Design of Iron(II) Phthalocyanine‐Derived Oxygen Reduction Electrocatalysts for High‐Power‐Density Microbial Fuel Cells

Abstract Iron(II) phthalocyanine (FePc) deposited onto two different carbonaceous supports was synthesized through an unconventional pyrolysis‐free method. The obtained materials were studied in the oxygen reduction reaction (ORR) in neutral media through incorporation in an air‐breathing cathode structure and tested in an operating microbial fuel cell (MFC) configuration. Rotating ring disk electrode (RRDE) analysis revealed high performances of the Fe‐based catalysts compared with that of activated carbon (AC). The FePc supported on Black‐Pearl carbon black [Fe‐BP(N)] exhibits the highest performance in terms of its more positive onset potential, positive shift of the half‐wave potential, and higher limiting current as well as the highest power density in the operating MFC of (243±7) μW cm−2, which was 33 % higher than that of FePc supported on nitrogen‐doped carbon nanotubes (Fe‐CNT(N); 182±5 μW cm−2). The power density generated by Fe‐BP(N) was 92 % higher than that of the MFC utilizing AC; therefore, the utilization of platinum group metal‐free catalysts can boost the performances of MFCs significantly.


Introduction
Microbial fuel cells (MFCs) are very attractive bioelectrochemical systems capable of degrading and removing organic pollutants and generating electricity. [1][2][3] To be competitive with existing wastewater-treatment approaches, the pollutant-removal efficiency has to be increased considerably;t herefore, the kinetics of water purification should be accelerated. In general, the power/current produced in MFCsi sq uite low,t he electrochemicalp rocesses require substantial optimization, and, in parallel, the losses associatedw ith thesep rocesses have to be reduced. [4,5] To date, several successful prototypes and scaledup systems have been presented to demonstrate the potential of MFC technology for both wastewater treatment and cleanenergy generation. [1,2,6,7] It also should be mentioned that major problemsa ssociated with the performances of MFC systemso riginate from the poor kinetics for the oxygen reduction reaction (ORR) and the high overpotentialso ft he cathode operating in neutral media. [4,8,9] Oxygen is primarily used as an electron acceptor in the majority of fuel cells as it is naturallya vailablei nt he atmosphere (and, consequently, has al ow cost) and has ah igh redox potential. An additional performance dropo fM FCs is related to the fact that oxygen electroreduction requires H + or OH À ions as reagents (depending on the mechanism of the ORR), and their concentration is lowest at neutralp H ( % 10 À7 m).T herefore, the ORR is severelyh ampered, which negativelya ffects the overall MFC performance. The general practicet or educe the overpotentials and accelerate the kinetics is utilization of electrocatalysts on the cathode. [10][11][12][13][14] Al iterature review indicates three main types of catalysts that can be integrated into the cathodic structures of MFCs: (i)platinum-group metals (PGMs), [10][11][12] (ii)carbonaceous metalfree catalysts, [10][11][12] and (iii)PGM-free materials. [10][11][12] The first type of catalystu tilizes platinum/PGMn anoparticles dispersed on carbonaceous supports and can be used conventionally as anode and cathode catalysts in hydrogen-airo rd irect-alcohol fuel cells. [15][16][17][18][19] Several issues are related to the employmento f Pt as ac athode catalyst in MFCs. First, Pt is ar are and very expensive metal, and large-scale deployment for practical applications seemsu nviable and cost-prohibitiveo wing to the low power produced by the MFCs. [20] Second, MFCs work in harsh and polluted environments in which platinum will interact with strongly adsorbed charged or neutral species, which will lead to ad ecrease in ORR catalytic activity. [21][22][23][24] For example, Cl À and S 2À ions are such species, and small concentrations of Iron(II)p hthalocyanine (FePc)d eposited onto two different carbonaceous supports was synthesized through an unconventional pyrolysis-free method. The obtained materials were studied in the oxygen reduction reaction (ORR) in neutral media through incorporationi na na ir-breathing cathode structure and tested in an operating microbial fuel cell (MFC) configuration. Rotating ring disk electrode (RRDE) analysisr evealed high performances of the Fe-based catalysts compared with that of activatedc arbon (AC). The FePc supported on Black-Pearl carbon black [Fe-BP(N)] exhibits the highest performance in terms of its more positive onset potential, positive shift of the half-wavep otential, and higherl imiting currenta sw ell as the highestp ower density in the operating MFC of (243 AE 7) mWcm À2 ,w hich was 33 %h ighert han that of FePc supported on nitrogen-doped carbon nanotubes (Fe-CNT(N);1 82 AE 5 mWcm À2 ). Thep ower density generated by Fe-BP(N) was 92 %h ighert han that of the MFC utilizing AC;t herefore, the utilization of platinum group metal-free catalysts can boost the performances of MFCs significantly. phthalocyanines with preformedF e ÀN x centers. [39,44,52,57,[60][61][62][63] This method of ORR catalyst design has been discussed in several articles [39,44] ;t he metal centers bound to the nitrogen atomsi nt he complex act as the active sites, whereas the conductive carbonaceous support improves the electron transfer. [39,44] These materials are very active towards the ORRb ut do not show high stability under different pH conditions. To increase their stabilities, theseh ybridsc an be pyrolyzed, and this approachisc lose to the first methodd escribed above.
In this study,F e-based catalysts weref abricated from commerciallya vailablei ron(II) phthalocyanine, which was tethered to the surfaces of two different carbonaceous supports, that is, (i)nitrogen-doped multi-walled carbon nanotubes and( ii)nitrogen-doped carbon black (Black Pearls)t op roduce Fe-CNT(N) and Fe-BP(N), respectively.T he surface chemistry and morphology of the obtained materials was studied comprehensively. The electrocatalytic activities of these materials in the ORR were studied by employing the RRDEmethod. Finally,t he catalysts were incorporatedi nto air-breathing cathodes and tested in operating MFCs. Thep erformances of the new catalysts were compared to that of activated carbon (AC) as ac arbonaceous benchmark.

