Combustion and Flame 155 (2008) 431–439www.elsevier.com/locate/combustflame

Homogeneous charge compression ignition of binary fuelblends

Gaurav Mittal ∗, Chih-Jen Sung

Department of Mechanical and Aerospace Engineering, Case Western Reserve University, Cleveland, OH 44106, USA

Received 8 January 2008; received in revised form 2 May 2008; accepted 14 May 2008

Abstract

The present experimental investigation aims to understand the homogeneous combustion chemistry associ-ated with binary blends of three surrogate components for practical fuels, including toluene, isooctane, anddiisobutylene-1 (DIB-1). Specifically, high-pressure autoignition characteristics of the three neat fuel componentsas well as the fuel blends of toluene + isooctane and toluene + DIB-1 are studied herein. Experiments are con-ducted in a rapid compression machine at compressed pressures varying from 15 to 45 bar and under low tointermediate temperatures. To obtain insights into interactions among fuels, the relative proportion of the two neatfuels in the reactive mixtures is systematically varied, while the total fuel mole fraction and equivalence ratio arekept constant. Experimental results demonstrate that ignition delays for neat toluene are more than an order ofmagnitude longer than those for neat isooctane. Whereas DIB-1 has ignition delays shorter than those for isooc-tane at higher temperatures, at temperatures lower than 820 K DIB-1 shows a longer ignition delay. Althoughthe ignition delays of binary blends lie in between the two extremes of neat components, the variation of ignitiondelay with the relative fuel proportion is seen to be highly nonlinear. Especially, a small addition of isooctaneor DIB-1 to toluene can result in greatly enhanced reactivity. In addition, the effect of DIB-1 addition to tolueneis more significant than the effect of isooctane addition. Furthermore, in the compressed temperature range from820 to 880 K, ignition delay of the toluene + isooctane blend shows greater sensitivity to temperature than that ofisooctane.© 2008 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

Keywords: Homogeneous charge compression ignition; Isooctane; Toluene; Diisobutylene-1; Fuel blends; Rapid compressionmachine

1. Introduction

Practical liquid fuels contain hundreds of hydro-carbons, which complicates the development of a ki-netic mechanism of reasonable size for large-scalecomputation. The use of surrogate fuels is a viable ap-

* Corresponding author.E-mail address: gaurav.mittal@case.edu (G. Mittal).

0010-2180/$ – see front matter © 2008 The Combustion Institute.doi:10.1016/j.combustflame.2008.05.003

proach to make the development of chemical kineticmechanisms for practical fuels tractable. A surrogatefuel model that consists of a small number (usuallyless than 10) of fuel components can be developed torepresent the real fuel and predict the desired charac-teristics of the actual fuel. One approach to develop-ing a fuel surrogate is to use a single component fromeach hydrocarbon class in the practical fuel so that theunique molecular structure of each class is included.Therefore, the reaction mechanisms of individual fuel

Published by Elsevier Inc. All rights reserved.

432 G. Mittal, C.-J. Sung / Combustion and Flame 155 (2008) 431–439

components typically provide the basis for the devel-opment of surrogate kinetic models, and the accuracyof a surrogate model is directly dependent on the val-idation of the neat component models over a broadrange of experimental conditions.

The present experimental investigation focuses onthree fuel components, isooctane (2,2,4-trimethylpen-tane, C8H18), toluene (methylbenzene, C7H8), anddiisobutylene-1 (2,4,4-trimethyl-1-pentene, C8H16).In particular, isooctane and toluene are anticipatedas the primary components respectively represent-ing alkanes and aromatics in gasoline surrogates, andare also recognized as components for jet fuel surro-gates [1]. In addition, diisobutylene-1 (DIB-1) is iden-tified as a potential candidate for representing alkenesin gasoline surrogates. Recognizing that these classesof hydrocarbons have vast differences in the overallreactivity and autoignition chemistry, and that theirbehaviors in various blends are not properly under-stood, we aim to conduct a systematic investigationfor neat fuels and their binary blends to characterizethe nature of interactions amongst them, with specialemphasis on autoignition kinetics at low to interme-diate temperatures and high pressures.

