Relevance of the Interaction between the M ... - ACS Publications

Sep 29, 2017 - and cobalt-phthalocyanines (FePc and CoPc) and carbon nanotubes (CNTs) used as ... conventional technologies: high energy density, zero...
3 downloads 0 Views 3MB Size
Subscriber access provided by UNIV OF ESSEX

Article

THE RELEVANCE OF THE INTERACTION BETWEEN THE M-PHTHALOCYANINES AND CARBON NANOTUBES IN THE ELECTROACTIVITY TOWARDS ORR Carolina González-Gaitán, Ramiro Ruiz-Rosas, Emilia Morallon, and Diego Cazorla-Amoros Langmuir, Just Accepted Manuscript • DOI: 10.1021/acs.langmuir.7b02579 • Publication Date (Web): 29 Sep 2017 Downloaded from http://pubs.acs.org on October 3, 2017

Just Accepted “Just Accepted” manuscripts have been peer-reviewed and accepted for publication. They are posted online prior to technical editing, formatting for publication and author proofing. The American Chemical Society provides “Just Accepted” as a free service to the research community to expedite the dissemination of scientific material as soon as possible after acceptance. “Just Accepted” manuscripts appear in full in PDF format accompanied by an HTML abstract. “Just Accepted” manuscripts have been fully peer reviewed, but should not be considered the official version of record. They are accessible to all readers and citable by the Digital Object Identifier (DOI®). “Just Accepted” is an optional service offered to authors. Therefore, the “Just Accepted” Web site may not include all articles that will be published in the journal. After a manuscript is technically edited and formatted, it will be removed from the “Just Accepted” Web site and published as an ASAP article. Note that technical editing may introduce minor changes to the manuscript text and/or graphics which could affect content, and all legal disclaimers and ethical guidelines that apply to the journal pertain. ACS cannot be held responsible for errors or consequences arising from the use of information contained in these “Just Accepted” manuscripts.

Langmuir is published by the American Chemical Society. 1155 Sixteenth Street N.W., Washington, DC 20036 Published by American Chemical Society. Copyright © American Chemical Society. However, no copyright claim is made to original U.S. Government works, or works produced by employees of any Commonwealth realm Crown government in the course of their duties.

Page 1 of 35

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

THE

RELEVANCE

OF

THE

INTERACTION

BETWEEN

THE

M-

PHTHALOCYANINES AND CARBON NANOTUBES IN THE ELECTROACTIVITY TOWARDS ORR Carolina González-Gaitán, Ramiro Ruiz-Rosas, Emilia Morallón, Diego Cazorla-Amorós Materials Institute of Alicante (IUMA), University of Alicante, Ap. 99, 03080, Alicante, Spain ABSTRACT In this work, the influence of the interaction between the iron and cobalt-phthalocyanines (FePc and CoPc) and carbon nanotubes (CNTs) used as support in the electroactivity towards oxygen reduction reaction (ORR) in alkaline media has been investigated. A series of thermal treatments were performed on these materials in order to modify the interaction between the CNTs and the phtalocyanines. The FePc-based catalysts showed the highest activity, with comparable performance to the state-of-the-art Pt-Vulcan catalyst. A heat treatment at 400 ºC improved the activity of FePc-based catalysts, while the use of higher temperatures or oxidative atmosphere rendered to the decomposition of the macrocyclic compound and consequently the loss of the electrochemical activity of the complex. CoPc-based catalysts performance was negatively affected for all the tested treatments. Thermogravimetric analyses demonstrated that the FePc was stabilized when loaded onto CNTs, while CoPc did not show such feature, pointing to a better interaction of the FePc instead of the CoPc. Interestingly, electrochemical measurements demonstrated an improvement of the electron transfer rate in thermally treated FePc-based catalysts. They also allowed us to assess that only 15% of the iron in the catalyst was available for direct electron transfer. This is the same iron amount that remains on the catalyst after a strong acid washing with concentrated HCl (ca. 0.3 wt.%), which is enough to deliver a comparable ORR activity. Durability tests confirmed that the catalysts deactivation occurs at slower rate in those catalysts where FePc is strongly attached to the CNT surface. Thus, the highest ORR activity seems to be provided by those FePc

ACS Paragon Plus Environment

Langmuir

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

molecules that are strongly attached to the CNT surface, pointing out the relevance of the interaction between the support and the FePc in these catalysts.