Surface morphologiesoft he catalysts
The morphological features of the two catalysts used in this work were studied by electron microscopy (Figures 1a nd 2). The SEM image (Figure 1a

Surfacechemistry of the catalysts
The surface chemistry of the two catalysts was investigated through X-ray photoelectrons pectroscopy (XPS). The elemental compositions and chemical speciationsa re shown in Ta ble 1. The catalysts mainly consist of carbon, the concentra-tion of which varies from 79.1 %t o8 1.9 %. The oxygen content was noticeably high, with total percentage varying from 9.3 % to 11 %. The nitrogen content varies from 5.8 %f or Fe-BP(N) to 10.0 %f or Fe-CNT(N).
Nitrogen has been identified to have ap ositive effect on the ORR owing to its ability to improve the electronic properties. [64][65][66] However,e xcess nitrogen concentrations may be harmful to the overall performance, as it decreases the conductivity of the material. Very high Fe contents between 1.3 %a nd 1.5 %werealso identified.
In ap reviouss tudy,p yridinic nitrogen and transition-metalcoordinated nitrogen atoms showed ap ositive relationship with performance. [42] The high-resolutionN 1s spectra for the two catalysts are shown in Figure 3a nd were fitted with four peaks, namely,p yridinic nitrogen atoms (N pyr. )a tabinding energy of 398.5 eV,F e-coordinated nitrogen atoms (NÀFe) at 399.5 eV,h ydrogenated nitrogen atoms (NÀH, pyrrolic nitrogen and hydrogenatedp yridine) at 401 eV,a nd graphitic nitrogen atoms (N gr. )a t4 02.3 eV.F or the two materials tested,t he relative percentage of pyridinicn itrogen atoms was the highest of all types of nitrogen atoms and varied from 47.4 %t o6 2.2 %,   and the Fe-BP(N) sample contained the largest amount.I mportantly,p yridinic nitrogen atoms represent an edge defect within the carbon matrix, and edge sites are more favorable for oxygen reduction. The Fe-CNT(N) catalysthad the larger relative percentages of hydrogenated nitrogen andF e-coordinated Na toms of 18.0 %a nd 28.3 %, respectively.T he peak at 399.5 eV is assigned to Fe-coordinated nitrogen atoms and may also have ac ontribution from amines. The analysiso ft he high-resolution Fe 2p spectra reveals that the well-resolved peak for the FeÀNa ctive centers is larger for the Fe-BP(N) sample. The Fe-CNT(N) samples has as ubstantially larger amount of hydrogenated nitrogen atoms. The negative effect of hydrogenated nitrogen atoms on the ORR due to its significant contribution to the reduction of oxygen to hydrogen peroxide was shown previously. [67] The Fe-BP(N) catalysta lso has the smallest amount of graphitic carbon atoms with the majority being amorphous with surface oxides,w hich are an important marker for al arge number of defect sites within the carbon network, andt his is related to ah igher density of active ORR sites. [67] The combination of higher amounts of nitrogen edge defects, higher amount of iron-coordinated nitrogen atoms, and highera mount of aliphatic carbon atoms and surface oxides should make the Fe-BP(N) sample more effective in the ORR.