It is noted that all three fuel components investi-gated in the present study have high octane numbersand are relatively resistant to autoignition. At lowtemperatures, isooctane shows a region of negativetemperature coefficient and dominance of peroxida-tion reactions [2], resulting in faster ignition than fortoluene. On the other hand, toluene is well known asbeing resistant to low-temperature oxidation due tothe stability of the benzyl radical. Unlike most alkylradicals and aromatics with side chains, the benzylradical has no easy reaction with O2 to give a con-jugate alkene and the HO2 radical [3]. Due to the en-hanced stability of benzyl as a result of delocalizationof electrons, H abstraction is also uncompetitive. Asa consequence, radical–radical recombination leadingto bibenzyl becomes important. In the rapid com-pression machine (RCM) experiments of Roubaud etal. [4], a stoichiometric mixture of toluene did not au-toignite below 916 K at 17 bar. However, oxidation oftoluene may take place at lower temperatures when itis present in a blended fuel mixture due to the devel-opment of a radical pool formed by the other easilyoxidized fuel components in the mixture. DIB-1 hasa molecular structure similar to that of isooctane, andtherefore its kinetics provides insight into the effect ofincluding a double bond in the carbon skeletal struc-ture of isooctane. Investigations [5] have shown thatthe inclusion of a double bond and its position candrastically alter the fuel combustion characteristics.

During the combustion of blended fuels, it isalso unclear how the interaction of different com-ponents can affect the system response. Klotz et

al. [6] observed that independently developed modelsof toluene and butane, when merged together, couldpredict the kinetics of blended fuels. While this as-pect can greatly simplify the development of reactionmechanism of blended fuels, it is not necessarily ageneral conclusion. In particular, there can be a possi-bility of chemical interactions between different fuelcomponents and their radicals, and exclusion of suchinteractions can place undue restrictions on the re-action pathways. Andrae et al. [7] investigated theautoignition of primary reference fuels (PRFs) andtoluene/n-heptane blends. In their model validation,interaction reactions between the individual fuel com-ponents were found to be important in predictinghomogeneous charge compression ignition (HCCI)experiments [7].

Due to chemical interactions in blended fuels, dif-ferences in sensitivity of fuels are also recognized.For example, Andrae et al. [7] attributed the differ-ence in sensitivity of a PRF and a toluene/n-heptaneblend with the same octane number to the effect ofinteraction reactions. Tanaka et al. [8] found bet-ter agreement between model predictions and experi-mental data by including a coupling reaction betweenn-heptane and isooctane. Therefore, systematic devel-opment of surrogate mechanisms and understandingof chemical kinetic processes leading to autoignitionrequires studies of kinetic interactions in blended fu-els.

Although blends of the primary reference fuels,n-heptane and isooctane, have been studied exten-sively, investigations on blends of toluene with isooc-tane and DIB-1 are meager. Recently, Vanhove etal. [9] conducted experiments focusing on selectivebinary blends, including toluene + isooctane, in arapid compression machine at temperatures below900 K and pressures less than 16 bar. For toluene +isooctane blends, in particular, a large change in thetemperature dependences of both the first stage delaysand the total ignition delays was noted with additionof toluene to isooctane [9].

This work focuses on the autoignition of toluene+isooctane and toluene + DIB-1 blends at low to inter-mediate temperatures and elevated pressures, becauseautoignition chemistry under such conditions has notbeen well established. Experiments are conducted forvarious homogeneous mixtures in a rapid compres-sion machine. The specific objective is to obtain ex-tensive experimental data for autoignition of neat fuelcomponents and binary blends under conditions rel-evant to engine combustion, thereby providing vali-dation targets for model development and developingan understanding of kinetic interactions in the fuelblends. In the following, we shall sequentially presentthe experimental specifications and discuss our exper-imental results.

G. Mittal, C.-J. Sung / Combustion and Flame 155 (2008) 431–439 433

Fig. 1. RCM schematic and creviced piston head configuration. Dimensions (in mm): A = 0.50, B = 4.0, C = 0.15, D = 20.0,and E = 1.50.