ACS Paragon Plus Environment

Page 2 of 35

Page 3 of 35

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

INTRODUCTION Fuel cells are energy production devices that constitute a promising technology for promoting clean energy generation. These devices have several advantages compared to other conventional technologies: high energy density, zero emission of pollutants, high efficiency, and, in some of their configurations, as the proton exchange membrane and alkaline fuel cells, low working temperature. This last mentioned feature also leads to an increase in the cathode overpotential for the oxygen reduction reaction (ORR), which drives the necessity of using highly active catalysts [1–3]. In the case of alkaline fuel cells, the reaction rate of ORR is higher than in acid medium, and the reaction can be catalyzed by a wider range of catalysts, which are more stable at high pH values [4]. The improved kinetic rate is related to the fact that alkaline medium confers reduced adsorption energies of anions, positively shifting the onset potential for electrochemical reduction of oxygen [5]. The most commonly used electro-catalysts are based on platinum and noble-metals as active phase, which have shown the highest activities to date [6]. However, their high cost, limited availability, high metal loading requirement and low resistance to catalyst poisoning in case of cell cross-over greatly hamper their performance [3,7]. Therefore, it is necessary to develop new catalysts, which must have a similar activity as noble-metal ones, while showing lower cost and higher chemical and electrochemical stability. Nanostructured carbon materials have high surface area that is readily accessible to the reagents, along with high electrical conductivity, resistance to electro-oxidation and a lower cost than current state-of-the-art catalysts [8,9]. Nevertheless, they have a low catalytic activity towards ORR, showing slow kinetics and low water selectivity. In this sense, nitrogen doped carbon materials loaded with metals (M-N/C) have been proposed as candidates to replace the noblemetal ORR catalysts and have become a major focus of the fuel cell research [10–21]. This interest arises from their outstanding improvement in the ORR performance regarding activity, selectivity to water and resistance against poisoning at the working conditions. M-N/C catalysts can be prepared following different synthetic routes, which include synthesis of non-noble metal nanoparticles (usually from transition metals as Fe, Co, Cu, Mn, etc.) and subsequently

ACS Paragon Plus Environment

Langmuir

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

supporting

them

on

N-doped

carbon

materials

Page 4 of 35

[7,15,16,22],

the

pyrolysis

of

metal/nitrogen/carbon compounds [12,13,20,21,23,24] and the use of M-N4 complexes supported on carbon materials [25–27]. In the first route, the strong interaction between the metal-containing particles and the support enhances the catalyst efficiency, reduces the loss of active sites and controls the charge transfer. The catalyst performance relies on the nanoparticles size, their distribution and dispersion on the support [17]. For the second route, the synthesis usually consists in the heat treatment under inert atmosphere in the presence of ammonia of a carbon material impregnated with a metal precursor [24]. Modifications of this protocol include the use of a solid or liquid nitrogen source [28], or employing M-N2 or M-N4 complexes as the nitrogen and metal source [12,13,23,29]. The active sites obtained through these synthesis have been identified as 1 metal atom coordinated by either 2 or 4 N atoms, with the former being regarded as the most active one [30,31]. In fact, only small metal loadings (2 wt% and even lower) are required to achieve ORR activity comparable to that of Pt-based catalysts. The enhancement in activity depends on the formation of these sites, which seems to be favored by the pyrolysis temperature, while the selection of the carbon support, the metal precursor and Ncontaining ligands is usually done trying to reduce the preparation costs [21]. In the case of the last synthesis route, transition metal complexes with N4 macrocycles are supported on carbon materials. Since the macrocycles include π-systems, these complexes are capable of undergoing fast redox processes, with minimal reorganization energies. They can act as mediator in electron transfer processes, which enhances the catalytic activity in several electrochemical reactions [18]. In particular, these compounds have a noticeable activity as ORR catalyst, being first reported in 1964 by Jasinski, who found that a complex formed by a N4-chelate with cobalt was electrochemically active for this reaction [14]. Since then, a large number of macrocyclic transition-metal compounds have been synthesized and successfully tested as ORR catalysts [32]. In this sense, phthalocyanines (Pc) are one of the most utilized compounds in the synthesis of M-N/C catalysts, being directly used as catalyst, but also as the M-N source in the M-N/C preparation routes based on pyrolysis of compounds adsorbed on

ACS Paragon Plus Environment

Page 5 of 35

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

carbon materials. For instance, Cao and co-workers have found high activities compared to Pt in alkaline media for an ORR catalyst using pyridine-functionalized carbon nanotubes (CNTs) in order to anchor FePc molecules providing an axial ligand for the iron center [33]. Pc are macrocyclic compounds combining eight N atoms in its structure that are able to coordinate different metal elements (MPc). MPc such as Fe and Co macrocycles have shown a suitable activity and remarkable selectivity compared to the Pt-based catalysts [18,19]. In addition, they show a high resistance to poisoning with alcohols which is a big concern in noble metal catalysts [20,21]. The major drawback of these materials is the low stability at the working conditions in the fuel cell [34]. This fact has been studied and there are several hypothesis about the deactivation of the catalyst: it can be related to the decomposition of the compound via hydrolysis in the electrolyte and loss of the conjugation in the macrocycle, or an attack by the hydrogen peroxide formed during the ORR, which causes the oxidation of the nitrogen atoms, losing the coordination with the metal [20,21,34]. It has been found that, depending on the carbon support, the metal content and the heat treatment, the electrocatalyst stability can be improved. However, the mechanism is not fully understood yet [20] and the stability and durability is still poor for practical use. In the present work, cobalt (CoPc) and iron (FePc) phthalocyanines supported on multiwall carbon nanotubes were submitted to different treatments in order to modify the interaction between the active phase and the support. The prepared materials have been studied as electrocatalysts towards ORR in alkaline medium. Special emphasis has been made on determining the importance of the strength of the interaction between the phthalocyanine and the support, using acid washes for the removal of weakly immobilized FePc. The electrochemical behavior, thermal stability and surface chemistry of the resulting materials have been determined, while their electrocatalytic performance towards ORR in alkaline medium, including their stability and resistance to methanol poisoning under potentiostatic conditions has been assessed.

