- Email: [email protected]
XPS study of surface oxidation of carbon-supported Pt catalysts
MATERIALS CHEM;STRYtiD ELSEVIER
Materials Chemistry and Physics 41 (1995) 9-14
XPS study of surface oxidation of carbon-supported
Pt catalysts
V. Alderucci ‘, L. Pino ‘, P.L. Antonucci b, W. Roh ‘, J. Cho ‘, H. Kim ‘, D.L. Cocke d, V. Antonucci a aCNR Insiitute for Transformation and Storage of Energy, Salita S. Lucia, 39, S. Lucia, Messina, Italy h Faculty of Engineering, University of Reggio Calabria, Institute of Chemistry, Via E. Cuzzocrea 48, Reggio Calabria, Italy ’ Department of Chemistry, Seoul National University, Seoul, South Korea ’ Chemistry Department, Lamar University, Beaumont, TX, USA Received 16 August 1993
Abstract Pt/C catalysts for phosphoric acid fuel cell (PAFC) applications have been studied by X-ray photoelectron spectroscopy (XPS) and potentiometric titration in aqueous suspensions. Catalysts activated at temperatures in the range 800-950 “C have shown a significant amount of C edge and Ca functionalities which generate a variable amount of pyrone-type complexes after exposure to air. The relative amount of such basic groups has been found to affect the amount of Pt oxidized and hence the electrochemically active surface area of the catalyst. Keywords: Pt/C catalysts; Phosphoric acid fuel cells; Surface groups
2. Experimental
1. Introduction
The role of the surface functionalities of carbon formation of a dispersed metal phase in Pt/C catalysts recognized [ l-31. As reported in a number of previous
in the is well inves-
tigations [4,5], the level of metal-support interaction is greatly affected by their acid-base properties, mainly resulting from the activation temperature and gas atmosphere, under equal preparation conditions such as metal precursor, impregnation method and loading. It has been shown that the acidic and basic surface functional groups of the catalyst pass progressively, upon activation in N, flow from 110 to 900 “C, from a strongly acidic to a basic behaviour, with varying degrees of dispersion and the chemical nature of supported Pt [ 61. More particularly, we have recently observed [ 71 that a high concentration of C basal plane groups in the carbon support gives rise to high metal surface area. Since such a high CT group concentration resulted, under controlled preparative conditions, from a heat treatment temperature able to induce basic behaviour at the catalyst surface [ 71, we have here investigated materials activated in a narrow temperature range, i.e., 800-950 “C, with the aim of finding some relationships between the surface characteristics of the samples examined and their final properties in terms of electrocatalytic activity for the oxygen reduction reaction in phosphoric acid fuel cell (PAFC) conditions. 0254-0584/95/$09.50
0 1995 Elsevier Science S.A. All rights reserved
SSDIO254-0584(95)01497-I
The electrocatalysts, prepared from a carbon black slurry in distilled water ( 10 g of carbon per liter of H,O), consisted of platinum (20% w/w) supported on Ketjenblack EC from Akzo Chemie. The carbon black was 0.1% ash with a BET surface area of 950 m’/g. An aqueous solution of H,PtCl, (Engelhard, 99.9%) was added to the carbon black slurry. The carbon-supported Pt was reduced at 45 “C, by adding an excess amount of Na2S,04 solution dropwise. The electrocatalysts were repeatedly washed with distilled water, filtered and dried overnight at 70 “C and thermally activated in a quartz tube with flowing nitrogen under reducing conditions at temperatures in the range 800-950 “C (see Table 1). The electrodes for electrochemical characterizations were manufactured according to a screen-printing based procedure described elsewhere [ 81. Cathodic polarizations were carried out in a half-cell apparatus with 98% w/w phosphoric acid at 170 “C. The Pt surface area was determined by cyclic voltammetry at a sweep rate of 50 mV/s. The hydrogen adsorbed in the potential range 0.054.4 V (vs. RHE) on flooded electrodes ( 10% PTFE) was measured in 50% w/w phosphoric acid at room temperature [ 81. Surface characterization of the materials has been carried out by X-ray photoelectron spectroscopy (XPS) and potentiometric titration (ZPC) .
