Synthesis and characterization of Carbon/Nitrogen/Iron based nanoparticles by laser pyrolysis as non-noble metal electrocatalysts for oxygen reduction

: This paper reports original results on the synthesis of Carbon/Nitrogen/Iron-based ORR 15 electrocatalysts by CO 2 laser pyrolysis. Precursors consisted in two different liquid mixtures 16 containing FeOOH nanoparticles or iron III acetylacetonate as iron precursors, being fed to the 17 reactor as an aerosol of liquid droplets. Carbon and nitrogen were brought by pyridine or a 18 mixture of pyridine and ethanol depending on the iron precursor involved. The use of ammonia 19 as laser energy transfer agent also provided a potential nitrogen source. For each liquid precursor 20 mixture, several syntheses were conducted through the step-by-step modification of NH 3 flow 21 volume fraction, so-called R parameter. We found that various feature such as the synthesis 22 production yield or the nanomaterial iron and carbon content, showed identical trends as a 23 function of R for each liquid precursor mixture. The obtained nanomaterials consisted in 24 composite nanostructures in which iron based nanoparticles are, to varying degrees, encapsulated 25 by a presumably nitrogen doped carbon shell. Combining X-Ray diffraction and Mossbauer 26 spectroscopy with acid leaching treatment and extensive XPS surface analysis allowed the difficult 27 question of the nature of the formed iron phases to be addressed. Besides metal and carbide iron 28 phases, data suggest the formation of iron nitride phase at high R values. Interestingly, 29 electrochemical measurements reveal that the higher R the higher the onset potential for the ORR, 30 what suggests the need of iron-nitride phase existence for the formation of active sites towards 31 the ORR.

carried out at room temperature with 57 Fe isotope and using a commercial Co * :Rh γ-ray source and 86 an electromagnetic drive with linear velocity signal. The different components present in the 87 spectra were characterized by their hyperfine field and their isomer shift, which were obtained 88 from the experimental data using a least-squares fitting procedure. The iron content was measured 89 by X-ray fluorescence using a previously described procedure [20] that requires tiny amount of 90 materials : first, a collection of porous deposit of known loading was prepared on carbon felts 91 C 2018, 4, x FOR PEER REVIEW 3 of 23 (Freudenberg H2415 I2C3) using carbon nanotubes (CNT's) whose Iron content was previously 92 determined by thermogravimetric analysis. Then a calibration curve reporting the iron K line [20] 93 as a function of the iron content in the CNT's electrodes was drawn. Second, porous deposit of 94 known loading were prepared on carbon felts with powders synthesized by laser pyrolysis. The 95 calibration curve was then used to convert the recorded iron K line intensity into an iron content.

96
The carbon content was measured using a Horiba carbon analyzer. Specific

132
The aerosol obtained from the liquid precursor mixture is driven by argon as carrier gas and 133 is further mixed with ammoniac. It is important to note that in this configuration, even if the NH3 134 flow does not directly take part to the aerosol extraction from the generator, it contributes to the 135 total gas flow once mixed into the mainstream in the nozzle. This means that increasing NH3 flow

139
All the gas flows are controlled by mass flowmeters. This allows parameter R defined by equation

140
(1) to be set and varied over a controlled range.

154
The amount of liquid precursor mixture was weighted before and after the end of the synthesis 155 in order to determine the precursor consumed mass. Then, taking into account the weight of 156 washed collected powders, chemical yield was estimated. Figure 2 reports the trends observed for 157 this chemical yield as a function of R for both FeOOH (LPM-1) and Fe(acac)3 (LPM-2) precursor 158 systems. It is seen that chemical yields are ranging from 0.15% to  5%, and that the higher R the 159 lower the yield. The powders were systematically characterized by Scanning Electron Microscopy (SEM) and 161 to a lesser extent by Transmission Electron Microscopy (TEM). The most important changes in the 162 morphology of the nanomaterials were found for those produced from LPM-1. As shown in figure   163 3, modification of the R parameter resulted in strong modification of the particle sizes.

169
The material also contains smaller particles, but it was impossible to estimate the ratio between

179
Elongated structures of these latter layers are also clearly seen in some part of the sample, with

206
It is seen that the higher the R parameter the higher the iron content, while the carbon content 207 exhibits the reverse trend (the higher R the lower the carbon content). As expected, the sum of

232
XRD was then recorded on powders. It is well known that information provided by XRD 233 analysis might be limited by poor crystallization and small crystallite size, which broaden 234 diffraction peaks. It is worth noting that materials obtained from LPM-2 gave XRD patterns with 235 no clearly defined peak, completely impeding phase identification. Therefore, the XRD data relate 236 here to materials obtained from LPM-1. Figure 7 shows two XRD patterns recorded on powders

244
In our case, higher concentration of NH3 provides a more nitriding atmosphere and also increases 245 the temperature through CO2 laser absorption.

