Impact of engine operating cycle, biodiesel blends and fuel impurities on soot production and soot characteristics

Combustion and Flame 198 (2018) 1–13

Contents lists available at ScienceDirect

Combustion and Flame journal homepage: www.elsevier.com/locate/combustflame

Impact of engine operating cycle, biodiesel blends and fuel impurities on soot production and soot characteristics Julie Schobing a, Valerie Tschamber a,∗, Alain Brillard a, Gontrand Leyssens a, Eduard Iojoiu b, Vincent Lauga b a b

Université de Haute-Alsace, LGRE EA 2334, F-68100 Mulhouse, France Renault Trucks – Volvo Group Trucks Technology, Powertrain Engineering, 99 route de Lyon, 69806 Saint-Priest Cedex, France

a r t i c l e

i n f o

Article history: Received 19 March 2018 Revised 5 June 2018 Accepted 31 August 2018

Keyword: Soot Biodiesel Oxidation Phosphorus Potassium

a b s t r a c t The impact of engine operating cycle, Biodiesel blends and fuel impurities on soot production and soot properties are evaluated in the present work. To this end, soot were produced on engine test bench and then collected inside a Diesel Particulate Filter (DPF). Two engine cycles (a Natural Loading and an Accelerated Loading) were tested. A standard Euro VI fuel blended with 7% of Biodiesel (B7) and a pure Biofuel (B100 RME EN 14214) were used. This latter was additivated with potassium and phosphorus at a low (B100+ ) or at a high (B100++ ) concentration. Soot characterization through elemental analyses, nitrogen adsorption, Raman spectroscopy, TGA and TPO experiments show that the engine operating cycle impact the soot reactivity through modifications of their texture and structure. Test bench experiments also show that increasing Biodiesel blend from B7 to B100+ divides by five the soot production. Moreover, soot obtained with B100+ are more reactive because of higher oxygen and ash content. When the inorganic content of the fuel is increased, few effects on the soot production are observed but the soot reactivity is significantly increased. In fact, analyses highlight that impurities present in the fuel are retrieved inside the soot composition and then catalyze their oxidation. K has a beneficial effect on both passive and active regenerations. On the contrary, P inhibits the active regeneration but has a significant catalytic impact on the C–NO2 –H2 O reaction. Finally, a numerical simulation allows to extract the kinetic constants of real B7- and B100+ -soot, whose values confirm the differences of the soot reactivity. © 2018 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

1. Introduction Nowadays, improving air quality is attracting more and more attention. In fact, the increase in industrialization and transportation has a negative impact on the daily environment because of the emission from mobile sources of several pollutants like greenhouse gases and particles. To limit such emissions, standards are becoming more and more stringent. Euro VI, which currently regulates Diesel heavy-duty trucks exhaust gas emissions, limits, among others, the emissions of NOx at 0.46 g/kWh and of particles in mass at 0.01 g/kWh but also in number at 6.0x1011 particles per kWh for a cold/warm WHTC (World Harmonized Transient Cycle) cycle [1]. To comply with these standards, vehicles are now equipped with a complex exhaust gas post-treatment line including an oxidation catalyst (DOC) for CO and hydrocarbons oxidation as well as NO oxidation, a Diesel particulate filter (DPF) and a catalyst for the selective catalyzer reduction of NOx (SCR). These devices have fur-

∗

Corresponding author. E-mail address: [email protected] (V. Tschamber).

ther to prove their durability over 7 years or 70 0,0 0 0 km in the case of heavy duty trucks. Another way to decrease pollutant emissions from Diesel engines is the use of Biodiesel. This fuel is obtained by transesterification of triglycerides contained in vegetable oils or animal fats and is characterized by its high oxygen content [2,3]. Besides complying with the European Directive 2009/28/CE promoting renewable energies [4], the use of Biodiesel as vehicle fuel has globally a positive impact on pollutant emissions, as it reduces the overall carbon footprint and soot production [2,3,5–7]. Salamanca et al. [6], indeed observed that B100 produces three times less soot than conventional Diesel. Shukla et al. [3] showed that a 20% Biodiesel fuel (B20) can be sufficient to decrease soot production by 50% in comparison to B0. Studying a DPF cross-section, Liati et al. [7] showed that, in contrast with B0 and B20 fuels, no soot cakes were formed inside DPF channels with B100 fuel. The low soot production of Biodiesel is generally attributed to its high oxygen content which allows a better combustion of the fuel but also of the soot produced [2,3,8–10].

https://doi.org/10.1016/j.combustflame.2018.08.025 0010-2180/© 2018 The Combustion Institute. Published by Elsevier Inc. All rights reserved.

2

J. Schobing et al. / Combustion and Flame 198 (2018) 1–13

Biodiesel is also known to produce more reactive soot than conventional Diesel [9,11–13]. Thus the regeneration of DPF could be easier and quicker when Biodiesel is used [14]. Song et al. [11,12], studying B0 and B20 soot, showed that B20 soot are more reactive under air than B0 soot because of their more amorphous carbon structure. Lamharess et al. [13], studying B0 and B30 particles, and Yehliu et al. [15], using B15 and B100 soot, also concluded that soot reactivity is linked to their structure. According to Yehliu et al. [15], soot oxidation reactivity is not dominated by the abundance of surface oxygen content but rather by the disorder of the carbonaceous nanostructure and accessible carbon atoms on the edge sites which are the most active sites. On the contrary, LöpezSuarez et al. [9] attributed the highest reactivity of B100 soot to their highest oxygen and ash contents. In fact, inorganic elements, and in particular alkali metals, present in the ash can act as catalysts toward carbon oxidation [16–19]. Biodiesel is distinguishable from conventional Diesel fuel by its inorganic content. These impurities, mainly alkali metals, are coming from catalysts used for the transesterification. Although the presence of these elements inside the Biofuel are regulated by European standards [20], when Biodiesel is used on long driving distance, the exhaust gas line can be perturbed by high amounts of inorganic elements. Captive fleets of buses and trucks currently operate exclusively on Biodiesel (B100). In this case, based on an average consumption of 35 L/100 km for a truck exclusively running with Biodiesel, the exhaust gas post-treatment system will be subjected to about 1 kg of alkali metals and 1 kg of phosphorus after 70 0,0 0 0 km. These very high quantities generate a high challenge in achieving the durability, requested by the regulation of post-treatment systems (and in particular of the DPF), even for lower levels of alkali and phosphorus. In this context, some authors studied the impact of Biodiesel impurities on DPF properties in accelerated aging by fuel-doping [21,22]. Brookshear et al. [21] tried to simulate a 70 0,0 0 0 km driving by doping the fuel by Na and S (50 0 0 ppm Na + 50 0 0 ppm S) with dioctylsulfosuccinate. After test bench operating, they observed some ash clogging of DPF channels and the penetration of Na inside the wall. Williams et al. [22] doped B20 fuel with either Ca, Na or K (with respectively calcium napthenate, dioxtyl sulfonate sodium salt or potassium dodecylbenzene) to represent a 240,0 0 0 km driving exposition. They observed no evidence of supplementary cracking or corrosion due to Ca, Na or K. These studies focused on the impact of inorganic elements on DPF operating but not on soot properties. In another study [23], A. Williams et al. investigated which Biodiesel properties impact the soot reactivity. They tested different Biofuels, one of them being a B20 fuel doped with sodium oleate at 7 ppm. After soot production on a test bench, the soot reactivity was tested by TGA under an air flow. These authors concluded that neither lower aromatic levels nor the presence of alkali metals are determinant contributors to faster DPF regeneration when using Biodiesel. On the contrary, the presence and form of fuel oxygen were proved to be determinant contributors. Using 20% Biodiesel blended into ultra-low sulfur Diesel (B20) doped with 14 ppm Na, Lance et al. [24] intended to simulate a 70 0,0 0 0 km aging. They concluded that B20-Na resulted in 50% more ash into the DPF. However, the Na did not diffuse into the cordierite DPF nor degrade its mechanical properties. So far, no paper discussing the impact of given elements present in the fuel on soot characteristics and reactivity under passive regeneration conditions seems to be published. The purpose of the present work is to evaluate the impact of Biodiesel blends, and especially Biodiesel impurities, on soot properties. Three fuels are thus tested: commercial 7% and 100% Biodiesel and the 100% Biodiesel doped with K and P to simulate an accelerating aging of soot. The impact of the engine operating cycle is also investigated testing two different cycles: a representative real driving cy-