V. Alderucci et al. /Materials Chemistry and Physics 41 (1995) 9-14
10
Table 1 Catalyst characteristics Catalyst no.
Heat treatment
Pt loading (%)
T (“C)
Amount of basic functional groups
Mass activity
MSA by CV
(mA/mg pt) at 0.9 V in Oz
(m’/g)
39 40 36 n.d. 41 57 33 35 46 44 n.d.
71 60 58 59 44 110 52 49 82 60 30
(meq/g) 1 2 3 4 5 6 7 8 9 10 11
800 850 850 900 950 900 900 900 900 900 900
16.4 17.6 16.1 17.1 15.4 13.7 13.8 15.2 14.5 16.1 19.3
The catalyst’s acid-base properties were evaluated by potentiometric titration in an aqueous suspension of Pt/C catalysts according to Parks and de Bruyn [9]. The experimental apparatus [lo] consisted of a Pyrex@ glass cell with an external jacket for circulation of thermostatic liquid, a glass electrode for pH measurement (Orion model 9 l-92) and a probe for temperature control (Orion ATC probe 917002). The two probes were connected to an Orion mode1 311 pH/mV meter. Titration of 500 ml of 0.1 M KNO, electrolyte solution containing the catalyst powder (2 g) was carried out under stirring. A 0.1 N HNO, solution was introduced (in increments of 5 ml to 20 ml) to the suspension in the electrolytic solution and the titrant (0.1 N KOH) was added via an automatic burette (Analytical Control Method). The XPS measurements were performed using a VSW Scientific Instruments (Manchester, England) apparatus; the X-ray source was Mg Ka at a power of 150 W. Spectra were obtained with pass energy of 90 eV for wide scans and 44 eV for individual elements. All of the spectra were obtained under identical conditions. The pressure of the spectrometer was 5 X 10-i’ mbar (Ti sublimation pump) and 5 X 10e9 mbar during the measurements. The XPS samples were prepared by making multiple coatings of a slurry consisting of catalyst with distilled water ( 1:5 by weight) on sample holders (stainless steel 304)) dried at 100 “C in a low-pressure (0.1 atm) oven for 1 h. These samples showed none of the Fe signals from stainless steel. The sample was introduced into the spectrometer using a separate differentially pumped fast entry chamber, then into a desorption chamber in ultra high vacuum for desorption of highly volatile species adsorbed on the carbon support, and finally into the analyzer chamber. The Pt 4f,,* and 4f,,* peaks of a Pt foil were taken each time after argon sputtering before sample observation for calibration of the spectrometer. The area analyzed was 0.51.O cm*. For data acquisition and processing, an IBM-compatible personal computer with RS232C was employed for data transport. The background correction by the Shirley
0.01 0.08 0.12 0.08 0.19 0.15 0.23 0.09 0.13 0.12 0.13
method [ 1 l] and curve fitting was performed by Gauss/ Lorentz routines. The quantitative evaluation of each peak was obtained by dividing the integrated peak area by atomic sensitivity factors, which were calculated from the ionization cross sections, the mean free electron escape depth and the measured transmission functions of the spectrometer expressed as the relative ratio with respect to carbon [ 1214].
3. Results and discussion In addition to chemical state information, XPS provides a considerable amount of information about the electronic properties of certain samples depending on the nature of the materials under examination. Electronic effects associated with the local and long-range charge transfer phenomena can be gleaned from the XPS spectra. In addition, differential charging and variations in electrical contact between the sample components with the spectrometer can cause asymmetric peak broadening and peak shifts from which more information can be drawn with careful analysis. The integral nature of the spectroscopy prevents differentiation between spectra1 changes due to chemical and structural effects. Therefore, additional microscopy data is often needed to better interpret XPS spectral changes. Thus the present XPS data are not intended to convey absolute values, but comparison of observations on similarly prepared Pt-C catalysts has demonstrated some significant differences worth elaborating. X-ray photoelectron spectra of the catalysts examined shows Pt 4f, 0 1s and C Is signals, and they are summarized in Table 2. The binding energies (BE) of all peaks are referenced to a C 1s value of 284.6 eV. Depending on the oxidation state of Pt, there are 2 or 3 pairs of platinum peaks. The most intense ones, located around 71 and 74 eV, are due to metallic platinum: Pt 4f,,, and Pt 4f,,,. The positions of these peaks are affected by the interactions between Pt and C. The second pair of Pt signals
V. Alderucci et al. /Materials Chemistry and Physics 41 (1995) 9-14
11
Table 2 XPS data of Pt/C catalysts Catalyst no.
