- Email: [email protected]
Chemistry of carbon polymer composite electrode – An X-ray photoelectron spectroscopy study
Journal of Power Sources 274 (2015) 1217e1223
Contents lists available at ScienceDirect
Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour
Chemistry of carbon polymer composite electrode e An X-ray photoelectron spectroscopy study Shuang Ma Andersen*, Rajnish Dhiman, Eivind Skou Department of Chemical Engineering, Biotechnology and Environmental Technology, University of Southern Denmark, Niels Bohrs All e 1, DK-5230 Odense M, Denmark
h i g h l i g h t s Surface oxidation state of PEMFC components and different electrodes were compared. The surface property of the catalyst layer is influenced by the electrode structure. Lamination can induce structure change in electrode.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 21 June 2014 Received in revised form 10 October 2014 Accepted 21 October 2014 Available online 29 October 2014
Surface chemistry of the electrodes in a proton exchange membrane fuel cell is of great importance for the cell performance. Many groups have reported that electrode preparation condition has a direct influence on the resulting electrode properties. In this work, the oxidation state of electrode components and the composites (catalyst ionomer mixtures) in various electrode structures were systematically studied with X-ray photoelectron spectroscopy (XPS). Based on the spectra, when catalyst is physically mixed with Nafion ionomer, the resulting electrode surface chemistry is a combination of the two components. When the electrode is prepared with a lamination procedure, the ratio between fluorocarbon and graphitic carbon is decreased. Moreover, ether type oxide content is decreased although carbon oxide is slightly increased. This indicates structure change of the catalyst layer due to an interaction between the ionomer and the catalyst and possible polymer structural change during electrode fabrication. The surface of micro porous layer was found to be much more influenced by the lamination, especially when it is in contact with catalysts in the interphase. Higher amount of platinum oxide was observed in the electrode structures (catalyst ionomer mixture) compared to the catalyst powder. This also indicates a certain interaction between the functional groups in the polymer and platinum surface. © 2014 Elsevier B.V. All rights reserved.
Keywords: XPS PEMFC Electrode structure Interaction Lamination Interphase
1. Introduction Many advances have been achieved since the first demonstration of proton exchange membrane fuel cell (PEMFC). Non-precious metal catalysts [1], corrosion resistant catalyst supports [2] and improvements of catalyst e support attachment [3] are important milestones in the history of PEMFCs. Meanwhile, understanding and optimization of the electrode structure: water management [4], catalyst ionomer ratio [5], electrode fabrication conditions [6,7], are also of crucial importance for the advancement of the technology. Surface chemistry and morphology of the electrode, though still not well understood, are essential parameters influencing the
* Corresponding author. E-mail address: [email protected] (S.M. Andersen). http://dx.doi.org/10.1016/j.jpowsour.2014.10.162 0378-7753/© 2014 Elsevier B.V. All rights reserved.
cell performance [8,9]. The delicate three-phase boundary (TPB), where proton, electron and gas meet and initiate the electrochemical reactions, has great impact on the catalyst utilization and cell lifespan. Though the thickness of the catalyst layer (CL) is only around 20 mm, which is less than 5% of the total thickness of a membrane electrode assembly (MEA), CL is the heart of PEMFC, and is responsible for hydrogen oxidation reaction (HOR) and oxygen reduction reaction (ORR). State of the art PEMFC catalyst layer consists of platinum supported by carbon and impregnated with proton conducting Nafion ionomer. The loading of the ionomer is typically in a range of 30e50% by weight [10,11], depending on the surface property of catalyst. Since the density [12] of Nafion is around 1.5 g/cm3, a composite electrode normally contains 40e60% ionomer by volume, depending also on hydration. This implies that the final morphology of the electrode has a significant contribution
1218
S.M. Andersen et al. / Journal of Power Sources 274 (2015) 1217e1223
from the polymer phase. The interaction between the catalyst and the proton conductor is one of the most important factors affecting the electrode morphology. Moreover, Nafion thin film in the catalyst layer (less than 10 nm) is of a significantly different structure and property comparing to the bulk membrane as reported by many groups [13,14]. Due to its low molecular weight and minute thickness, Nafion ionomer in CL is nearly transparent to the most microscopic techniques [15,16]. Identification of the polymer requires normally tedious preparation of the samples for the microscopy such as staining with heavy metal ions [17], while the genuineness might be compromised. More convenient methods are various spectroscopies such as infrared [18], Raman [19], nuclear magnetic resonance [20,21] and X-ray photoelectron spectroscopy (XPS). Among those, XPS has the advantage of multi element analysis, nondestruction, high sensitivity, low sample demand and simple sample preparation. There is a rich literature of chemical composition and oxidation state of Nafion membrane [22,23] and carbon supported platinum catalyst [24e26]. However, a fundamental study of various PEMFC components and electrode structure is currently missing in the literature. In this work, a systematic examination of carbon, platinum and Nafion polymer in various electrode structures was carried out using XPS. The spectra were investigated on various types of electrode designs. The oxidation state of carbon, oxygen, fluorine and platinum were discussed. 2. Experiment Pristine Nafion 212 membrane (M) was used as received. The protective films on both sides were peeled off right before the XPS measurement. Thermally treated membrane (M-L) was hot pressed at 140 C, 7 bar for 3 min (the same condition as lamination for electrode preparation). Commercially available catalyst Hispec 9100 (Johnson Matthey) with 57 wt.% platinum supported on high surface area carbon black was used for electrode preparation. A catalyst powder (CP) electrode was prepared by drop coating of catalyst water suspension on a piece of carbon paper (Toray Industries). A catalyst ionomer electrode (CIE) sample was prepared following a standard PEMFC electrode preparation recipe [2], where a catalyst e water/alcohol suspension of 30% Nafion ionomer (w/w) was coated onto a gas diffusion layer (GDL) Sigracet 35DC® (SGL Group). CIE-L was prepared by laminating CIE. The micro porous layer (MPL) is made by a mixture of active carbon and 20 wt.% polytetrafluoroethylene (PTFE), SIGRACET®. The same lamination treatment was applied to MPL, and the resulting
product is labeled as MPL-L. The MEA was constructed by laminating CIE onto membrane Nafion 212. MEA-I, the interphase between CIE-L and MPL-L in a MEA, was prepared by physically remove GDL from MEA by peeling. A list of the samples is shown in Table 1. XPS analysis was performed using a SPECS® system. The spectrometer was equipped with a hemispherical analyzer and a monochromator and all XPS data presented in this study were acquired using Mg Ka (1253.6 eV). The binding energies of the C 1s, O 1s, F 1s, and Pt 4f of the samples were calibrated with respect to C 1s: graphitized carbon peak at 284.5 eV. Survey spectra were collected at pass energy (PE) of 50 eV, resolution 2.5 eV over the binding energy range 0e1250 eV. High resolution multiplex data for the individual elements were collected at a PE of 25 eV, resolution 1.5 eV. Between 60 and 80 min X-ray exposure was applied on each sample. X-ray induced sample degradation and reproducibility were systematically studied. XPS data was analyzed using CasaXPS™. The background was subtracted using the non-linear, Shirley method. The position of the peaks fitting is within resolution of 0.2 eV. 3. Result and discussion 3.1. Survey PEMFC components and electrodes of different structures were examined with XPS. Due to the low electron conductivity of the polymer, all membrane samples (M and ML) were found to charge and this led to a shift of 3e4 eV towards higher binding energy (BE). Such charging effects were not observed for any of the electrodes (CIE, MPL and MEA etc.) containing 20e30 wt.% polymer. This indicates that the electrodes are of good electron conductivity, since the polymer is rather homogeneously mixed among the catalyst/ carbon, and the electrons can conduct through tunneling [27].
Table 1 A list of samples. Acronym Description M M-L CP CIE CIE-L MPL MPL-L MEA-I
Preparation
Membrane Laminated membrane
Use as received Hot pressing M at 140 C, 7 bar for 3 min Catalyst power Use as received Catalyst ionomer electrode Mixture of ionomer (30 wt.%) and catalyst Laminated catalyst ionomer Hot pressing CIE at 140 C, 7 bar electrode for 3 min Micro porous layer Mixture of PTFE (20 wt.%) and active carbon Laminated micro porous layer Hot pressing MPL at 140 C, 7 bar for 3 min Interphase between CIE-L Physically removing GDL from MEA, and MPL-L which was prepared by hot pressing CIE with Nafion 212 membrane at 140 C, 7 bar for 3 min
Fig. 1. Survey spectra of PEMFC components and electrodes.
