoline sensor Octane number RON Ethanol ratio

a b s t r a c t In order to optimize the combustion process, engine performance and meeting the EPA emission standards, information should be gained in real time about properties of gasoline, which is supplied nowadays in a wide range with different compositions. This study presents a new sensor platform based on a phononic crystal (PnC) sensor comprising a cavity and its application as in-line real-time measuring system to determine gasoline properties. The method is based on the analysis of the transmission spectrum of a phononic crystal sensor filled with the liquid gasoline blend. We could reveal a strong correlation between gasoline properties and the frequency of maximum transmission. Obtained experimental results show that the phononic crystal sensors can be considered as a prospective, competitive and inexpensive device specifically for ethanol in gasoline detection and distinguishing fuels with different octane numbers. © 2013 Elsevier B.V. All rights reserved.

1. Introduction A number of gasoline parameters are crucial for automobile engine compatibility. Octane number, for systematic investigations one typically uses the research octane number (RON), is one of the key parameters for normal operation of automobile engines and generally defined by fuel composition. The typical composition of gasoline hydrocarbons is regulated in certain documents [1–3]. At the same time, petrol industry is currently on a way to massively use ethanol as a gasoline additive. The motor fuel content of ethanol in gasoline is defined as 5% (E5) or 10% (E10) by volume at present time. Therefore, in order to secure engine reliability in terms of material compatibility with ethanol-blended gasoline, its real-time control becomes extremely important. There is a number of well-established methods for gasoline analysis [4] which are not applicable under field conditions. Other methods that can be utilized for real time monitoring which apply determination of dielectric constant or density of the fuel [5] can provide sufficient accuracy only in combination with each other. Here we introduce a new prospective sensor platform based on a phononic crystal. Phononic crystals, the acoustic version of a band gap material with artificial properties, have recently been introduced as a new platform for liquid sensing purposes [6–8]. It has been shown that they respond to speed of sound of a liquid mixture. This material property may vary in a distinct manner when changing the composition of the liquid mixture. For sensor applications it is

∗ Corresponding author. Tel.: +49 391 671 8310; fax: +49 391 671 2609. E-mail address: [email protected] (R. Lucklum). 0925-4005/$ – see front matter © 2013 Elsevier B.V. All rights reserved.

advantageous introducing a defect into an otherwise regular structure to create a defect mode. In [7] we have realized a slit cavity for that purpose. The great advantage of applying a defect is the confinement of acoustic energy in the defect at an appropriate frequency. The cavity mode appears as a sharp transmission peak in the transmission spectrum of the phononic crystal. Change in speed of sound of the liquid in the cavity moves the resonance frequency of the cavity mode which in first approximation is jointly determined by the width of the slit cavity and speed of sound of the material therein. The sensor concept primarily relies on the capability to determine the frequency of maximum transmission, shortly called the peak frequency, with high accuracy. One major factor limiting the accuracy is the peak half bandwidth: The sharper the peak, the better the resolution. However, the transmission peak must also be well separated from other transmission features, elevated to the background and survive under the broad environmental conditions a car is used. We here exploit a sensor featuring two 4-row regular phononic crystals separated by a slit (Fig. 1a). The frequency of the cavity mode has been designed to lie close to the upper band gap of the phononic crystal. The band gap of the regular structure in air, i.e., 8 rows of holes without slit, extends from 590 kHz to 944 kHz taking the −3 dB-level [7]. Since speed of sound is usually a less common value in liquid property determination, the value of interest must secondly have a beneficial relation to speed of sound. Note, that the variation of speed of sound can be much larger than the variation of other liquid properties like the refractive index as used for many optical sensors including photonic crystal sensors. The sensitivity of the phononic crystal sensor with respect to speed of sound is defined by the design of the phononic crystal, the sensitivity of the sensor to the value of interest is a material-dependent issue and can vary by

