Structure and Oxidation Activity Correlations for Carbon Blacksand Diesel SootLakshitha Pahalagedara,† Hom Sharma,‡,§ Chung-Hao Kuo,† Saminda Dharmarathna,† Ameya Joshi,∥

Steven L. Suib,† and Ashish B. Mhadeshwar*,‡,§,#

†Department of Chemistry, University of Connecticut, 55 North Eagleville Road, Unit 3060, Storrs, Connecticut 06269, United States‡Department of Chemical, Materials, and Biomolecular Engineering, University of Connecticut, 191 Auditorium Road, Unit 3222,Storrs, Connecticut 06269, United States§Center for Clean Energy Engineering, University of Connecticut, 44 Weaver Road, Unit 5233, Storrs, Connecticut 06268,United States∥Modeling and Simulation, Corning Incorporated, One Science Center Road, SP TD 01-1, Corning, New York 14831, United States

*S Supporting Information

ABSTRACT: This work focuses on a comprehensive investigation of structure−activity relationships for a diesel engine sootsample (Corning) and 10 commercially available carbon black samples. Particle sizes were determined using scanning electronmicroscopy (SEM) and transmission electron microscopy (TEM). Specific surface area was determined by nitrogen sorptionstudies, while the microstructure was investigated by X-ray diffraction (XRD) peak profile analysis, Raman spectroscopy, andTEM. Oxidation activity of these samples was studied using thermogravimetric analysis (TGA) under an oxidative (10% O2)environment consistent with the typical oxygen levels in the diesel engine exhaust. Various structural parameters, such as theaverage particle size, specific surface area, degree of organization, and average crystallite stacking height, were correlated with theTGA oxidation activity data. In general, samples with low particle size, high surface area, highly amorphous nature (low degree oforganization), and low crystallite stacking height showed high oxidation activity. A second diesel engine soot sample (Corning),which was collected at different operating conditions, was used to validate the obtained structure−activity correlations. Overall,our rigorous analysis for a large number of samples with multiple techniques indicated unique and novel correlations/trendsbetween soot structure and reactivity.

1. INTRODUCTION

Atmospheric aerosol particles are considered as air pollutants1,2

because of their impact on the environment, climate, and publichealth, including in vitro mammalian cell chromosomal and DNAdamage activities.3 Among these pollutants, diesel particulatematter (DPM) or soot, primarily composed of carbon, has asignificant adverse impact on global warming and health issues.1,4

DPM is typically trapped using diesel particulate filters (DPFs),which are periodically regenerated to oxidize the accumulated sootparticulates.1 Because of the highly complex nature,5,6 ambiguity,and unpredictability of the multicomponent soot structure, as wellas varying diesel engine operating conditions, optimized DPFoperation and regeneration is a challenging task.7 Because the DPFregeneration behavior is highly dependent upon the oxidationreactivity of soot, several studies have been conducted to explorethe effect of the structure on the soot oxidation activity.1,8,9

In previous studies, structure and activity of diesel soot andother carbonaceous materials have been investigated usingvarious characterization techniques. Su and colleagues usedhigh-resolution transmission electron microscopy (HRTEM)and thermogravimetric analysis (TGA) to study the relationbetween the microstructure and oxidation behavior of sootfrom exhausts of different heavy-duty diesel engines and discoveredthe microstructure-controlled oxidation behavior of diesel soot.10

Boehman et al. also used transmission electron microscopy(TEM) and HRTEM along with TGA and differential scanning

calorimetry (DSC) to study the relationship between the sootnanostructure and oxidation reactivity and, thereby, the DPFregeneration behavior.11 In addition to the TEM/HRTEM andTGA studies, Muller et al. used diffuse reflectance infrared Fouriertransform spectroscopy (DRIFTS) to correlate the reactivity withoxygen-containing functional groups and the nanostructure ofspark discharge soot, soot from a heavy-duty diesel engine, sootfrom a diesel engine in black smoking, and furnace carbon black.12

They concluded that the amount of defects and types of functionalgroups are important in determining the reactivity. Muller andcolleagues also studied the relation between the oxidative behaviorand the microstructure of black smoke soot using a diesel enginesoot sample (P1 soot) and a commercial carbon black sample(lamp black) and discovered that P1 soot is more aromatic andcontains higher surface functionality with lower oxidationtemperatures than lamp black.9 Song et al. used variouscharacterization techniques, such as TEM/HRTEM, X-raydiffraction (XRD), electron energy loss spectroscopy (EELS),Fourier transform infrared (FTIR) spectroscopy, Raman spectros-copy, and TGA, and determined correlations between somestructural properties, such as the degree of organization andnumber of stacking, and the oxidation rate constants. They also

Received: August 10, 2012Revised: October 10, 2012Published: October 11, 2012

Article

pubs.acs.org/EF

© 2012 American Chemical Society 6757 dx.doi.org/10.1021/ef301331b | Energy Fuels 2012, 26, 6757−6764

studied the oxidation mechanism of one of the soot samplesdescribing the importance of the presence of surface oxygengroups in addition to the initial structure and pore sizedistribution.8 Both Knauer et al. and Schmid et al. studied thesoot structure and reactivity correlations mainly with Ramanspectroscopic studies and temperature-programmed oxidationstudies using various soot samples from different engineconditions as well as commercial carbon black samples.4,13