1
2 3
4
5
6 7
8
9 10 11
0 1s
c Is
Pt 4f7/2
Binding energy
Relative ratio
Binding energy
Relative ratio
(eV)
(%)
(eV)
(%)
5.20
287.4 289.0
76.8 23.2
533.3 535.0
78.3 21.7
0.4
2.96
0.4
4.43
286.9 288.5 286.8 288. I
86.9 13.1 81.4 18.6
533.4 534.7 532.2 534.7
31.7 68.3 69.9 30.1
0.5
4.62
286.8 287.6
86.6 13.4
534.6 536.9
86.5 13.5
0.6
6.45
85.8 14.2
2.58
1.1
6.87
533.3 535.1 533.7 535.4
38.2 61.8 65.6 34.4
0.4
4.53
66.6 23.4 10.0 86.9 13.4 79.9 12.7 7.4 88.1 11.9
533.3 534.7
0.1
286.9 288.5 291.0 286.9 288.6 287.5 289.2 292.1 287.0 288.5
533.2 534.9
91.2 8.8
0.2
2.00
0.5
2.43
0.4
5.45
287.3 289.1 286.7 288.1 286.9 288.3
86.7 13.3 87.9 12.1 81.5 18.5
533.3 535.2 533.0 534.7 533.2 534.8
66.5 33.5 75.4 24.6 39.9 60.1
Binding energy
Relative ratio
A& (vs. Pt ref.)
(eV)
(%I
(eW
73.5 75.1
68.3 21.5
0.8
77.0 73.3 75.2 73.6 75.1
10.2 83.2 16.8 72.5 17.2
77.0 73.4 75.0
10.3 73.0 15.6
77.0 73.5 76.0 78.3 73.0 74.4 74.5 76.9 78.8 73.7 75.8 78.2 73.1 74.7 73.4 75.5 73.7 75.0 77.0
11.4 58.1 30.7 11.2 81.2 18.8 50.2 26.2 23.6 70.2 19.6 10.2 86.2 13.8 84.9 15.1 72.8 15.6 11.6
appears around 1.5-2.8 eV higher than that for metallic Pt, and the third pair is observed at 3.4-5.3 eV higher than that for metallic Pt. The intensities of the third doublet are quite low, especially for the samples with two Pt 4f BE listed in the table, and an attempt to obtain deconvoluted data was made. The second set of doublets of PtO or Pt (OH) 2 was determined to be 2.6 eV higher than that of metallic Pt [ 1.51, and a few of our samples agreed with this value. Many of the samples, however, showed less than 2.0 eV difference, probably due to the chemical nature of the Pt in the dispersion because of the varying degree of electronic interaction of Pt with the carbon support [7,16,17]. Similar binding energy shifts due to Pt( II) were observed on graphite-supported Pt [ 18,191, Pt/CO,-treated Vulcan XC-72 [ 20-231, Pt single crystals [ 24,251, and on Pt group bimetallic alloys on carbon [26]. A small binding energy shift due to the presence of PtO,,, should not be ruled out [ 271. However, Dijkgraaf et al. [ 281 pointed out the possibility of oxygen diffusion into the Pt lattice without formation of Pt-0, based on the great differences between the heat of chemisorption of oxygen (71 kcal/mol) and the heat of formation of PtO, (32.2 kcal/mol). Because of the reasons
ptx (%)
mentioned above, the signal from the second set of doublets can be from either or both of the partially oxidized Pt and/or Pt-0 moieties. Signals observed 3.4-5.3 eV higher have been reported to be due to the presence of PtO, .xH,O or Pt( OH), [ 281. Two C 1s signals can be assigned as carbon (graphitic) for 284.6 eV and -C=O for 285.4-286.2 eV. The third peak, observed around 288.5 eV, seems due to either -COO or carbonate. These assignments agree very well with the extensive XPS study made on commercially available carbon used as supports by Albers et al. [ 291. The highly oxidized carbon signal (around 288 eV) was also observed by Goodenough et al. [ 231, but no precise identification of the species could be made; Bowker and Madix [ 301 have suggested that this highly oxidized species is M-O-CH-O-M after XPS and EELS investigation of the adsorption of formic acid on Cu. The oxygen signal observed around 530.9 eV is in between the values 530 eV for Pt-0 and 53 1.5 for Pt-OH,,, [ 241. It is very difficult to distinguish the Pt-0 signal in the presence of a large amount of oxygenated groups, because the samples were exposed to the air. Thus no quantitative ratio of Pt 4f,,, to 0 1s can be made for the same reason. A similar binding energy of 0 Is is obtained using electrochemically
12
V. Alderucci et al. /Materials
Chemistry and Physics 41 (1995) 9-14