ACS Paragon Plus Environment

Langmuir

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 6 of 35

EXPERIMENTAL Reagents Multiwall carbon nanotubes (CNTs) were purchased from Cheap Tubes Inc. (Brattleboro, Vt, USA) with a 95% of purity and they were used without further purification. N,Ndimethylformamide (DMF), potassium hydroxide (KOH), cobalt phthalocyanine (97% purity), iron phthalocyanine (90% purity) and platinum on Vulcan (20% loading) were purchased from Sigma-Aldrich. Methanol (99.8%) was purchased from VWR-Chemicals Prolabo. All the solutions were prepared using ultrapure water (18 MOhm cm Elga Labwater Purelab water system). The gases N2 (99.999%), O2 (99.995%) and H2 (99.999%) were provided by Air Liquide and were used without any further purification or treatment. Synthesis of N-metal modified CNTs Scheme 1 presents a summary of all the N-metal modified CNTs used along this work. N-metal modified CNTs with a metal loading of around 2 wt% were prepared using the incipient wetness impregnation method. The pristine CNTs were used as supports. First, 50 mg of the CNTs were dried in a vacuum oven at 80 °C. Next, 15.5 mg of FePc and 13.7 mg of CoPc were dissolved in 1.8 ml of DMF. These solutions were added separately to 50 mg of CNTs previously outgassed at 80 ºC under vacuum. The mixture was dried in an oven at 200 °C for 12 h, resulting in NT_FePc and NT_CoPc samples.

Impregnation CoPc 21.5% CoPc, 2.2% Co

NT_CoPc

400°C Heat treatment 800°C 100% N2

NT_CoPc_400 NT_CoPc_800

CNTs Impregnation FePc 23.7% FePc, 2.3% Fe

NT_FePc

400°C Heat treatment 800°C 100% N2 Heat treatment 500°C 3125 ppm O2, 99.7% N2

NT_FePc_400

Acid wash 37% HCl

NT_FePc_400_HCl37%

NT_FePc_800 NT_FePc_500O Acid wash 37% HCl

NT_FePc_500O_HCl37%

Scheme 1. Summary of the N-metal modified CNTs used along this work.

ACS Paragon Plus Environment

Page 7 of 35

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

These samples were subsequently heat treated in a tubular furnace under nitrogen atmosphere at 400 and 800 ºC for 30 min using a heating rate of 20 ºC min-1 in order to check the effect of thermal treatments on their activity. These samples are denoted according to the temperature of treatment as NT_MPc_T, where M is the metal and T is the heat-treated temperature, respectively. NT_FePc samples were also treated under a slightly oxidizing mixture of gases (3125 ppm O2 in N2) at 500 ºC for 30 minutes, resulting in NT_FePc_500O sample. In order to study the interaction between the phthalocyanine compound and the support in the catalytic activity in the FePc samples series, 20 mg of the catalysts were washed in 150 ml HCl (37%) under continuous stirring. After the first 2 h, the acid was replaced by a fresh one and left under stirring overnight. Then the samples were rinsed with water until neutral pH. Finally, the samples were dried in an oven at 120 °C overnight, resulting in NT_FePc_400_HCl37% and NT_FePc_500O_HCl37% samples. Physicochemical characterization The CNTs were characterized by Transmission Electron Microscopy (TEM) coupled to energy dispersive X-Ray analysis (EDX) with a JEOL JEM-2010 microscope operating at 200 kV with a spatial resolution of 0.24 nm. The TEM images (Figure S1) revealed that the diameter of the nanotubes varies from 6 to 10 nm. They have a multiwall structure, formed by 3 – 5 concentric layers, and a smooth and homogeneous surface. The surface area (SBET) of the CNTs was calculated from the N2 adsorption isotherm at −196 °C, which was determined in an automatic adsorption system (Autosorb-6, Quantachrome). Prior to the measurements, the samples were degassed at 250 °C for 4 h. Thermogravimetric analyses were carried out in a thermobalance (SDT 2960 instrument, TA). After a purging time of 1 hour, the samples were heated up to 800 ºC at 20 ºC min-1 in nitrogen atmosphere. The surface composition and oxidation states of the species in the catalysts were studied by XPS, using a VG-Microtech Mutilab 3000 spectrometer and Al Kα radiation (1253.6 eV). The deconvolution of the N1s and metal XPS spectra were done by least squares fitting using Gaussian-Lorentzian curves, while a Shirley line was used for the background determination.