These two groups discussed how the dispersive character ofRaman “D mode” can be used in structural analysis ofcarbonaceous material and how the multiwavelength Ramanmicroscopy (MWRM) can be applied in the prediction of dieselsoot oxidation behavior, respectively. Lapuerta et al. studied theeffect of the different engine conditions on soot structuralcharacteristics, such as primary particle size, Raman band arearatios (AD1/AG and AD3/AG), intensity ratios (ID1/IG and ID3/IG),crystallite stacking height (Lc), and crystallite length (La), using abiodiesel soot and a diesel soot with three different loading modes.Their studies revealed that biodiesel soot is higher in oxidationactivity than diesel soot and suggested that the internal graphiticstructure of biodiesel soot alone does not describe its oxidationbehavior and other analyses are needed for a satisfactoryexplanation.14

The aforementioned literature studies have focused on limitedstructural characterization of diesel soot formed under variousdiesel engine operating conditions,4,8,14,15 as well as commerciallyavailable carbon black materials.9,10,13 However, a fundamentalunderstanding about the direct correlations between the importantstructural parameters and the reactivity under oxidative environ-ments using a large number of samples and multiple character-ization methods is lacking. To address this gap, we focus on thestructural characterization of 2 different diesel engine soot samplesand 10 commercial carbon black samples using six techniques, viz.,scanning electron microscopy (SEM), TEM/HRTEM, nitrogensorption, Raman spectroscopy, XRD, and TGA.Use of microscopic studies, such as field emission scanning

electron microscopy (FE-SEM) and TEM, is quite common instudies on soot and other carbonaceous materials. Thesetechniques have been used to study the aggregation patternsand morphology of primary particles, to calculate the averageprimary particle size, to analyze the microstructure of primaryparticles, etc.4,6,16,17 Nitrogen sorption methods, such as totalsurface area, pore structure,18 and pore size distribution,19

morphological changes upon chemical modification,20 anddefects on carbon surfaces21 have been widely used in structure-related studies of diesel soot and commercial carbon blacks.

Raman spectroscopy is a promising and sensitive technique forthe analysis of crystalline long-range order (e.g., graphiticmaterials22), as well as molecular structures with a short-rangeorder (e.g., amorphous carbon or disturbed graphitic lattices23).In the previous studies on soot using Raman spectroscopy, thestructure has been described considering either the relative Dband (stands for “defect”) and G band (stands for “graphitic”)intensities or the relative D band and G band areas.8,13 Theseratios have been quantitatively analyzed, because they reflectthe structural defects in the basal planes of graphene layers.14

XRD profile analysis for carbon blacks has been used in earlierstudies to obtain two structural parameters: crystallite stackingheight (Lc) and interplanar distance.24 The bands observed in theXRD patterns for diesel soot and other carbonaceous materialshave been identified as (002), (100), (004), and (110), and the(002) band has been used to calculate both the interplanar layerspacing (d002) of crystallites composed of stacked graphene layersusing Bragg’s law14 and the average crystallite height (Lc) using theDebye−Scherrer equation.25 Finally, TGA has been widely appliedin determining structural stability and direct studies on the kineticsof diesel soot oxidation based on the change in the weight of thecarbonaceous material.6,7,9

On the basis of SEM, TEM/HRTEM, nitrogen sorption,Raman spectroscopy, XRD, and TGA, we present fourstructure−activity correlations in this work, viz., (i) light-offtemperature versus SEM particle size, (ii) light-off temperatureversus surface area, (iii) light-off temperature versus RamanD1/G peak ratio, and (iv) light-off temperature versus XRDcrystallite height. These structure−activity correlations could becritical for the improved design and operation of DPFs.

2. EXPERIMENTAL SECTION2.1. Materials. A total of 2 diesel engine soot samples (diesel soot-1

and diesel soot-2) were provided by Corning (see Table S1 of theSupporting Information for engine operating conditions), whereas 10carbon black samples (Mogul-E, Monarch 280, Monarch 1300, Monarch1400, Printex-G, Printex-U, Printex-XE2B, Regal 330R, Regal 400R, andVulcan XC 72R) were obtained from various manufacturers/suppliers.Table 1 shows the physical properties and oxidation activity data for allof the samples based on the characterization techniques described next.

2.2. Characterization Techniques. 2.2.1. SEM. Morphologicaland particle size analyses of the carbon black materials and diesel sootsamples were carried out using Zeiss DSM 982 Gemini FE-SEM with abeam current of 1 mA and a Schottky emitter operating at 2 kV. FE-SEM sample preparation was performed by suspending the samples inabsolute ethanol, and then a drop of the suspension was dispersed onAu-coated silicon chips formerly mounted onto stainless-steel sample

Table 1. Structural Properties and Oxidation Activity Data for Carbon Black and Diesel Soot Samples

SNa name particle size and standard deviation (nm) surface area (m2/g) Raman D1/G intensity ratio Lc (Å) T10 (°C) T50 (°C)