oxidized Pt [ 311. The peak at 532.5 eV is due to -t&O. Again, the presence of signals from Pt-HZ0 (533 eV) made it difficult to correlate the peak area to that from -C=O. The shift of the 0 1s signal from that of partially oxidized single crystal Pt to lower binding energy by heat treatment was observed by Peuckert [24] and confirms the presence of oxyhydrogen, which results from reduction of oxide, to give a lower BE. The binding energy shift (the difference between the BE of Pt 4f,,, of the sample and that of the platinum metal used as an external standard) is taken into account in the magnitude of the interaction between platinum and the carbon support. This represents the actual magnitude of the interaction on an absolute scale, because the BE of C 1s is influenced by the presence of Pt dispersion, as reported by Albers et al. [ 291. If the BE shift is measured against the BE of C 1s at 284.6 eV, the observed BE of Pt 4f will represent the difference from the already shifted C Is, depending on the nature of the charge transfer through the carbon support. This argument is acceptable as long as the energy calibration (work functions in a broad sense) of the instrument does not change during the measurements of the sample and platinum foil. It is believed to be so in this case, because the peaks of the external standard were obtained each time just before running each sample. Data derived from potentiometric titrations show that heat treatments varying between 800 and 950 “C induce marked changes in the basic properties of the catalysts samples, with the dissociation constants of surface functional groups in the range 3.4-3.8 and a concentration between 0.01 and 0.23 meq/g. Since the catalysts were prepared following the same experimental procedure (except for the final heat treatment temperature), the observed differences in the amount of oxygenated basic groups can be reconciled and explained in terms of different times of storage in air. It is well known, in fact, that upon exposure to air, oxygen fixation occurs on the active sites generated during heat treatment, giving rise to the formation of pyrone-type groups [ 61. By plotting the amounts of such basic functionalities (including both Crrand oxygenated pyrone-type groups) versus the relative fraction of oxidized Pt (derived from the deconvoluted Pt 4f spectra), it appears that higher Pt,, amounts are associated with the samples characterized by a high concentration of basic groups (Fig. 1 (a)). The scatter of the data is likely due to the fact that potentiometric titration does not allow distinction between CT and oxygenated groups. Such a differentiation can be made possible by analysis of the C 1s spectra, which show a noticeable asymmetry at high binding energies. This is caused by the oxidized carbon species at the surface, which have been separated, identified by curve fitting and attributed to the specific species by referring to Albers et al. [ 291. The less scattered trend of the data shown in Fig. 1 (b) seems to confirm the expectations, according to which increasing amounts of oxygen groups parallel the higher levels of surface Pt oxidation determined.
Fig. 2 shows a linear relationship between the Pt BE shift and the amount of Pt oxidized. This shows not only indirect evidence of the effect of oxidized carbon functionality in the catalyst, but also the ineffective screening of holes by oxidized Pt in the final state (after ejection of photoelectrons [ 321) . When all Pt atoms are oxidized and remain as Pt( II), the BE of Pt( II) will be about 2.6 eV higher than the values (4 : 0 q
0.05
0.1
Amount of baw
0.15
0.2
0.25
groups by ZPC, meqlg
(b)1
d 0
0
0
0’
5
10
15
20
I
/
1
J
25
30
35
40
Amount of oxidized C by XPS, %
Fig. 1. (a) Amount of oxidized Pt vs. concentration of basic groups detected by ZPC; (b) amount of oxidized Pt vs. oxygenated carbon functional groups detected by XPS. 10
0’ 0
I
’
’
0.1
0.2
0.3
0.4
’ 0.5
I
/
’
0.6
0.7
0.8
0.9
I
1
1
1.1
Pt binding energy shift, eV Fig. 2. Amount of Pt oxidized vs. Pt binding energy shift.