S.M. Andersen et al. / Journal of Power Sources 274 (2015) 1217e1223
The samples can be generally categorized into polymer only (M and M-L), non-polymer (CP), physically mixed polymer (CIE and MPL) and laminated polymer (CIE-L, MPL-L and MEA-I). The corresponding survey spectra are shown in Fig. 1. The atomic concentrations derived from the areas of the characteristic photoelectron lines are reported in Table 2. The sample composition is dominated by carbon, fluorine (for polymer containing samples), oxygen and additionally small extents of platinum (for catalyst containing samples) and sulfur (from Nafion). The presence of nitrogen is most likely due to gas adsorption. Nafion membrane shows characteristic elements corresponding to the chemical composition of the fluorocarbon backbone and ether containing side chain with sulfonate end groups. Comparing the two types of membrane used, thermally treated Nafion shows higher oxygen and lower fluorine than the pristine Nafion. This indicates that the polymer is sensitive to the thermal treatment. The lower nitrogen presence in the sample may reflect a more compact structure, as shown in the case of M-L, CIE-L MPL-L and MEA-I in comparison to the non-laminated equivalents. For the pure catalyst powder, the signals are mainly attributed to carbon and platinum. The observation of sulfur signal is most likely due to the carbon black support from its natural content, as we observed previously [20,21]. Moreover, the binding energy of the sulfur signal in the catalyst is significantly lower than the one in the polymer. This indicates that sulfur is in a low oxidation state (e.g. 0) in the catalyst and in a high oxidation state (e.g. 6þ) in the membrane. The content of platinum is below the theoretical value (57%). This is probably due to the high porosity of the supporting carbon, which leads to part of the catalyst trapped in the deeper volume and not detected by the XPS [28]. Besides, the catalyst powder often has an adventitious carbon layer which would skew the C/Pt ratio to higher C and lower Pt. When the catalyst and Nafion ionomer are mixed with water/ alcohol, there is a strong interaction between the materials [20,21]. Moreover, as confirmed by microscopy [17], catalyst aggregates are extensively covered with the polymer and the thickness of the Nafion thin film is in the range of a few nanometers [17]. Because the inelastic electron mean free path in solids is a few nanometers (1e2 nm), XPS detects, in different degree, Nafion for ionomer containing samples. As a result, CIEs show much lower percentage of carbon and platinum and strong signal from fluorine. Membrane electrode assembly is produced by hot pressing catalyst layer with the membrane. The condition of the fabrication has shown to have a great influence on the performance of the catalyst [6,7]. This is due to the fact that fabrication has a direct impact on the electrode interface, carbon e ionomer interaction/ distribution. This phenomenon is also reflected by the XPS survey spectra. As shown in Table 2, CIE-L shows slightly higher carbon and platinum, but lower oxygen and fluorine content than that of CIE. The micro porous layer shows similar dominating elements (C and F) as CIE, though there is large difference in oxygen content. MPL contains significantly lower oxygen than that of CIE. This is
Table 2 Element content based on the survey spectra. Sample id
M
M-L
CP
CIE
CIE-L
MPL
MPL-L
MEA-I
C 1s F 1s O 1s Pt 4f S 2p N 1s F/C O/C
47.34 47.93 3.93
57.13 38.08 4.21
87.07
68.19 30.32 1.50
0.37 0.22 0.67 0.07
60.09 32.61 3.87 2.86 0.35 0.22 0.54 0.06
63.15 36.21 0.44
0.36 0.44 1.01 0.08
54.42 37.46 4.96 2.22 0.61 0.33 0.69 0.09
67.77 27.27 4.41 0.12 0.28 0.15 0.40 0.07
3.31 8.38 0.65 0.58 0.04
0.2 0.57 0.007
0.09 0.44 0.02
1219
mainly due to the coverage of PTFE, which is oxygen free. However, the oxygen content was seen to be considerably increased after the hot pressing; concurrently, an increase of C to F ratio. All above evidence indicates an electrode structure different from the one before the lamination step. MEA-I, the interphase between CIE-L and MPL-L, shows a mixed character of CIE-L and MPL-L. Based on the platinum content, MEA-I does not represent CIE or CIE-L. At the same time, based on its oxygen content, MEA-I does not reflect MPL or MPL-L. The interphase shows higher oxygen and lower fluorine content than both CIE-L and MPL-L. This might indicate reactions occur only at the interphase of the MEA components. Further investigations were carried out on deconvolution of the high resolution spectra. 3.2. High resolution spectra High resolution XPS studies were carried out at lower passing energy and more frequent signal acquisition. Damage of organic polymer under X-ray irradiation is a well know phenomenon. Therefore, minimal irradiation exposure and optimized X-ray power (with respect to signal to noise level) were applied to the samples in this study. Furthermore, a series of reproducibility and stability study on the composite electrode (containing both catalyst and polymer) were carried out. The corresponding spectra can be found in the Supplementary Materials. The signal (count per second) was found to increase with X-ray power (150 W, 200 W and 250 W), Fig. S1. Signals from different functional groups were found to vary rather proportionally. There was no detectable degradation due to the X-ray radiation up to 250 W. Continuous scan were carried out at the same spot, Fig. S2. There was no clear sign of degradation either within the applied time (up to 3 h). Dominating signals form the 5 spectra were found within 5% relative random deviation. Nevertheless, long time X-ray exposure was avoided during the study. 3.2.1. The C 1s XPS spectra The C 1s spectra are shown in Fig. 2. The corresponding deconvolution data are presented in Table 3. The pristine Nafion membrane shows one dominating peak at BE 291.3 eV. The signal corresponds to CF2 of the polymer backbone. CF3 and CF are assigned at BE 293.5 eV and 288.9 eV respectively. A minor peak at BE 286.6 eV is related to the side chain group. A small contribution at 282.7 eV may due to the carbon e carbon bonds. A substantially different signal was observed for thermally treated membrane (ML). The small signal at low BE is seen to decrease; instead, a peak at BE 285.0 eV becomes dominating. It may be related to partially oxidized carbon-carbon bonds. The difference between pristine and treated membrane may due to oxidation of the polymer during the hot pressing. Meanwhile, it could also be a self-organizing or reorientation of the polymer in such a way that more side chains relative to backbones are exposed to the surface, as observed by other groups [29,30]. It is also observed that the CF2 signal in Nafion membranes or thin films (of electrodes) is around 0.8 eV lower in BE than that of polytetrafluorethylene (PTEF) [31] or the binder used in MPL (see later). A possible explanation could be that the side chain group may donate electrons to the backbone, which results in a significant peak shift to a lower energy level. The catalyst powder shows single dominating peak at BE 284.5 eV due to the crystalline CeC bonds. The long tail at higher BE corresponds to non-perfect carbon bonds of different states (defect, CeO, C]O, carbonate and p-p*). When the catalyst is physically mixed with the Nafion ionomer (CIE), two dominating peaks at BE 284.5 eV and 291.3 eV are observed. They are due to the graphitic carbon from the catalyst support and the CF2 bonds from the polymer backbone respectively. The CIE spectrum corresponds well
1220
S.M. Andersen et al. / Journal of Power Sources 274 (2015) 1217e1223
Fig. 2. The C 1s deconvolution spectra (selected spectra are enlarged to the right).
The MPL shows similar C 1s spectra as CIE at first glance: two dominating peaks due to CeC and CF2 are observed. However, MPL shows significantly lower oxide content (CeO and C]O) than that of CIE. As mentioned in the survey spectra, this is due to the binder used in MPL is oxygen free PTFE rather than Nafion. Moreover, the CF2 signal position in MPL corresponds well with that of PTFE from the literate. After lamination treatment, MPL-L showed significantly higher ratio of oxide to graphitic carbon than that of MPL. In addition, CF2 peak position shifted slightly towards lower BE after the hot press. These indicate oxidation of the PTFE associated with a degradation of the carbonecarbon chains. Based on the ratio between CF2 and CeC, considerably decrease of the PTFE coverage on active carbon was observed after the lamination (from 0.67 ± 0.02 to 0.50 ± 0.01). From all this evidence, oxidation reaction happened in the MPL during the lamination. The surface morphology is probably affected as well. For the interphase (MEA-I) study, if the interphase is a simple mixture of the two components (CIE-L) and (MPL-L), we should expect that the observed spectrum will be a sum of the spectra of the two components. However, higher oxide signal in MEA-I than CIE-L and MPL-L indicates further oxidation at the interphase. Moreover, a new moiety appearing at the exceptionally high BE 295.2 eV (labeled as extra in Table 3) in C 1s spectrum was
with an overlay of CP and M spectra. Due to a close BE overlap (with fluorocarbon) and their minor contribution (less than 4% each), carbonate and p-p* from the carbon are not included in the deconvolution analysis for the ionomer containing electrodes (CIE and MEA). Due to the same reason, eCOF moiety from the polymer (BE 289.8 eV) is not included either. The catalyst layer in a laminated electrode (CIE-L) shows two major peaks at BE 284.5 eV and 291.3 eV respectively. Comparing to non-laminated electrode (CIE), CIE-L contains slightly a higher amount of graphitic carbon (CeC). This may due to the fact that dehydration of the polymer allows more carbon to be exposed to the surface, and the polymer may reorient, as is the case for the laminated membrane. Moreover, slightly higher amounts of defect and oxidized carbon were observed. This might be due to oxidation of the carbon substrate or the polymer component due to the hot pressing. CF2 signal in CIE-L showed less relative intensity than that of CIE. This indicates again reorganization of the polymer after the lamination. Based on the simplified estimation of the peak area and the corresponding sample preparation, the ratio between the polymer (indicated by CF2) and the catalyst (indicated by CeC) is 0.95 ± 0.14 in CIE and 0.73 ± 0.08 in CIE-L within the effective volume of the X-ray penetration depth. Therefore, slight decrease of polymer relative to carbon after the lamination step was noticed.