ACS Paragon Plus Environment

Langmuir

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 8 of 35

The quantification of the metal content of the prepared catalysts based on iron and cobalt were studied by ICP – OES. A Perkin Elmer (Optima 4300DV) spectrometer was used for the analysis. The samples were treated in acid aqueous solutions (HNO3 and HCl in a molar ratio of 1:3) in an ultrasound bath for 15 mins in order to extract the metals loaded in the catalysts. After this treatment, the solutions were diluted to have the appropriate concentration for the analysis. Electrochemical measurements The electrochemical characterization of the electrodes was performed in an Autolab PGSTAT302 (Metrohm, Netherlands) potentiostat using a standard three-electrode cell configuration. A rotating ring-disk electrode (RRDE, Pine Research Instruments, USA) equipped with a glassy carbon disk (5.61 mm diameter) and an attached platinum ring was used as the working electrode, a platinum wire was used as the counter electrode and a reversible hydrogen electrode (RHE) as the reference electrode. The glassy carbon disk was modified with the samples using 76 µl of a 0.25 mg ml-1 dispersion (50 % isopropanol, 0.02 % Nafion®), obtaining a catalyst charge of 0.08 mg cm-2. The electrochemical behavior was studied by cyclic voltammetry (CV) and linear sweep voltammetry (LSV) in 0.1 M KOH between 0.0 and 1.0 V (vs. RHE). The former was done in a N2-saturated atmosphere at 50 mV s-1, while the later measurements were performed in an O2saturated atmosphere at a rotation rate of 1600 rpm and at a scan rate of 10 mV s-1, while the potential of the Pt ring was held constant at 1.5 V (vs. RHE). The onset potential was measured at a current density of 0.1 mA cm-2 for all samples. The electron transfer number of the reaction was calculated from the hydrogen peroxide oxidation in the Pt ring as follows:  =

     ⁄

Eq. 1

Where Ir and Id stand for the currents measured at the ring and the disk, respectively, and N is the collection efficiency of the ring, which was experimentally determined to be 0.37. Chronoamperometric experiments were performed at 0.65 V (vs. RHE) in order to study the stability of the electrodes. The tests lasted for 2 hours, and the currents in the disk and ring were

ACS Paragon Plus Environment

Page 9 of 35

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

tracked during the duration of the analyses. The crossover effect of methanol in the catalyst was also studied by the addition of methanol to achieve a concentration of 2.5 M during the measurements. The surface concentration (Γ) was calculated from the redox processes related to MPc observed in the cyclic voltammetry after subtraction of the double layer charge.

RESULTS AND DISCUSSION Electrochemical characterization Fig. 1 shows the cyclic voltammetry in 0.1 M KOH for all samples. The CV of CNTs shows a typical rectangular shape, characteristic of capacitive behavior corresponding to the double layer formation. In contrast, the MPc supported on CNTs (NT_CoPc and NT_FePc samples) show different redox processes depending on the metal. In the case of cobalt sample (NT_CoPc, Fig. 1a), a redox process is observed at 0.37V, which is related to the Co(I)/Co(II) redox process of the adsorbed complex [35]. On the other hand, NT_FePc sample shows two redox processes at 0.25 and 0.80 V, corresponding to the Fe(I)/Fe(II) and Fe(II)/Fe(III) couples from coordinated metal in the phthalocyanine complex, respectively (Fig. 1b), [36]. The amounts of electroactive cobalt and iron have been estimated from the electrical charge measured between 0.25 and 0.55 V and between 0.65 and 0.95 V, respectively, after correcting the double layer contribution. Values of 4.45 and 0.98 C g-1 have been obtained, which correspond to 0.29 and 0.06 wt.% of metal content for NT_CoPc and NT_FePc, respectively. From the ICP determinations, the amounts measured are 2.4 and 2.1 wt.% for CoPc and FePcbased catalysts, respectively, pointing out that most of the loaded metal is not electrochemically active in these samples.

ACS Paragon Plus Environment

60 40 20 0 -20 -40 -60 -80

60 40 20 0 -20 -40 -60 -80

Page 10 of 35

(a)

CNTs NT_CoPc NT_CoPc_400 NT_CoPc_800

-0.1 0.1

C / F g-1

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

C / F g-1

Langmuir

0.3 0.5 0.7 0.9 1.1 E vs. RHE / V

(b)

CNTs NT_FePc NT_FePc_400 NT_FePc_800

-0.1 0.1

0.3 0.5 0.7 0.9 E vs. RHE / V

1.1

Fig. 1 Cyclic voltammetry of (a) Co-based samples and (b) Fe-based samples in N2-saturated 0.1 M KOH at 50 mV s-1.

Different behaviors have been found after the heat treatments of FePc and CoPc-based catalysts. After the treatment at 400 °C both samples still show the redox processes associated to the metal center of the N4-chelate, though slight changes in the potentials can be seen. In the case of NT_CoPc_400, the redox peak of cobalt is broader and less defined after the heat treatment. However, the currents of the redox processes associated to NT_FePc_400 sample are much higher, pointing out a remarkable enhancement of the interaction between the FePc and the CNTs after the treatment, that leads to a better electron transfer. Thus, the amount of iron electrochemically active in this sample increases to 0.26 wt%. Interestingly, the position of the redox peaks shifted in a different direction after the 400 °C treatment. For the Co-based samples, the peak potential shifts to less positive values than the initial one, while the Fe-based materials show a positive shift. A positive shift has been seen before in other studies of heat