1 diesel soot-1 68.5 (16.9) 155 1.39 11.7 541 6402 Mogul-E 69.7 (17.0) 49 0.99 14.0 548 6303 Monarch 280 73.0 (22.3) 46 0.80 13.3 617 6744 Monarch 1300 21.5 (2.8) 342 2.13 10.4 482 5635 Monarch 1400 20.9 (3.0) 443 1.52 11.5 513 6166 Printex-G 74.0 (14.0) 45 0.75 18.5 637 6967 Printex-U 68.6 (28.9) 96 2.07 10.9 543 6158 Printex-XE2B 26.7 (3.0) 1005 1.71 11.2 479 5479 Regal 330R 32.4 (7.6) 98 1.36 13.1 587 65910 Regal 400R 46.1 (8.4) 91 0.99 12.6 565 64111 Vulcan XC 72R 39.3 (5.1) 213 1.32 12.1 631 69212 diesel soot-2 40.5 (11.6) 351 1.67 7.7 548 620

aSN = serial number (the same sequence number is used throughout this paper).

Energy & Fuels Article

dx.doi.org/10.1021/ef301331b | Energy Fuels 2012, 26, 6757−67646758

holders with double-sided carbon tape. The average primary particlesize of carbon black and diesel soot samples was determined usingFE-SEM images obtained under 50 000 and 100 000 magnifications, bymeasuring the diameter of nearly 200 particles for each sample. Thestandard deviation for each sample is listed in Table 1.2.2.2. TEM. The HRTEM studies were carried out using a JEOL

2010 instrument with an accelerating voltage of 200 kV. The samples wereprepared by dispersing the material in ethanol. A drop of a homogeneous-like dispersion was loaded onto a carbon-coated copper grid and allowedto dry before analysis. The obtained TEM images were used to analyze themicrostructure of carbon black and diesel soot samples.2.2.3. Nitrogen Sorption Studies. The nitrogen sorption experi-

ments were conducted using a Quantachrome Autosorb iQ2 surfacearea system. Prior to the experiments, all samples were degassed at 200 °Cfor 12 h. The Brunauer−Emmett−Teller (BET) method was used todetermine the specific surface area of carbon black samples from dataobtained at P/Po between 0.05 and 0.30. The pore size distribution andpore volume were calculated from the desorption data using the Barrett−Joyner−Halenda (BJH) method.2.2.4. Raman Spectroscopy. Raman measurements were taken

at room temperature on a Renishaw 2000 Ramascope attached to acharge-coupled device (CCD) camera, with an Ar+ ion laser (514.4 nm)as the excitation source. Before each measurement was taken, thespectrometer was calibrated with a silicon wafer. Curve fitting for thedetermination of spectral parameters was performed with the softwareprogram GRAMS/32.2.2.5. XRD. XRD studies were performed on the carbon black and

diesel soot samples to investigate the structure. The patterns wereanalyzed with a Rigaku Ultima IV diffractometer using Cu Kα (λ =0.154 06 nm) radiation. The phases were identified using theInternational Center for Diffraction Data (ICDD) database. Theaverage crystallite height (Lc) of the carbon black and diesel sootsamples was determined using the Debye−Scherrer equation based onthe (002) peak.2.2.6. TGA. TGA experiments26 were performed in a 10% O2

atmosphere (balance Ar) using a TGA Q5000 IR analyzer from TAInstruments. The temperature ramp rate was 5 °C/min. The TGAcrucibles were made out of Pt, which was confirmed to be inert inthese experiments. Short and wide TGA crucibles (h/D = 0.2) wereused to avoid mass-transfer limitations. The initial mass of the carbonblack samples was ∼10−12 mg. Because of the low density of dieselsoot, an initial mass of only ∼2.5 mg was used. Light-off temperaturescorresponding to 10 and 50% conversion are denoted as T10 and T50,respectively. The TGA data were highly reproducible (within 5 °C), asreported in our earlier work.26

3. RESULTS

3.1. Effect of the Particle Size on Oxidation Activity.FE-SEM images (see Figure S1 of the Supporting Information) ofsome carbon blacks (Monarch 1400, Monarch 280, and Printex-U)and diesel soot-1 show the agglomerates, which are composed oftheir fundamental units called primary particles.7 The averageparticle size (diameter) was estimated by analysis of nearly 200particles in the FE-SEM images. The average particle size of all ofthe samples varied between 20 and 74 nm. Monarch 1400 (seeFigure S1a of the Supporting Information) and Monarch 1300(not shown) had the smallest average particle sizes (20.9 and21.5 nm, respectively). On the other hand, Monarch 280 (seeFigure S1b of the Supporting Information) and Printex-G (notshown) had the largest particle sizes (73 and 74 nm, respectively).Printex-U (see Figure S1c of the Supporting Information) anddiesel soot-1 (see Figure S1d of the Supporting Information) hadalmost the same average particle sizes (68.5 and 68.6 nm,respectively), supporting the idea of morphological similaritybetween the two materials.27

Because the samples selected in our work had a broaddistribution of average particle sizes, the oxidation activity data

obtained from TGA were correlated with the average particlesize, as shown in Figure 1. The data did not follow a clear

correlation (low R ∼ 0.4 with all points). Vulcan XC 72R andPrintex-XE2B were identified as outliers, because their T10values were different from the trend line by at least 10%).Nonetheless, a general trend (R ∼ 0.5 after excluding theoutliers) that the oxidation activity decreases (i.e., the light-offtemperatures T10 and T50 increase) as the average particle sizeincreases was observed, consistent with the literature studies.1

The average particle size of diesel soot-2 (see Figure S2 ofthe Supporting Information) was 40.5 nm, composed of smallerparticles than diesel soot-1. To validate the structure−activitycorrelations in Figure 1, we also included the diesel soot-2sample in the plots, which showed that the diesel soot-2 samplewas very close to the correlation trend lines.