cle and an accelerated load one. After test bench production, soot were collected inside the DPF and then characterized by elemental analyses (FAAS, ICP and CHONS), nitrogen physisorption, thermogravimetric analyses (TGA) and temperature programmed oxidation (TPO). Finally, a model of carbon oxidation is proposed to extract the kinetic constants associated to the reaction. This work is part of a bigger project named AppiBio which intends to determine the impact of Biodiesel impurities on the whole exhaust gas line posttreatment system [19]. 2. Experimental conditions 2.1. Fuel doping Two types of fuel: 7% Biodiesel (standard Euro VI fuel) and 100% Biodiesel (B100 RME EN 14214 from TOTAL) were used to study the impact of the Biodiesel blends on soot production and their characteristics. One aim of this work is also to study the impact of Biodiesel impurities. But fuel analyses proved that the initial B100 used in this study is very clean (Table 1). On a short time soot production, these inorganic traces did not allow to appreciate their impact on soot properties. It was thus decided to enrich the fuel in inorganic elements by an external doping to simulate an accelerated aging. The challenge was to find precursors which are soluble in Biodiesel, which do not react between them and finally which are free from sulfur and chlorine. Taking into account these restrictions, it was only possible to dope simultaneously the fuel by phosphorus and potassium, using triethylphosphate ((Et)3 PO4 ) and a KOH solution in methanol (1 M KOH). Two concentrations were chosen: a low one corresponding to 0.06 times lower than the high limit of the standard for K and equal to the high limit of the standard for P and a high concentration equals to, respectively, 6 and 18 times the high limit of the standard for K and P. Elemental analyses were performed after the fuel doping to control the obtained concentrations. Results are given in Table 1. It was also verified that the density and viscosity of the doped fuels are good. 2.2. Soot production The production of real Diesel soot was carried out at Renault Trucks on an engine bench equipped with 8 L Euro VI engine. Soot were collected from the DPF by air blowing with a protocol allowing the recovery of two thirds of the total soot mass. Two types of operating engine cycles were tested: a Natural Loading cycle and an Accelerated Loading cycle. The first one is a low loading cycle operating at low temperatures which is representative of very severe cold real driving cycles. To keep real conditions, a DOC was placed upstream the DPF. Because this cycle produced low amounts of soot, a smoking patch (increased soot engine out) was sometimes applied to ensure the collection of sufficiently large amounts of soot. The smoking patch consists in a specific engine calibration which artificially increases the engine out soot emissions. The smoke increase induced by this specific calibration consists in increasing the EGR (exhaust gas recirculation) rate, decreasing the rail pressure, using no post injection and for the transient phases using less fuel limitation and no smoke limitation. The second cycle is an Accelerated Loading cycle which produced large amounts of soot. This Accelerated Loading cycle was used for fuels which are known to produce low amounts of soot (100% Biodiesel). To promote the soot production, no DOC was placed upstream the DPF for this cycle. Thus, four soot samples were produced on the engine test bench with these two engine cycles and three different fuels. Their denominations and associated operating conditions are summa-

J. Schobing et al. / Combustion and Flame 198 (2018) 1–13

3

Table 1 Fuel standards EN 14,214 and B100 inorganic composition before and after doping. Elements

High standard EN 14214 (ppm) [20]

Raw B100 fuel (ppm)

Doped B100+ fuel (ppm)

Doped B100++ fuel (ppm)

Phosphorus Potassium Sodium Calcium Magnesium

4,0 5,0 (K + Na) 5,0 (Ca + Mg)

0,1 < 1,0 < 1,0 < 1,0 < 1,0

4 0,3 < 1,0 < 1,0 < 1,0

72 28 < 1,0 < 1,0 < 1,0

Table 2 Soot produced on the test bench. Soot B7-Natural Loading B100+ -Natural Loading B100+ -Accelerated Loading B100++ -Accelerated Loading

DOC √ √ ✗ ✗

Fuel

K and P doping

Cycle

B7 B100+ B100+ B100++

✗ Low Low High

Low Temperature Low Temperature Accelerated Loading Accelerated Loading

rized in Table 2. Soot collections were realized, for the four samples, after the same engine operation duration equals to 20 h. During the soot production with the low temperature cycle, the average exhaust gas temperature upstream the DPF was about 160 °C. These conditions are inadequate to allow any efficiency of the DOC towards oxidation of gaseous compounds and especially the oxidation of NO into NO2 [19]. With the Accelerated Loading cycle, this temperature went up to 300 °C. In agreement with its nature, more soot are naturally produced with the Accelerated Loading cycle than with Natural Loading cycle, with the same fuel. The use of B100+ , instead of B7 on the Natural Loading cycle, divides by 5 the production of soot in mass. This behavior is in agreement with Biodiesel properties generally described in the literature [3,6,7]. Under the accelerated load cycle, the soot masses obtained with the two doped fuels are similar.

2.3. Soot characterization The elemental composition of the different samples was measured by CHONS (note that Oxygen content was measured and not calculated). Flame Atomic Absorption Spectroscopy (FAAS) after microwave digestion at 260 °C was used for sodium (Na), potassium (K), calcium (Ca), zinc (Zn), iron (Fe) and magnesium (Mg) contents. Inductively Coupled Plasma-Optical Emission Spectroscopy (ICP-OES), after microwave digestions was used for phosphorus (P) content. The entire protocol was described in [18]. The textural soot properties such as specific surface area, micropore surface area and micropore volume were determined by N2 adsorption-desorption isotherms which are obtained using a Micromeritics ASAP2020 device. BET method was used to calculate surface specific area and t-plot method for microporous volume. Raman spectra were recorded on a Horiba Labram 300 apparatus using a 532 nm laser. A scanning range of 2200 cm−1 was chosen. The spectrometer includes a grating with 600 grooves/mm and a CCD detector with 50x magnification objective lens. Decomposition of the spectra was done according to Sadezky et al. [25] method. Thermogravimetric Analyses (TGA) were performed on a TA Q500 thermobalance under air or nitrogen up to 700 °C. All TGA were performed with about 5 mg of soot and under a heating ramp of 5 °C/min. TGA experiments under air intend to determine ash content in comparison with TGA experiments under nitrogen. TGA experiments under nitrogen attempt to measure volatiles content as described by Zinbo et al. [26]. With ash and volatiles content, the carbon content of real Diesel soot can be calculated as shown in Fig. S1 of the supplementary data [26,27]. This carbon content

allows the calculation of the carbon mass (mcini ) initially introduced in the reactor during reactivity test. 2.4. Soot reactivity Soot reactivity under active regeneration was evaluated through TGA experiments under air (protocol described above). Soot reactivity under passive regeneration conditions was studied performing Temperature Programmed Oxidation (TPO) tests under different atmospheres: 400 ppmv NO2 in N2 (noted NO2 ), 400 ppmv NO2 + 10 vol% O2 in N2 (noted NO2 + O2 ), 400 ppmv NO2 + 10 vol% H2 O in N2 (noted NO2 + H2 O) or 400 ppmv NO2 + 10 vol% O2 + 10 vol% H2 O in N2 (noted NO2 + O2 + H2 O). TPOs were performed using a tubular fixed bed reactor (internal diameter 16 mm) following the protocol described in [18]. The total flow rate through the reactor was 90 Nl/h. 15 mg of carbon sample were diluted in 200 mg of SiO2 to avoid heat transfer and reactor clogging. Reactive gas flow was injected at ambient temperature. Then temperature was increased up to 700 °C under a 5 °C/min temperature ramp. The outlet gases were analyzed using an infrared Rosemount Xstream analyzer to quantify the outlet NO, NO2 , CO and CO2 molar fractions. The reproducibility of the TPO profiles was verified (not shown here). From each experiment, the carbon specific oxidation rate (Vspe in mg/s/gcini ) was calculated from CO and CO2 emissions (XCO and XCO2 in ppmv) through equation below:

Vspe =

(XCO + XCO2 ) ∗ D ∗ MC 106 ∗ 3600 ∗ VM ∗ mCini

(1)

where Vspe is expressed per gram of carbon initially introduced in the reactor (mcini ), D corresponds to the flow rate in Nl/h and VM is the molar volume (VM = 22.4 L/mol). 2.5. Modeling The model which has been used to extract the kinetic constants in the present context is based on a model developed in previous works [28–30], initially developed for isothermal conditions. The model was adapted to the ramp temperature conditions of TPO experiments for the oxidation of model Diesel soot doped with inorganic elements [18]. As the model only considers passive regeneration conditions, it was applied on the temperature range 150– 450 °C. The NO/NO2 equilibrium was not included in the model as it has been verified by performing a blank experiment that until 500 °C no NO2 is spontaneously converted into NO. The bed, which lies between 0 and h > 0, is decomposed in Nz thin layers of equal thinness dz = h/Nz .

