Optics Communications 282 (2009) 3785–3788

Contents lists available at ScienceDirect

Optics Communications

journal homepage: www.elsevier .com/locate /optcom

Rapid and quantitative detection of ethanol proportion in ethanol–gasolinemixtures by Raman spectroscopy

Qing Ye a,*, Qinfeng Xu a, Yongai Yu b, Ronghui Qu a, Zujie Fang a

a Lab of Information Optics, Shanghai Institute of Optics and Fine Mechanics, Chinese Academy of Sciences, Shanghai 201800, PR Chinab Ocean Optics Asia, Shanghai 200336, PR China

a r t i c l e i n f o

Article history:Received 5 January 2009Received in revised form 9 June 2009Accepted 12 June 2009

PACS:33.20.Fb78.30.�j92.82.Cp

Keywords:Raman spectroscopyEthanol–gasoline mixtureRapid and quantitative detection

0030-4018/$ - see front matter � 2009 Elsevier B.V. Adoi:10.1016/j.optcom.2009.06.034

* Corresponding author.E-mail address: yeqing@mail.siom.ac.cn (Q. Ye).

a b s t r a c t

The paper reports a Raman spectral in-situ detection technique to measure the ethanol proportion in anethanol–gasoline mixture rapidly and quantitatively. The transformed information of Raman spectra fordifferent mixtures of gasoline and ethanol have been observed and recorded. The relative intensity ratioof two typical Raman peaks for ethanol and gasoline satisfies a linear relation with a variation in volumeratio which could be used as a quantitative measure. This method, compared with the direct intensitymeasurement of a Raman peak, may not only eliminate the random measurement error due to dark cur-rent, but also confirm the accurate the proportion of ethanol and gasoline.

� 2009 Elsevier B.V. All rights reserved.

1. Introduction

With the development of world economy, the requirement ofpetroleum resources have increased sharply with reserves reducedrapidly. Excessive usage of petroleum and associated hydrocarbonproducts has induced the shortage of petroleum and increased glo-bal warming. This also became one of the important problemswhich influence the development of human life. The search forclean, renewable energy, such as bio-fuel [1–3] or solar energy[4–6], has become a challenge to sustainable human development.Ethanol and methanol, two types of alcohol bio-fuels, are preferredbecause of relatively low cost and non-contaminated. However, theuse of pure alcohol in internal combustion engines is possible onlyif the engine is designed or modified for that purpose. Usually theycan be mixed with gasoline in various ratios for use in unmodifiedor minor modified gasoline engines. Typically, ethanol is usedwider than methanol in this manner because methanol is toxic.Currently, the ethanol–gasoline mixtures are described by ‘‘E”numbers which is the percentage of ethanol in the mixture by vol-ume. For example, E50 is 50% anhydrous ethanol and 50% gasoline.Low ethanol mixtures, from E5 to E25, are also known as gasohol,though internationally the most common use of the term ‘gasohol’

ll rights reserved.

refers to the E10 mixture. Moreover, some other fuel additives arealso added to improve ignition. For different kinds of vehicles, aspecially designed ethanol–gasoline ratio is required, for example,E85, an alternative motor fuel as defined by the US Department ofEnergy [7]. Thus some rapid and accurate detection methods areneeded to confirm the ethanol proportion in ethanol–gasoline mix-tures. This may not only protect the benefit of the consumer fromfake productions, but also keep the performance of vehicle andprolong its life-span.

In recent years, there has been increasing interest in using opti-cal methods to identify ethanol–gasoline content and proportion[8–10]. However, a quantitative detection of ethanol/methanolproportion in ethanol/methanol–gasoline mixtures is usuallytime-consuming. The specialized fluorescence methods are notonly expensive, but also that their bulk set-up and power require-ment are not at all suited for in-field diagnostics. Raman spectros-copy [11] is considered as the fingerprint of molecule and has beenused widely in the qualitative analysis of chemical component, forexample, drug [12] and gemstone identification [13,14], melaminedetection in milk [15] and so on. In this paper, an in-situ Ramanspectral detection technique is implemented to measure theethanol proportion in ethanol–gasoline mixtures rapidly and quan-titatively. The transformed information of Raman spectra for differ-ent-proportioned gasoline–ethanol mixtures are observed andrecorded. The ratio of two typical Raman peaks is used to confirm

Fig. 2. Spectrum of 785 nm diode laser.