3.2. Effect of the Surface Area on Oxidation Activity.Panels a and b of Figure 2 show the variation of light-offtemperatures T10 and T50, respectively, with the specific surfacearea. In the correlation plots, Vulcan XC 72R was againidentified as an outlier. The general trend (R ∼ 0.6 with allpoints, whereas R ∼ 0.8 after excluding the outlier) was that theoxidation activity increased (i.e., T10 and T50 decreased) as the

Figure 1. Correlation for oxidation activity with the average initialparticle size. Panels a and b correspond to the light-off temperaturesT10 and T50, respectively. Outliers are based on at least 10% deviationfrom the linear trend line based on all of the points. R values areprovided to indicate the qualitative linear trends after excluding theoutliers. R values with all points were (a) 0.38 and (b) 0.44.

Energy & Fuels Article

dx.doi.org/10.1021/ef301331b | Energy Fuels 2012, 26, 6757−67646759

specific surface area increased. The correlations for T10 and T50were equivalent, even though we expect that, after a 50% mass loss,the significance of the surface area should be lower because of lesscontribution by the inner coagulated crystallites to the total surfacearea.28 The specific surface area of diesel soot-1 and carbon blacksranged between 45 and 1005 m2/g, with Printex-G and Printex-XE2B showing the lowest and highest surface areas, respectively.Diesel soot-1 had a moderate specific surface area of 155 m2/g, andcorrespondingly, it had a moderate oxidation activity.The specific surface area of diesel soot-2 was 351 m2/g,

which was greater than that of diesel soot-1. To validate thesestructure−activity correlations in panels a and b of Figure 2, wealso included diesel soot-2 in the plots, which showed thatdiesel soot-2 was reasonably close to the correlation trend lines.3.3. Effect of the Degree of Organization on Oxida-

tion Activity. Raman spectra (λ = 514 nm) shown in Figure S3of the Supporting Information for carbon blacks and dieselsoot-1 consisted of two overlapping bands, one around 1600 cm−1

(G band) and the other around 1350 cm−1 (D band). Thecurve-fitted spectra (not shown) showed the presence of fivebands (G, D1, D2, D3, and D4 around 1580, 1345, 1620, 1500,and 1200 cm−1, respectively). From these spectra, a ratio of the

intensity of the D1 and G band peaks could be used as ameasure of the relative degree of organization, i.e., amorphousnature. Figure 3 shows the variation of T50 with the D1/G

intensity ratios of carbon blacks and diesel soot-1. The fairlylinear relationship (R = 0.64 with all points, whereas R = 0.65after excluding the outlier Printex-XE2B) indicated a strongdependence of the oxidation activity upon the degree oforganization of the carbonaceous material. Samples withhigher D1/G intensity ratios had lower T50 and vice versa.Here, the D1/G intensity ratio varied between 0.75 and2.13, with Printex-G and Monarch 1300 having the lowestand highest ratios, respectively. The D1/G intensity ratio fordiesel soot-1 was 1.39, which was considered a moderatevalue of the series, corresponding to moderate oxidationactivity.Raman spectrum of diesel soot-2 (see Figure S4 of the

Supporting Information) had a D1/G intensity ratio of 1.67. Tovalidate the structure−activity correlation in Figure 3, we alsoincluded diesel soot-2 in the plot, which showed that dieselsoot-2 was very close to the correlation trend line.

3.4. Effect of the Crystallite Stacking Height on Oxi-dation Activity. Figure S5 of the Supporting Informationshows the XRD patterns observed for commercial carbon blacksand diesel soot-1. The patterns consisted of two broad diffusepeaks, one indexed as the crystalline reflection (002) and theother as two-dimensional lattice reflections (10l), which could beconsidered as a poorly crystalline graphite X-ray pattern.Figure 4 shows the variation of T50 with the crystallite

stacking height (Lc). Samples with a lower Lc showed a higheroxidation activity and vice versa (R = 0.38 with all points,whereas R = 0.46 after excluding the outliers Vulcan XC 72Rand Printex-XE2B). Here, Lc ranged between 10.4 and 18.5 Å,with Monarch 1300 and Printex-G showing the lowest andhighest Lc, respectively. Diesel soot-1, which showed goodagreement with the observed trend, had crystallites with anaverage height of 11.7 Å and might be composed of 4 graphenelayers (d002 = 3.62 Å).

Figure 2. Correlation for oxidation activity with the specific surfacearea. Panels a and b correspond to the light-off temperatures T10 andT50, respectively. An outlier is based on at least 10% deviation from thelogarithmic trend line based on all of the points. R values are providedto indicate the qualitative logarithmic trends after excluding the outlier.R values with all points were (a) 0.58 and (b) 0.64.