4

J. Schobing et al. / Combustion and Flame 198 (2018) 1–13

According to the global mechanism proposed by Jacquot et al. in [28], the following empirical kinetic law for soot oxidation by NO2 with O2 and H2 O present in the gas phase is written as below equation:

−

1 d (δ m ) = kCO2 PNO2 + kCO PNO2 + δ m dt   kO2 CO2 PNO2 XO02.3 + kO2 CO PNO2 XO02,3 ,

(2)

where: - δ m is the local soot mass (kg), - kCO2 , kCO , kO2 CO2 and kO2 CO are the kinetic constants (Pa−1 s−1 ) which have to be determined, - PNO2 is the local partial NO2 pressure (Pa) defined as PNO2 = XNO2 Ptot , where XNO2 is the NO2 mole fraction and Ptot is the total pressure in the fixed bed reactor: Ptot ≈ 1,2 × 105 Pa. In each layer of the bed, the balance for each gaseous species leads to the following equation:

⎧ ∂ XNO2 Vm 2r + r + r + r ⎪ ⎪ (z, t ) = − (T ) 1 2 3 4 XNO2 (z, t ) m(z, t ), ⎪ ∂ z D Mc ⎪ V ⎪ ⎪ ⎪ ∂ XNO z, t = Vm T 2r1 + r2 + r3 X ⎨ ( ) ( ) NO2 (z, t ) m (z, t ), ∂z DV Mc ∂ X V r + r ⎪ ⎪ CO2 (z, t ) = m (T ) 1 3 XNO2 (z, t ) m(z, t ), ⎪ ⎪ ∂z DV Mc ⎪ ⎪ ⎪ ⎩ ∂ XCO (z, t ) = Vm (T ) r2 + r4 XNO2 (z, t ) m(z, t ), ∂z DV Mc (3) with

r1 r2 r3 r4





= kCO2 Ptot 1 + aXHq 2 O  μ  = kCO Ptot 1 + bXH2 O   = kO2 CO2 Ptot XO02,3 1 + cXHq 2 O   = kO2 CO Ptot XO02,3 1 + aXHq 2 O

XH2O here means the catalytic effect of water on carbon oxidation. a, b and c are subgrade coefficients reaction and γ , q, μ are reaction orders. Systems (2) and (3) are completed with the following initial conditions :

XNO2 (z, 0 ) = 0, 0,

XCO2 (z, 0 ) = 0, 0,

XNO (z, 0 ) = 0, 0, m0 m(z, 0 ) = , Nz

XCO (z, 0 ) = 0, 0, (4)

where m0 is the initial soot mass contained in the bed. The boundary conditions at the entrance of the bed (z = 0) are :

XNO2 (0, t ) = XNO2 , 0 , XNO (0, t ) = 0, 0

XCO2 (0, t ) = 0, 0, XCO (0, t ) = 0, 0,

(5)

where XNO2 , 0 represents the imposed NO2 concentration at the entrance of the bed, which is a constant expressed in ppm and which depends on the experiments. Problems (1)–(4) are solved using an explicit Euler method in an alternate way with initial values of the kinetic constants. An optimization procedure is used which consists to minimize the difference (hereafter called error) between the experimental and simulated data at regularly disposed computational times tj . For example, under a NO2 atmosphere, this error is taken as equation below: error

⎛

 ⎞ 

       2      2   t j  ⎟ ⎜ XNO2 exp t j − XNO2 sim t j  +  XCO2 exp t j − X C = sup⎝ O2 ⎠. sim      j 2 +mexp t j − msim t j 

(6)

3. Results and discussion 3.1. Soot characteristics 3.1.1. Ash and volatile contents Ash and volatile contents of the soot samples, determined through TGA experiments under air and nitrogen, are given in Table 3. For a given fuel (B100+ ), the Natural Loading cycle produces soot with higher volatile and ash contents when compared to the Accelerated Loading cycle. One may note that the exhaust gas temperature upstream DPF was about 160 °C with the Natural Loading cycle and went up to 300 °C with the Accelerated Loading cycle. In this last condition, in-situ oxidation of trap soot could occur. As it is known that the first step of soot oxidation, at around 300 °C, consists to the volatilization of condensables [31–33], one may attribute the lower volatile content of the B100 + -Accelerated Loading soot sample, compared to the B100 + -Natural Loading one, to a partial in-situ oxidation during the collection process. Considering the impact of the use of Biodiesel, for the Natural Loading cycle, an increase in Biodiesel blends in the fuel leads to an increase of the soot ash percentage. In fact, B100+ -Natural Loading soot contains four times more ashes than B7-Natural Loading soot. As the engine operating cycle and the exhaust gas temperature are the same for these two soot collections, this increase of ash content may be attributed to the nature of the fuel used. This is in agreement with the presence of inorganic elements in Biodiesel fuel [2,7,9]. Also with this cycle, B100+ produces soot with higher volatile content. Even though it is generally admitted that Biodiesel soot are less harmful according to their low concentration of adsorbed HAPs [2,14], some authors have already observed high volatile contents in Biodiesel soot composition [8,34]. As observed for the Natural Loading cycle, for the Accelerated one, an increase in the inorganic element contents of the fuel leads to the production of soot with more ash: B100++ -Accelerated Loading sample contains 4.62 wt% ash while B100+ -Accelerated Loading soot only 1.53 wt%. Slightly higher volatile content is also observed when B100++ fuel is used. Thus, whatever the production cycle, increasing the inorganic concentration of the fuel leads to an increase of the ash and volatile contents of the soot produced. 3.1.2. Elemental composition Elemental composition of soot determined by CHONS, AA and ICP analyses are given in Table 3. Soot are mainly composed of carbon (> 80 wt%) and C content measured by CHONS analyses is in agreement with the percentage determined by TGA (Table 3). The four soot samples contain about 1 wt% of hydrogen and no nitrogen was detected. When the Biodiesel fraction in the fuel increases, for a given cycle, the oxygen content of the soot increases. In fact, B7-Natural Loading soot contains 3.8 wt% of O and B100+ -Natural Loading sample, 6.2 wt%. This is in agreement with the oxygenated character of Biodiesel [2]. Increasing the inorganic part of the fuel also increases the oxygen content of the soot. Soot produced with the Natural Loading cycle do not seem to contain any traces of sulfur. On the other side, soot produced with the accelerated cycle contain about 1 wt% of sulfur. As B100+ Natural Loading soot does not contain sulfur, this element does not seem to come from the fuel. It appears that the presence or not of sulfur is linked to the presence or not of the DOC device in the exhaust gas line. In fact, if a DOC is placed upstream the DPF, it may retain sulfur compounds. The four soot samples contain similar sodium contents. Surprisingly, the highest sodium concentration is observed for the B7 soot (0.51 wt%). This sodium element may come from the lubricant oil.

J. Schobing et al. / Combustion and Flame 198 (2018) 1–13 Table 3 Ash, Volatile and Elemental composition of soot (1 TGA under air, +

2

TGA under N2 ,

3

CHONS,

5

4

AA,

5

+

ICP, n.d. = non detected).