Table 3 C 1s deconvolution data. Sample id
~eV
BE CeC Defect CeO C¼O Carbonate p- p* CF CF2 CF3 Extra CF2/CeC CO/CeC
284.5 285.7 286.6 287.5 289.3 291.2 288.9 291.3 293.5 295.2
M
ML
CP
CIE
CIE-L
MPL
MPL-L
%
%
%
%
%
%
%
%
3.02@ 282.7eV
22.20@ 284.9eV
20.25
17.40
76.15 5.61 4.91 5.61 3.86 3.85
36.38 2.75 7.23 5.33
37.88 5.00 9.77 5.68
47.08 3.06 6.84 2.78
48.26 5.02 8.85 3.64
44.34 6.39 9.19 6.85
5.16 64.73 6.84
16.60 34.60 9.30
9.06 34.47 4.78
10.77 27.51 3.38
4.43@ 289.6eV 31.70@ 292.1eV 4.11@ 294.1eV
6.23@ 289.5eV 24.37@ 291.9eV 3.62@ 294.0eV
21.43 6.71
1.56 0.78
0.95 0.35
0.73 0.41
0.67 0.20
0.50 0.26
4.34@ 289.8eV 22.02@ 292.1eV 3.78@ 294.0eV 4.09 0.50 0.36
0.24
MEA-I
S.M. Andersen et al. / Journal of Power Sources 274 (2015) 1217e1223
repeatedly observed. The closest match to this energy scale is the carbon directly connected to one oxygen and three fluorine atoms e oxidized tri-fluorocarbon (OTFC), as reported for Fomblin Y™ (FOM) by Beamson and Briggs [31]. OTFC is not part of the pristine structure of any of the polymers. Its appearance is directly related to the interphase lamination procedure to fabricate the membrane electrode assembly. Moreover, it was not observed for individual components (M-L, CIE-L or MPL-L), though the same lamination procedure was applied. Based on the position of CF2 in C 1s spectrum and the content of platinum, the interphase locates at the top layer of MPL-L (where the layer separation is the easiest). This indicates a partial oxidation of MPL component occurred during hot pressing under the presence of platinum catalyst in the interphase. The presence of the extra moiety is possibly electrode structure and component dependent. The nature of it requires further investigation. 3.2.2. The O 1s XPS spectra The O 1s spectra are shown in Fig. 3. The corresponding deconvolution data are available in Table 4. Oxygen in Nafion has two binding states [22,31]: the sulfonate group (at lower BE) and the ether environment (at higher BE). The ether group assigned at high BE is probably due to the strong electronegativity of the neighboring fluorine. Based on the chemical formula of Nafion, sulfonate and ether of ratio 3:2 are expected. However, we detect the ratio of 1:2 for M and 2:2 for ML. The discrepancy between the theoretical prediction and the experimental value might be due to the preferred orientation of the side chains. The increased amount of sulfonate group on the surface after lamination may be due to the re-orientation of the polymer side chains in the surface region sampled by XPS. CP contains oxidized platinum and oxidized carbon moieties. In the CIE, due to the strong ionomer catalyst interaction, the polymer occupies the majority of the surface. A peak at high BE
Fig. 3. The O 1s deconvolution spectra.
1221
Table 4 O 1s deconvolution data. Sample id BE Pt Oxide S]O, C]O CeO Ether
~eV
M
M-L
CP
CIE
CIE-L
MPL
MPL-L MEA-I
%
%
%
%
%
%
%
531.4 32.92 26.18 28.39 532.5 33.15 51.2 43.66 16.79 18.71 75.01 67.15 533.5 23.43 14.68 17.83 24.99 32.85 534.6 66.85 48.8 42.36 35.07
% 12.95 10.25 30.05 46.74
corresponding to the ether group in the polymer dominates the CIE O 1s signal. Oxidized platinum and oxidized carbon show smaller contribution. In the laminated electrode (CIE-L), the spectrum is still dominated by the polymer signal, though the oxidized carbon and oxidized platinum seems more prominent. This implies that the decrease of the oxygen content as indicated by the survey spectra is probably due to a reduction of the ether group despite carbon oxidation. This agrees well with the information obtained from C 1s analysis. Besides, the slightly decreased platinum oxide signal in CIE-L compare to the CIE's, which also indicates that the catalyst is less covered by the polymer. MPL-L contains more single bonded carbon oxygen species than that of non-laminated MPL. This might indicate that further oxidation has happened during lamination for MPL as well. MEA-I shows a high contribution from ether group at high BE, and a low contribution from the catalyst. 3.2.3. The F 1s XPS spectra The F 1s spectra are shown in Fig. 4. The corresponding deconvolution data are available in Table 5. F1s deconvolution signal consists of dominating CFn corresponding to back bone and minor OCF from the side chains. Since the binding energies of CF, CF2 and CF3 are very close, they are not clearly distinguishable from each other, and therefore presented collectively as CFn. The CFn signal from Nafion samples (M, M-L, CIE & CIE-L) appears at BE
Fig. 4. The F 1s deconvolution spectra.
1222
S.M. Andersen et al. / Journal of Power Sources 274 (2015) 1217e1223
Table 5 F 1s deconvolution data. Sample id
~eV
BE
M
ML
CIE
CIE-L
MPL
MPL-L
MEA-I
%
%
%
%
%
%
%
84.10@ 689.4eV 11.21 4.68
86.9@ 689.2eV 9.32 3.78
68.73@ 689.3eV 15.13 16.14
CFn
688.6
97.95
98.43
95.51
97.71
OCF Extra
691.1 692.2
2.05
1.57
4.49
2.29
688.6 eV; while, the CFn signal from PTFE samples (MPL, MPL-L & MEA-I) appears at 0.6e0.8 eV higher BE. This agrees well with the observation in C 1s spectra (see 3.2.1): electron sharing between backbone and side chain groups leads to CFn peak shift to the lower BE. Based on the F 1s deconvolution data, there is a general trend that laminated samples (M-L, CIE-L & MPL-L) show lower content of OCF than the non-laminated equivalents (M, CIE & MPL). Many studies [32,33] have shown that degradation happening at the sulfonate group is the initial step of the polymer degradation. Though thermogravimetry [33] indicates that the functional group is stable up to 320 C, the polymer thin film structure in the catalyst layer might be more venerable than the bulk membrane, especially in the presence of platinum catalysts. Moreover, OCF contribution in CIE is more apparent than that of membrane (M). This might be due to the unique Nafion thin film structure in the catalyst layer, where the ether, sulfonate group containing side chains interact with catalyst aggregates and expose more towards the surface, rather than forming the hydrophilic domain in the bulk membrane. An extra component in MPL and MPL-L was observed at BE 692.2 eV (labeled as extra in Table 5). It is probably related to a more oxidized fluorocarbon moiety (MOFCM), and its content was also seen slightly reduced after the hot pressing process. Based on the position of the signal, it could be assigned to SF6 [34]. Though the origin of such compound is unclear, it is probably related to the fabrication of the layer. In the interphase study (MEA-I), both OCF and MOFCM were found of higher contribution. This might indicate higher degree of oxidation was promoted by the catalyst in the interphase during lamination.