Figure 3. Correlation for oxidation activity with the intensity ratio ofthe Raman D1 and G peaks. An outlier is based on at least 10%deviation from the linear trend line based on all of the points. The Rvalue is provided to indicate the qualitative linear trend after excludingthe outlier. The R value with all points was 0.64.

Energy & Fuels Article

dx.doi.org/10.1021/ef301331b | Energy Fuels 2012, 26, 6757−67646760

The crystalline reflection (002) of the XRD pattern of dieselsoot-2 (see Figure S6 of the Supporting Information) showedLc of 7.7 Å (d002 = 3.60 Å), which was the smallest of the seriesand outside the range for the 10 carbon blacks and diesel soot-1.To validate the structure−activity correlation in Figure S5 ofthe Supporting Information, we also included diesel soot-2 inthe plot, which showed that diesel soot-2 did not lie close to thecorrelation trend line. This indicates that the correlation ofoxidation activity with Lc is the poorest of the four correlationspresented in this work, as also seen from the low R values.

4. DISCUSSION

4.1. Particle Size. The morphology and microstructure ofcarbon black and diesel soot samples, along with their correla-tions with the oxidation behavior in Figure 1 can be explainedwith a careful analysis of their TEM and HRTEM images. As ageneral trend, smaller particles show lower T10 and T50 valueswith higher oxidation activities, whereas the larger particlesshow higher T10 and T50 values with lower oxidation activities.Spillover of the oxidant onto the soot surface followed byadsorption at the active carbon sites are found to be importantin many oxidation mechanisms proposed for soot oxidation.29

Hence, smaller particles, having a larger surface area/volumeratio, should have better contact with the oxidant, resulting inhigher oxidation activity. The average particle size influencesthe oxidation activity, but it is not a very clear descriptor in suchcorrelations; therefore, additional structural features are alsostudied and described next.

4.2. Surface Area. Even though the carbon gasificationmechanism is quite complex,30 the influence of the surface area onthe oxidation behavior of carbon black and diesel soot samples isexpected to provide important insights, because the chemisorptionof the oxidant on the active surface sites and the subsequentformation of oxygen surface complexes are expected to stronglydepend upon the specific surface area. In Figure 2, we proposed acorrelation for oxidation activity versus BET surface area. Toexplain the trend and low versus high surface areas of certainsamples (e.g., Printex-G versus Printex-XE2B), we propose thatthe primary particles are composed of two major parts, as shownin Figure 5: (i) an inner core made out of several fine particles andseveral carbon layers with a distorted structure and (ii) an outershell made out of microcrystallites with periodic orientation ofplanar graphene layers.17

The HRTEM image of Printex-G (see Figure S7 of theSupporting Information) shows the microstructure of individual

Figure 4. Correlation for oxidation activity with the crystallite stackingheight. Outliers are based on at least 10% deviation from the lineartrend line based on all of the points. The R value is provided toindicate the qualitative linear trend after excluding the outliers. The Rvalue with all points was 0.38.

Figure 5. Proposed structure of carbon black.

Energy & Fuels Article

dx.doi.org/10.1021/ef301331b | Energy Fuels 2012, 26, 6757−67646761

primary particles. The lattice fringe contrast from the stacking ofthe graphene units is clearly visible in the shell of the primaryparticle. Both the shell and outer surface of the particle show long-range ordered contrast, indicating a highly dense shell and a lessdefected surface, hence, a very low surface area. The low surfacearea of Printex-G makes it one of the samples with very lowoxidation activity.On the other hand, the overview HRTEM image of Printex-

XE2B (see Figure S8a of the Supporting Information) revealsthe microstructure of the primary particle shell, in which thelattice fringes are clearly visible but are with weak long-rangeorder. The curved graphene layers form a less dense shell,whereas shells of the neighboring particles appear to be merged,forming a continuous surface with a large number of smallcrystallites (see Figure S8b of the Supporting Information). Ahigher amount of crystallites confirms the presence of a highernumber of crystallite edges, which, in turn, results in a higherdensity of surface-active sites. The high surface area of Printex-XE2B is responsible for the highest oxidation activity.4.3. Degree of Organization. The agglomerated primary

particles are comprised of graphite-like crystalline and amorphousdomains,23 and carbon blacks change into coagulations ofcrystallites as oxidation proceeds.28 Therefore, it is important tounderstand the microstructure of the carbonaceous materials. Asmentioned earlier, Raman spectroscopy can be used for the analysisof crystalline long-range order (e.g., graphitic materials22) as well asmolecular structures with a short-range order (e.g., amorphouscarbon or disturbed graphitic lattices23). The curve-fitted Ramanspectra in Figure S3 of the Supporting Information for carbonblacks and diesel soot-1 contain one first-order band at around1585 cm−1 (G band) and four additional bands (D1, D2, D3, andD4 bands, represented by D band), which are characteristic for thedisordered graphite structure. The G band stands for the in-planebond-stretching motions of the sp2-hybridized carbon atoms of thegraphene plane, which has E2g symmetry.