Soot

B7-Natural Loading

B100 -Natural Loading

B100 -Accelerated Loading

B100++ -Accelerated Loading

Ash (wt%)1 Volatile (wt%)2 Carbon (wt%) C (wt%)3 H (wt%)3 O (wt%)3 N (wt%)3 S (wt%)3 Na (wt%)4 K (wt%)4 Ca (wt%)4 Zn (wt%)4 Fe (wt%)4 Mg (wt%)4 P (wt%)5

1.84 5.71 92.45 89.5 ± 0.8 1.2 ± 0.1 3.8 ± 0.1 n.d. n.d. 0,51 0,10 1,57 0,13 0,04 0,09 0,05

7.78 8.67 83.55 83.1 ± 0.3 1.3 ± 0.1 6.2 ± 0.1 n.d. n.d. 0,21 0,10 1,62 0,26 0,42 0,49 0,19

1.53 7.67 90.80 84.4 ± 0.7 0.7 ± 0.1 7.3 ± 0.1 n.d. 0.9 ± 0.4 0,30 0,03 1,16 0,04 0,03 0,19 0,05

4.62 8.20 87.18 81.3 ± 0.3 0.8 ± 0.1 9.4 ± 0.5 n.d. 1.2 ± 0.5 0,33 1,17 0,24 0,05 0,02 n.d. 0,89

Table 4 Textural properties of soot. Soot

Specific surface area (m²/gc )

Microporous specific surface area (m²/gc )

Microporous volume (cm3 /gc )

B7-Natural Loading B100+ -Natural Loading B100+ -Accelerated Loading B100++ -Accelerated Loading

254 251 278 291

– 109 229 62

– 4.72.10−2 9.11.10−2 2.81.10−2

B7 and B100+ soot contain less than 0.1 wt% of potassium. Thus, it seems that the %K of B100+ fuel doping (0.3 ppm, Table 1) is not sufficient to impact the soot composition on short-time engine test bench soot production. On the other hand, for the higher doped fuel (B100++ ), potassium enrichment is high enough to impact soot composition. Indeed, B100++ -Accelerated Loading soot contain 1.17 wt% of K. Fuel phosphorus doping is also retrieved in soot composition. In fact, for the Natural Loading cycle, B7 soot contains 0.05 wt% of P while this P content rises to 0.19 wt% in B100+ soot. For the Accelerated Loading cycle, P content ratio in soot composition is almost the same than P content ratio in fuel. In fact, B100++ fuel contains 20 times more phosphorus than B100+ fuel, while B100++ -Accelerated Loading sample contains 18 times more P than B100+ -Accelerated Loading soot. It has been shown in the previous work [18] that when impregnating model soot by phosphorus, P is present in the form of phosphate linked at the soot surface by C–O–P binding. The higher %O measured in the soot, when P is also present, are in agreement with the phosphate species hypothesis. Calcium concentration is higher than 1 wt% for all samples. Only B100++ -Accelerated Loading soot shows a lower Ca content (0.24 wt%). As this element preferentially originates from lubricant oil [2], it seems that B100++ fuel has an impact on oil consumption. Likewise, lower Zn content is also found in soot produced with the accelerated cycle. Finally, Fe and Mg contents are low for all samples, except for B100++ -Accelerated Loading sample which contains more than 0.4 wt% of these elements. 3.1.3. Textural properties The total and microporous specific surface areas and the microporous volume of soot samples, determined by nitrogen physisorption, are given in Table 4. The values are normalized per gram of carbon present in the sample (mcini ). The BET specific surface area of the different samples vary from 251 to 291 m2 /g which is in agreement with values given in the literature [32,35,36]. An impact of the operating engine cycle on the textural properties of soot is observed. In fact, higher specific surface areas are obtained with the accelerated load cycle: 251 m²/gc for B100+ -Natural Loading soot against 278 m²/gc for

B100+ -Accelerated Loading sample. The part of microporous specific area is also doubled with the Accelerated Loading cycle. These behaviors are again in agreement with a partial insitu oxidation during the collection process. Indeed some authors found evidence for increases in soot surface area during oxidation [31,35,37]. According to Kandas et al. [31], the greatest increase in surface area of Diesel soot occurs during the initial volatilization step of condensables and this increase can reach 50% of the initial surface area. Moreover, while soot is generally considered as a nonporous material, several authors reported the existence of porosity and its development during the oxidation process [31,32,36,38]. According to Ishiguro et al. [32], porous structure is formed by the release of SOF occluded within the soot particles. Furthermore, Wicke and Grady [38] observed an increase in the micropore structure of soot during the thermal desorption of adsorbed oxygen leading to the conclusion that oxygen atom adsorption occurs throughout the soot particles and not just on the surface. In a more recent work, Kandas et al. [31] indicated that during soot combustion in air, the structural evolution is the consequence of firstly the volatilization of condensables providing access to the porous structure then oxidation further opens up porosity leading to complete access to the internal surfaces. In the present study, the total surface area of B100+ -Accelerated Loading soot is only 10% higher than that of B100+ -Natural Loading soot. Thus, it seems that although a partial in-situ oxidation of soot may occur during the Accelerated cycle occurs leading to a modification of soot composition and properties, this oxidation however seems to be relatively limited. For the Natural Loading cycle, increasing the Biodiesel fraction does not impact the total surface specific area but increases the microporous area in the soot. Indeed, no micropores were detected in the B7-Natural Loading soot whereas B100+ -Natural Loading sample has a microporous volume of 4.72.10-2 m3 /gc . Considering the impact of the presence of inorganic compounds in the fuel, one may observe that when K and P contents increase in the fuel, a significant decrease in the microporous volume is observed. In a previous study on impregnated model soot, the surface specific area of carbon black doped with phosphorus was

6

J. Schobing et al. / Combustion and Flame 198 (2018) 1–13 Table 5 ID /IG ratios of soot. Ratio

B7-Natural Loading

B100+ -Natural Loading

B100+ -Accelerated Loading

B100++ -Accelerated Loading

ID1 /IG ID2 /IG ID3 /IG ID4 /IG

1.66 ± 0.04 0.76 ± 0.01 0.54 ± 0.03 0.18 ± 0.01

1.68 ± 0.12 0.66 ± 0.10 0.50 ± 0.04 0.19 ± 0.01

1.41 ± 0.07 0.76 ± 0.09 0.41 ± 0.03 0.11 ± 0.03

1.48 ± 0.07 0.65 ± 0.14 0.35 ± 0.03 0.14 ± 0.02

determined at a low level because of carbon surface recovery by the phosphate form of P [18]. Thus by analogy, as B100++ Accelerated Loading soot contains 0.89 wt% of P, the low microporous specific surface area of this sample is probably due to pore blocking by phosphate species. 3.1.4. Structural properties Table 5 presents ID /IG ratios of the soot calculated from Raman analyses. Soot obtained with the Natural Loading cycle exhibit slightly higher ID /IG ratios than soot produced with the Accelerated Loading cycle. Hence, the accelerated cycle leads to slightly more graphitic soot which is again in agreement with a partial insitu oxidation during collection and thus linked to the difference of temperatures measured during the test bench experiments. Indeed, the residence time in the DPF may lead to an oxidation of the more amorphous carbon within the soot present behind a globally more graphitic material. However, as observed in Table 5, the changes in soot structure are relatively low. As already described par Dippel and Heintzenberg [39], no impact of the type of fuel on the soot structure is observed. 3.1.5. Soot reactivity TG analyses under air allow to investigate soot reactivity toward active regeneration. TGA profiles (Fig. 1) exhibit few differences of the reactivity between the soot samples for the C–O2 reaction. Only B7 soot, with a Tpeak of 624 °C, has a lower reactivity. The shoulder between 200 and 400 °C on the TGA profiles of Natural Loading soot samples is attributed to their higher volatile contents. No correlation between ash and volatile contents and soot reactivity for C–O2 reaction can be established here. Seong and Choi [40] determined that soot combustion by O2 is conducted via an internal combustion of the crystallite. This internal combustion is limited by oxygen penetration inside the soot, hence by its carbon structure and organization [7,11,41]. Raman spectroscopy has shown few structure differences between the soot explaining the slight difference of reactivity of the soot toward active regeneration. Microporous surface area developed during the oxidation process may initiate oxygen penetration inside the crystallite [31,32,36,38]. Thus, the lower reactivity of B7Natural Loading can be attributed to the absence of micropores detected by nitrogen physisorption. In agreement with Williams et al. [23], no effect of the presence of the inorganic compounds is observed on the soot reactivity under an air flow. The reactivity of soot toward passive regeneration was then investigated in the fixed bed reactor under several gas mixtures containing NO2 . Figure 2 gathers the temperature at which 5% of conversion is obtained for all the soot samples under the different effluents. The order of activity of the gas mixture given in the literature (NO2 < NO2 + O2 < NO2 + O2 + H2 O) is found with respect to previous works [28,29,42,43] for the real Diesel soot samples. Under an effluent containing only 400 ppm of NO2 (Fig. 3(a)), the four soot samples present a similar shape for the evolution of the carbon specific oxidation rate toward temperature: a unique peak with a maximum between 550 and 600 °C. B100++ Accelerated Loading soot however differ with a large peak centered at 500 °C. As already seen in TGA profiles, the evolution of the carbon oxidation rate of B100+ -Natural Loading soot presents a little