A. Oseev et al. / Sensors and Actuators B 189 (2013) 208–212


Fig. 1. Pictures of the phononic crystal (a), the experimental setup (b) and the coupling to the ultrasonic transducers (c).

a large amount. Here we specifically analyze gasoline blends with different composition and investigate how ethanol presence in a gasoline blend is reflected in the overall sensor response. Sensing of gasoline has to consider severe safety regulation. Since the phononic crystal sensor device can be separated via (long) waveguides from piezoelectric transducers, which generate and detect sound and which have the electrical contacts, the application of the phononic crystal sensor in explosive environment is much more relaxed and does not require expensive protection measures. Furthermore, a defect structure like a slit can be employed as a component of a fluidic system and measurement can be performed in a (micro)fluidic measurement cell meeting safety requirements. Phononic crystal structures can hence be designed for a great diversity of complete sensor systems with fluidic components. Dimensions can be scaled giving access also to applications allowing also for small analyte volumes. In this sense the gasoline sensor may be understood as a phononic crystal sensor demonstrator as well.

2. Experimental 2.1. Measurement setup The 2-dimensional phononic crystal sensor in Fig. 1a that has been exploited in experiments consists of a steel plate with holes in square arrays and a slit cavity in the center of the structure. The lattice constant is 3.0 mm, the thickness of the plate is 15 mm, the diameter of the hole is 1.8 mm and the width of the cavity is 1.5 mm. These geometric values are in close correlation to the probing ultrasonic frequency range which, for technical reasons, has been set around 1 MHz. The phononic crystal sensor is only acoustically coupled to the rest of the sensor system, including the electrical circuit. The PnC sensor is part of a fluidic system as well. A number of injectors have been used for filling the whole structure (holes and slit cavity) with the liquid to be analyzed as shown in Fig. 1b.

Clamp-on contact piezoelectric transducers Panametrics V103RB having a central frequency of 1.0 MHz have been reversibly coupled to PnC sensor with a coupling fluid, Fig 1c. No matching circuits have been applied. Longitudinal sound waves contributing to transmission propagate parallel to slit cavity, side lobes do not disturb the measurement. A network analyser (Agilent 4395A) together with an S-parameter test set (Agilent 87511A) (100 kHz–500 MHz) has been applied for electrical determination of the transmission properties of the setup (Fig. 1d). The transmission amplitude has always been normalized by the amplitude of equivalent setup without the sensor in place. For further details see [7]. 2.2. Materials According to requirements of typical gasoline composition, a number of gasoline blends have been prepared with components given in Table 1 (Roth and Sigma–Aldrich). Table 1 also provides the acoustically most relevant properties. Gasoline blends have been obtained by mixing alkane components (n-heptane, nhexane), isoalkane component (isooctane), cycloalkane component (cyclopentane) with aromatic additive (ethylbenzene) and octane number enhancer (ethanol). Ethanol has been added in proportion of 10% by volume for regularly blended gasoline mixtures.

Table 1 Gasoline components and their properties at 20 ◦ C [16–18]. Component

Speed of sound (m/s)

Density (mg/ml)

Viscosity (mPa*s)

n-Heptane n-Hexane Isooctane (2,2,4-trimethylpentane) Cyclopentane Ethanol Ethylbenzene

1162 1083 1111 1182 1162 1338

626 649 692 751 789 866.5

0.387 0.300 0.473 0.413 1.074 0.631


A. Oseev et al. / Sensors and Actuators B 189 (2013) 208–212

Table 2 Composition of gasoline blends E10 with different octane numbers; concentration in vol% [6]. Component/octane number





n-Heptane n-Hexane Isooctane (2,2,4-trimethylpentane) Cyclopentane Ethanol Ethylbenzene