4 The D1 band, whichappears at around 1345 cm−1, is associated with the graphitic latticemotions with A1g symmetry of carbon atoms from graphene layersclose to the lattice disturbances.23

In the correlation for oxidation activity versus degree oforganization (Figure 3), Monarch 1300 had the highest D1/Gintensity ratio and, hence, the lowest degree of organization. TheHRTEM image of Monarch 1300 in Figure S9 of the SupportingInformation shows twisted ribbon-like molecular units in additionto the basic structural units of graphene; they are strongly curved,indicating a less dense and highly disordered crystalline (highlyamorphous) structure. In some areas, the crystallites are not clearlyobserved because of the presence of disordered, amorphouscarbon. This structural feature of Monarch 1300 brings about afaster gasification mechanism and, hence, a high oxidation activity.On the other hand, Printex-G had the lowest D1/G intensity

ratio and, hence, the highest degree of organization. The HRTEMimage of Printex-G in Figure S7 of the Supporting Informationconsists of more pronounced long-range ordered onion-likegraphene structures with very low curvature for the graphiticsheets. The highly dense, less spaced, and ordered structure limitsthe gasification process and leads to low oxidation activity.Both diesel soot-1 and diesel soot-2 have moderate D1/G

intensity ratios (1.4 and 1.67, respectively); hence, they showmoderate oxidation activities. The HRTEM image of dieselsoot-1 in Figure S10 of the Supporting Information shows thepresence of primary soot particles with different shapes as wellas dark particles of metal oxides with clearly observed latticefringes. Overall, the D1/G intensity ratio based on Raman

spectroscopy is an excellent descriptor for the oxidation activityof the carbon black and diesel soot samples.

4.4. Crystallite Stacking Height. The carbon blackparticles are composed of densely packed stacks of almost perfectgraphite layers,24 and it is important to identify the effect ofstructural parameters of a single crystallite on the oxidationbehavior. XRD profile analysis for carbon blacks has been used inthe past studies to obtain two structural parameters: crystallitestacking height (Lc) and interplanar distance.

24 The XRD patternsconsist of two broad diffuse peaks, one indexed as the crystallinelattice reflection (002) and the other as two-dimensional latticereflections (100)/(101).31 The absence of (hkl) peaks indicatesrandom orientation around the layer normal and lack of three-dimensional order.24

Diesel soot-2 had the lowest Lc, whereas Printex-G had thehighest Lc. The HRTEM image of diesel soot-2 in Figure S11 ofthe Supporting Information shows that the primary particleshave highly curved graphene layers, which are loosely stackedon each other. The curvature, formed because of a less sp2

character of the graphene layers, discourages the formation of ahigher number of stacks per crystallite, which increases itsoxidation activity by inducing a faster gasification mechanism.Similar features are observed from the HRTEM image ofMonarch 1300 in Figure S9 of the Supporting Information.On the other hand, the HRTEM image of Printex-G in Figure S7

of the Supporting Information shows the closely packed graphenelayers with a very low curvature, indicating a higher sp2/sp3-hybridized carbon ratio. Overall, a larger crystallite stacking heightis associated with a smaller amount of surface carbons beingexposed to the oxidant; therefore, more energy is needed todissociate the graphene planes from the crystallite, resulting inelevated oxidation temperatures (i.e., low oxidation activity).

4.5. Outliers and Anomalies in the Structure−ActivityCorrelations. In the aforementioned structure−activitycorrelations, Vulcan XC 72R and Printex-XE2B were determinedas the most common outliers. These were determined on the basisof at least 10% deviation from the trend lines. Here, we provide apossible explanation for the different behavior of these two samples.Vulcan XC 72R (the first outlier) is much more difficult to

oxidize compared to the predicted oxidation activity based onthe correlations. The HRTEM images of Vulcan XC 72R inFigure S12 of the Supporting Information confirm the presenceof three types of primary particles: the first being the smallest(∼10 nm) and composed of curved, more spaced graphenelayers and agglomerated to form chain-like structures, thesecond being the most abundant and middle-sized (∼39 nm)(see Figure S12a of the Supporting Information), and the thirdbeing the largest (∼84 nm) and the least abundant (see FigureS12b of the Supporting Information). Both middle-sized and largeparticles are composed of planar, closely packed graphene layers.Although the relative abundance of middle-sized (∼39 nm)particles in the sample is the highest, the smallest (∼10 nm) andmore amorphous particles may have been more observed inRaman and XRD studies. Because of its amorphous nature, thesmaller primary particles may have a higher surface area, whichresults in a moderate surface area value (213 m2/g) for Vulcan XC72R. Because both the XRD and Raman peaks become narrowerwith the development of crystallographic structures,32 the peaksoriginating from larger, more crystalline primary particles mayhave been obscured inside the broad peaks, arising because of thesmallest (∼10 nm), more amorphous particles. Hence, a largerD1/G intensity ratio and a lower Lc value are observed for VulcanXC 72R.