ACS Paragon Plus Environment

Page 11 of 35

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

treated M-phthalocyanines, which could be a factor that favors the O2 reduction [37,38]. On the contrary, the heat treatment at 800 °C leads to a major decrease of the redox processes of these samples, and the CVs are similar to that of bare CNTs. This result is probably related to the decomposition of the macrocyclic complex, which was confirmed in the TG measurements shown in following sections. Electrocatalytic activity towards ORR The electroactivity of the catalysts towards ORR was studied in O2-saturated 0.1 M KOH electrolyte. The analysis was performed by LSV using a RRDE at 1600 rpm. Fig. 2 shows the LSV curves at 1600 rpm for all Fe and Co samples. Tafel plots have been directly calculated from LSV curves and the slopes in the kinetic control region (low overpotentials) have been calculated and reported in Table 1. Measurements corresponding to bare CNTs and a commercial sample of 20% Pt-Vulcan are included for comparison purposes. Table 1 also compiles the onset potential and number of electrons transferred derived from the RRDE experiments. When compared to bare CNTs, all the tested electrocatalysts displayed an enhanced activity towards ORR. All of them show higher onset potential values and limiting current density, confirming that MPc has an important role in the ORR activity. It can be also seen that CoPcloaded catalysts show a lower performance than FePc-loaded ones, which is in consonance with experimental findings already reported in literature [7,39], where the higher onset potential of FePc over CoPc catalysts has been addressed and connected to their different redox potentials [18]. The onset potential for the ORR is close to the reduction of the metal center M(III)/M(II) in the case of the FePc-based catalysts as can be seen the recorded voltammograms in Fig. 1. Additionally, in the LSV for these catalysts, the limiting currents are not flat and show a minimum at ~0.25V, which is close to the redox potential of the Fe(II)/Fe(I) shown in Fig. 1. As it was proposed in previous studies, this could be an indicator that at this point, the oxidation state of the catalyst is not active for the ORR, being related to a weaker binding energy between Fe(I) and O2, leading to H2O2 formation at lower potentials [40,41]. As for the effect of the heat

ACS Paragon Plus Environment

Langmuir

treatment, it seems that there exists a relationship between an improved ORR catalytic activity and the current of the redox processes (which was affected by the heat treatments) for the metal center registered in O2-free CV measurements (Fig. 1). In this sense, recent studies have related the ORR activity and onset potential of Fe-N4/C sites to the Fe(II) oxidation state [42].

0

-2

CNTs NT_FePc NT_FePc_400 NT_FePc_800 Pt-Vulcan

-1 j / mA cm-2

j / mA cm-2

0

CNTs NT_CoPc NT_CoPc_400 NT_CoPc_800 Pt-Vulcan

-1

-3 -4 -5

-2 -3 -4 -5

(a) -6

(b)

-6 -0.1 0.1

0.3 0.5 0.7 E vs. RHE / V

0.9

1.1

-0.1 0.1

4

4

3.5

3.5

3

3

2.5

2.5

n

n

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 12 of 35

2

2

CNTs NT_CoPc NT_CoPc_400 NT_CoPc_800 Pt-Vulcan

1.5 1 -0.1

1.5 1 0.7

0.9

1.1

CNTs NT_FePc NT_FePc_400 NT_FePc_800 Pt-Vulcan

(c)

0.1 0.3 0.5 E vs. RHE / V

0.3 0.5 0.7 E vs. RHE / V

-0.1

0.1 0.3 0.5 E vs. RHE / V

0.7

Fig. 2 Linear sweep voltammetry (a,b) and electron transfer number (c,d) calculated from RRDE experiments of (a, c) Co samples and (b, d) Fe samples in an O2-saturated 0.1 M KOH at 10 mV s-1 and 1600 rpm. Bare CNTs and 20%Pt-Vulcan catalyst are also included for comparison purposes.

The NT_CoPc and NT_CoPc_400 catalysts display similar ORR activity, the limiting current is not reached and they show a two-wave processes, which could be related to a 2 + 2 electron mechanism for ORR. First, a 2-electron reduction reaction would generate H2O2 with a subsequent 2-electron reduction to form H2O (OH- in alkaline media), which occurs at different

ACS Paragon Plus Environment

Page 13 of 35

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

potentials. This was confirmed following the registered current in the ring (Fig. 2c). At 0.7 V the electron transfer number was 2.13 and 2.31 respectively for NT_CoPc and NT_CoPc_400 catalysts, and at 0.0 V the electron transfer number was 3.04 and 2.83, for NT_CoPc and NT_CoPc_400, respectively. Table 1 Electrochemical parameters calculated from the RRDE experiments of the different electrocatalysts in O2-saturated 0.1 M KOH at 10 mV s-1 and 1600 rpm.

Eonset at 0.1 mA cm-2 V vs. RHE

Tafel slope mVdec-1

Γ µmol cm-2

--

--

CNTs

0.74

n at 0.4 V 2.23

NT_CoPc

0.83

2.46

27

46

NT_CoPc_400

0.83

2.42

26

62

NT_CoPc_800

0.85

2.34

49

1

NT_FePc

0.92

3.51

27

10

NT_ FePc_400

0.94

3.86

23

33

NT_FePc_800

0.88

2.81

32

1

20% Pt-C

0.97

3.92

--

--

Sample

The preferential occurrence of a 2 e- mechanism instead of a 4 e- ORR mechanism in CoPcbased catalysts is in agreement with DFT simulations of the reaction mechanisms in FePc and CoPc [43,44]. These studies reported that the O-O bond of an adsorbed O2 molecule could be weaken (and therefore the 4 e- ORR pathway would be favored) depending on the adsorption configuration, with side-on configurations being more effective than end-on configurations for a 4 e- pathway [43,45]. DFT calculations demonstrated that end-on configurations seem to be energetically stable for O2 on CoPc, while side-on is preferred on FePc. Therefore, a 2 + 2 ereaction pathway at low potentials is proposed for ORR in these CoPc-based catalysts, in which the occurrence of hydrogen peroxide reduction is probably necessary to achieve the second pair of electrons [46]. Conversely, the NT_CoPc_800 sample reaches a limiting current and does not display a twowave process. The electron transfer number slightly changes during all potential range and is