shoulder around 250 °C, which is attributed to the oxidation of the volatile fraction. When oxygen is added to the gas mixture (NO2 + O2 , Fig. 3(b)), the TPO profile shapes are significantly modified. Soot produced with an accelerated cycle present two distinct peaks: the first one between 475 and 510 °C and the second one beyond 600 °C. Soot originated from the Natural Loading cycle present one main peak between 515 and 550 °C and two shoulders around 250 and 600 °C. The low temperature shoulders are attributed to the combustion of the volatile part of the soot. Other peaks are attributed to the presence of several types of carbon sites with different reactivities [11,12,35]. It could be noted that complete soot oxidation is obtained at higher temperatures for B100++ -Accelerated Loading soot (maximum at 670 °C) than B100+ -Accelerated Loading sample (650 °C) and B100+ -Natural Loading soot (maximum at 640 °C). This behavior and the presence of the second peak above 600 °C for Accelerated Loading soot samples are consistent with the textural and structural properties presented in Section 3.1. Higher graphitic is the sample, higher is the oxidation peak which further shifts towards higher temperatures. When water is present in the effluent (Fig. 3(c)), the profiles obtained under NO2 + O2 are retrieved with more pronounced shapes. In fact, the shoulders which can be observed at high temperatures on the Natural Loading soot profiles under NO2 + O2 become peaks under NO2 + O2 + H2 O. However, under this gas mixture, the B100++ -Accelerated Loading soot reactivity is significantly modified. This sample is now the most reactive one on the whole temperature range of the combustion process and its profile exhibits a large peak at low temperatures (200–450 °C) and a main peak at 550 °C. Whatever the gas mixture, when NO2 is present in the effluent (Fig. 3), the order of reactivity of the soot is as follows: B100++ -Accelerated Loading > B100+ -Accelerated Loading > B100+ Natural Loading > B7-Natural Loading. As observed under air, the B7-Natural Loading sample is still the less reactive soot. No relation between ash or volatile contents and reactivity toward passive regeneration can be observed, as already observed for active regeneration, i.e. soot oxidation under air [15,33]. When the gas effluent contains NO2 , soot produced with Accelerated Loading cycle are more reactive than one obtained with the representative real driving cycle (Natural Loading cycle). The better reactivity of B100+ -Accelerated Loading soot in comparison to B100+ -Natural Loading sample can be attributed to its higher specific surface area (global and microporous) and its higher oxygen content. This conclusion is consistent with those of Williams et al. [23]. For the Natural Loading cycle, the higher reactivity of the B100+ soot compared to that of + B7 soot confirms that increasing the Biodiesel blends in the fuel leads to the formation of more reactive soot [9,11–13]. Indeed, B100+ produces more oxygenated soot which are more easily oxidized. B100+ -Natural Loading sample also contains four times more ash than B7-Natural Loading soot. These inorganic elements can act as catalysts in soot oxidation [9]. Moreover, for the accelerated cycle, increasing the impurities concentration of the fuel also leads to the production of more reactive soot. Elemental analyses have highlighted that potassium and phosphorus contained in the fuel are also present in soot

J. Schobing et al. / Combustion and Flame 198 (2018) 1–13

7

Fig. 1. TGA experiments under air.

composition. The activity of alkali elements in the gasification of carbonaceous materials has received many attention in the past [21,22,24,44–47]. All these studies agree with a catalytic effect of alkali metals towards carbon oxidation in air. It is generally proposed that alkali metals facilitate oxygen retention and its transfer to carbon surface. According to Wigmans et al. [46], alkali metals have intercalating properties and the maximum activity of alkali metals well correlates with the oxygen content of carbon and this intercalation depends on the degree of graphitization of carbon. In the recent work [18], it was shown, studying the reactivity of carbon black doped by Na, K or P in the same conditions as those of the present study, that potassium has a beneficial effect on carbon oxidation on the whole temperature range through passive and active regenerations. This beneficial effect was attributed to the capacity of K to act as O2 and NO2 transporter on the carbon surface. On the contrary, phosphorus exhibited an inhibiting effect on the C–O2 reaction due to phosphate species covering the surface but a huge catalytic effect on C–NO2 reaction in presence of water at low temperatures. This latter was attributed to the formation of phosphoric acid which catalyzes the C-NO2 reaction in a similar way as nitric acid. Thus, under NO2 and NO2 + O2 effluents, the better reactivity of B100++ -Accelerated Loading soot can be attributed to the beneficial effect of potassium. Under NO2 + O2 and O2 effluents, the delay in the complete combustion of B100++ -Accelerated Loading sample at high temperatures (Fig. 3b) is attributed to the inhibiting effect of P on the C–O2 reaction. Finally, the high reactivity of this soot under NO2 + O2 + H2 O between 200 and 450 °C seems to be linked to the catalytic effect of phosphorus on the C–NO2 reaction assisted by water. Over 450 °C, the oxidation of B100++ -Accelerated Loading soot is helped by potassium.

3.2. Kinetics 3.2.1. TReal soot oxidation mechanism Under an oxidizing environment containing only NO2 , the carbon oxidation takes place through direct C–NO2 reactions (reactions 7 and 8) [28,29].

C + 2 NO2 → CO2 + 2 NO

(7)

C + NO2 → CO + NO

(8)

When there is no other oxidation mechanism, the Q ratio defined through below equation

Q=

2[CO2 ] + [CO] [NO]

(9)

is equal to 1. It was established in a previous work that the model Carbon Black soot, oxidized in the same conditions, presents Q ratios between 0.8 and 1.1 and that the presence of inorganic elements (Na, K or P) does not modify these ratios [18]. Q ratios calculated from the gas emissions measured during NO2 -TPO of the soot samples are given in Table 6. Table 6 shows that all Q ratios are between 0.7 and 1.1. Thus, it can be assumed that real soot oxidation mechanism only goes through the direct C–NO2 oxidation for temperatures lower than 450 °C. Hence, the Biodiesel blends in fuel or its inorganic elements contents do not imply another oxidation mechanism for the soot produced. When an excess of oxygen is added to NO2 in the gas mixture, co-operative reactions simultaneously involving O2 and NO2 , take

8

J. Schobing et al. / Combustion and Flame 198 (2018) 1–13

Fig. 2. Temperature at which 5% of conversion is obtained for the different under the four gas mixture. Table 6 Soot Q ratio at different temperature under NO2. Soot

T = 250 °C

T = 300 °C

T = 350 °C

T = 400 °C

T = 450 °C

B7-Natural Loading B100+ -Natural Loading B100+ -Accelerated Loading B100++ -Accelerated Loading

0.7 0.8 1.0 1.0

0.9 1.0 1.1 1.0

1.0 1.1 1.1 1.1

1.1 1.1 1.1 1.1

1.1 1.1 1.1 1.1

(11)

Same ratios have been calculated for NO2 + O2 + H2 O-TPO experiments, whose values are given in Table 7. [CO2 ]/[NO] ratios calculated in presence of water in the gas effluent are globally lower than that obtained in absence of water. This later is known to have a catalytic effect on the direct oxidation reactions of carbon [28,43]. Thus, the ratios are smaller as the direct oxidation reactions are favored by H2 O.