Electrocatalytic activities of the catalysts in neutral media
RRDE measurements were performed using the three catalysts in an O 2 -saturated electrolyte to evaluate the oxygen reduction activity as well as the total number of electrons transferred in the electrocatalytic process [Eq. (1)] and the hydrogen peroxide generated [Eq. (2)].T wo different catalyst loadings (0.1 and 0.6 mg cm À2 )w ere used for the experiments. It was shown previously that an increasei nl oading could hindert he catalyst kinetics. [38,68,69] At hick catalytic layer of porous carbonaceous materialc an trap the intermediate peroxidei nside the pores, where the H 2 O 2 is consumed (eitherb yc hemical decomposition or by electrochemical reduction to water) without being detected on the ring. [38,68,69] As the material loading on the disk increased, the peroxide detected decreased significantly. [38,68,69] Therefore, two extreme loadings, one low (0.1 mg cm À2 )a nd one high (0.6 mg cm À2 ), were selected to obtain reasonable data fort he ORR with these catalysts. The Fe-BP(N)c atalyste xhibits a better performance than Fe-CNT(N) as it has ahigher onset potential, more positive half-wavep otential, and higherl imiting current (Figure 4a and b). Both of the Fe-based catalysts exhibited ap erformance superior to that of AC. As imilar trend was observed for both catalyst loadings. The peroxide percentage generatedi nt he ORR with the three catalysts is consistent with the findings of the disk-current measurements (Figure 5a and b). The Fe-BP(N) catalystp roduced the least peroxide ( % 1%,F igure 5a), and the Fe-CNT(N) catalystp roduced slightly more peroxide ( % 1-2 %, Figure 5a and b). Interestingly,t he overall peroxide productioni sq uite lowf or the FePc-derived catalysts. The AC produced am uch higher peroxide yield, in agreement with previously reported data (Figure 5a and b). It was proveni nt his study that an increased loading leads to a significant decrease in detected peroxide (Figure 5a andb ). The Fe-based catalysts exhibit af our-electron reduction process independentofthe loading (Figure 5c and d). On the contrary,A Cf ollowed at wo-electron-transfer mechanism at al ow loading of 0.1 mg cm À2 (Figure 5c). The thick layer at 0.6 mg cm À2 loading masked the AC behavior,a nd the electron transfer could be considered to be closert oaf our-electron process. It is important to mention here that the FePc catalysts greatly outperform commercial activated carbon in terms of ORR activity and, therefore, are suitable substitutes forA Cf or the ORR in neutralmedia.

Powerg eneration of operating MFCs
Polarization curves were recorded for the differentF e-based catalysts incorporatedi nto an air-breathing cathode composed of ac arbonaceous matrix made of AC, CB, and polytetrafluoroethylene (PTFE) as the binder (Figure 6a). The open-circuit voltages (OCVs) measured at the beginning of the polarization curvesw ere different for the materials investigated. The AC had the lowest OCV of (635 AE 2) mV.T he Fe-BP(N) and Fe-CNT(N)h ad higher OCV values of (688 AE 12) and (684 AE 8) mV, respectively (Figure 6a). In the polarization curves, three distinctive trends can be noticed:A Ch as the poorestp erformance, Fe-CNT(N) performs better than AC but worse than Fe-BP(N), and Fe-BP(N) has the best performance of the cata-   (Figure 6a). The highest power density measured in this investigation wast hat produced by Fe-BP(N) of (243 AE 7) mWcm À2 (Figure 6b). Fe-CNT(N) produced al ower power density of (182 AE 5) mWcm À2 (Figure6b). The AC sample had the lowest power density of (127 AE 1) mWcm À2 (Figure 6b). The Fe-BP(N) performed 33 %b etter than Fe-CNT(N) and 92 % better than AC. The separate anode (Figure 6c)a nd cathode (Figure 6d)p rofiles show similara nodic performances, which underlines that the difference in the overall polarization curve was caused substantially by the cathode behavior.