ACS Paragon Plus Environment

Langmuir

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

lower in this case, while the Tafel slope increases, all of which is in agreement with literature claims about prejudicial effect of thermal treatment above 600ºC of supported phthalocyanines [7]. The surface concentration decreases importantly after heat treatment at 800ºC due to either agglomeration or partial loss of CoPc during heat treatment. Previous studies about the effect of the heat treatment of CoPc shows that at 800 ºC, peroxide formation is maximized, a feature that could be related to the destruction of the Co-N4 complex and the formation of Co-N2 active sites [47]. The increase in the onset potential could be related to the incorporation of nitrogen groups in the CNTs structure, that could improve electrical conductivity and serve as ORR catalysts [48,49]. This is the case of pyridinic functions, being known to be active sites for ORR [50–52], although not necessarily selective towards water formation [53]. The FePc-based samples show an excellent activity towards ORR, reaching in some cases values very close to the Pt-based catalyst. These results are in agreement with previous studies that showed the remarkable activity of FePc complex supported on carbon nanotubes [13,39], graphene or Vulcan [25,45]. As previously mentioned, the higher activity can arise from the side-on oxygen adsorption mode in the vicinity of the Fe center, which is preferred in case of FePc [43–45], easing the breaking of the O-O bond, a prerequisite for enabling the 4 e- pathway. The activity towards ORR changes depending on the performed heat treatment in the samples, following the order of activity: NT_FePc_400 > NT_FePc > NT_FePc_800. NT_FePc shows an onset potential close to the Pt-based catalyst, nonetheless, the limiting current density is lower owing to the lower number of electron transferred. This fact seems to be overcome when the sample is heat treated at 400°C, which shows an ORR performance comparable to that of the PtVulcan catalyst. The enhanced activity is probably related to the improved electrochemical interaction between the FePc and the CNTs (Fig. 1). The heat treatment of NT_FePc at 800 ºC rendered to a decrease in the ORR performance, and an increase in the Tafel slope (Table 1), a feature similar to that seen for NT_CoPc_800 catalysts and that can be explained by the breakage of the macrocycle structure.

ACS Paragon Plus Environment

Page 14 of 35

Page 15 of 35

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

It is interesting to note that the ORR activity achieved with NT_FePc_400 outmatches what was reported in previous works, where a loading of 1:1 or higher FePc/CNT ratios were employed [13,39,54]. This is probably related to the poor electrical conductivity of phthalocyanines, which is considered to be one of their drawbacks as electrocatalysts, making necessary to disperse them on highly conductive surfaces in order to improve their performance [44]. Considering the surface area of CNTs used in this work (SBET of 405 m2 g-1, as calculated from the N2 adsorption isotherm) and the area covered by a FePc molecule (that is 1.15 nm2, calculated from the side length of FePc, 1.07 nm and considering square geometry), a monolayer of supported FePc over the carbon nanotubes employed in this work would be achieved at a 0.34:1 FePc/CNT weight ratio, which is close to the 0.31:1 ratio employed in this work and also in others where a similar ORR performance of FePc/C catalysts was observed [45]. Excess of FePc would be undesirable, not only because the FePc that is not in direct contact with the surface of CNT is not active (an effect already found in the catalysts herein reported, as previously discussed in the previous section), but also due to the oxygen diffusional constrains that it would render, making it less accessible to the FePc molecules located over the surface of CNT, which are expected to be the most active centers for ORR. This idea will be addressed in more detail later. Interestingly, the highest surface concentration is obtained for the material prepared after heat treatment at 400ºC (Table 1). Surface chemistry and thermogravimetric analyses

Table 2 shows the atomic composition for all samples calculated by XPS. The results confirm the incorporation of the phthalocyanine in all the prepared catalysts (N and Fe/Co are found in atomic ratios close to 8:1 for all samples), although the metal content determined by this technique was much lower than expected. Given the surface character of this technique, the metal content of the MPc over CNTs catalysts has been also determined by ICP-OES, obtaining values between 2.1 – 2.4 wt% (Table 2), which are in agreement with the formulation of the samples. The lower amount of metal detected by XPS can be related to the preferential stacking of phtalocyanines molecules in several layers due to a strong intermolecular interaction [55].

ACS Paragon Plus Environment

Langmuir

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 16 of 35

After the heat treatment at 400 ºC, the metal and nitrogen content barely changed, whereas a much different behavior was observed when the samples were treated at 800 ºC. Both the nitrogen and metal content decreased, a feature that is attributable to the thermal decomposition of the Pc. ICP analyses are in agreement with this finding, yielding iron and cobalt amount around 1.0 wt.% after the heat treatment at 800 ºC. Table 2 Chemical composition calculated by XPS and ICP-OES

XPS Sample

ICP

C1s / at %

N1s / at%

O1s / at%

M 2p / at% (wt%)

M / wt%

CNTs

98.1

--

1.9

--

--

NT_CoPc

93.9

3.5

2.3

0.3 (1.7)

2.1

NT_CoPc_400

91.5

3.7

4.4

0.4 (1.7)