Co-operative reactions take place simultaneously with the direct ones. The study of [CO2 ]/[NO] ratios allows to determine which reaction is favored. If the oxidation only goes through direct reactions, the [CO2 ]/[NO] ratio is close to 0.5, while in the case of an oxidation occurring through co-operative reactions, the [CO2 ]/[NO] ratio tends to 1 [43]. Table 7 gives the ratios calculated between 300 and 450 °C from the NO2 + O2 -TPO of the soot. At 300 °C, soot oxidation preferentially goes through the direct oxidation reactions. Indeed, the [CO2 ]/[NO] ratios are close to 0.5 for all soot samples. When the temperature increases, these ratios increase too, indicating a growing participation of the co-operative reactions in the oxidation mechanism. Real soot behaviors here observed are similar to that of model soot as described in the previous work [18].

3.2.2. Modeling the carbon oxidation As it is well established that real soot oxidation goes through the same global mechanism than model soot oxidation, the same mathematic model can be applied to real soot samples to extract the associated kinetic parameters. Figure 4 shows the evolution of carbon mass for real Diesel soot in the temperature range 150– 400 °C in which direct and co-operative reactions occur. The evolution of NO2 and CO2 emissions can be found in the supplementary data (Figs. S2 and S3). These figures show that a good correlation is obtained between experimental and simulated data for all the samples. Thus, the model seems to describe in a satisfying way the oxidation of real Diesel soot under TPO conditions. Pre-exponential factors (A) and activation energies (Ea) calculated from the TPO under NO2 are given in Table 8 for the dif-

part in the carbon oxidation mechanism, at temperatures lower than 450 °C, through the following reactions [43]:

C + NO 2 +

C+

1 O2 → CO2 + NO 2

1 NO2 O2 → CO 2

(10)

J. Schobing et al. / Combustion and Flame 198 (2018) 1–13

Fig. 3. TPO profiles of soot under (a) NO2 , (b) NO2 + O2 and (c) NO2 + O2 + H2 O.

9

10

J. Schobing et al. / Combustion and Flame 198 (2018) 1–13 Table 7 [CO2 ]/[NO] ratio at different temperatures under NO2 + O2 and NO2 + O2 + H2 O. Soot Under NO2 + O2 B7-Natural Loading B100+ -Natural Loading B100+ -Accelerated Loading B100++ -Accelerated Loading Under NO2 + O2 + H2 O B7-Natural Loading B100+ -Natural Loading B100+ -Accelerated Loading B100++ -Accelerated Loading

T = 300 °C

T = 350 °C

T = 400 °C

T = 450 °C

0.53 0.56 0.57 0.55

0.60 0.63 0.63 0.60

0.63 0.66 0.66 0.64

0.71 0.73 0.73 0.70

0.48 0.54 0.56 0.55

0.57 0.59 0.60 0.57

0.61 0.62 0.64 0.61

0.72 0.65 0.65 0.72

Fig. 4. Mass loss evolution versus temperature during TPO under NO2 of Diesel soot: solid line – experimental data / dotted line – simulated data. Table 8 Kinetic constants determined by the model from NO2 -TPO.

ACO2 (Pa−1 s−1 ) EaCO2 (kJ/mol) kCO2 300 (Pa−1 s−1 ) kCO2 400 (Pa−1 s−1 ) ACO (Pa−1 s−1 ) EaCO (kJ/mol) kCO 300 (Pa−1 s−1 ) kCO 400 (Pa−1 s−1 )

B7-Natural Loading

B100+ - Natural Loading

B100+ -Accelerated Loading

B100++ -Accelerated Loading

1.49.10−4 27.1 5.07.10−7 1.18.10−6 7.40.10−5 29.0 1.67.10−7 4.13.10−7

1.25.10−4 24.7 7.00.10−7 1.51.10−6 2.39.10−5 22.7 2.02.10−7 4.11.10−7

7.47.10−3 45.9 4.89.10−7 2.05.10−6 5.82.10−5 26.2 2.39.10−7 5.40.10−7

6.71.10−2 55.8 5.46.10−7 3.12.10−6 8.19.10−3 49.3 2.62.10−7 1.22.10−6

ferent real soot samples. Reaction constants (k) of C–NO2 reaction calculated at 300 and 400 °C are also given in Table 8. Soot produced with the Natural Loading cycle show, under NO2 , low activation energies for the predominant reaction forming CO2 . EaCO2 is indeed equal to 27.1 kJ/mol for B7-Natural Loading soot and to 24.7 kJ/mol for B100+ -Natural Loading sample, whereas this constant is above 45 kJ/mol for soot produced with an accelerated cycle. On the other hand, for this reaction, Accelerated Loading soot

have a greater pre-exponential factor (ACO2 ), which may be associate to the number of active sites, than Natural Loading soot. Reaction constants kCO2 calculated at 30 0 and 40 0 °C are in agreement with the soot order of reactivity established in TPO at these temperatures. It appears that soot (B100++ -Natural Loading sample) with highest reactivity at low temperatures (T < 350 °C) under NO2 are the ones with the lowest activation energies and soot (B100++ -

J. Schobing et al. / Combustion and Flame 198 (2018) 1–13

11

Table 9 Kinetic constants determined by the model from NO2 + O2 -TPO.

AO2CO2 (Pa−1 s−1 ) EaO2CO2 (kJ/mol) kO2CO2 300 (Pa−1 s−1 ) kO2CO2 400 (Pa−1 s−1 ) AO2CO (Pa−1 s−1 ) EaO2CO (kJ/mol) kO2CO 300 (Pa−1 s−1 ) kO2CO 400 (Pa−1 s−1 )

B7-Natural Loading

B100+ - Natural Loading

B100+ -Accelerated Loading

B100++ -Accelerated Loading

3.75.101 92.0 1.53.10−7 2.70.10−6 5.82.10−1 89.0 4.50.10−9 7.22.10−8

1.10.101 83.8 2.56.10−7 3.49.10−6 1.10.101 88.8 9.19.10−8 1.46.10−6

3.87.101 91.9 1.63.10−7 2.86.10−6 5.81.102 110 5.88.10−8 1.79.10−6

4.53.107 174 6.37.10−9 1.44.10−6 1.21.109 198 1.00.10−9 4.86.10−7

Fig. 5. Mass loss evolution versus temperature during TPO experiments under NO2 + O2 + H2 O of Diesel soot: solid line – experimental data/dotted line – simulated data.

Accelerated Loading sample) with highest reactivity at high temperatures (T > 350 °C), those with highest pre-exponential factors. Thus, Natural Loading soot need few energy intake (lower temperatures) to initiate their oxidation. Above a given temperature, energy is high enough to initiate combustion of all soot samples, and those with the greatest number of active sites burn faster. For the secondary reaction producing CO, B7-Natural Loading, B100+ -Natural Loading and B100+ -Accelerated Loading soot show kinetic constants (EaCO and ACO ) with the same order of magnitude. B100++ -Accelerated Loading sample differ with an activation energy close to 50 kJ/mol and a pre-exponential factor 100 times higher than that of the other soot. Pre-exponential factors, activation energies and reaction constants at 300 and 400 °C, for the C–NO2 –O2 reaction calculated from the TPO experiments under NO2 + O2 for the different real soot samples are given in Table 9. Under NO2 + O2 , B7-Natural Loading, B100+ -Natural Loading and B100+ -Accelerated Loading soot have kinetic constants with the same order of magnitude for the preponderant reaction (EaO2CO2 and kO2CO2 ). Concerning the reaction giving CO, the three samples differ by their number of actives sites (AO2CO ). In

agreement with soot reactivity observed in TPO under NO2 + O2 , B100+ -Natural Loading has the highest reaction constants at 300 °C (kO2CO2 300 and kO2CO 300 ). In presence of O2 in the gas effluent, B100++ -Accelerated Loading sample again exhibit very different kinetics in comparison to the other samples. In fact, its activation energies (EaO2CO2 and EaO2CO ) are about twice higher than other soot. Their preexponential factors (AO2CO2 and AO2CO ) are about 106 –107 times higher. Moreover the reaction constants kO2CO2 and kO2CO deduced from these kinetic constants disagree with experimental B100++ -Accelerated Loading soot reactivity observed in TPO under NO2 + O2 . Thus, it seems that the model is not totally adapted for this sample. Figure 5 presents the experimental (solid line) and simulated (dotted line) mass losses for real Diesel soot oxidation under TPO experiments under NO2 + O2 + H2 O. The evolution of CO2 and NO2 emissions are given by Figs. S4 and S5 of the supplementary data. Coefficients linked to the water catalytic effect were determined as follows: a = b = c = 0.75 and γ = q = μ = 0.9. The presence of water in the gas effluent leads to slightly degraded quality measures, hence the correlation between