3.2.4. The Pt 4f XPS spectra The Pt 4f spectra are shown in Fig. 5. The corresponding deconvolution data are available in Table 6. The presence of Pt in XPS is indicated by photoelectrons originating from the 4f orbital, which appears as a doublet due to the spin-orbit coupling. The doublet peaks occur three times here, indicating metallic Pt and two oxidized species. The lower BE values of 71.3 and 74.6 eV are assigned to metallic platinum (Pt), while the next doublet's two values of 72.6 and 76.4 eV at middle BEs are assigned to oxidized platinum (Pt2þ), and the last doublet's two values of 74.5 and 77.9 eV at higher BEs are assigned to oxidized platinum (Pt4þ). The collective contribution from Pt2þ and Pt4þ are noted as PtO. As indicated in the survey spectra (Fig. 1, Table 2), the platinum contribution decreases following the sequence PtCP > PtCIE-L > PtCIE. The lower platinum signal in CIE and CIE-L than CP is due to the surface coverage by the polymer. The slightly higher platinum signal in CIE-L than CIE may be attributed to the polymer structure change after the lamination. As shown in the deconvolution spectra (Fig. 5, Table 6), the platinum oxide contribution increase following the sequence PtOCP < PtOCIE-L < PtOCIE (2e5% deviation), though identical metal catalyst was used in the three electrode structures. If we assume that no oxidation reaction happened to the noble metal during fabrication, the higher percentage of platinum oxide in CIE and CIEL is an indication of a strong interaction between platinum catalyst
Fig. 5. The Pt 4f deconvolution spectra.
Table 6 Pt 4f deconvolution data. Sample id
~eV
CP %
%
%
71.3 74.6 72.6 76.4 74.5 77.9
40.27 40.01 8.08 6.35 1.73 3.56 80.28 19.72
37.05 35.06 12.66 7.63 4.24 3.37 72.11 27.89
39.76 37.06 9.99 9.16 2.02 2.02 76.81 23.19
BE Pt(0) 7/2 Pt(0) 5/2 Pt(2) 7/2 Pt(2) 5/2 Pt(4) 7/2 Pt(4) 5/2 Pt Pt Oxide
CIE
CIE-L
and oxygen containing groups in the polymer [35]. Such interaction may benefit catalyst performance due to higher proton accessibility and durability due to better attachment. 4. Conclusion and outlook Characteristic XPS spectra of Nafion membrane (M), laminated membrane (M-L), catalyst powder (CP), catalyst ionomer electrode (CIE), laminated electrode (CIE-L), micro porous layer (MPL), laminate MPL (MPL-L) and CIE-L/MPL-L interphase (MEA-I) were presented and compared. The surface oxidation state of Nafion membrane was found sensitive toward thermal treatment. The surface of CIE corresponds well with a combination of CP and M. CIE-L surface shows a lower contribution from the polymer and side chain group than that of CIE, which relates to possible polymer structure change. MPL-L was found to be more oxidized and less covered by the binders than MPL. Higher degree of oxidation and extra moieties were observed in MEA-I. Increased ratio between platinum oxides to platinum metal in the electrodes indicates interaction between the platinum and the ionomer. XPS is highly surface sensitive. It may be employed to further investigate electrode fabrication condition and the resulting electrode structure and cell performance. Moreover, XPS may also provide enlightening information on durability of the electrode components, electrode structures and interphases, after being applied to various accelerated stress tests. These are the subjects of our further studies. Acknowledgment This work was financially supported by DuraPEM III 2013-112064, 4M center and European Commission, INTERREG IVA,
S.M. Andersen et al. / Journal of Power Sources 274 (2015) 1217e1223
Southern Denmark-Schleswig-K.E.R.N, Project#111-1.2-12. IRD Fuel Cells A/S is greatly appreciated for providing samples. Ole Wernberg is appreciated for language editing. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.jpowsour.2014.10.162. References [1] R. Othman, A.L. Dicks, Z.H. Zhu, Int. J. Hydrogen Energy 37 (2012) 357e372. [2] S.M. Andersen, M. Borghei, P. Lund, Y.R. Elina, A. Pasanen, E. Kauppinen, et al., Solid State Ionics 231 (2013) 94e101. [3] N. Cheng, S. Mu, X. Chen, H. Lv, M. Pan, P.P. Edwards, Electrochim. Acta 56 (2011) 2154e2159. [4] H. Li, Y.H. Tang, Z.W. Wang, Z. Shi, S.H. Wu, D.T. Song, et al., J. Power Sources 178 (2008) 103e117. [5] S. Choi, D. Jung, S. Yoon, S. Park, E. Oh, K. Junbom, Met. Mater. Int. 17 (2011) 811e816. [6] D.W. Fultz, P.Y.A. Chuang, J. Fuel Cell. Sci. Tech. 8 (2011), 041010e1-041010-6. [7] O. Okur, C.L. Karada, F.J.B. San, E. Okumus, G. Behmenyar, Energy 57 (2013) 574e580. [8] T. Suzuki, S. Tsushima, S. Hirai, Int. J. Hydrogen Energy 36 (2011) 12361e12369. [9] K. Kim, K. Lee, H. Kim, E. Cho, S. Lee, T. Lim, S. Yoon, I. Hwang, J. Jang, Int. J. Hydrogen Energy 35 (2010) 2119e2126. [10] X. Zhao, W. Li, Y. Fu, A. Manthiram, Int. J. Hydrogen Energy 37 (2012) 9845e9852. [11] W. Li, M. Waje, Z. Chen, P. Larsen, Y. Yan, Carbon 48 (2010) 995e1003. [12] L.A. Zook, J. Leddy, Anal. Chem. 68 (1996) 3793e3796. [13] K.P. Devproshad, F. Andrew, P. Joshua, K. Kunal, ECS Trans. 41 (2011) 1393e1406.
1223
[14] A. Kongkanand, J. Phys. Chem. C. 115 (2011) 11318e11325. [15] J. Yu, Z. Jiang, M. Hou, D. Liang, Y. Xiao, M. Doub, Z. Shao, B. Yi, J. Power Sources 246 (2014) 90e94. [16] S. Ma, C. Solterbeck, M. Odgaard, E. Skou, Appl. Phys. A 96 (2009) 581e589. [17] F. Scheiba, N. Benker, U. Kunz, C. Roth, H. Fuess, J. Power Sources 177 (2008) 273e280. [18] J.I. Eastcott, E.B. Easton, J. Power Sources 245 (2014) 487e494. [19] S. Radicea, C. Oldani, L. Merlo, M. Rocchi, Polym. Degrad. Stabil. 98 (2013) 1138e1143. [20] S.M. Andersen, M. Borghei, R. Dhiman, H. Jiang, V. Ruiz, E. Kauppinen, E. Skou, Carbon 71 (2014) 218e228. [21] S.M. Andersen, M. Borghei, R. Dhiman, V. Ruiz, E. Kauppinen, E. Skou, J. Phy. Chem. C. 118 (2014) 10814e10823. [22] M. Schulze, M. Lorenz, N. Wagner, E. Giilzow, E. Fresenius, J. Anal. Chem. 365 (1999) 106e113. [23] C. Chen, G. Levitin, D. Hess, T. Fuller, J. Power Sources 169 (2007) 288e295. [24] X. Ma, F. Zaera, J. Am. Chem. Soc. 128 (51) (2006) 16414e16415. [25] A.K. Santra, D.W. Goodman, Electrochim. Acta 47 (2002) 3595e3609. l, R. Schlo €gl, Appl. Surf. Sci. 47 (1991) 281e285. [26] M. Muhler, Z. Paa [27] I. Balberg, Phy. Rev. Lett. 59 (1987) 1305e1308. [28] S.N. Stamatin, M. Borghei, S.M. Andersen, S. Veltze, V. Ruiz, E. Kauppinen, E.M. Skou, Int. J. Hydrogen Energy 39 (2014) 8215e8224. [29] D.K. Paul, A. Fraser, J. Pearce, K. Karan, ECS Trans. 41 (1) (2011) 1393e1406. [30] X. Li, F. Feng, K. Zhang, S.U. Ye, D.Y. Kwok, V. Birss, Langmuir 28 (2012) 6698e6705. [31] G. Briggs, D. Briggs, High Resolution XPS of Organic Polymers: the Scienta ESCA300 Database, 1992, pp. 234e235. [32] L. Ghassemzadeh, M. Marrony, R. Barrera, K.D. Kreuera, J. Maier, K. Müller, J. Power Sources 186 (2009) 334e338. [33] S.M. Andersen, L. Grahl-Madsen, E.M. Skou, Solid State Ionics 192 (2011) 602e606. [34] G.B. Fisher, N.E. Eirikson, T.E. Madey, J.T. Yates Jr., Surf. Sci. 65 (1977) 210e228. [35] S. Ma, Q. Chen, F. Jøgensen, P. Stein, E. Skou, Solid State Ionics 178 (2007) 1568e1575.