5.5 23.6 52.7 8.2 10.0 0

3.9 17.1 42.1 6.6 10.0 20.3

3.4 14.9 36.7 5.7 10.0 29.3

3.1 13.5 33.1 5.3 10.0 35.0

The gasoline blend with octane number 98 has been prepared with maximal percentage of aromatic additive (35%). Gasoline blends with octane number 95, 91 and 80 have been obtained with the same composition as 98 octane number blend by reducing the percentage of aromatic additives. The respective composition of the gasoline blends is shown in Table 2. Finally a number of gasoline blends overloaded by ethanol with different concentration has been proposed for experiments in order to show the ability distinguishing between regular gasoline and gasoline blends boosted with ethanol. Octane numbers RON91, RON95 and RON98 blends starting initially from 80 RON mixtures were obtained by adding ethanol in concentrations that exceed 10%. The respective values are given in Table 3.

Fig. 2. Transmission spectrum through phononic crystal sensor filled with different RON gasoline (E10). (For interpretation of the references to color in the text, the reader is referred to the web version of the article.)

and alcohols can be neglected here as long as the ethanol concentration is constant.

vmix =


xi ∗ vi



3. Experimental results 3.1. Octane number determination Since standard concentration of ethanol in Europe has been assessed to 10% (E10), we have measured standard composition blends with RON80, RON91, RON95 and RON98 with this ethanol concentration. Transmission spectra of the phononic crystal sensor filled with different octane number gasoline blends are depicted in Fig. 2. All transmission spectra show well distinguishable transmission peaks with amplitude considerably above the noise floor. Peak frequency separation is larger than peak half band width of about 4–10 kHz. The dependence of the transmission peak frequency on the octane number of gasoline is evident. It correlates approximately linear with octane number of analyzed gasoline blend as shown in Fig. 3. We have taken a value of 2 kHz as safe estimate for peak frequency resolution achievable under real conditions. We have supposed that the frequency of maximum transmission is attributed to the frequency of the longitudinal cavity mode confined in a slit defect of sensor structure. Indeed, the maxima of transmission for each blend approximately correspond to 3/2 of wave length, , in each gasoline blend i.e., to the third order cavity mode. This estimation assumes that speed of sound of the gasoline blend is proportional to sum of speed of sound of its components weighted by the percentage of composition (1) [9]. Non-linearity of speed of sound in liquid mixture as known from e.g., hydrocarbons

Here vmix is the resulting speed of sound of gasoline mixture, xi is volume ratio of mixture component and vi is its speed of sound. The relation between frequency of maximum transmission and speed of sound of gasoline blend constituents can be described as follows: fmax =

vmix (2/3) ∗ wc




x i=1 i

2 ∗ wc

∗ vi


where wc is a cavity width. We could not observe the second order cavity mode. It should appear within the band gap. This finding agrees with [7] where

Table 3 Composition of gasoline blends overloaded by ethanol octane number enhancer; concentration in vol%. Component/octane number





n-Heptane n-Hexane Isooctane (2,2,4-trimethylpentane) Cyclopentane Ethanol

5.5 23.6 52.7 8.2 10

3.4 14.9 41.1 5.7 34.9

2.8 12.0 33.3 4.6 47.2

2.3 9.9 27.5 3.8 56.5

Fig. 3. Reduced frequency shift for gasoline blends with standard composition including 10% ethanol concentration (blue dashed line ( )) and mixtures overloaded by ethanol (solid line, green ( )). Two single dots (red ( )) show the response of gasoline blends without ethanol as a reference. The reduced frequency has been obtained with respect to the transmission peak frequency of gasoline RON80 without ethanol. (For interpretation of the references to color in the text, the reader is referred to the web version of the article.)