12

J. Schobing et al. / Combustion and Flame 198 (2018) 1–13

experimental and simulated results for B7-Natural Loading and B100+ soot under NO2 + O2 + H2 O cannot be as good as observed under other gas effluents. Nevertheless, this correlation is still quite satisfactory and proves that the model is able to simulate real soot oxidation under the different gas mixtures which are here considered. On the other hand, simulated data for B100++ Accelerated Loading soot highly differ for experimental ones, confirming that the model is not totally adapted for this sample. In the previous work [18], simulations of oxidation of Carbon Black doped with inorganic elements of up to 0,5 wt% have been presented. In comparison with this model soot, real Diesel soot B100++ -Accelerated Loading contain an important quantity of K and P (1.17 wt% and 0.89 wt%, respectively). The study of Q and [CO2 ]/[NO] ratios seems to reveal that the global oxidation mechanism of B100++ -Accelerated Loading soot is the same than that of other soot samples. However, a high quantity of P or K can result in the formation of new reaction intermediate species like potassium nitrate or phosphoric acid. The above-described model, which is only based on global reactions, shows its limits. Thus a new model taking into account all the reaction steps of the global mechanism and the reaction intermediates needs to be developed. 4. Conclusions The study of the impact of two different operating engine cycles on soot properties showed that soot produced with an accelerated cycle are more reactive. The operating engine cycle indeed modifies the soot reactivity mainly by impacting the soot structure. When Biodiesel blend is increased in the fuel, less soot are produced and they are more reactive. Characterizations have shown that increasing Biodiesel blend in the fuel increases oxygen and ash contents in the soot, which leads to a better oxidation of the soot particles. Experiments have shown that impurities present in the fuel are retrieved inside the soot composition. Thus, when the inorganic elements concentrations are increased in the fuel, soot naturally contain more of these elements. Oxygen content is also increased. TPO experiments have shown that soot produced with K- and Pdoped fuel are very reactive, due to their inorganic contents which act as catalysts for soot oxidation toward passive and/or active regenerations. Thus the use of Biodiesel is a good way to avoid soot clogging of DPF and to limit active regeneration. The determination of the kinetic parameters by a global model allows to link low-temperature soot reactivity to the activation energy (Ea) and high-temperature soot reactivity to the preexponential (A) for B7 and B100+ soot. The development of a new model is necessary to extract the kinetic parameters of soot containing high amounts of inorganic elements acting as catalysts toward C-oxidation (B100++ soot). Acknowledgments The authors gratefully acknowledge the French National Agency for Research for its financial support (Appibio Project, Ref. ANR-14CE22-0 0 03). Supplementary materials Supplementary material associated with this article can be found, in the online version, at doi:10.1016/j.combustflame.2018.08. 025. References [1] Direction générale des Infrastructures, des transports et de la mer – des véhicules aux normes pour réduire la pollution de l’air, (2011).

[2] A.K. Agarwal, T. Gupta, P.C. Shukla, A. Dhar, Particulate emissions from biodiesel fuelled CI engines, Energy Convers. Manag. 94 (2015) 311–330, doi:10.1016/j.enconman.2014.12.094. [3] P.C. Shukla, T. Gupta, N.K. Labhsetwar, A.K. Agarwal, Trace metals and ions in particulates emitted by biodiesel fuelled engine, Fuel 188 (2017) 603–609, doi:10.1016/j.fuel.2016.10.059. [4] Directive 2009/28/CE du Parlement Européen et du Conseil, (2009). [5] J. Xue, T.E. Grift, A.C. Hansen, Effect of biodiesel on engine performances and emissions, Renew. Sustain. Energy Rev. 15 (2011) 1098–1116, doi:10.1016/j.rser. 2010.11.016. [6] M. Salamanca, F. Mondragón, J.R. Agudelo, P. Benjumea, A. Santamaría, Variations in the chemical composition and morphology of soot induced by the unsaturation degree of biodiesel and a biodiesel blend, Combust. Flame 159 (2012) 1100–1108, doi:10.1016/j.combustflame.2011.10.011. [7] A. Liati, A. Spiteri, P. Dimopoulos Eggenschwiler, N. Vogel-Schäuble, Microscopic investigation of soot and ash particulate matter derived from biofuel and diesel: implications for the reactivity of soot, J. Nanopart. Res. 14 (2012) 1224. http://link.springer.com/article/10.1007/s11051- 012- 1224- 7. (accessed November 28, 2016). [8] H. Omidvarborna, A. Kumar, D.-S. Kim, Recent studies on soot modeling for diesel combustion, Renew. Sustain. Energy Rev. 48 (2015) 635–647, doi:10. 1016/j.rser.2015.04.019. [9] F.-E. Löpez Suárez, A. Bueno-López, M.-J. Illán-Gómez, B. Ura, J. Trawczynski, Study of the uncatalyzed and catalyzed combustion of diesel and biodiesel soot, Catal. Today 176 (2011) 182–186, doi:10.1016/j.cattod.2010.11.094. [10] B. Guan, R. Zhan, H. Lin, Z. Huang, Review of the state-of-the-art of exhaust particulate filter technology in internal combustion engines, J. Environ. Manag. 154 (2015) 225–258, doi:10.1016/j.jenvman.2015.02.027. [11] J. Song, M. Alam, A.L. Boehman, U. Kim, Examination of the oxidation behavior of biodiesel soot, Combust. Flame 146 (2006) 589–604, doi:10.1016/j. combustflame.2006.06.010. [12] J. Song, M. Alam, A.L. Boehman, Characterization of diesel and biodiesel soot, Prepr. Pap. – Am. Chem. Soc. Div. Fuel Chem. 49 (2) (2004) 767. [13] N. Lamharess, C.-N. Millet, L. Starck, E. Jeudy, J. Lavy, P. Da Costa, Catalysed diesel particulate filter: study of the reactivity of soot arising from biodiesel combustion, Catal. Today 176 (2011) 219–224, doi:10.1016/j.cattod.2011.01.011. [14] J.P. Szybist, J. Song, M. Alam, A.L. Boehman, Biodiesel combustion, emissions and emission control, Fuel Process. Technol. 88 (2007) 679–691, doi:10.1016/j. fuproc.20 06.12.0 08. [15] K. Yehliu, R.L. Vander Wal, O. Armas, A.L. Boehman, Impact of fuel formulation on the nanostructure and reactivity of diesel soot, Combust. Flame 159 (2012) 3597–3606, doi:10.1016/j.combustflame.2012.07.004. [16] E. Aneggi, C. de Leitenburg, G. Dolcetti, A. Trovarelli, Diesel soot combustion activity of ceria promoted with alkali metals, Catal. Today 136 (2008) 3–10, doi:10.1016/j.cattod.2008.01.002. [17] L. Castoldi, R. Matarrese, L. Lietti, P. Forzatti, Intrinsic reactivity of alkaline and alkaline-earth metal oxide catalysts for oxidation of soot, Appl. Catal. B Environ. 90 (2009) 278–285, doi:10.1016/j.apcatb.2009.03.022. [18] J. Schobing, V. Tschamber, A. Brillard, G. Leyssens, Impact of biodiesel impurities on carbon oxidation in passive regeneration conditions: influence of the alkali metals, Appl. Catal. B Environ. 226 (2018) 596–607, doi:10.1016/j.apcatb. 2017.12.011. [19] E. Iojoiu, V. Lauga, J. Abboud, G. Legros, J. Bonnety, P.D. Costa, J. Schobing, A. Brillard, G. Leyssens, V. Tschamber, P. Anguita, J.G. Vargas, L. Retailleau, S. Gil, A. Giroir-Fendler, M.-L. Tarot, F. Can, D. Duprez, X. Courtois, Biofuel impact on diesel engine after-treatment: deactivation mechanisms and soot reactivity, Emiss. Control Sci. Technol. (2017) 1–18, doi:10.1007/ s40825- 017- 0079- x. [20] Texte de norme EN 14214 (2008). [21] D.W. Brookshear, K. Nguyen, T.J. Toops, B.G. Bunting, W.F. Rohr, J. Howe, Investigation of the effects of biodiesel-based Na on emissions control components, Catal. Today 184 (2012) 205–218, doi:10.1016/j.cattod.2011.12.001. [22] A. Williams, J. Burton, R. McCornick, T. Toops, Impact of fuel metal impurities on the durability of a light-duty diesel aftertreatment system, SAE Technical Paper 2013-01–0513 (2013). [23] A. Williams, S. Black, R.L. McCormick, Biodiesel fuel property effects on particulate matter reactivity, Conference paper NREL/CP-540-47286 (2010). [24] M. Lance, A. Wereszczak, T.J. Toops, R. Ancimer, H. An, J. Li, L. Rogoski, P. Sindler, A. Williams, A. Ragatz, R.L. McCormick, Evaluation of fuel-borne sodium effects on a DOC-DPF-SCR heavy-duty engine emission control system: simulation of full-useful life, SAE Int. J. Fuels Lubr. 9 (2016) 683–694, doi:10.4271/2016-01-2322. [25] A. Sadezki, H. Muckenhuber, H. Grothe, R. Niessner, U. Pöschl, Raman microspectroscopy of soot and related carbonaceous materials: spectral analysis and structural information, Carbon 43 (2005) 1731–1742, doi:10.1016/j.carbon. 2005.02.018. [26] M. Zinbo, L.M. Skewes, C.E. Hunter, D. Schuetzle, Thermogravimetry of filterborne diesel particulates, Thermochim. Acta 166 (1990) 267–275, doi:10.1016/ 0040- 6031(90)80187- 4. [27] G. Neri, L. Bonaccorsi, A. Donato, C. Milone, M.G. Musolino, A.M. Visco, Catalytic combustion of diesel soot over metal oxide catalysts, Appl. Catal. B Environ. 11 (1997) 217–231, doi:10.1016/S0926-3373(96)0 0 045-8. [28] F. Jacquot, V. Logie, J.F. Brilhac, P. Gilot, Kinetics of the oxidation of carbon black by NO2 : influence of the presence of water and oxygen, Carbon 40 (2002) 335–343, doi:10.1016/S0 0 08-6223(01)0 0103-8.