’
1.2
V. Alderucci et al. /Materials
Chemistry and Physics 41 (1995) 9-14
13
the necessity of controlling the preparation and activation procedures in order to lower the deactivation effects resulting from a too high population of oxygen-containing surface groups.
4. Conclusions
OL
0
1
2
3
4
Amount
5
6
of oxidized
Fig. 3. Pt surface area by cyclic voltammetry
7
1
2
3
4 Amount
I 10
9
Pt, 96
vs. the amount of Pt oxidized.
1
0
6
5 of oxidized
6
7
1
1
6
9
10
Pt, %
Fig. 4. Mass activity in oxygen vs. the amount of Pt oxidized.
expected from metallic Pt [ 151. In the presence of a large amount of oxidizing functionalities, however, the electronic states of Pt will be disturbed in such a way that the electron density is less than that of a neutral metal atom. Also, oxidized Pt near Pt atoms (2-3 atomic neighbors) cannot screen effectively compared with the metal. These, in turn, make the BE of Pt metal species shift toward higher BE. We have previously reported [ 7] a decrease in Pt dispersion (measured by CV) with increasing number of oxygenated surface groups (either basic or mildly acidic); this has been explained by the higher electron withdrawing character of the carbon-oxygen complexes as compared to C edge and CT functionalities. The Pt MSA/Pt,, plot shown in Fig. 3 clearly reflects the effects exerted by such complexes on the stabilization of oxidized Pt states; it appears that highly electroactive Pt areas are found in samples having low amounts of Pt,, and low amounts of basic oxygenated groups. The high Pt-C bond strength and the resonance stabilization of the aromatic ring of the electron pair are able to anchor the Pt particles without any concurrent deactivation (i.e., oxidation) effect. This view seems to be confirmed by the mass activity in the oxygen reduction reaction/Pt,, relationship shown in Fig. 4. Accordingly, the trend of the data explains
Our results clearly show that: ( 1) the amount of Pt-0 increases with increasing amount of basic groups (Fig. 1 (a) ) ; (2) the oxidation of C is associated with higher Pt-0 amounts (Fig. 1 (b) ) ; (3) increasing amount of oxidized Pt produces more Pt BE shift (Fig. 2); (4) the amount of electroactive Pt area decreases with increasing amount of Pt-0 (Fig. 3) ; (5) the activity decreases as the amount of Pt-0 increases (Fig. 4). The high activation temperatures (800-950 “C) of Pt/C catalysts lead to a high concentration of basic carbon groups. Most of them are C edge and Cn- functionalities, but upon exposure to air, oxygen fixation occurs to different extents, generating pyrone-type groups. Both the Pt binding energy shift and the amount of Pt oxidized, which are measured by XPS, show that the differences arising in the electrochemical properties of Pt/C catalysts for the oxygen reduction reaction are related to the relative amount of such basic groups. Particularly, an optimal interaction between the supported Pt crystallites and the carbon surface appears to be associated with the edge sites and Z- sites of carbon, reflecting a positive effect on the electroactive surface area of the catalysts. On the other hand, the oxygenated groups having a lower electron donating character (in comparison with C edge and Crfunctionalities) interact less with zero-valent Pt, affecting the physical and electronic state of the metal in such a way as to produce negative effects on its electrochemical properties.
Acknowledgements One of the authors (HK) is grateful to the HAN Project for financial support through the New Energy Development Project (component and cell technology development for PAFC).