A. Oseev et al. / Sensors and Actuators B 189 (2013) 208–212

Fig. 4. Reduced frequency shift for 98 octane number gasoline blends with different ethanol concentration. The reduced frequency has been obtained with respect to the transmission peak frequency of ethanol-free gasoline RON98. (For interpretation of the references to color in the text, the reader is referred to the web version of the article.)

we have noticed that the amplitude of the transmission peak corresponding to the second order cavity mode decreases while the respective transmission peak moves further into the band gap region. Enhanced liquid viscosity at high octane ratio gasoline blends due to the increasing content of high absorptive aromatic components is not a significant reason for the disappearance of the second order cavity mode since it can be found in simulation with idealized conditions as well. It is rather a consequence of multiple reflection and interference in the phononic crystal which causes phase mismatch or impedance mismatch at the phononic subcrystal–cavity interface. In other words, the geometric solid–liquid interfaces of the slit cavity defining the cavity width, wc , do not act like an interface to a semi-infinite solid as assumed in case of the second order cavity mode n = 2 wc /n with n = 2. One may understand our phononic crystal slit cavity sensor as a series arrangement of 3 band-pass filters, the two 4-row phononic semi-crystals and the slit cavity, where transmission is only possible in the overlap region of the filters. The respective first order cavity mode transmission peak lies below 600 kHz in the first pass band of the structure and is hardly to identify. 3.2. Ethanol in gasoline In order to demonstrate distinguishability between ordinary composed gasoline (E5) and ethanol blended fuels (E10) we have studied the sensor response for RON98 gasoline blend with ethanol in the concentration range between 0% and 12.5%. The results are depicted in Fig. 4. The reduced frequency, here taken as the ratio of the transmission peak frequencies and the peak frequency of ethanol-free RON98 blend, decreases non-linearly with increasing ethanol concentration. Significant changes in frequency shift were determined in a range of low ethanol concentrations up to 5% by volume whereas the slope tends to zero between 10% and 12.5% ethanol concentration. The region 0–12.5% of ethanol concentration for RON98 initial gasoline blend can be empirically described with relation to ethanol volume ratio as follows: fmax = 1.206 + 0.018 ∗ exp





Here fmax is frequency of maximum transmission in MHz and xeth is a volume ratio of ethanol in RON98 gasoline blend.


Non-linear frequency response can be explained from consideration of properties of binary mixtures of ethanol with selected gasoline components separately. Binary mixtures of hydrocarbons with ethanol have been studied extensively in recent years. Ethanol is well-known for its tendency for molecule association through hydrogen bonds into complex aggregates and clusters with other liquids. Thermodynamic properties of alcohols with non-polar solvents have been studied and a number of interpretations in terms of association of alcohol molecules have been provided [10]. It was shown that enthalpy of dilution of ethanol in non-polar solvents such as n-hexane and cyclohexane demonstrate strong non-linearity in a range of low ethanol concentrations. Thermodynamic excess properties for ethanol-n-heptane mixtures were investigated in [11] and it was shown that thermodynamic parameters such as heat capacity, enthalpy, excess entropy and Gibbs function show nonlinear response to ethanol concentration. Volumetric properties of binary mixtures of ethanol–hydrocarbons mixtures have been investigated in [12,13] and it was found that ethanol mixtures with non-polar solvents have non-linear isentropic compressibility and excess molar volume dependences from ethanol concentration with considerably rapid changes in a range of low ethanol concentrations. Isentropic compressibility and speed of sound are correlated, i.e., nonlinear behavior of thermodynamic characteristics and excess properties of ethanol–hydrocarbons binary mixtures are reflected in speed of sound of mixtures thereof [14]. Correlation between association constants and sound velocity for binary liquid mixtures has been determined within certain models [15]. Speed of sound dependences for binary mixtures of ethanol and hydrocarbons have been calculated with consideration of association and have been confirmed by experimentally obtained data. From this point we can assume that the nonlinearity of the frequency response obtained for ethanol with RON98 gasoline blend has thermodynamic origin which directly affects volumetric liquid properties and is reflected in speed of sound for certain gasoline blends. This is finally the key for the ability of the sensor to distinguish between E5 and E10 gasoline. In practice the sensor can be used for signaling the presence of ethanol above 5% in order to ensure safety of vehicles not approved for E10. The minimum in the ethanol sensitivity around 10% is the necessary feature that allows octane number determination of E10 gasoline as reported in the previous chapter.