Energy & Fuels Article

dx.doi.org/10.1021/ef301331b | Energy Fuels 2012, 26, 6757−67646762

On the other hand, Printex-XE2B (the second outlier) iseasier to oxidize compared to the predicted oxidation activitybased on the correlations. The TEM image of Printex-XE2B inFigure S8b of the Supporting Information shows that the shellof the primary particle is very narrow and the core has anamorphous nature. Although the Raman and XRD dataanalyses give information about the crystallinity of the shell,they do not represent depth analyses. Hence, the T50 value ofPrintex-XE2B, which includes both shell and amorphous coreoxidation, is lower than expected. Also, the very high surfacearea for Printex-XE2B (1005 m2/g), observed because of thehigh concentration of surface-active sites, might be responsiblefor its low T10 and T50 values.Besides the two outliers, an opposite trend is observed in

Figure 1a (T10 versus particle size) and Figure 2a (T10 versussurface area) for the two diesel soot samples. A number of volatileorganic materials and a layer of sulfuric acid droplets are typicallyadsorbed on the primary particle surface of diesel soot.1,5,6 Thismakes it challenging to correlate the initial oxidation activity (orT10) with the structural characteristics for diesel soot. Diesel soot-1had higher initial activity than that of diesel soot-2, even thoughthe primary particle size of diesel soot-1 was larger than that ofdiesel soot-2 (Figure 1a). However, as oxidation proceeded, bothparticles followed the expected trend at moderate conversion(Figure 1b); i.e., diesel soot-1 had lower oxidation activity (higherT50) than that of diesel soot-2. This might be due to the higheramount of volatile organic matter and sulfuric acid droplets boundto the primary particles of diesel soot-1. The same justificationcould explain the unexpected trend for T10 versus surface area forthe diesel soot samples (Figure 2a). The samples followed theexpected trend at moderate conversion (T50; Figure 2b). Futurework will focus on rigorous validation of the correlations/trendsbased on additional carbon black and diesel engine soot samples.4.6. General Trend. We combine the aforementioned

structure−activity correlations in Figure 6, which presents an

overall summary schematic of the dependence of oxidation activityupon the structural characteristics for all of the samples consideredin this work. On the basis of this summary schematic, a generaltrend could be explained as follows. Samples with lower D1/Gratios (<0.8) showed a lower oxidation activity (T50 > 670 °C),whereas samples with higher D1/G ratios (>1.5) showed a higheroxidation activity (T50 < 620 °C). Because none of the samples fellin the left bottom or right top regions, the Raman D1/G intensityratio is a good descriptor for the oxidation activity. In general,samples with low activity (high T50) are associated with low surfacearea, large particle diameter, and large crystallite stacking height(many graphene layers). On the other hand, samples with highactivity (low T50) are associated with high surface area, small particlediameter, and small crystallite stacking height (few graphene layers).

5. CONCLUSIONOur comprehensive investigation of 10 commercial carbon blacksamples and 2 diesel engine soot samples has provided novel andclear insights that explain the origin of their oxidation activitylinked to their structural characteristics. Four structure−activitycorrelations were investigated by analyzing their structuralcharacteristics, such as initial primary particle size, specific surfacearea, degree of organization (amorphous nature), and crystallitestacking height, and their oxidation activity using a diesel enginesoot sample (diesel soot-1) and 10 commercial carbon blackmaterials. Validity of these correlations/trends was proven by thestructure and reactivity analysis of the second diesel engine sootsample (diesel soot-2). Such a fundamental understandingcombined with the unique soot structure−activity correlationswill be key to the improved design and operation of DPFs.

■ ASSOCIATED CONTENT*S Supporting InformationEngine operating conditions (Table S1) and structuralcharacterization images/patterns (Figures S1−S12). This

Figure 6. Summary of the variation of oxidation activity as correlated to particle size, surface area, degree of organization, and crystallite stackingheight (Lc). The x axis represents light-off temperature T50 at moderate conversion, whereas the y axis represents the Raman D1/G intensity ratio.The average particle sizes of the carbon black and diesel soot samples are represented by the relative diameters of the circles. Three colors are used toindicate different surface area ranges (red, <50 m2/g; orange, 50−250 m2/g; and yellow, >250 m2/g). Lc values are represented by the relative heightsof the boxes inside the circles. On the basis of Lc, the numbers of graphene layers per crystallite are estimated and shown with four colors (gray, 3layers; green, 4 layers; blue, 5 layers; and black, 6 layers).

Energy & Fuels Article

dx.doi.org/10.1021/ef301331b | Energy Fuels 2012, 26, 6757−67646763

material is available free of charge via the Internet at http://pubs.acs.org.

■ AUTHOR INFORMATION

Corresponding Author*Telephone: 1-443-523-8609. Fax: 1-860-486-2959. E-mail:ashish.mhadeshwar@gmail.com.

Present Addresses#ExxonMobil Research & Engineering, 1545 US Hwy 22,Annandale, New Jersey 08801.

NotesThe authors declare no competing financial interest.

■ ACKNOWLEDGMENTS

This work was supported by Corning Incorporated and theDepartment of Energy (DOE) through Award DE-EE0003226(“Improving Reliability and Durability of Efficient and CleanEnergy Systems”). Hom Sharma acknowledges a Department ofEducation Graduate Assistance in Areas of National Need(GAANN) Fellowship for funding.