Contents lists available at ScienceDirect
Journal of Power Sources journal homepage: www.elsevier.com/locate/jpowsour
Chemistry of carbon polymer composite electrode e An X-ray photoelectron spectroscopy study Shuang Ma Andersen*, Rajnish Dhiman, Eivind Skou Department of Chemical Engineering, Biotechnology and Environmental Technology, University of Southern Denmark, Niels Bohrs All e 1, DK-5230 Odense M, Denmark
h i g h l i g h t s Surface oxidation state of PEMFC components and different electrodes were compared. The surface property of the catalyst layer is influenced by the electrode structure. Lamination can induce structure change in electrode.
a r t i c l e i n f o
a b s t r a c t
Article history: Received 21 June 2014 Received in revised form 10 October 2014 Accepted 21 October 2014 Available online 29 October 2014
Surface chemistry of the electrodes in a proton exchange membrane fuel cell is of great importance for the cell performance. Many groups have reported that electrode preparation condition has a direct influence on the resulting electrode properties. In this work, the oxidation state of electrode components and the composites (catalyst ionomer mixtures) in various electrode structures were systematically studied with X-ray photoelectron spectroscopy (XPS). Based on the spectra, when catalyst is physically mixed with Nafion ionomer, the resulting electrode surface chemistry is a combination of the two components. When the electrode is prepared with a lamination procedure, the ratio between fluorocarbon and graphitic carbon is decreased. Moreover, ether type oxide content is decreased although carbon oxide is slightly increased. This indicates structure change of the catalyst layer due to an interaction between the ionomer and the catalyst and possible polymer structural change during electrode fabrication. The surface of micro porous layer was found to be much more influenced by the lamination, especially when it is in contact with catalysts in the interphase. Higher amount of platinum oxide was observed in the electrode structures (catalyst ionomer mixture) compared to the catalyst powder. This also indicates a certain interaction between the functional groups in the polymer and platinum surface. © 2014 Elsevier B.V. All rights reserved.
Keywords: XPS PEMFC Electrode structure Interaction Lamination Interphase
1. Introduction Many advances have been achieved since the first demonstration of proton exchange membrane fuel cell (PEMFC). Non-precious metal catalysts [1], corrosion resistant catalyst supports [2] and improvements of catalyst e support attachment [3] are important milestones in the history of PEMFCs. Meanwhile, understanding and optimization of the electrode structure: water management [4], catalyst ionomer ratio [5], electrode fabrication conditions [6,7], are also of crucial importance for the advancement of the technology. Surface chemistry and morphology of the electrode, though still not well understood, are essential parameters influencing the
* Corresponding author. E-mail address: [email protected] (S.M. Andersen). http://dx.doi.org/10.1016/j.jpowsour.2014.10.162 0378-7753/© 2014 Elsevier B.V. All rights reserved.
cell performance [8,9]. The delicate three-phase boundary (TPB), where proton, electron and gas meet and initiate the electrochemical reactions, has great impact on the catalyst utilization and cell lifespan. Though the thickness of the catalyst layer (CL) is only around 20 mm, which is less than 5% of the total thickness of a membrane electrode assembly (MEA), CL is the heart of PEMFC, and is responsible for hydrogen oxidation reaction (HOR) and oxygen reduction reaction (ORR). State of the art PEMFC catalyst layer consists of platinum supported by carbon and impregnated with proton conducting Nafion ionomer. The loading of the ionomer is typically in a range of 30e50% by weight [10,11], depending on the surface property of catalyst. Since the density [12] of Nafion is around 1.5 g/cm3, a composite electrode normally contains 40e60% ionomer by volume, depending also on hydration. This implies that the final morphology of the electrode has a significant contribution
1218
S.M. Andersen et al. / Journal of Power Sources 274 (2015) 1217e1223
from the polymer phase. The interaction between the catalyst and the proton conductor is one of the most important factors affecting the electrode morphology. Moreover, Nafion thin film in the catalyst layer (less than 10 nm) is of a significantly different structure and property comparing to the bulk membrane as reported by many groups [13,14]. Due to its low molecular weight and minute thickness, Nafion ionomer in CL is nearly transparent to the most microscopic techniques [15,16]. Identification of the polymer requires normally tedious preparation of the samples for the microscopy such as staining with heavy metal ions [17], while the genuineness might be compromised. More convenient methods are various spectroscopies such as infrared [18], Raman [19], nuclear magnetic resonance [20,21] and X-ray photoelectron spectroscopy (XPS). Among those, XPS has the advantage of multi element analysis, nondestruction, high sensitivity, low sample demand and simple sample preparation. There is a rich literature of chemical composition and oxidation state of Nafion membrane [22,23] and carbon supported platinum catalyst [24e26]. However, a fundamental study of various PEMFC components and electrode structure is currently missing in the literature. In this work, a systematic examination of carbon, platinum and Nafion polymer in various electrode structures was carried out using XPS. The spectra were investigated on various types of electrode designs. The oxidation state of carbon, oxygen, fluorine and platinum were discussed. 2. Experiment Pristine Nafion 212 membrane (M) was used as received. The protective films on both sides were peeled off right before the XPS measurement. Thermally treated membrane (M-L) was hot pressed at 140 C, 7 bar for 3 min (the same condition as lamination for electrode preparation). Commercially available catalyst Hispec 9100 (Johnson Matthey) with 57 wt.% platinum supported on high surface area carbon black was used for electrode preparation. A catalyst powder (CP) electrode was prepared by drop coating of catalyst water suspension on a piece of carbon paper (Toray Industries). A catalyst ionomer electrode (CIE) sample was prepared following a standard PEMFC electrode preparation recipe [2], where a catalyst e water/alcohol suspension of 30% Nafion ionomer (w/w) was coated onto a gas diffusion layer (GDL) Sigracet 35DC® (SGL Group). CIE-L was prepared by laminating CIE. The micro porous layer (MPL) is made by a mixture of active carbon and 20 wt.% polytetrafluoroethylene (PTFE), SIGRACET®. The same lamination treatment was applied to MPL, and the resulting
product is labeled as MPL-L. The MEA was constructed by laminating CIE onto membrane Nafion 212. MEA-I, the interphase between CIE-L and MPL-L in a MEA, was prepared by physically remove GDL from MEA by peeling. A list of the samples is shown in Table 1. XPS analysis was performed using a SPECS® system. The spectrometer was equipped with a hemispherical analyzer and a monochromator and all XPS data presented in this study were acquired using Mg Ka (1253.6 eV). The binding energies of the C 1s, O 1s, F 1s, and Pt 4f of the samples were calibrated with respect to C 1s: graphitized carbon peak at 284.5 eV. Survey spectra were collected at pass energy (PE) of 50 eV, resolution 2.5 eV over the binding energy range 0e1250 eV. High resolution multiplex data for the individual elements were collected at a PE of 25 eV, resolution 1.5 eV. Between 60 and 80 min X-ray exposure was applied on each sample. X-ray induced sample degradation and reproducibility were systematically studied. XPS data was analyzed using CasaXPS™. The background was subtracted using the non-linear, Shirley method. The position of the peaks fitting is within resolution of 0.2 eV. 3. Result and discussion 3.1. Survey PEMFC components and electrodes of different structures were examined with XPS. Due to the low electron conductivity of the polymer, all membrane samples (M and ML) were found to charge and this led to a shift of 3e4 eV towards higher binding energy (BE). Such charging effects were not observed for any of the electrodes (CIE, MPL and MEA etc.) containing 20e30 wt.% polymer. This indicates that the electrodes are of good electron conductivity, since the polymer is rather homogeneously mixed among the catalyst/ carbon, and the electrons can conduct through tunneling [27].
Table 1 A list of samples. Acronym Description M M-L CP CIE CIE-L MPL MPL-L MEA-I
Preparation
Membrane Laminated membrane
Use as received Hot pressing M at 140 C, 7 bar for 3 min Catalyst power Use as received Catalyst ionomer electrode Mixture of ionomer (30 wt.%) and catalyst Laminated catalyst ionomer Hot pressing CIE at 140 C, 7 bar electrode for 3 min Micro porous layer Mixture of PTFE (20 wt.%) and active carbon Laminated micro porous layer Hot pressing MPL at 140 C, 7 bar for 3 min Interphase between CIE-L Physically removing GDL from MEA, and MPL-L which was prepared by hot pressing CIE with Nafion 212 membrane at 140 C, 7 bar for 3 min
Fig. 1. Survey spectra of PEMFC components and electrodes.