Outlookand comparisonw ith existingliterature
In MFC systems, the ORR is often identified as the most problematic aspect;and therefore solutions have to be investigated and considered. [4,8,9] PGM-free catalysts seem to be interesting and appropriate for furtheri nvestigations. Once again,i nt his study,t he utilization of PGM-free catalysts resulted in as ubstantiali ncrease in the power produced compared with that for AC. Among the earth-abundant metals, Fe was selected for the catalyst because it was previously identified to be more active than other earth-abundant metals such as Mn, Cu, Co, and Ni. [10][11][12][13] Thisw ork confirms that PGM-free catalysts based on iron can boost the performance considerably.T he power density was doubled for Fe-BP(N) compared with that of bare AC [(243 AE 7) and (127 AE 1) mWcm À2 ,r espectively].T he results were consistent for the data obtainedd uring the RRDE tests and the data acquired during the MFCs tests. Therefore, it is demonstrated once more that RRDE data can be used to predict the performance of ac atalysti ncorporated into an airbreathing cathode,i na greement with previously reported data. [46] DifferentP GM-free catalysts incorporated into air-breathing cathodes were reported previously. [55,[70][71][72][73] Yang et al. integrated Co/NÀCn anoparticles in an air-breathing cathode and obtained am aximum power density of 251 mWcm À2 . [70] The same research group also utilized aN iCo 2 O 4 -modified activatedcarbon cathode and obtained alower maximum power density of 173 mWcm À2 . [71] Fu et al., [72] Pan et al., [73] and Yang and Logan [55] decorated activated-carbon-based cathodes with Febased catalystsa nd obtained maximum powerd ensities of 143, 244, and2 60 mWcm À2 ,r espectively,w ith an electrolyte containing 50 mm phosphate buffer,a cetate as bacterial food, and MFCs operatinga t308C.
Compared to previously reported studies, the main feature of this work is the utilization of Fe-based catalysts that are fabricated through the attachment of commercially available iron(II) phthalocyanine onto high-surface-area carbon-nitrogen-doped supports insteado fahigh-temperature pyrolysis approach. The two carbonaceous supports selected were nitrogen-doped carbon nanotubes [CNT(N)] and nitrogen-doped Black Pearls [BP(N)].T he positive effect of nitrogen on the ORR performances was elucidated previously. [22,23,46] BP(N) was the support with the highest BET surface area among the materials used in this study;t he previously quantified BET surfacea rea of 1317 m 2 g À1 is almost four times higher than that for CNT(N) of 359 m 2 g À1 . [57] The BP(N) also showed the best performances in the RRDE and MFC tests [(243 AE 7) mWcm À2 ].
To be suitable for large-scale applications,F e-based catalysts must also be cheap and durable. Many parameters will affect the final price of PGM-free electrocatalysts, and economical price analysis is generally quite complicated.I naprevious study,t he cost of the catalystp roduced through the sacrificialsupport methoda nd fabricated through ap yrolysis technique was quantified as approximately3 .5 US$ g À1 . [25] The estimation only considered the materials utilized;u nfortunately,t he gas utilized during pyrolysis wasn ot included and, most importantly,t he cost of the electricity utilized during the heat treatment was not considered. High-temperature processes are very energy consuming as electricity is used to keep the temperaturew ithin the furnacec onstant during the pyrolysis process. The costs of the catalysts on the basis of the lab-scale procedure utilized were estimated to be 8a nd 5US$ g À1 for Fe-CNT(N) and Fe-BP(N), respectively.T his estimation seems to be more realistic, as the procedure presented in this manuscript does not have high-temperature processes, and this leads to an overall reduction of the preparation costs that should be taken into serious consideration. Moreover,a st he heat treatment is avoided, the fabrication of the catalysts is simpler andm ore affordable.
The maximum powerd ensity obtainedi nt his work was (243 AE 7) mWcm À2 for the BP(N) support. Ad irect comparison with previously reported materials is difficult owing to the differento perating conditions utilized. The performance is effected dramatically by the MFC design, [1] operating temperature, [74] altitudea bove sea level, electrolyte utilized (different solution conductivity), [43] bacteria utilized (e.g.,s ingle or mixed culture), [1] organic compounds utilized (e.g.,l actate, acetate, fumarate, etc.), [3,75] organics concentration, [74] and the presence or absence of membranes. [66] Several others tudies demonstrated the utilization of Fe-based catalysts with excellent performances and superiority over platinum-based cathodes [22,23] or AC-based cathodes. [43] The power densities usually varyb etween approximately 100 and 600 mWcm À2 . [76,77] The highest values were obtained with an electrolyte with ah ighs olutionc onductivity and large anodes. [43,55,78,79] Even higher power densitieso fu pt o 600 mWcm À2 were reportedf or Pt-based cathodes but for very short times. [80] Ad irect comparison can instead be made with previously presented work in which an identical (i)MFC configuration (single-chamber MFC), (ii)operating temperature [(22 AE 2) 8C], (iii)altitude above sea level [experiments conducted in Albuquerque, New Mexico, at 1500m above mean sea level (AMSL)],( iv) electrolyte (50 %a ctivated sludge, 50 %0 .1 m K-PB (potassium phosphate buffer), and 0.1 m KCl (potassium chloride)), (v) mixed-cultureb acteria, (vi)organic compound (sodium acetate), and (vii)membrane-less configuration were utilized. [22,23,38,43,46] The obtained maximump owerd ensity achieved [(243 AE 7) mWcm À2 ]i sone of the highestp ower densities recordedu nder the same operating conditions and is second only to that of (251 AE 2) mWcm À2 for Fe-AAPyr (AAPyr = aminoantipyrine). [43] The difference was just 3% and, therefore, the results can be considered to be comparable. This result is interesting because it indicates that Fe-BP(N)h as ah igh activity towards the ORR that is comparable with those of catalysts fabricated through high-temperature pyrolysis. Further investigations should focus on the stabilities and durabilities of these catalysts under long-term operation. This will certainly be an aspect that will be considered in future studies.