■ REFERENCES(1) van Setten, B. A. A. L.; Makkee, M.; Moulijn, J. A. Catal. Rev. Sci.Eng. 2001, 43 (4), 489−564.(2) Poschl, U. Angew. Chem., Int. Ed. 2005, 44 (46), 7520−7540.(3) Risom, L.; Dybdahl, M.; Bornholdt, J.; Vogel, U.; Wallin, H.;Møller, P.; Loft, S. Carcinogenesis 2003, 24 (11), 1847−1852.(4) Knauer, M.; Schuster, M. E.; Su, D.; Schlogl, R.; Niessner, R.;Ivleva, N. P. J. Phys. Chem. A 2009, 113 (50), 13871−13880.(5) Palotas, A. B.; Rainey, L. C.; Feldermann, C. J.; Sarofim, A. F.;Vander Sande, J. B. Microsc. Res. Tech. 1996, 33 (3), 266−278.(6) Clague, A. D. H.; Donnet, J. B.; Wang, T. K.; Peng, J. C. M.Carbon 1999, 37 (10), 1553−1565.(7) Stanmore, B. R.; Brilhac, J. F.; Gilot, P. Carbon 2001, 39 (15),2247−2268.(8) Song, J.; Alam, M.; Boehman, A. L.; Kim, U. Combust. Flame2006, 146 (4), 589−604.(9) Muller, J.-O.; Su, D. S.; Jentoft, R. E.; Wild, U.; Schlogl, R.Environ. Sci. Technol. 2006, 40 (4), 1231−1236.(10) Su, D. S.; Jentoft, R. E.; Muller, J.-O.; Rothe, D.; Jacob, E.;Simpson, C. D.; Tomovic, Z.; Mullen, K.; Messerer, A.; Poschl, U.;Niessner, R.; Schlogl, R. Catal. Today 2004, 90 (1−2), 127−132.(11) Boehman, A. L.; Song, J.; Alam, M. Energy Fuels 2005, 19 (5),1857−1864.(12) Muller, J.-O.; Su, D. S.; Jentoft, R. E.; Krohnert, J.; Jentoft, F. C.;Schlogl, R. Catal. Today 2005, 102−103, 259−265.(13) Schmid, J.; Grob, B.; Niessner, R.; Ivleva, N. P. Anal. Chem.2011, 83 (4), 1173−1179.(14) Lapuerta, M.; Oliva, F.; Agudelo, J. R.; Boehman, A. L. Combust.Flame 2012, 159 (2), 844−853.(15) Kirchner, U.; Vogt, R.; Natzeck, C.; Goschnick, J. J. Aerosol. Sci.2003, 34 (10), 1323−1346.(16) Cain, J. P.; Gassman, P. L.; Wang, H.; Laskin, A. Phys. Chem.Chem. Phys. 2010, 12 (20), 5206−5218.(17) Ishiguro, T.; Takatori, Y.; Akihama, K. Combust. Flame 1997,108 (1−2), 231−234.(18) Darmstadt, H.; Roy, C.; Kaliaguine, S.; Joo, S. H.; Ryoo, R.Microporous Mesoporous Mater. 2003, 60 (1−3), 139−149.(19) Ustinov, E. A.; Do, D. D.; Fenelonov, V. B. Carbon 2006, 44(4), 653−663.(20) Borah, D.; Satokawa, S.; Kato, S.; Kojima, T. Appl. Surf. Sci.2008, 254 (10), 3049−3056.(21) Darmstadt, H.; Roy, C. Carbon 2001, 39 (6), 841−848.(22) Jawhari, T.; Roid, A.; Casado, J. Carbon 1995, 33 (11), 1561−1565.

(23) Sadezky, A.; Muckenhuber, H.; Grothe, H.; Niessner, R.; Poschl,U. Carbon 2005, 43 (8), 1731−1742.(24) Peng, W. G.; Strauss, M.; Pieper, T.; Kilian, H. G. Mol. Phys.1993, 80 (2), 419−429.(25) Fujimoto, H. Carbon 2003, 41 (8), 1585−1592.(26) Sharma, H. N.; Pahalagedara, L. R.; Joshi, A.; Suib, S. L.;Mhadeshwar, A. B. Energy Fuels 2012, 26, 5613−5625.(27) Hansen, C. W.; Kamp, C. J.; Skoglundh, M.; Andersson, B;Carlsson, P.-A. J. Phys. Chem. C 2011, 115 (32), 16098−16108.(28) Kamegawa, K.; Nishikubo, K.; Yoshida, H. Carbon 1998, 36 (4),433−441.(29) Krishna, K.; Bueno-Lopez, A.; Makkee, M.; Moulijn, J. A. Appl.Catal., B 2007, 75 (3−4), 189−200.(30) Jeguirim, M.; Tschamber, V.; Brilhac, J. F.; Ehrburger, P. Fuel2005, 84 (14−15), 1949−1956.(31) Saenko, N. S. Phys. Procedia 2012, 23, 102−105.(32) Xue, P.; Gao, J.; Bao, Y.; Wang, J.; Li, Q.; Wu, C. Carbon 2011,49 (10), 3346−3355.

Energy & Fuels Article

dx.doi.org/10.1021/ef301331b | Energy Fuels 2012, 26, 6757−67646764