2. Experimental specifications

Fig. 1 shows a schematic of the present RCM sys-tem that consists of a driver piston, a reactor cylinder,a hydraulic motion control chamber, and a driving airtank. The driver cylinder has a bore of 5 in. (12.7 cm)and that of the reactor cylinder is 2 in. (5.08 cm).The machine is pneumatically driven and hydrauli-cally stopped. Stroke and clearance can be varied toprovide a range of compression ratios. The cylindricalreaction chamber is equipped with sensing devices formeasuring pressure and temperature, gas inlet/outletports for preparing the reactant mixture, and quartzwindows for optical access. Dynamic pressure dur-ing compression is measured using a Kistler 6125Btransducer with a 5010B charge amplifier. The ma-chine incorporates an optimized creviced piston headdesign to promote the homogeneous reaction zone.Homogeneous mixture of reactants is prepared mano-metrically inside a mixing tank equipped with a mag-netic stirrer. When preparing fuel/oxidizer mixturesat room temperature, the maximum fuel partial pres-sure is kept less than 75% of the corresponding sat-uration pressure to prevent condensation. In addition,by monitoring the pressure of individually filled re-actants, it is ascertained that there is no reactant ad-sorption on the wall inside the mixing tank. Furtherdetails of the present rapid compression machine canbe found in Mittal and Sung [10,11].

Autoignition investigations for isooctane, DIB-1,toluene, and various blends of toluene + DIB-1 andtoluene + isooctane are conducted with a constantoverall equivalence ratio of 0.75, over a compressedtemperature range of 740–1060 K and a compressedpressure range of 15–45 bar. Here, “compressed pres-sure” and “compressed temperature” refer to the pres-sure and the temperature at the end of the compres-sion stroke (top dead center, TDC), respectively. Theoverall equivalence ratio is defined here by consid-ering each fuel component to be part of a binaryfuel mixture, so that the individual equivalence ra-

Table 1Relative molar compositions of gas mixtures investigated,with the overall equivalence ratio being kept constant at 0.75

Mixture#

RI Isooc-tane

Tolu-ene

O2 N2 Ar

1 1 0.5 0 8.333 0.000 50.6002 0.75 0.375 0.125 7.750 2.400 48.7843 0.5 0.25 0.25 7.166 4.667 47.1004 0.25 0.125 0.375 6.583 6.990 45.3605 0.05 0.025 0.475 6.117 8.900 43.9176 0 0 0.5 6.000 9.323 43.6107 1 0.5 0 8.333 16.850 33.7508 0.75 0.375 0.125 7.750 19.183 32.000

Mixture#

RDIB-1 DIB-1 Tolu-ene

O2 N2 Ar

9 1 0.5 0 8.000 1.343 49.59010 0.5 0.25 0.25 7.000 5.423 46.51011 0.05 0.025 0.475 6.100 9.063 43.77012 0 0 0.5 6.000 9.323 43.61013 1 0.5 0 8.000 18.300 32.63314 1 1 0 16.000 7.500 52.660

tio of the neat fuel is the same as the overall equiv-alence ratio. All of the neat fuels are supplied byFisher Scientific, and isooctane, DIB-1, and tolueneare 99.5, 99, and 99.8% pure, respectively. To system-atically characterize the chemical interactions, exper-iments are conducted by varying the relative propor-tion of the fuel components while keeping the com-bined fuel mole fraction of the two-component fuelmixture (toluene+ isooctane or toluene+DIB-1) con-stant at 0.0084. As such, fuel loading is fixed for themixtures tested. The molar compositions of variousblended mixtures investigated herein are listed in Ta-ble 1. RDIB-1 is defined as the fraction of DIB-1 inthe combined toluene + DIB-1 fuel mixture, and isexpressed as RDIB-1 = FDIB-1/(FDIB-1 +FT), whereFDIB and FT are, respectively, the mole fractions ofDIB-1 and toluene in the fuel mixture. RI is definedin a similar manner as the fraction of isooctane in