Conclusions
New FeÀNÀCc atalysts were obtained throught he deposition of iron(II) phthalocyanine on two different high-surface-area carbonaceous materials. These catalysts were fabricated without the utilization of high-temperature pyrolysis methods and deposited on (i)carbon black (Black Pearls) doped with nitrogen [Fe-BP(N)] and (ii)multi-walled carbon nanotubes doped with nitrogen [Fe-CNT(N)].T he catalystk ineticsw as studied using ar otatingr ing disk electrode (RRDE), and the results showedt he superiority of Fe-BP(N) in terms of its onset potential, half-wave potential, and limiting current. Fe-CNT(N) showedm uch higherp erformances compared with that of activated carbon (AC), which was used as ac ontrol. The catalysts were then integrated into air-breathing cathodes and tested in microbial fuel cells. Fe-BP(N) had the highest powerd ensity output of (243 AE 7) mWcm À2 ,w hich was over 90 %h ighert han that of AC [(127 AE 1) mWcm À2 ].

Experimental Section Preparation of N-doped carbonaceous supports
Multi-walled carbon nanotubes (CNTs) were purchased from Sigma-Aldrich, and Black Pearls 2000 (BP) were purchased from Cabot Corporation. The CNTsa nd BP were modified by at wo-step treatment with nitric acid and ammonia gas. In the first treatment, the materials were heated in concentrated HNO 3 (65 wt %) under reflux at 90 8Cf or 16 h. Then, the materials were collected by filtration and washed with distilled water until neutral pH was obtained. The materials were then dried in an oven at 70 8Co vernight and ground with an agate mortar and pestle. In the second treatment, af low of anhydrous ammonia was fed into in at ubular oven at T = 400 8C( heating rate 5 8Cmin À1 )f or 4h.T he obtained products were labeled as CNT(N) and BP(N).

Deposition of the Fe catalyst onthe carbonaceous supports
Iron(II) phthalocyanine (FePc, Aldrich;0.5 g) was dispersed in methanol (30 mL), and CNT(N) or BP(N) (0.5 g) was added. The mixture was stirred for 30 min in aw ater bath at 70 8Ct oe vaporate the methanol, and the resulting powder was dried completely in a vacuum oven at 70 8Cf or 3h to obtain samples labeled as Fe-CNT(N) and Fe-BP(N).