References [ 11 S. Mukerjee, J. Appl. Electrochem., 20 ( 1990) 537. [2] F. Coloma, C. Prado-Burguete and F. Rodriguez-Reinoso, Proc. ZOth ht. Congress on Catalysis, Budapest, 19-24 July 1992, p. 301. [3] P. Birke, S. Engels, K. Becker and H.D. Nenbaner, Chem. Tech. (Leipzig), 32 (1980) 245. [4] P. Ehrburger, Adv. Colloid Interface Sci., 21 ( 1984) 275. [5] C. Prado-Burguete, A. Linares-Solano, F. Rodriguez-Reinoso and C. Salinas-Martinez De Lecea, J. Catal., 125 ( 1989) 98.
14
V. Alderucci et al. /Materials
Chemistry and Physics 41 (1995) 9-14
[6] AS. Aricb, V. Antonucci, L. Pino, P.L. Antonucci and N. Giordano, Carbon, 28 ( 1990) 599. [7] P.L. Antonucci, V. Alderucci, D.L. Cocke, H. Kim and N. Giordano, J. Appl. Electrochem.. 24 (1994) 58. [S] N. Giordano, E. Passalacqua, V. Alderucci, P. Staiti, L. Pino, H. Mirzaian, E.J. Taylor and G. Wilemski, Electrochim. Acra, 36 ( 1991) 1049. [ 91 G.A. Parks and P.L. de Bruyn, J. Phys. Chem., 66 ( 1962) 967. [lo] A.S. Aricb, V. Antonucci, M. Minutoli and N. Giordano, Carbon, 27 (1989) 337. [ 111 D.A. Shirley, Phys. Rev. B, 5 ( 1972) 4709. [ 121 C.D. Wagner, J. Electron Spectrosc. Relat. Phenom., 32 (1983) 99. [ 131 J.J. Yeh and I. Lindau, At. Data Nucl. Data Tables, 32 (1) (1985). [ 141 L.C. Feldman and J.W. Mayer, Fundamentals of Sur$ace and Thin Film Analysis, North-Holland, Amsterdam, 1986, pp. 226-228. [ 151 K.S. Kim, N. Winograd and R.E. Davis, J. Am. Gem. Sot., 93 (1974) 6296. [ 161 A.Q. Contractor and J. O’M. Bockris, Electrochim. Acta, 29 (1984) 427. [ 171 P.N. Ross, K. Kinoshita and P. Stonehart, J. Caral., 32 ( 1974) 163. [ 181 S. Wangand P.S. Fedkiw,J. Electrochem. Sot., 139 (1992) 3151. [ 191 A.K. Shulka, K.V. Ramesh, R. Manoharan, P.R. &rode and V. Vasudevan, Eler. Bunsenges. Phys. Chem., 89 (1985) 1261.
1201 B.J. Kennedy and A. Hamnett,J. 1211 J.B. Goodenough,
Electroanal. Chem., 283 ( 1990) 271.
A. Hamnett,
Electrochim. Acta, 32 (1987)
B.J. Kennedy
and S.A. Weeks,
1233.
1221 J.B. Goodenough, A. Hamnett, B.J. Kennedy, R. Manoharan Weeks, Electrochim. Acta, 35 ( 1990) 199.
and S.A.
[231 J.B. Goodenough, A. Hamnett, B.J. Kennedy, R. Manoharan Weeks, J. Electroanal. Chem., 240 (1988) 133.
and S.A.
[Xl M. Peuckert, Electrochim. Acta, 29 (1984) 1315. [251 M. Peuckert and H.P. Bonzel, Surf: Sci., 145 (1984) 239.
[261K.V. Ramesh,
P.R. Sarode,
S. Vasudevan
and A.K.
Shulka,
J.
Electronnal. Chem., 223 ( 1987) 91. 1271 C. Allen, P.M. Tucker,
A. Capon and R. Parsons,
J. Electroanal.
Chem., 50 (1974) 335. 1281 P.J.M. Dijkgraaf,
H.A.M.
Wiele, J. Cutal., 112 (1988)
Duisters,
B.F.M. Kuster and K. van der
337.
(291 P. Albers, K. Deller, B.M. Despeyroux,
A. Schafer and K. Seibold, J.
Catal., 133 (1992) 467. 1301 M. Bowker and R.J. Madix, Surf: Sci., 102 (1981) 542. [311 T. Dickinson, A.F. Prey and P.M.A. Sherwood, J. Chem. Sot., Faraday Trans. 1, 7i (1975) 298. 1321 B. Johansson and N. Martensson,
Phys. Rev. B. 21 ( 1980) 4427.