2.1

NT_CoPc_800

97.3

1.1

1.5

0.1 (0.7)

1.1

NT_FePc

92.6

2.8

4.3

0.3 (1.1)

2.4

NT_ FePc_400

95.6

2.0

2.2

0.2 (0.7)

2.1

NT_FePc_800

97.5

0.5

1.9

0.1 (0.4)

0.9

The absence of changes in the molecular structure of CoPc and FePc at 400 ºC was corroborated by thermogravimetric analyses of bare and CNT supported phthalocyanines (Fig. 3), where only small mass losses (that are slightly larger in the case of the raw FePc, which is expected given its lower purity) are detected up to 550 ºC, confirming that the macrocyclic compound remains unaffected. Attending to the TG profiles of bare Pcs, a first decomposition stage occurs between 550 and 650 ºC. The resulting pyrolyzed products can undergo further decomposition reactions, as pointed out by the weight loss registered at temperatures higher than 750 ºC. Interestingly, when FePc is supported on the CNTs (Fig. 3a), the thermal decomposition seems to be delayed to higher temperature (630-650 ºC), and a much lower weight loss than expected is attained (only 3.5 %, while 11.8 % is expected considering the amount of Pcs in the catalyst, 23.7 %). An additional weight loss at temperatures between 300 and 450 ºC can be also seen in the case of the NT_FePc sample, which can be ascribed to the desorption of chemisorbed DMF employed in the impregnation step – corroborated by TG-MS experiments, following the m/z

ACS Paragon Plus Environment

Page 17 of 35

lines of 44 and 73 associated to the DMF presence –. Differently, in the case of CoPc loaded on the CNTs the thermal decomposition is not delayed, and the process also shows the expected weight loss in the corresponding temperature range (Fig. 3b). The high thermal stability of FePc and its tendency to catalyze nitrogen fixation under thermal treatment up to 600 ºC has been previously reported for the preparation of carbon alloy catalysts using FePc/phenolic resin mixtures [56]. The huge impact of the carbon support as a driving agent of the pyrolysis mechanism of CoPc and FePc has been also proposed by Bambagioni et al., who detected the formation of different Pc gaseous and solid fragments during the pyrolysis of Pc supported on carbon black, but only found sublimated phthalocyanines as the product of quartz-supported CoPc and FePc [57]. The TG results support that there exists a strong interaction between the FePc and the CNTs, while this interaction is weaker in the case of CoPc.

1

(a)

W/W0

0.9 0.8 0.7 0.6 0.5 0.4

FePc NT_FePc NT_FePc_400 100 200 300 400 500 600 700 800 T / ºC

1

(b)

0.9 W/W0

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

0.8 0.7 0.6 0.5

CoPc NT_CoPc

0.4 100 200 300 400 500 600 700 800 T / ºC Fig. 3 Weight loss during TG analyses of (a) FePc, NT_FePc and NT_FePc_400, (b) CoPc and NT_CoPc

ACS Paragon Plus Environment

Langmuir

The changes in the oxidation state of iron and cobalt-based catalysts have been analyzed by XPS (Fig. 4). In the case of CoPc-loaded CNTs, Fig 4a shows two broad and asymmetric peaks with a spin-orbit splitting of about 15 eV. The main Co 2p3/2 component is located at ~780.5 eV and is accompanied by a low-intensity shake-up satellite shifted by 9-10 eV to higher binding energies. The XPS spectrum corresponds to the Co(II) species [58]. This peak is not shifted either when the CoPc is supported on the CNTs or after the heat treatment at 400º C (red and green spectra in Fig. 4a, respectively), pointing out that the N-Co bond and probably the structure of the coordination complex is not affected. The oxidation state seems to be unaltered even after the heat treatment at 800 °C, where the maximum of the spectrum is again found at the same binding energy (although a much lower intensity is recorded due to the loss of cobalt after the heat treatment).

NTCoPc800

(a)

(b)

NTFePc800

Counts / a.u

NTCoPc400

Counts / a.u

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 18 of 35

NTCoPc

CoPc

802

797

NTFePc400

NTFePc

FePc

792 787 782 B.E. / eV

777

772

734

729

724 719 714 B.E. / eV

709

704

Fig. 4 (a) Co 2p XPS spectra and (b) Fe 2p XPS spectra for all samples.

On the other hand, the Fe samples show a different behavior. The determination of the Fe metal species using Fe 2p3/2 XPS region is difficult because it has a complex multiplet structure, due to the coupling of the core hole and the open valence shell of the Fe atom [13,59]. However, from Fig. 4b, the peak at ~710.1 eV found at the FePc, which must be related to Fe(II) of the phthalocyanine complex, is slightly shifted to more positive binding energies in the NT_FePc