S.M. Andersen et al. / Journal of Power Sources 274 (2015) 1217e1223
The samples can be generally categorized into polymer only (M and M-L), non-polymer (CP), physically mixed polymer (CIE and MPL) and laminated polymer (CIE-L, MPL-L and MEA-I). The corresponding survey spectra are shown in Fig. 1. The atomic concentrations derived from the areas of the characteristic photoelectron lines are reported in Table 2. The sample composition is dominated by carbon, fluorine (for polymer containing samples), oxygen and additionally small extents of platinum (for catalyst containing samples) and sulfur (from Nafion). The presence of nitrogen is most likely due to gas adsorption. Nafion membrane shows characteristic elements corresponding to the chemical composition of the fluorocarbon backbone and ether containing side chain with sulfonate end groups. Comparing the two types of membrane used, thermally treated Nafion shows higher oxygen and lower fluorine than the pristine Nafion. This indicates that the polymer is sensitive to the thermal treatment. The lower nitrogen presence in the sample may reflect a more compact structure, as shown in the case of M-L, CIE-L MPL-L and MEA-I in comparison to the non-laminated equivalents. For the pure catalyst powder, the signals are mainly attributed to carbon and platinum. The observation of sulfur signal is most likely due to the carbon black support from its natural content, as we observed previously [20,21]. Moreover, the binding energy of the sulfur signal in the catalyst is significantly lower than the one in the polymer. This indicates that sulfur is in a low oxidation state (e.g. 0) in the catalyst and in a high oxidation state (e.g. 6þ) in the membrane. The content of platinum is below the theoretical value (57%). This is probably due to the high porosity of the supporting carbon, which leads to part of the catalyst trapped in the deeper volume and not detected by the XPS [28]. Besides, the catalyst powder often has an adventitious carbon layer which would skew the C/Pt ratio to higher C and lower Pt. When the catalyst and Nafion ionomer are mixed with water/ alcohol, there is a strong interaction between the materials [20,21]. Moreover, as confirmed by microscopy [17], catalyst aggregates are extensively covered with the polymer and the thickness of the Nafion thin film is in the range of a few nanometers [17]. Because the inelastic electron mean free path in solids is a few nanometers (1e2 nm), XPS detects, in different degree, Nafion for ionomer containing samples. As a result, CIEs show much lower percentage of carbon and platinum and strong signal from fluorine. Membrane electrode assembly is produced by hot pressing catalyst layer with the membrane. The condition of the fabrication has shown to have a great influence on the performance of the catalyst [6,7]. This is due to the fact that fabrication has a direct impact on the electrode interface, carbon e ionomer interaction/ distribution. This phenomenon is also reflected by the XPS survey spectra. As shown in Table 2, CIE-L shows slightly higher carbon and platinum, but lower oxygen and fluorine content than that of CIE. The micro porous layer shows similar dominating elements (C and F) as CIE, though there is large difference in oxygen content. MPL contains significantly lower oxygen than that of CIE. This is
Table 2 Element content based on the survey spectra. Sample id
M
M-L
CP
CIE
CIE-L
MPL
MPL-L
MEA-I
C 1s F 1s O 1s Pt 4f S 2p N 1s F/C O/C
47.34 47.93 3.93
57.13 38.08 4.21
87.07
68.19 30.32 1.50
0.37 0.22 0.67 0.07
60.09 32.61 3.87 2.86 0.35 0.22 0.54 0.06
63.15 36.21 0.44
0.36 0.44 1.01 0.08
54.42 37.46 4.96 2.22 0.61 0.33 0.69 0.09
67.77 27.27 4.41 0.12 0.28 0.15 0.40 0.07
3.31 8.38 0.65 0.58 0.04
0.2 0.57 0.007
0.09 0.44 0.02
1219
mainly due to the coverage of PTFE, which is oxygen free. However, the oxygen content was seen to be considerably increased after the hot pressing; concurrently, an increase of C to F ratio. All above evidence indicates an electrode structure different from the one before the lamination step. MEA-I, the interphase between CIE-L and MPL-L, shows a mixed character of CIE-L and MPL-L. Based on the platinum content, MEA-I does not represent CIE or CIE-L. At the same time, based on its oxygen content, MEA-I does not reflect MPL or MPL-L. The interphase shows higher oxygen and lower fluorine content than both CIE-L and MPL-L. This might indicate reactions occur only at the interphase of the MEA components. Further investigations were carried out on deconvolution of the high resolution spectra. 3.2. High resolution spectra High resolution XPS studies were carried out at lower passing energy and more frequent signal acquisition. Damage of organic polymer under X-ray irradiation is a well know phenomenon. Therefore, minimal irradiation exposure and optimized X-ray power (with respect to signal to noise level) were applied to the samples in this study. Furthermore, a series of reproducibility and stability study on the composite electrode (containing both catalyst and polymer) were carried out. The corresponding spectra can be found in the Supplementary Materials. The signal (count per second) was found to increase with X-ray power (150 W, 200 W and 250 W), Fig. S1. Signals from different functional groups were found to vary rather proportionally. There was no detectable degradation due to the X-ray radiation up to 250 W. Continuous scan were carried out at the same spot, Fig. S2. There was no clear sign of degradation either within the applied time (up to 3 h). Dominating signals form the 5 spectra were found within 5% relative random deviation. Nevertheless, long time X-ray exposure was avoided during the study. 3.2.1. The C 1s XPS spectra The C 1s spectra are shown in Fig. 2. The corresponding deconvolution data are presented in Table 3. The pristine Nafion membrane shows one dominating peak at BE 291.3 eV. The signal corresponds to CF2 of the polymer backbone. CF3 and CF are assigned at BE 293.5 eV and 288.9 eV respectively. A minor peak at BE 286.6 eV is related to the side chain group. A small contribution at 282.7 eV may due to the carbon e carbon bonds. A substantially different signal was observed for thermally treated membrane (ML). The small signal at low BE is seen to decrease; instead, a peak at BE 285.0 eV becomes dominating. It may be related to partially oxidized carbon-carbon bonds. The difference between pristine and treated membrane may due to oxidation of the polymer during the hot pressing. Meanwhile, it could also be a self-organizing or reorientation of the polymer in such a way that more side chains relative to backbones are exposed to the surface, as observed by other groups [29,30]. It is also observed that the CF2 signal in Nafion membranes or thin films (of electrodes) is around 0.8 eV lower in BE than that of polytetrafluorethylene (PTEF) [31] or the binder used in MPL (see later). A possible explanation could be that the side chain group may donate electrons to the backbone, which results in a significant peak shift to a lower energy level. The catalyst powder shows single dominating peak at BE 284.5 eV due to the crystalline CeC bonds. The long tail at higher BE corresponds to non-perfect carbon bonds of different states (defect, CeO, C]O, carbonate and p-p*). When the catalyst is physically mixed with the Nafion ionomer (CIE), two dominating peaks at BE 284.5 eV and 291.3 eV are observed. They are due to the graphitic carbon from the catalyst support and the CF2 bonds from the polymer backbone respectively. The CIE spectrum corresponds well
1220
S.M. Andersen et al. / Journal of Power Sources 274 (2015) 1217e1223
Fig. 2. The C 1s deconvolution spectra (selected spectra are enlarged to the right).
The MPL shows similar C 1s spectra as CIE at first glance: two dominating peaks due to CeC and CF2 are observed. However, MPL shows significantly lower oxide content (CeO and C]O) than that of CIE. As mentioned in the survey spectra, this is due to the binder used in MPL is oxygen free PTFE rather than Nafion. Moreover, the CF2 signal position in MPL corresponds well with that of PTFE from the literate. After lamination treatment, MPL-L showed significantly higher ratio of oxide to graphitic carbon than that of MPL. In addition, CF2 peak position shifted slightly towards lower BE after the hot press. These indicate oxidation of the PTFE associated with a degradation of the carbonecarbon chains. Based on the ratio between CF2 and CeC, considerably decrease of the PTFE coverage on active carbon was observed after the lamination (from 0.67 ± 0.02 to 0.50 ± 0.01). From all this evidence, oxidation reaction happened in the MPL during the lamination. The surface morphology is probably affected as well. For the interphase (MEA-I) study, if the interphase is a simple mixture of the two components (CIE-L) and (MPL-L), we should expect that the observed spectrum will be a sum of the spectra of the two components. However, higher oxide signal in MEA-I than CIE-L and MPL-L indicates further oxidation at the interphase. Moreover, a new moiety appearing at the exceptionally high BE 295.2 eV (labeled as extra in Table 3) in C 1s spectrum was
with an overlay of CP and M spectra. Due to a close BE overlap (with fluorocarbon) and their minor contribution (less than 4% each), carbonate and p-p* from the carbon are not included in the deconvolution analysis for the ionomer containing electrodes (CIE and MEA). Due to the same reason, eCOF moiety from the polymer (BE 289.8 eV) is not included either. The catalyst layer in a laminated electrode (CIE-L) shows two major peaks at BE 284.5 eV and 291.3 eV respectively. Comparing to non-laminated electrode (CIE), CIE-L contains slightly a higher amount of graphitic carbon (CeC). This may due to the fact that dehydration of the polymer allows more carbon to be exposed to the surface, and the polymer may reorient, as is the case for the laminated membrane. Moreover, slightly higher amounts of defect and oxidized carbon were observed. This might be due to oxidation of the carbon substrate or the polymer component due to the hot pressing. CF2 signal in CIE-L showed less relative intensity than that of CIE. This indicates again reorganization of the polymer after the lamination. Based on the simplified estimation of the peak area and the corresponding sample preparation, the ratio between the polymer (indicated by CF2) and the catalyst (indicated by CeC) is 0.95 ± 0.14 in CIE and 0.73 ± 0.08 in CIE-L within the effective volume of the X-ray penetration depth. Therefore, slight decrease of polymer relative to carbon after the lamination step was noticed.