434 G. Mittal, C.-J. Sung / Combustion and Flame 155 (2008) 431–439

Fig. 2. A typical pressure history for autoignition of DIB-1.Molar composition: DIB-1/O2/N2/Ar = 0.5/8/18.3/32.633.Initial conditions: T0 = 297 K and P0 = 785 Torr. Condi-tions at TDC: TC = 869 K and PC = 35.1 bar.

the combined toluene + isooctane fuel mixture. Mix-tures #1 to #6 vary RI from 0 to 1, while keeping themixture specific heat ratio constant through adjust-ment in relative compositions of inert gases—N2 andAr. Mixtures #7 and #8 have somewhat lower valuesof specific heat ratio to enable experiments at lowercompressed temperatures. Mixtures #9 to #12 varyRDIB-1 while keeping the same specific heat ratio asfor mixtures #1 to #6, whereas mixtures #13 and #14have lower specific heat ratios. In addition, mixtures#1 to #13 have the same total fuel mole fraction of0.0084, and mixture #14 has a higher fuel mole frac-tion of 0.0130.

For a given mixture composition, the compressedgas temperature at the end of compression is variedby altering the compression ratio, whereas the desiredpressure at TDC is obtained by varying the initialpressure of the reactive mixture. The temperature atTDC, TC, is determined by the adiabatic core hypoth-esis according to the relation of

TC∫

T0

γ

γ − 1

dT

T= ln[PC/P0],

where P0 is the initial pressure, T0 is the initial tem-perature, γ is the specific heat ratio, which is a func-tion of temperature, and PC is the measured pressureat TDC.

3. Results and discussion

Fig. 2 shows an example of pressure traces forthe ignition of a DIB-1/O2/N2/Ar mixture (mixture#13) with an initial temperature of T0 = 297 K and aninitial pressure of P0 = 785 Torr. The measured pres-sure and the deduced temperature at the end of com-pression (t = 0) are PC = 35.1 bar and TC = 869 K,

respectively. It is also seen from Fig. 2 that the com-pression stroke is ∼ 25 ms and approximately a 40%pressure rise occurs in the last 2 ms of the compres-sion stroke. Such a rapid pressure rise is desirable sothat the extent of chemical reaction during compres-sion is minimized. Fig. 2 further demonstrates that theignition delay is defined as the time from the end ofthe compression stroke, where the pressure peaks att = 0, to the instant of 50% of the total pressure risedue to ignition. Because of the rapid rise in pressureduring ignition, as seen in Fig. 2, other definitions ofignition delay, such as the inflection point in the pres-sure history, yield the similar results.

As discussed in [12], the overall uncertainty inthe reported ignition delays obtained from the currentRCM experiments is estimated to be in the range of 3–10%, by considering the uncertainties in the mixturepartial pressure measurements, the measured pressurehistory, and the deduced compressed gas tempera-ture. We further note that the experimental raw datain terms of pressure history and the associated oper-ating conditions, including reactive mixture composi-tion, initial temperature/pressure, compression strokespecification, and compressed temperature/pressure,are detailed and tabulated in [13]. In addition, theempirical effective volume parameters characterizingthe effect of heat loss (cf. [10,12]), which are neededfor modeling the specific RCM experiments, are alsoavailable in [13]. For kinetic modeling of RCM exper-iments, it is important to simulate the entire processesoccurring in each experiment, including the compres-sion stroke.

Fig. 3 presents the measured ignition delays forthe three neat fuels in different temperature and pres-sure regimes. In Fig. 3a, significant differences in theindividual reactivity of neat fuels can be noted, whilethe fuel mole fraction in the mixture is kept constantat 0.0084. Under the conditions of Fig. 3a, the igni-tion delays of isooctane are an order of magnitudelower than those of toluene, whereas DIB-1 igniteseven faster than isooctane. However, Fig. 3b showsthat the ignition delays of DIB-1 can become muchlonger than those of isooctane at lower temperatures.It is further noted in Fig. 3b that while isooctane ex-hibits a discernible negative temperature coefficient(NTC) behavior, DIB-1 does not show such behav-ior. Even at a higher fuel loading of XDIB-1 = 0.0130,where XDIB-1 is the mole fraction of DIB-1 in the re-active mixture, only an inflection point is noted.