Catalysts surface chemistrya nd morphology
The surface chemistry of the catalyst was identified through highresolution XPS with aK ratos Axis Ultra DLD spectrometer.T hree separate areas of the same sample were analyzed, and the average values are presented. The average values have an error of less than 0.1 %. The high-resolution O1s, N1s, C1s, and Fe 2p spectra were obtained without the need for charge neutralization with a2 25 W AlK a monochromatic X-ray source. The acquired spectra were then processed with the CASAxps software.
The surface morphologies of the catalysts were investigated through SEM and TEM. The SEM images were recorded using aH itachi S-800 instrument at different magnifications. The TEM imaging was performed with aJ EOL 2010 instrument with samples on a copper grid. For simplicity,a ni mage for each sample at one magnification was shown as representative of the sample.

Cathode preparation
An air-breathing cathode configuration was adopted in this study. Am ixture of activated AC, CB, and PTFE was blended in ag rinder in aw eight ratio of 70:20:10 %, respectively.T he mixture was placed in ad ie pellet and pressed over as tainless-steel mesh (McMaster,USA), which was used as the current collector.The loading of the mixture was 40 mg cm À2 .Apure AC cathode was fabricated by the above method and used as ac ontrol. The cathodes containing Fe-BP(N) and Fe-CNT(N) were prepared by the same method with acatalyst loading of 2mgcm À2 .

RRDE analysis
The RRDE technique was used to study the catalyst kinetics of Fe-BP(N), Fe-CNT(N), and AC. The catalyst (5 mg) was mixed with 0.5 wt %N afion solution (FuelCellStore, USA;1 50 mL) and deionized water/isopropyl alcohol (DI/IPA) in a1 :1 ratio (850 mL). The obtained suspension was sonicated at least three times to obtain a uniform dispersion. The obtained ink was drop cast on the disk of ag lassy carbon working electrode. Twol oadings (0.1 and 0.6 mg cm À2 )were used for each catalyst. The tests were performed in as olution containing 0.1 m potassium phosphate and 0.1 m KCl. The solution simulates an electrolyte with circumneutral pH value. Before the experiments, the electrolyte was saturated with oxygen for at least 30 min. Linear sweep voltammetry was then run from + 0.5 to À0.7 V( vs. Ag/AgCl) at as can rate of 5mVs À1 .T he RRDE setup guarantees the possibility of measuring the current density produced by the disk (j D )b ut also the current density of the ring (j R )t oq uantify the intermediate (H 2 O 2 )p roduced during the ORR. From j D and j R ,t he number of electrons transferred (n)d uring the ORR can be calculated with Equation (1): Therefore, the percentage of H 2 O 2 produced during the ORR can be calculated using Equation (2):

MFC construction and operation
After the catalyst had been incorporated into the air-breathing cathode, the cathode was screwed on al ateral hole of am odified Pyrex bottle with av olume of 125 mL. [22,23] The cathode part containing the AC, CB, and PTFE pellet faced the liquid, and the current collector faced the air side. The geometric area of the cathode was 2.85 cm 2 .T he chamber was filled with as olution containing 50 vol %0 .1 m K-PB and 50 vol %a ctivated sludge from the Albuquerque Southeast Water Reclamation Facility located in Albuquer- que, New Mexico, USA. Precolonized and well-working anodes were moved from existing and running MFCs into MFCs with new and fresh cathodes. The anodes consisted of two carbon brushes with titanium cores (Millirose, USA) and ad iameter and height of 3cme ach. The anode area was decided to be much higher than the cathode area as the latter was the subject of the study.T he MFCs were left at the OCV for at least 3h until the output stabilized. Polarization curves were then recorded using two potentiostats (Biologic-USA, USA). The first potentiostat was connected in a two-electrode mode with the anode as the working electrode and the cathode as the counter electrode short-circuited with the reference channel. The second potentiostat was set up just to read the potentials of the anode and cathode versus the reference electrode (Ag/AgCl 3 m KCl). The polarization curves gave voltagecurrent curves as output. The power was calculated as the product of voltage and current. The current and power densities were shown as af unction of the geometric area of the cathode, which was 2.85 cm 2 .