0 500 1000 1500 2000

0

50

100

150

200

250

300

350

481cm-1 1637cm-1

1407cm-1

1479cm-1

759cm-1251cm-1

1030cm-1In

tens

ity (a

.u.)

Raman Shift (cm-1)

Pure gasoline

Fig. 3. Raman spectrum of pure gasoline.

3786 Q. Ye et al. / Optics Communications 282 (2009) 3785–3788

ethanol proportion quantitatively. The method may eliminate theeffect of the dark current and obtain rapidly an accurate result.

2. Principle and experimental set-up

Raman spectroscopy is a spectroscopic technique used in con-densed matter physics and chemistry to study vibrational, rota-tional, and other low-frequency modes in a system [16,17]. It hasbeen used as a relatively simple analytical method for identifica-tion of molecules in gases, liquids and solids by scattering of laserlight. Through measuring the frequency and intensity of in-elasti-cally scattered light (Raman light) from the sample, the moleculesin the sample can be qualitatively and quantitatively measured. Itenables real-time, non-intrusive multi-component measurement.In contrast to Fourier transform infrared (FT-IR), visible or nearIR, Raman wavelengths can be efficiently transported by optical fi-ber. This has made Raman spectroscopy suitable for remote mea-surement in the harsh and difficult environment [10]. Fig. 1shows the completed optical fiber Raman experimental set-up. Awavelength 785 nm diode laser, with the maximum output powerof 300 mW and FWHM of 0.3 nm as shown in Fig. 2, was used asthe excitation light source. The laser was output by a 200 lmcore-diameter fiber and collimated by a lens. A 3 nm band-pass fil-ter at 785 nm was placed between the collimated lens and focuslens. This filter was tunable for its band-pass by adjusting the inci-dent angle onto the filter. When the incident angle was 24� asshown in the right side of Fig. 1, the filter had a maximum trans-mittance of about 85% at 785 nm and 100% reflection for otherwavelengths. The test sample was exited with a laser through a fo-cus lens. The back scattered Raman signal from the sample was re-flected and collected to the spectrometer, by the same focal lens,via a second band-pass filter. This filter may remove the Rayleighscattering which improved the signal-to-noise (SNR). The high-sensitivity spectrometer of QE65000-Raman-785 (supplied byOcean Optics Asia) had a 1200 mm�1 grating (covering 780–1100 nm wavelength range) along with a 1024 � 58 pixel chargecoupled device (CCD) attached to the exit of the spectrographand was interfaced with a PC via an USB cable. In order to separatethe Stokes light from the back scattered light, a larger 6-OD low-ripple long-pass filter with started wavelength of 794.5 nm (sup-plied by Iridian Inc.) was used.

3. Results and analysis

In our experiment, the commercial anhydrous ethanol with apurity of 99.7% and 93# gasoline were mixed in glass bottles withknown proportions. In order to obtain good Raman spectrum infor-mation, a 20-s integration time was set. The spectrometer wascooled to �20 �C with the thermoelectric Cooler (TEC) in order toachieve a 20� SNR improvement. Before measuring the sample,the dark current has been collected and stored for the same inte-gration time and temperature. This will be subtracted from the testsample. Fig. 3 shows the Raman spectrum of pure gasoline. A largenumber of Raman peaks, for example, at 251 cm�1, 481 cm�1,759 cm�1, 1030 cm�1, 1407 cm�1, 1479 cm�1 and 1637 cm�1, were

band-passfilter1

785nm laser

Spectrograph

Long-passfilter(OD>6)