The significant differences between the ignitionresponses of isooctane and DIB-1 are introduced bythe presence of a double bond. Comparison of theirRCM data seems to suggest that the presence ofdouble bond has significant impact on the peroxida-tion reactions. In the low-to-intermediate-temperaturerange, the alkenyl chains of alkenes undergo perox-

G. Mittal, C.-J. Sung / Combustion and Flame 155 (2008) 431–439 435

Fig. 3. Comparison of ignition delays for neat toluene, isooc-tane, and DIB-1 with equivalence ratio of 0.75: (a) highertemperature regime with PC = 45 bar and (b) lower temper-ature regime with PC ∼ 35 bar.

idation reactions that are typical of alkanes. In ad-dition, alkenes have specific pathways related to theaddition of hydroxyl radicals and oxygenated radi-cals to the double bond. Thus, the overall kinetic be-havior of alkenes is a result of competition betweenthese two channels, namely the addition reactions tothe double bond and the reactions through hydrogenabstraction of the alkyl chain [5]. When chemistryis dominated by the reactivity of a double bond, thephenomenology of autoignition is plain and does notexhibit pronounced features of NTC [5]. In a study byRibaucour et al. [14] on autoignition and kinetic mod-eling of n-pentane and 1-pentene, the importance ofthe reactions involving direct addition of OH and HO2to the double bond in predicting ignition characteris-tics was highlighted. Additionally, in their work [14],1-pentene showed slower reactivity and longer igni-tion delays than n-pentane in the NTC regime, as with

Fig. 4. Effect of percentage of isooctane (RI) on autoigni-tion in toluene + isooctane blends (mixtures #1–#6). Com-bined fuel mole fraction = 0.0084, overall equivalence ratioφ = 0.75, and P0 = 930 Torr. Conditions at TDC: TC =992–996 K and PC = 44.8 bar.

isooctane and DIB-1, demonstrated in the presentstudy. A similar role of addition reactions at the dou-ble bond was also highlighted by Touchard et al. [15].The less pronounced features of thermokinetic inter-action and NTC behavior for DIB-1 imply that theaddition reactions at the double bond dominate overthe peroxidation reactions of the alkenyl chain.

Fig. 4 compares the pressure traces for various RI,showing the effect of the percentage of isooctane inthe binary toluene + isooctane blends on ignition de-lay. Again, in these experiments the total fuel molefraction and global equivalence ratio are kept con-stant at 0.0084 and 0.75, respectively. By keeping themixture specific heat ratio constant (mixtures #1–#6),it can be seen from Fig. 4 that all the mixtures arecompressed to the same pressure of 44.8 bar at TDC.However, there is a very minor difference in the tem-perature at TDC (TC = 992–996 K) due to the varia-tion of initial room temperature. It is observed that theignition delay of neat toluene (RI = 0) is more than anorder of magnitude greater than that of neat isooctane(RI = 1). Although the ignition delays of the blendedmixtures are in between those of the two neat compo-nents, the variation of ignition delay with RI is seento be highly nonlinear. It is also noted that a smallamount of isooctane addition to toluene will greatlyreduce the ignition delay, while the effect of tolueneaddition to isooctane is very weak.

Figs. 5a and 5b respectively compare the effects ofRI (mixtures #1–#6) and RDIB-1 (mixtures #9–#12)on ignition delay at PC = 45 bar, with compressedtemperatures of 994 K for the former and 964 K forthe latter. Similar to the toluene + isooctane blends,the toluene + DIB-1 blends also show a nonlinearvariation of ignition delay. It is further seen in Fig. 5that the effect of small amount of DIB-1 addition to

436 G. Mittal, C.-J. Sung / Combustion and Flame 155 (2008) 431–439

Fig. 5. Effect of (a) RI (TC = 994 K and PC = 45 bar; mix-tures #1–#6) and (b) RDIB-1 (TC = 964 K and PC = 45 bar;mixtures #9–#12) on ignition delay. Total fuel mole frac-tion = 0.0084 and overall equivalence ratio φ = 0.75.

toluene is much more significant than the effect ofisooctane addition to toluene.