ACS Paragon Plus Environment

Page 19 of 35

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

and NT_FePc_400 samples, which are now located at 710.4 and 710.9 eV, respectively. This feature seems to be related to a stronger interaction between the carbon nanotubes and the FePc than in the case of the CoPc, thus, leading to a decrease in the electron density of the Fe atom [25]. When the heat treatment temperature was increased at 800 ºC, the iron remaining in the NT_FePc_800 catalyst was found to be reduced, as pointed out by the negative shift of binding energy of the maximum of the XPS spectra, located at 707.3 eV, a value that corresponds to metal iron species. A similar result has been found for carbon nanofibers, where a heat treatment at 1000ºC of the FePc supported on carbon nanofibers leads to the formation of metal iron (or its carbide) [60]. The presence of Fe(II) species coordinated with the nitrogen atoms in the phthalocyanine structure are known to be necessary for an enhanced electrocatalytic activity [61], and the formation of metal iron may be related to the activity loss in the sample NT_FePc_800. Fig. 5 shows the N1s spectra for all samples. The spectra of the unsupported metal phthalocyanines show a peak at ~398.8 eV with small contributions at higher binding energies. Although metal phthalocyanines have two different N atoms in the molecule, it produces only one N1s peak at around 398.8 eV since both contributions are separated by only 0.3 eV in binding energy which is beneath the energy resolution of the spectra [62]. The contribution observed at higher binding energies (400.2 eV) can be related to the impurities in these metal phthalocyanines [63,64]. In supported CoPc samples, neither NT_CoPc nor the NT_CoPc_400 samples show any significant differences in the N1s spectrum compared to the initial CoPc. Different behavior is found for Fe-containing samples. The NT_FePc and NT_FePc_400 materials show a change in the position of the peaks from the initial iron phthalocyanine, which could be related to the enhanced interaction with the carbon support, leading to a shift in the position of the peaks. When the heat treatment is performed at 800 ºC notable changes in the N1s spectrum of both NT_CoPc_800 and NT_FePc_800 samples can be seen (Fig. 5). After the heat treatment, the spectra became wider in the BE region and the features characteristic of carbon materials

ACS Paragon Plus Environment

Langmuir

appear. Thus, a new peak is seen at 401.0 eV, that could be attributed to the formation of quaternary nitrogen species, where N atoms from the macrocycle could be incorporated to the graphene layer [65]. The peak at 398.7 eV can either be due to remaining Fe-N4 sites or to the formation of pyridine groups from N incorporation into the carbon nanotubes [65]. The peak at 400 eV can be assigned to positively charged N species like pyrrole or pyridine groups [65]. These results are in agreement with previous studies, where it was found that the presence of Co during the heat treatment of nitrogen-containing polymers and molecules induces the formation of higher amount of pyridinic and quaternary nitrogen groups that are active toward the ORR [66].

NT_CoPc_800

NT_FePc_800

x6

x6

NT_CoPc_400

NT_FePc_400

x3

NT_CoPc

Counts / a.u

Counts / a.u

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Page 20 of 35

x3 NT_FePc

x3

x3

FePc

(a)

CoPc

402

400 398 B.E. / eV

396

(b)

402

400 398 B.E. / eV

396

Fig. 5 N1s XPS spectra of (a) CoPc samples, and (b) FePc samples.

The effect of acid washing in ORR activity

The role of the amount of Fe in FePc-based catalysts towards ORR was studied by testing the activity of acid washed FePc-based catalysts. This study was done for NT_FePc_400 and NT_FePc_500O samples. NT_FePc_500O was prepared by submitting the FePc loaded to the CNTs to a heat treatment using a slightly oxidant atmosphere. This treatment is performed in

ACS Paragon Plus Environment

Page 21 of 35

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60

Langmuir

order to deliver the oxidation of the macrocycle. As a result, the ICP-OES analysis shows a complete iron removal in NT_FePc_500O after the acid washing. Nonetheless, the sample NT_FePc_400 still showed a small amount of Fe (0.3 % wt). This fact can be attributed to the good interaction of the Fe atoms in the phthalocyanine compound, which, as discussed above, remains intact after the heat treatment at 400 ºC. On the other hand, the oxidative treatment at 500 ºC leads to the partial decomposition of the macrocyclic structure and the oxidation of iron. The XPS N1s spectrum of NT_FePc_500O shows an increase in the amount of oxidized nitrogen that may result in the breakage of the N4-chelate species (not shown). In addition, the Fe2p XPS spectrum shows a peak at higher binding energies (712.0 eV), attributed to the Fe(III) species [67]. A much lower nitrogen content is also observed in the catalyst (0.9 at.%) due to the oxidative decomposition of the phthalocyanine compound, which facilitates the Fe removal during the acid washing process. Very interestingly, the iron content remaining in the sample NT_FePc_400 is close to the calculated amount in previous section from the cyclic voltammetry of these samples (ca. 0.26%). It seems that the amount of remaining iron is the one that is on Pc directly interacting with the CNT surface and therefore detectable by electrochemical measurements. Fig. 6 shows the LSV curves at 1600 rpm for the acid washed and parent FePc-based catalysts. As expected, the NT_FePc_500O seems to lose most of its ORR activity when iron is removed, presenting similar activity than the pristine CNTs. In fact, its ORR performance is comparable to that found for CNTs heat treated using a similar oxidant atmosphere [49]. It seems that the presence of iron within the Pc is necessary for promoting the catalytic activity. Contrarily, the activity of acid washed NT_FePc_400 is still high, with the onset potentials scarcely shifted to a lower value, but still close to the Pt-based catalyst. Consequently, the limiting current density only slightly decreased when compared to the parent samples with a much higher amount of Fe. This is in agreement with previous studies by Lefèvre and Dodelet, who reported that Fe content as low as 0.5% is enough for achieving a 4 e- pathway (