Lens1

Lens2

ffiber1

fiber2

Fig. 1. Schematic diagram o

observed. The most intense and typical Raman peaks appeared at1030 cm�1 and 1479 cm�1 which could be use to measure the sam-ple concentration or proportion. Fig. 4 shows the Raman spectrumof pure ethanol. Five Raman peaks at 469 cm�1, 912 cm�1,1082 cm�1, 1122 cm�1, 1479 cm�1 were also visible. After compar-ing the Raman spectra of gasoline and ethanol, it was found thatthey had a common Raman shift at 1479 cm�1. In order to improvethe tested results, the two most intense and typical Raman peaks,i.e., the gasoline at 1030 cm�1 and the ethanol at 912 cm�1, wereselected to measure quantitatively the ethanol proportion in etha-nol–gasoline mixtures for the experiments to follow.

The test sample was prepared by mixing ethanol and gasoline.The associated Raman spectra are shown in Fig. 5 for different vol-ume ratios of ethanol and gasoline 1:10 (E10), 3:10 (E30), 5:10(E50) and 7:10 (E70). With the incremention of the concentration

24

24

Sample

ocus Lens

band-passfilter2

band-passfilter1

band-passfilter2

f experimental set-up.

0 500 1000 1500 2000

0

50

100

150

200

250

300

350

1479cm-1

1122cm-1

1082cm-1

912cm-1

469cm-1Inte

nsity

(a.u

.)

Raman Shift (cm-1)

Pure ethanol

Fig. 4. Raman spectrum of pure ethanol.

0.0 0.2 0.4 0.6 0.80.0

0.3

0.6

0.9

1.2

1.5

1.8

R SD N P-------------------------------------------------0.99907 0.031 4 9.26024E-4

Ram

an p

eak

inte

nsity

ratio

Ethanol-gasoline ratio

Experimental result Linear fit (y=2.2765x+0.00365)

Fig. 6. Detecting the ethanol ratio in an ethanol–gasoline mixture using Ramanspectroscopy.

Table 1Relative Raman peak intensities at 912 cm�1 and 1030 cm�1.

Ethanol Raman peakat 912 cm�1

Gasoline Raman peakat 1030 cm�1

Ratio

E10 47 186 0.253E30 76 117 0.650E50 115 100 1.150E70 146 91 1.604

Q. Ye et al. / Optics Communications 282 (2009) 3785–3788 3787

of ethanol, the intensity of Raman peak at 912 cm�1 increased cor-respondingly while the intensity of Raman peak at 1030 cm�1 re-duced. Anand et al. used the direct intensity of Raman peak at730 cm�1 to measure quantitatively the concentration of methanolin the methanol–gasoline mixtures [18] and this was easy to gen-erate a relatively random measurement error due to the dark cur-rent and other factors. For example, the Raman spectra in Figs. 3and 4 had different direct current background noise and they werevibrational with the time. Moreover, this method was also difficultto confirm the accurate ratio of methanol and gasoline because ofother adulterations. In this investigation, the relative intensity ra-tio of two typical Raman peaks for ethanol at 912 cm�1 and gaso-line at 1030 cm�1 was used to eliminate the above effects andobtained an accurate ratio of ethanol and gasoline.

Table 1 shows the relative Raman peak intensities at 912 cm�1

and 1030 cm�1 with the backgrounds removed. These were ob-

Fig. 5. Raman spectra for various ethanol–gasoline m

tained by selecting the maximum and minimum values for eachreferenced peak in a small local range. The ratio of two relativeRaman peaks was related with both the concentrations of ethanoland gasoline and it embodied the real-time state of sample. For theE10, E30, E50 and E70 samples, the related ratios were 0.253,

ixtures: (a) E10; (b) E30; (c) E50; and (d) E70.

3788 Q. Ye et al. / Optics Communications 282 (2009) 3785–3788

0.650, 1.150 and 1.604, respectively. Fig. 6 shows the ratio varia-tion vs. the ethanol proportion. This result satisfied a very good lin-ear relation with a correlation coefficient (R) of 0.99907 andstandard deviation (SD) of 0.031. Using this linear relation, an un-known proportion ethanol–gasoline mixture will be identified. Themethod used also has a relatively rapid response time (less than20 s) compared with other methods, for example, high perfor-mance liquid chromatographic (HPLC) (about 30 min). A high per-formance TEC spectrometer may also reduce this integration timefurther and accelerate the measuring process.