Fig. 6a summarizes the measured ignition delaytimes of toluene + isooctane mixtures (mixtures #1and #4–#6) at a pressure near 45 bar at TDC as a func-tion of the inverse of the compressed temperature. It isseen that ignition delay follows the Arrhenius depen-dence quite well in the compressed temperature rangeinvestigated (TC = 920–1060 K). Over this range ofcompressed temperatures, the effective activation en-ergies, reflected by the slope in Fig. 6a, are seen to befairly similar at varying RI. Fig. 6b further shows thesummary of the present measurements for the binarytoluene + DIB-1 blends investigated (mixtures #9–#12). As noted earlier, the addition of 5% DIB-1 totoluene, RDIB-1 = 0.05, yields a substantial reductionin ignition delay. It is also demonstrated in Fig. 6b thatthe effective activation energies at varying RDIB-1 arefairly similar.

Fig. 7 shows a similar comparison at a rela-tively lower compressed pressure of PC = 25 barfor both binary blends of toluene + isooctane and

Fig. 6. Effect of (a) RI (mixtures #1, #3–#6) and (b) RDIB-1(mixtures #9–#12) on ignition delay. Lines represent thebest fits for the Arrhenius dependence. Combined fuel molefraction = 0.0084, overall equivalence ratio φ = 0.75, andPC = 44.8 bar.

toluene + DIB-1. The trends are similar to the higherpressure results in Fig. 6. Experiments conducted at acompressed pressure near 15.5 bar (not shown here)also yield similar results. It is therefore concluded thatfor compressed pressures ranging from 15 to 45 bar,the autoignition response of toluene + isooctane(toluene + DIB-1) mixtures with the variation of RI(RDIB-1) is similar in the compressed temperaturerange of Figs. 6 and 7.

Experiments are also conducted at relatively lowercompressed temperatures, TC = 800–905 K, to ex-amine the behaviors of the toluene + isooctane mix-tures as temperature approaches the region of nega-

G. Mittal, C.-J. Sung / Combustion and Flame 155 (2008) 431–439 437

Fig. 7. Effect of (a) RI (mixtures #1, # 3–#6) and (b) RDIB-1(mixtures #9–#12) on ignition delay. Lines represent thebest fits for the Arrhenius dependence. Combined fuel molefraction = 0.0084, overall equivalence ratio φ = 0.75, andPC = 25 bar.

tive temperature coefficient. Experimental results fora compressed pressure of PC = 36 bar are shown andcompared in Fig. 8 for mixtures of RI = 1.0 (mix-ture #7) and 0.75 (mixture #8). The ignition delaysshow bending behavior as the lower end compressedtemperature is close to the end of the NTC regionseen in Fig. 3b, where TC ∼ 800 K. In addition, atlower compressed temperatures, the difference in ig-nition delays between RI = 0.75 and RI = 1.0 in-creases with decreasing TC, thereby demonstratingan inhibition effect of toluene on autoignition. Ascompressed temperature increases, ignition delay forthe RI = 0.75 blend is also noted to be more sen-sitive to temperature than neat isooctane. This trend

suggests a change in the oxidation mechanism oftoluene + isooctane mixtures. Furthermore, it is seenfrom Figs. 6–8 that the change in temperature sen-sitivity of toluene + isooctane blends in comparisonto neat isooctane occurs only in the range of lowertemperatures, from 820 to 880 K. This observation isconsistent with the recent work by Vanhove et al. [9].

Modeling of the present experimental data for thebinary blends studied cannot be readily conductedfor the following reasons. Our recent work [12] ontoluene autoignition has highlighted the deficienciesof the literature kinetic models in predicting ignitiondelays at elevated pressures and low temperatures.Similarly, significant deficiencies in terms of predict-ing ignition characteristics near the NTC regime havebeen noted for isooctane [10]. Furthermore, except foran attempt by Metcalfe et al. [16] to develop a reac-tion mechanism of diisobutylene based on the isooc-tane mechanism of Curran et al. [2], there exists nomechanism for DIB-1 validated under the present ex-perimental conditions.