Table 3 C 1s deconvolution data. Sample id
~eV
BE CeC Defect CeO C¼O Carbonate p- p* CF CF2 CF3 Extra CF2/CeC CO/CeC
284.5 285.7 286.6 287.5 289.3 291.2 288.9 291.3 293.5 295.2
M
ML
CP
CIE
CIE-L
MPL
MPL-L
%
%
%
%
%
%
%
%
3.02@ 282.7eV
22.20@ 284.9eV
20.25
17.40
76.15 5.61 4.91 5.61 3.86 3.85
36.38 2.75 7.23 5.33
37.88 5.00 9.77 5.68
47.08 3.06 6.84 2.78
48.26 5.02 8.85 3.64
44.34 6.39 9.19 6.85
5.16 64.73 6.84
16.60 34.60 9.30
9.06 34.47 4.78
10.77 27.51 3.38
4.43@ 289.6eV 31.70@ 292.1eV 4.11@ 294.1eV
6.23@ 289.5eV 24.37@ 291.9eV 3.62@ 294.0eV
21.43 6.71
1.56 0.78
0.95 0.35
0.73 0.41
0.67 0.20
0.50 0.26
4.34@ 289.8eV 22.02@ 292.1eV 3.78@ 294.0eV 4.09 0.50 0.36
0.24
MEA-I
S.M. Andersen et al. / Journal of Power Sources 274 (2015) 1217e1223
repeatedly observed. The closest match to this energy scale is the carbon directly connected to one oxygen and three fluorine atoms e oxidized tri-fluorocarbon (OTFC), as reported for Fomblin Y™ (FOM) by Beamson and Briggs [31]. OTFC is not part of the pristine structure of any of the polymers. Its appearance is directly related to the interphase lamination procedure to fabricate the membrane electrode assembly. Moreover, it was not observed for individual components (M-L, CIE-L or MPL-L), though the same lamination procedure was applied. Based on the position of CF2 in C 1s spectrum and the content of platinum, the interphase locates at the top layer of MPL-L (where the layer separation is the easiest). This indicates a partial oxidation of MPL component occurred during hot pressing under the presence of platinum catalyst in the interphase. The presence of the extra moiety is possibly electrode structure and component dependent. The nature of it requires further investigation. 3.2.2. The O 1s XPS spectra The O 1s spectra are shown in Fig. 3. The corresponding deconvolution data are available in Table 4. Oxygen in Nafion has two binding states [22,31]: the sulfonate group (at lower BE) and the ether environment (at higher BE). The ether group assigned at high BE is probably due to the strong electronegativity of the neighboring fluorine. Based on the chemical formula of Nafion, sulfonate and ether of ratio 3:2 are expected. However, we detect the ratio of 1:2 for M and 2:2 for ML. The discrepancy between the theoretical prediction and the experimental value might be due to the preferred orientation of the side chains. The increased amount of sulfonate group on the surface after lamination may be due to the re-orientation of the polymer side chains in the surface region sampled by XPS. CP contains oxidized platinum and oxidized carbon moieties. In the CIE, due to the strong ionomer catalyst interaction, the polymer occupies the majority of the surface. A peak at high BE
Fig. 3. The O 1s deconvolution spectra.
1221
Table 4 O 1s deconvolution data. Sample id BE Pt Oxide S]O, C]O CeO Ether
~eV
M
M-L
CP
CIE
CIE-L
MPL
MPL-L MEA-I
%
%
%
%
%
%
%
531.4 32.92 26.18 28.39 532.5 33.15 51.2 43.66 16.79 18.71 75.01 67.15 533.5 23.43 14.68 17.83 24.99 32.85 534.6 66.85 48.8 42.36 35.07
% 12.95 10.25 30.05 46.74
corresponding to the ether group in the polymer dominates the CIE O 1s signal. Oxidized platinum and oxidized carbon show smaller contribution. In the laminated electrode (CIE-L), the spectrum is still dominated by the polymer signal, though the oxidized carbon and oxidized platinum seems more prominent. This implies that the decrease of the oxygen content as indicated by the survey spectra is probably due to a reduction of the ether group despite carbon oxidation. This agrees well with the information obtained from C 1s analysis. Besides, the slightly decreased platinum oxide signal in CIE-L compare to the CIE's, which also indicates that the catalyst is less covered by the polymer. MPL-L contains more single bonded carbon oxygen species than that of non-laminated MPL. This might indicate that further oxidation has happened during lamination for MPL as well. MEA-I shows a high contribution from ether group at high BE, and a low contribution from the catalyst. 3.2.3. The F 1s XPS spectra The F 1s spectra are shown in Fig. 4. The corresponding deconvolution data are available in Table 5. F1s deconvolution signal consists of dominating CFn corresponding to back bone and minor OCF from the side chains. Since the binding energies of CF, CF2 and CF3 are very close, they are not clearly distinguishable from each other, and therefore presented collectively as CFn. The CFn signal from Nafion samples (M, M-L, CIE & CIE-L) appears at BE
Fig. 4. The F 1s deconvolution spectra.
1222
S.M. Andersen et al. / Journal of Power Sources 274 (2015) 1217e1223
Table 5 F 1s deconvolution data. Sample id
~eV
BE
M
ML
CIE
CIE-L
MPL
MPL-L
MEA-I
%
%
%
%
%
%
%
84.10@ 689.4eV 11.21 4.68
86.9@ 689.2eV 9.32 3.78
68.73@ 689.3eV 15.13 16.14
CFn
688.6
97.95
98.43
95.51
97.71
OCF Extra
691.1 692.2
2.05
1.57
4.49
2.29
688.6 eV; while, the CFn signal from PTFE samples (MPL, MPL-L & MEA-I) appears at 0.6e0.8 eV higher BE. This agrees well with the observation in C 1s spectra (see 3.2.1): electron sharing between backbone and side chain groups leads to CFn peak shift to the lower BE. Based on the F 1s deconvolution data, there is a general trend that laminated samples (M-L, CIE-L & MPL-L) show lower content of OCF than the non-laminated equivalents (M, CIE & MPL). Many studies [32,33] have shown that degradation happening at the sulfonate group is the initial step of the polymer degradation. Though thermogravimetry [33] indicates that the functional group is stable up to 320 C, the polymer thin film structure in the catalyst layer might be more venerable than the bulk membrane, especially in the presence of platinum catalysts. Moreover, OCF contribution in CIE is more apparent than that of membrane (M). This might be due to the unique Nafion thin film structure in the catalyst layer, where the ether, sulfonate group containing side chains interact with catalyst aggregates and expose more towards the surface, rather than forming the hydrophilic domain in the bulk membrane. An extra component in MPL and MPL-L was observed at BE 692.2 eV (labeled as extra in Table 5). It is probably related to a more oxidized fluorocarbon moiety (MOFCM), and its content was also seen slightly reduced after the hot pressing process. Based on the position of the signal, it could be assigned to SF6 [34]. Though the origin of such compound is unclear, it is probably related to the fabrication of the layer. In the interphase study (MEA-I), both OCF and MOFCM were found of higher contribution. This might indicate higher degree of oxidation was promoted by the catalyst in the interphase during lamination.