3.3. Boosting gasoline Ethanol as a high octane ratio liquid can be applied as an octane number enhancer in order to obtain high RON gasoline from initially low octane ratio fuels. Such non-regular content of ethanol exceeding regularly defined values can cause significant damage in the automobile engine. The analysis of both regularly blended and ethanol boosted blends shows evident difference in the transmission spectrum. Fig. 3 shows the relation of a reduced frequency shift of transmission maximum vs. octane number of blends for standard composition as well as compositions boosted by ethanol. The transmission peak frequencies determined for RON80 without ethanol have been taken as reference. The dashed line ( ) shows experimental data obtained from regular mixtures of gasoline components including 10% ethanol. The solid line ( ) summarizes the data extracted from measurements with mixtures overloaded by ethanol. The reduced frequency plot of regular E10 gasoline lies considerably above the boosted mixtures. Increasing of octane number of gasoline blend by overloading initial RON80 mixture with ethanol affects the frequency change considerably less in comparison to regular gasoline that makes it possible to distinguish regularly blended gasoline from boosted ones.


A. Oseev et al. / Sensors and Actuators B 189 (2013) 208–212

The increase in frequency of regularly blended gasoline is explained by enlarging of content of high speed of sound aromatic components which finally increase speed of sound of the whole blend. In case of overloading initially low octane ratio blend with ethanol the mixture with the other gasoline components shows a reduced speed of sound in a range of low concentrations, despite the fact that ethanol itself has a considerably higher speed of sound. A minimum can be found around 40–50% molar fraction and only after this value speed of sound raises slightly with further ethanol content increase in the whole mixture. In Fig. 3 we furthermore have added two single measurement points ( ) for regular blends of RON80 and 98 without ethanol. Both points are above the respective E10 blends with almost the same difference in the reduced frequency. This feature in speed of sound behavior of ethanol blended gasoline confirms previously stressed point of major speed of sound effect on the PnC sensor response and affords to distinguish initially overloaded fuels from regularly blended ones. 4. Conclusions Phononic crystal sensors provide a new approach for real-time in situ measurement. Due to the absence of any electronic component at the place of measurement, the sensor offers significant advantages especially under harsh measurement conditions, e.g., in explosive environment. The new sensor platform based on a two-dimensional phononic crystal having a resonant cavity paves the road to a robust method of gasoline properties determination. Analysis shows the potential of phononic crystals to act as sensor for distinguishing gasoline with different octane numbers and as ethanol presence sensor in ethanol-gasoline blends specifically in the range between 0% and 10% by volume. Since the transduction scheme is based on one physical parameter, the speed of sound of the mixture, the result is ambiguous in a certain range. Ongoing work will therefore address the peak maximum value. The actual design is not optimized for this purpose since the peak amplitude is too much influenced by the band structure of the device. However, absorption of binary mixtures comprising alcohols and aromatics is known to be concentration dependent different to speed of sound [16–18]. Acknowledgments The work has been supported by a grant of the German Research Foundation (LU 605/12-2). Support by the European Commission Seventh Framework Program (233883, TAILPHOX) is gratefully acknowledged as well. References [1] Commission of the European communities, Proposal for a directive of the European parliament and of the council on the quality of petrol and diesel fuels and amending, Directive 98/70/EC, Brussel, 2001.