Based on the experimental results, the chemicalinteractions of the binary fuel components can be dis-cussed qualitatively by first considering the oxidationchemistry of individual neat fuels. Regarding the low-temperature chemistry of isooctane, alkyl radicalsreadily react via peroxidation and subsequent internalisomerization to form QOOH, where Q denotes theCnH2n structure. As such, chain branching is mainlythrough carbonylhydroperoxide species formed byperoxidation of QOOH. As temperature increases,chain propagation channels of QOOH, leading to for-mation of cyclic ethers, aldehydes, alkenes, and rad-icals of OH and HO2, become predominant [2]. Onthe other hand, at low temperatures toluene is mainlyconsumed by the reactions

C6H5CH3 + OH → C6H5CH2 + H2O (R1)

and

C6H5CH3 + O2 → C6H5CH2 + HO2. (R2)

Based on Ellis et al. [17], benzyl radical can undergoreactions with HO2 and O2. The reaction of benzylradical with HO2 proceeds via

C6H5CH2 + HO2 → OH + C6H5CH2O, (R3)

followed by

C6H5CH2O + O2 → C6H5CHO + HO2. (R4)

Ellis et al. [17] also considered that peroxidation ofthe benzyl radical and subsequent isomerization leadto a significant amount of benzaldehyde productionat low temperatures in addition to the benzyl + HO2channel. In addition, the overall reaction

C6H5CH2 + O2 → C6H5CHO + OH (R5)

438 G. Mittal, C.-J. Sung / Combustion and Flame 155 (2008) 431–439

Fig. 8. Comparison of ignition delays for neat isooctane(RI = 1.0, mixture #7) and toluene + isooctane blend ofRI = 0.75 (mixture #8) near the region of NTC. Combinedfuel mole fraction = 0.0084, overall equivalence ratio φ =0.75, and PC = 36 bar.

proceeds via the following sequence:

C6H5CH2 + O2 → C6H5CH2O2→ C6H5CHOOH → C6H5CHO + OH.

It was argued that although the equilibrium for per-oxidation of benzyl is highly unfavorable in com-parison to peroxidation of alkyl radicals, the inter-nal isomerization of benzyl peroxy is rapid due tothe transfer of a labile H atom, and at low tempera-tures the benzyl + O2 channel can compete with thebenzyl + HO2 channel [17].

In view of the above, at low temperatures oxida-tion of neat toluene is difficult due to the necessity ofOH radicals for reaction (R1) to occur. Addition ofisooctane or DIB-1 to toluene provides a pool of OHfor toluene and thus increasing the overall reactivityof toluene. However, if toluene participates predom-inantly through (R1) without the subsequent genera-tion of radical pool by the reactions of benzyl, the ox-idation of isooctane would be inhibited due to radicalscavenging by toluene. Such an effect is observed inthe present experiments at lower compressed temper-atures. At higher temperatures, contribution to radicalpool by reactions of benzyl, such as (R3) and (R5),could overcome the radical scavenging by (R1), asshown by the ignition delay results of Fig. 8.

In closing this section, it is noted that the ob-served nonlinear sensitization of toluene ignition tosmall amount of isooctane or DIB-1 addition is sim-ilar to the kinetic sensitization of methane ignitionwith addition of small quantities of ethane or propanedemonstrated in [18,19]. Because of the low reactivityof the methyl radical, methane is also highly resis-tant to autoignition. Addition of a small amount of

ethane or propane to methane was found to effectivelysensitize the fuel blend mixture, resulting in muchshorter ignition delay as compared to pure methane[18,19]. Recognizing the fact that the radicals pro-duced by H atom abstraction from toluene or methanedo not have a decomposition reaction leading to chainbranching, the similarity between the present sensiti-zation of toluene by isooctane or DIB-1 and the sen-sitization of methane by ethane or propane in [18,19]is a result of early radical generation by a small addi-tion of a more reactive fuel component, which in turnaccelerates the ignition of the less reactive fuel com-ponent.