J. Schobing et al. / Combustion and Flame 198 (2018) 1–13 [29] M. Jeguirim, V. Tschamber, J.F. Brilhac, Kinetics of catalyzed and non-catalyzed soot oxidation with nitrogen dioxide under regeneration particle trap conditions, J. Chem. Technol. Biotechnol. 84 (2009) 770–776, doi:10.1002/jctb.2110. [30] P. Moreau, P. Valero, A. Brillard, V. Tschamber, J.-F. Brilhac, Y. Hohl, R. Vonarb, L. Germanese, B. Courtalon, Determination of the kinetic parameters for the soot combustion through a dynamic numerical procedure, SAE Technical Paper 2015-24–2396 (2015). [31] A.W. Kandas, I. Gokhan Senel, Y. Levendis, A.F. Sarofim, Soot surface area evolution during air oxidation as evaluated by small angle X-ray scattering and CO2 adsorption, Carbon 43 (2005) 241–251, doi:10.1016/j.carbon.2004.08.028. [32] T. Ishiguro, N. Suzuki, Y. Fujitani, H. Morimoto, Microstructural changes of diesel soot during oxidation, Combust. Flame 85 (1991) 1–6. [33] A. Strzelec, T. Toops, C. Daw, D.E. Foster, C. Rutland, Diesel particulate oxidation model: combined effects of volatiles and fixed carbon combustion, SAE International, Warrendale, PA, 2010, doi:10.4271/2010-01-2127. [34] J.A. Soriano, J.R. Agudelo, A.F. López, O. Armas, Oxidation reactivity and nanostructural characterization of the soot coming from farnesane – a novel diesel fuel derived from sugar cane, Carbon 125 (2017) 516–529, doi:10.1016/j.carbon. 2017.09.090. [35] A. Strzelec, T.J. Toops, C. Stuart Daw, Oxygen reactivity of devolatilized diesel engine particulates from conventional and biodiesel fuels, Energy Fuels 27 (2013) 3944–3951. [36] K.J. Rockne, G.L. Taghon, D.S. Kosson, Pore structure of soot deposits from several combustion sources, Chemosphere 41 (20 0 0) 1125–1135, doi:10.1016/ S0 045-6535(0 0)0 0 040-0. [37] T.W. Zerda, X. Yuan, S.M. Moore, C.A. Leon y Leon, Surface area, pore size distribution and microstructure of combustion engine deposits, Carbon 37 (1999) 1999–2009, doi:10.1016/S0008- 6223(99)00068- 8. [38] B.G. Wicke, K.A. Grady, Porosity changes in soot resulting from oxygen atom adsorption at 298K, Carbon 25 (1987) 791–797, doi:10.1016/0 0 08-6223(87) 90153-9.

13

[39] B. Dippel, J. Heintzenberg, Soot characterization in atmospheric particles from different sources by NIR FT Raman spectroscopy, J. Aerosol Sci. (Supplement 1) (1999) S907–S908, doi:10.1016/S0021- 8502(99)80464- 9. [40] H. Seong, S. Choi, Oxidation-derived maturing process of soot, dependent on O2 –NO2 mixtures and temperatures, Carbon 93 (2015) 1068–1076. http://dx. doi.org/10.1016/j.carbon.2015.07.008. [41] R.L. Vander Wal, A. Yezerets, N.W. Currier, D.H. Kim, C.M. Wang, HRTEM study of diesel soot collected from diesel particulate filters, Carbon 45 (2007) 70–77, doi:10.1016/j.carbon.20 06.08.0 05. [42] N. Zouaoui, M. Labaki, M. Jeguirim, Diesel soot oxidation by nitrogen dioxide, oxygen and water under engine exhaust conditions: kinetics data related to the reaction mechanism, C. R. Chim. 17 (2014) 672–680, doi:10.1016/j.crci.2013. 09.004. [43] M. Jeguirim, V. Tschamber, J.F. Brilhac, P. Ehrburger, Oxidation mechanism of carbon black by NO2 : effect of water vapour, Fuel 84 (2005) 1949–1956, doi:10. 1016/j.fuel.2005.03.026. [44] G. Mul, F. Kapteijn, J.A. Moulijn, A DRIFTS study of the interaction of alkali metal oxides with carbonaceous surfaces, Carbon 37 (1999) 401–410, doi:10. 1016/S0 0 08-6223(98)0 0202-4. [45] J.P.A. Neeft, M. Makkee, J.A. Moulijn, Catalytic oxidation of carbon black – 1. Activity of catalysts and classification of oxidation profiles, Fuel 77 (1998) 111–119. [46] T. Wigmans, H. Haringa, J.A. Moulijn, Nature, activity and stability of active sites during alkali metal carbonate-catalysed gasification reactions of coal char, Fuel 62 (1983) 185–189, doi:10.1016/0016-2361(83)90195-3. [47] F. Kapteijn, J. Jurriaans, J.A. Moulijn, Formation of intercalate-like structures by heat treatment of K2 CO3 -carbon in an inert atmosphere, Fuel 62 (1983) 249– 251, doi:10.1016/0016- 2361(83)90210- 7.