[2] J.W. Weaver, L.R. Exum, L.M. Prieto, Gasoline Composition Regulations Affecting LUST Sites, Ecosystems Research Division, Athens, 2010. [3] C. Harper, J.J. Liccione, Toxicological Profile for Gasoline, U.S. Department of Health and Human Services, Atlanta, 1995107–111. [4] A.W. Drews, Manual on Hydrocarbon Analysis, 6th ed., ASTM International, Baltimore, 1998. [5] E.V. Shatokhina, Fast analysis of the quality and environmental safety of motor fuels, Chemistry and Technology of Fuels and Oils 43 (2007) 242–247. [6] R. Lucklum, J. Li, Phononic crystals for liquid sensor applications, Measurement Science and Technology 10 (2009) 124014. [7] R. Lucklum, M. Ke, M. Zubtsov, Two-dimensional phononic crystal sensor based on a cavity mode, Sensors and Actuators B 171/172 (2012) 271–277. [8] R. Olsson, I. El-Kady, Microfabricated phononic crystal devices and applications, Measurement Science and Technology 20 (2009) 012002. [9] J. Berryman, Analysis of ultrasonic velocities in hydrocarbon mixtures, Stanford Exploration Project 75 (1997) 479–486. [10] R. Stokes, K. Marsh, Solutions of nonelectrolytes, Annual Review of Physical Chemistry 2541 (1972) 65–92. [11] H. Van Ness, C. Soczek’, N. Kochar, Thermodynamic excess properties for ethanol-n-heptane, Journal of Chemical and Engineering Data 12/13 (1965) 346–351. [12] K. Marsh, C. Burfitt, Excess volumes for alcohols + non-polar solvents I. Ethanol + cyclohexane, +n-hexane, +benzene, +carbon tetrachloride, +cyclopentane, and +p-xylene, Journal of Chemical Thermodynamics 7 (1975) 955–968. [13] D. Papaioannou, D. Zlakas, C. Panaylotou, Volumetric properties of binary mixtures. 1. 2-Propanone 4-2,2,4-trimethylpentane and n-heptane + ethanol mixtures, Journal of Chemical and Engineering Data 36 (1991) 35–39. [14] O. Kiyohara, G. Benson, Ultrasonic speeds and isentropic compressibilities of n-alkanol + n-heptane mixtures at 298.15 K, Journal of Chemical Thermodynamics 11 (1979) 861–873. [15] J. Glinski, Determination of the conditional association constants from the sound velocity data in binary liquid mixtures, Journal of Chemical Physics 118 (2003) 2301. [16] W. Schaaffs, Landolt-Boernstein NS5, Springer, Berlin, 1967 (Chapter [17] W. Schaaffs, Landolt-Boernstein NS5, Springer, Berlin, 1967 (Chapter 3.2). [18] D.R. Lide, CRC Handbook of Chemistry and Physics, 84th ed., CRC press, Boca Raton, 2003–2004.

Biographies Aleksandr Oseev has been employed as Research Associate at the Otto-vonGuericke-University Magdeburg, Institute for Micro and Sensor Systems since 2011. He received his master degree in Microelectronic and Solid State Electronics in 2008 from the Faculty of Electronics at Saint-Petersburg Electrotechnical University, Russia. He presently works on his Ph.D. degree in the area of acoustoelectronic devices for sensing applications. Mikhail Zubtsov has been employed as Research Associate at the Otto-vonGuericke-University Magdeburg, Institute for Micro and Sensor Systems since 2010. He received his Diploma degree in physics, with specialization on electron–phonon interaction in disordered alloys, from Moscow State University, Moscow (Soviet Union) in 1979 and the Ph.D degree in physics of magnetic phenomena, with specialization on the quantum theory of non-local and non-linear exchange interactions in metals, from the same university in 1985. He currently works on phononic crystal sensors. Ralf Lucklum has been employed at the Otto-von-Guericke University, Magdeburg (Germany) at the Department of Electrical Engineering since 1986. In 1977 he received his Ph.D degree; in 2002 he habilitated at the Institute of Micro and Sensor Systems and is currently chairing the Sensor and Measurement Science group. He has been involved in several national and international sensor research projects. His present research activities include the development of ultrasonic sensor systems for process monitoring in fluidic systems based on phononic crystals, acoustic microsensors for chemical analysis and material science as well as application orientated sensor projects.