300
Finally the main points suggested by the tentative peak fitting is that at high R iron nitride 301 could be present in significant amount (37.0 At.%), as well as pyridinic nitrogen (18.4%). Graphitic

302
nitrogen content is about 5.5 At%. In contrast at low R, no iron nitride can be found. The pyridinic 303 nitrogen content is only 10.9 At.%. The major contribution is clearly seen at 399.3 eV, in an energy 304 range close to those reported for FeNx sites, but which can also correspond to amines, nitriles or 305 nitroso nitrogen atoms. The attribution of one of the components to graphitic nitrogen is unclear.

307
The N1s spectra shown in figure 8 and tentative peak fitting suggest that nitrided iron is     338 Figure 11 shows the N1s spectra for FeOOH and Fe(acac)3 precursors derived powders before     itself. Figure 12b) shows the data related to LPM-2, for which a marked trend is observed i.e. the 380 higher the R parameter the higher the ORR onset potential. Finally, these data strongly suggest a 381 direct influence of the R parameter on the performances of the ORR electrocatalysts synthesized by 382 the CO2 laser pyrolysis method.

384
This discussion session addresses the trends observed for the different feature of the materials, 385 particle morphology, yield and composition, and finally electrochemical performances, in 386 particular as a function of R. It must be mentioned that we are sometimes reduced to making 387 hypothesis since we have no direct indication that support a given explanation rather than another 388 one.

390
Particle Size And Morphology :

391
The particle size and morphology obtained is this work can be discussed in the frame of well-

407
-Surface reaction of monomers on primary particles can contribute to their growth, while 408 evaporation/sublimation contribute to particle size decrease and forming monomers again.

409
-Further coagulation and/or coalescence process between primary particles may also 410 contribute to nanoparticles growth.

412
Finally, depending on the temperature value and profile, the final morphology can consist in  Or, 3b) drying/melting allowing fragmentation/evaporation towards above mentioned gas-fed 424 flame particle formation scenario.

426
To summarize the main parameters influencing particle size and morphology, it can be 427 pointed that short resident time, high temperature and fast cooling rates reduce the particle size

432
We can now comment on the size and morphology of the products we obtained in this work, 433 although at this time only assumptions can be made, based on the above-mentioned pathways 434 described for gas/liquid flame synthesis methods. It is also important to remind that for laser 435 pyrolysis, the chemical composition of the effective precursor media through ammonia 436 concentration is not disconnected from the temperature conditions. Indeed, as mentioned before, 437 changing the ammonia flow modifies the CO2 laser absorption and therefore the energy 438 (temperature) provided to the precursors.

439
First, it is seen that materials produced from LPM-1 (FeOOH particles as iron precursor) show

451
For LPM-2 the trends observed as a function of R for averaged particle size is suspected to be

504
We showed that the most active materials are obtained for high R values. This is observed for both 505 kind of materials (LPM-1 and LPM-2), although the trend for ORR activity as a function of R is less 506 marked for LPM-1. It is worth reminding that the trends as a function of R for the whole set of 507 characterization performed on both material are exactly the same. Besides, the most active materials 508 towards the ORR have the highest initial iron content and they retain a significant iron content in 509 spite of acid leaching. Therefore, we believe that the active sites responsible for the ORR activity 510 are presumably related to the initial iron content in the materials. The iron nitride phases formed 511 at high R are not directly involved in the ORR activity because they are not stable in acidic media.

512
However, the X-Ray diffraction recorded after acid leaching (figure 10) suggests that iron nitride 513 particles are surrounded by a carbon shell which protects the core from acid leaching. Finally, it 514 can be tentatively assumed that the active site formation in our materials is related to iron nitride 515 phases, as already reported for the same family of materials addressed here, [55]. In this work 516 authors also show that acid leaching resulted in a moderate loss of activity.

517
The parameters that influence the activity are thus related to the nature and amount of the 518 active sites, but also to the specific surface area of the materials. Indeed, the higher the specific 519 surface area the higher the volume density of active sites in the catalyst layer and thus the ORR

554
The materials were characterized using different methods and surface analysis of the materials 555 advantageously completed the information recorded on bulk materials. In particular, the N1s core 556 level spectra as well as the semi quantitative analysis tend to confirm the higher R the higher the 557 amount of iron nitride-like phase formed. This point appears comforted by the comparison of the 558 XPS and XRD data recorded for as-prepared materials and acid leached ones. Preliminary 559 evaluation of the ORR by cyclic voltammetry indicates that the performance is improved when the 560 R parameter is increased. Based on these trends and those recorded regarding the whole set of 561 characterization, it can be tentatively suggested that iron nitride formation during the CO2 laser 562 pyrolysis favors the formation of active sites for the ORR in these non-noble electrocatalysts.

563
Current work is dealing with the development of this new approach consisting of using CO2 laser 564 pyrolysis for the synthesis of carbon/nitrogen/iron based non-noble electrocatalysts, and will be the 565 subject of coming reports.

582
Supporting information related to Mossbauer spectroscopy:

583
The table below reports the parameters involved in the fitting of the spectra shown in figure   584 6, for the identified phases, consistently with data reported in the literature [57][58][59]