4. Concluding remarks

Autoignition of toluene, isooctane, diisobutyl-ene-1 (DIB-1), and binary blends of toluene + iso-octane and toluene + DIB-1 is investigated in a rapidcompression machine at compressed pressures from15 to 45 bar and compressed temperatures from 740to 1060 K. To characterize the chemical interactionsbetween two fuel components, the relative proportionof isooctane or DIB-1 in toluene is systematicallyvaried. These experimental data are of direct rele-vance to the development of kinetic mechanisms ofsurrogates for gasoline and aviation fuels. At lowtemperatures, DIB-1 exhibits longer ignition delaysthan isooctane, whereas at higher temperatures, ig-nition delays of DIB-1 are shorter than those ofisooctane. Experimental results also show that the ad-dition of a small amount of DIB-1 or isooctane totoluene results in a drastic reduction in ignition de-lay, with the effect of DIB-1 addition being muchstronger. On the other hand, the change in ignitiondelay by adding small amount of toluene to isooctaneor DIB-1 is significantly less. In addition, a changein oxidation chemistry of toluene + isooctane blends,in comparison to that of isooctane, is noted in thecompressed temperature range of 820–880 K. Atthese lower temperatures, the overall reactivity oftoluene + isooctane blends is more sensitive to tem-perature than that of neat isooctane. At higher temper-atures, however, the binary blends of different relativeproportions show similar temperature sensitivity, asreflected by the closeness of their overall activationenergies.

Acknowledgments

This work has been supported by the NationalAeronautics and Space Administration under GrantNNX07AB36Z and by the Air Force Office of Sci-entific Research under Grant FA9550-07-1-0515.

G. Mittal, C.-J. Sung / Combustion and Flame 155 (2008) 431–439 439

References

[1] T. Edwards, L.Q. Maurice, J. Propul. Power 17 (2)(2001) 461–466.

[2] H.J. Curran, P. Gaffuri, W.J. Pitz, C.K. Westbrook,Combust. Flame 129 (2002) 253–280.

[3] R.W. Walker, C. Morley, in: M.J. Pilling (Ed.), Low-Temperature Combustion and Autoignition, vol. 35, El-sevier, Amsterdam, 1997, p. 1.

[4] A. Roubaud, R. Minetti, L.R. Sochet, Combust. Flame121 (2000) 535–541.

[5] G. Vanhove, M. Ribaucour, R. Minetti, Proc. Combust.Inst. 30 (2005) 1065–1072.

[6] S. Klotz, K. Brezinsky, I. Glassman, Proc. Combust.Inst. 27 (1998) 337–344.

[7] J. Andrae, D. Johansson, P. Bjornbom, P. Risberg, G.Kalghatgi, Combust. Flame 140 (2005) 267–286.

[8] S. Tanaka, F. Ayala, J.C. Keck, Combust. Flame 133(2003) 467–481.

[9] G. Vanhove, G. Petit, R. Minetti, Combust. Flame 145(2006) 521–532.

[10] G. Mittal, C.J. Sung, Combust. Sci. Technol. 179(2007) 497–530.

[11] G. Mittal, C.J. Sung, Combust. Flame 145 (2006) 160–180.

[12] G. Mittal, C.J. Sung, Combust. Flame 150 (2007) 355–368.

[13] http://www.mae.case.edu/facilities/cdl/projects/rapidcomp/rapiddatabase.

[14] M. Ribaucour, R. Minetti, L.R. Sochet, Proc. Combust.Inst. 27 (1998) 345–351.

[15] S. Touchard, R. Fournet, P.A. Glaude, V. Warth, F.Battin-Leclerc, G. Vanhove, M. Ribaucour, R. Minetti,Proc. Combust. Inst. 30 (2005) 1073–1081.

[16] W. Metcalfe, W.J. Pitz, H.J. Curran, J.M. Simmie, C.K.Westbrook, Proc. Combust. Inst. 31 (2007) 377–384.

[17] C. Ellis, M.S. Scott, R.W. Walker, Combust. Flame 132(2003) 291–304.

[18] C.K. Westbrook, Combust. Sci. Technol. 20 (1979) 5–17.

[19] C.K. Westbrook, W.J. Pitz, Combust. Sci. Technol. 33(1983) 315–319.