4. Conclusion

An in-situ Raman spectral detection technique to measure theethanol ratio in an ethanol–gasoline mixture quantitatively in ashort time frame was used. The transformed information of Ramanspectra for different mixtures of gasoline and ethanol have beenobserved and recorded. The ethanol proportion was obtained accu-rately by measuring the ratio of two typical ethanol and gasolineRaman peaks. This method, compared with the direct intensitymeasurement of a Raman peak, may not only eliminate a relativelylarge measurement error due to the dark current, but also confirmthe proportion accuracy of ethanol and gasoline. Moreover, a rapidand real-time measurement may be also realized.

Acknowledgements

The authors would like to thank the support of Ocean OpticsAsia for supplying the high-sensitivity spectrometer of QE65000-Raman-785 and helpful discussion.

References

[1] B. Hahn-Hägerdal, M. Galbe, M.F. Gorwa-Grauslund, G. Lidén, G. Zacchi, Trendsin Biotechnology 24 (2006) 549.

[2] F. Frusteri, S. Freni, Journal of Power Sources 173 (2007) 200.[3] Ben-Amotz Ami, Journal of Biotechnology 136 (2008) S523.[4] A. Barnett, C. Honsberg, D. Kirkpatrick, in: Proceedings of the Conference

Record of the 2006 IEEE Fourth World Conference on, vol. 2, 2006, p. 2560Available from: <http://ieeexplore.ieee.org/xpl/topAccessedArticles.jsp?punumber=4059527>.

[5] D. Feuermann, J.M. Gordon, Solar Energy 70 (2002) 423.[6] A. Mart, E. Antol, C.R. Stanley, Physical Review Letters 97 (2006) 247701.[7] B. West, K. Knoll, W. Clark, R. Graves, J. Orban, S. Przesmitzki, T. Theiss, Oak

Ridge National Laboratory and National Renewable Energy Laboratory,Retrieved on 2008-10-19, ORNL/TM-2008/117.

[8] K.B. Ozanyan, T.L. Yeo, F.P. Hindle, N.R.J. Poolton, H. McCann, K.L. Tan, IEEESensors Journal 4 (2004) 681.

[9] D. Steers, C. Gerrard, B. Hirst, W. Sibbett, M.J. Padgett, Journal of Optics A-Pureand Applied Optics 1 (1999) 680.

[10] S.K. Khijwania, V.S. Tiwari, Fang-Yu Yueh, J.P. Singh, Sensors and Actuators B125 (2007) 563.

[11] G. Herzberg, Infrared and Raman Spectra of Polyatomic Molecules, VanNostrand Reinhold, New York, NY, 1945.

[12] J.S. Daya, H.G.M. Edwards, S.A. Dobrowski, A.M. Voice, Spectrochimica ActaPart A 60 (2004) 1725.

[13] A.L. Jenkins, R.A. Larsen, Spectroscopy 19 (2004) 20.[14] G.H. Lamprecht, H.G.C. Human, L.W. Snyman, International Journal of Mineral

Processing 84 (2007) 262.[15] L.L. He, Y. Liu, M.S. Lin, J. Awika, D.R. Ledoux, H. Li, A. Mustapha, Sensing and

Instrumentation for Food Quality and Safety 2 (2008) 66.[16] D. Lin-Vien, N.B. Colthup, W.G. Fateley, J.G. Grasselli, The Handbook of Infrared

and Raman Characteristic Frequencies of Organic Molecules, Academic Press,Boston, 1991.

[17] R.A. Nyquist, C.L. Putzig, M.A. Leugers, R.O. Kagel, The Handbook of Infraredand Raman Spectra of Inorganic Compounds and Organic Salts, vol. 1–4,Academic Press, San Diego, 1997.

[18] K. Anand, V. Asundi, Rao Vasudeva, SAE Technical Paper Series, 2000-01-1336.