3.2.4. The Pt 4f XPS spectra The Pt 4f spectra are shown in Fig. 5. The corresponding deconvolution data are available in Table 6. The presence of Pt in XPS is indicated by photoelectrons originating from the 4f orbital, which appears as a doublet due to the spin-orbit coupling. The doublet peaks occur three times here, indicating metallic Pt and two oxidized species. The lower BE values of 71.3 and 74.6 eV are assigned to metallic platinum (Pt), while the next doublet's two values of 72.6 and 76.4 eV at middle BEs are assigned to oxidized platinum (Pt2þ), and the last doublet's two values of 74.5 and 77.9 eV at higher BEs are assigned to oxidized platinum (Pt4þ). The collective contribution from Pt2þ and Pt4þ are noted as PtO. As indicated in the survey spectra (Fig. 1, Table 2), the platinum contribution decreases following the sequence PtCP > PtCIE-L > PtCIE. The lower platinum signal in CIE and CIE-L than CP is due to the surface coverage by the polymer. The slightly higher platinum signal in CIE-L than CIE may be attributed to the polymer structure change after the lamination. As shown in the deconvolution spectra (Fig. 5, Table 6), the platinum oxide contribution increase following the sequence PtOCP < PtOCIE-L < PtOCIE (2e5% deviation), though identical metal catalyst was used in the three electrode structures. If we assume that no oxidation reaction happened to the noble metal during fabrication, the higher percentage of platinum oxide in CIE and CIEL is an indication of a strong interaction between platinum catalyst
Fig. 5. The Pt 4f deconvolution spectra.
Table 6 Pt 4f deconvolution data. Sample id
~eV
CP %
%
%
71.3 74.6 72.6 76.4 74.5 77.9
40.27 40.01 8.08 6.35 1.73 3.56 80.28 19.72
37.05 35.06 12.66 7.63 4.24 3.37 72.11 27.89
39.76 37.06 9.99 9.16 2.02 2.02 76.81 23.19
BE Pt(0) 7/2 Pt(0) 5/2 Pt(2) 7/2 Pt(2) 5/2 Pt(4) 7/2 Pt(4) 5/2 Pt Pt Oxide
CIE
CIE-L
and oxygen containing groups in the polymer [35]. Such interaction may benefit catalyst performance due to higher proton accessibility and durability due to better attachment. 4. Conclusion and outlook Characteristic XPS spectra of Nafion membrane (M), laminated membrane (M-L), catalyst powder (CP), catalyst ionomer electrode (CIE), laminated electrode (CIE-L), micro porous layer (MPL), laminate MPL (MPL-L) and CIE-L/MPL-L interphase (MEA-I) were presented and compared. The surface oxidation state of Nafion membrane was found sensitive toward thermal treatment. The surface of CIE corresponds well with a combination of CP and M. CIE-L surface shows a lower contribution from the polymer and side chain group than that of CIE, which relates to possible polymer structure change. MPL-L was found to be more oxidized and less covered by the binders than MPL. Higher degree of oxidation and extra moieties were observed in MEA-I. Increased ratio between platinum oxides to platinum metal in the electrodes indicates interaction between the platinum and the ionomer. XPS is highly surface sensitive. It may be employed to further investigate electrode fabrication condition and the resulting electrode structure and cell performance. Moreover, XPS may also provide enlightening information on durability of the electrode components, electrode structures and interphases, after being applied to various accelerated stress tests. These are the subjects of our further studies. Acknowledgment This work was financially supported by DuraPEM III 2013-112064, 4M center and European Commission, INTERREG IVA,
S.M. Andersen et al. / Journal of Power Sources 274 (2015) 1217e1223
Southern Denmark-Schleswig-K.E.R.N, Project#111-1.2-12. IRD Fuel Cells A/S is greatly appreciated for providing samples. Ole Wernberg is appreciated for language editing. Appendix A. Supplementary data Supplementary data related to this article can be found at http:// dx.doi.org/10.1016/j.jpowsour.2014.10.162. References [1] R. Othman, A.L. Dicks, Z.H. Zhu, Int. J. Hydrogen Energy 37 (2012) 357e372. [2] S.M. Andersen, M. Borghei, P. Lund, Y.R. Elina, A. Pasanen, E. Kauppinen, et al., Solid State Ionics 231 (2013) 94e101. [3] N. Cheng, S. Mu, X. Chen, H. Lv, M. Pan, P.P. Edwards, Electrochim. Acta 56 (2011) 2154e2159. [4] H. Li, Y.H. Tang, Z.W. Wang, Z. Shi, S.H. Wu, D.T. Song, et al., J. Power Sources 178 (2008) 103e117. [5] S. Choi, D. Jung, S. Yoon, S. Park, E. Oh, K. Junbom, Met. Mater. Int. 17 (2011) 811e816. [6] D.W. Fultz, P.Y.A. Chuang, J. Fuel Cell. Sci. Tech. 8 (2011), 041010e1-041010-6. [7] O. Okur, C.L. Karada, F.J.B. San, E. Okumus, G. Behmenyar, Energy 57 (2013) 574e580. [8] T. Suzuki, S. Tsushima, S. Hirai, Int. J. Hydrogen Energy 36 (2011) 12361e12369. [9] K. Kim, K. Lee, H. Kim, E. Cho, S. Lee, T. Lim, S. Yoon, I. Hwang, J. Jang, Int. J. Hydrogen Energy 35 (2010) 2119e2126. [10] X. Zhao, W. Li, Y. Fu, A. Manthiram, Int. J. Hydrogen Energy 37 (2012) 9845e9852. [11] W. Li, M. Waje, Z. Chen, P. Larsen, Y. Yan, Carbon 48 (2010) 995e1003. [12] L.A. Zook, J. Leddy, Anal. Chem. 68 (1996) 3793e3796. [13] K.P. Devproshad, F. Andrew, P. Joshua, K. Kunal, ECS Trans. 41 (2011) 1393e1406.
1223
[14] A. Kongkanand, J. Phys. Chem. C. 115 (2011) 11318e11325. [15] J. Yu, Z. Jiang, M. Hou, D. Liang, Y. Xiao, M. Doub, Z. Shao, B. Yi, J. Power Sources 246 (2014) 90e94. [16] S. Ma, C. Solterbeck, M. Odgaard, E. Skou, Appl. Phys. A 96 (2009) 581e589. [17] F. Scheiba, N. Benker, U. Kunz, C. Roth, H. Fuess, J. Power Sources 177 (2008) 273e280. [18] J.I. Eastcott, E.B. Easton, J. Power Sources 245 (2014) 487e494. [19] S. Radicea, C. Oldani, L. Merlo, M. Rocchi, Polym. Degrad. Stabil. 98 (2013) 1138e1143. [20] S.M. Andersen, M. Borghei, R. Dhiman, H. Jiang, V. Ruiz, E. Kauppinen, E. Skou, Carbon 71 (2014) 218e228. [21] S.M. Andersen, M. Borghei, R. Dhiman, V. Ruiz, E. Kauppinen, E. Skou, J. Phy. Chem. C. 118 (2014) 10814e10823. [22] M. Schulze, M. Lorenz, N. Wagner, E. Giilzow, E. Fresenius, J. Anal. Chem. 365 (1999) 106e113. [23] C. Chen, G. Levitin, D. Hess, T. Fuller, J. Power Sources 169 (2007) 288e295. [24] X. Ma, F. Zaera, J. Am. Chem. Soc. 128 (51) (2006) 16414e16415. [25] A.K. Santra, D.W. Goodman, Electrochim. Acta 47 (2002) 3595e3609. l, R. Schlo €gl, Appl. Surf. Sci. 47 (1991) 281e285. [26] M. Muhler, Z. Paa [27] I. Balberg, Phy. Rev. Lett. 59 (1987) 1305e1308. [28] S.N. Stamatin, M. Borghei, S.M. Andersen, S. Veltze, V. Ruiz, E. Kauppinen, E.M. Skou, Int. J. Hydrogen Energy 39 (2014) 8215e8224. [29] D.K. Paul, A. Fraser, J. Pearce, K. Karan, ECS Trans. 41 (1) (2011) 1393e1406. [30] X. Li, F. Feng, K. Zhang, S.U. Ye, D.Y. Kwok, V. Birss, Langmuir 28 (2012) 6698e6705. [31] G. Briggs, D. Briggs, High Resolution XPS of Organic Polymers: the Scienta ESCA300 Database, 1992, pp. 234e235. [32] L. Ghassemzadeh, M. Marrony, R. Barrera, K.D. Kreuera, J. Maier, K. Müller, J. Power Sources 186 (2009) 334e338. [33] S.M. Andersen, L. Grahl-Madsen, E.M. Skou, Solid State Ionics 192 (2011) 602e606. [34] G.B. Fisher, N.E. Eirikson, T.E. Madey, J.T. Yates Jr., Surf. Sci. 65 (1977) 210e228. [35] S. Ma, Q. Chen, F. Jøgensen, P. Stein, E. Skou, Solid State Ionics 178 (2007) 1568e1575.