Materials Chemistry and Physics 41 (1995) 9-14
XPS study of surface oxidation of carbon-supported
Pt catalysts
V. Alderucci ‘, L. Pino ‘, P.L. Antonucci b, W. Roh ‘, J. Cho ‘, H. Kim ‘, D.L. Cocke d, V. Antonucci a aCNR Insiitute for Transformation and Storage of Energy, Salita S. Lucia, 39, S. Lucia, Messina, Italy h Faculty of Engineering, University of Reggio Calabria, Institute of Chemistry, Via E. Cuzzocrea 48, Reggio Calabria, Italy ’ Department of Chemistry, Seoul National University, Seoul, South Korea ’ Chemistry Department, Lamar University, Beaumont, TX, USA Received 16 August 1993
Abstract Pt/C catalysts for phosphoric acid fuel cell (PAFC) applications have been studied by X-ray photoelectron spectroscopy (XPS) and potentiometric titration in aqueous suspensions. Catalysts activated at temperatures in the range 800-950 “C have shown a significant amount of C edge and Ca functionalities which generate a variable amount of pyrone-type complexes after exposure to air. The relative amount of such basic groups has been found to affect the amount of Pt oxidized and hence the electrochemically active surface area of the catalyst. Keywords: Pt/C catalysts; Phosphoric acid fuel cells; Surface groups
2. Experimental
1. Introduction
The role of the surface functionalities of carbon formation of a dispersed metal phase in Pt/C catalysts recognized [ l-31. As reported in a number of previous
in the is well inves-
tigations [4,5], the level of metal-support interaction is greatly affected by their acid-base properties, mainly resulting from the activation temperature and gas atmosphere, under equal preparation conditions such as metal precursor, impregnation method and loading. It has been shown that the acidic and basic surface functional groups of the catalyst pass progressively, upon activation in N, flow from 110 to 900 “C, from a strongly acidic to a basic behaviour, with varying degrees of dispersion and the chemical nature of supported Pt [ 61. More particularly, we have recently observed [ 71 that a high concentration of C basal plane groups in the carbon support gives rise to high metal surface area. Since such a high CT group concentration resulted, under controlled preparative conditions, from a heat treatment temperature able to induce basic behaviour at the catalyst surface [ 71, we have here investigated materials activated in a narrow temperature range, i.e., 800-950 “C, with the aim of finding some relationships between the surface characteristics of the samples examined and their final properties in terms of electrocatalytic activity for the oxygen reduction reaction in phosphoric acid fuel cell (PAFC) conditions. 0254-0584/95/$09.50
0 1995 Elsevier Science S.A. All rights reserved
SSDIO254-0584(95)01497-I
The electrocatalysts, prepared from a carbon black slurry in distilled water ( 10 g of carbon per liter of H,O), consisted of platinum (20% w/w) supported on Ketjenblack EC from Akzo Chemie. The carbon black was 0.1% ash with a BET surface area of 950 m’/g. An aqueous solution of H,PtCl, (Engelhard, 99.9%) was added to the carbon black slurry. The carbon-supported Pt was reduced at 45 “C, by adding an excess amount of Na2S,04 solution dropwise. The electrocatalysts were repeatedly washed with distilled water, filtered and dried overnight at 70 “C and thermally activated in a quartz tube with flowing nitrogen under reducing conditions at temperatures in the range 800-950 “C (see Table 1). The electrodes for electrochemical characterizations were manufactured according to a screen-printing based procedure described elsewhere [ 81. Cathodic polarizations were carried out in a half-cell apparatus with 98% w/w phosphoric acid at 170 “C. The Pt surface area was determined by cyclic voltammetry at a sweep rate of 50 mV/s. The hydrogen adsorbed in the potential range 0.054.4 V (vs. RHE) on flooded electrodes ( 10% PTFE) was measured in 50% w/w phosphoric acid at room temperature [ 81. Surface characterization of the materials has been carried out by X-ray photoelectron spectroscopy (XPS) and potentiometric titration (ZPC) .
V. Alderucci et al. /Materials Chemistry and Physics 41 (1995) 9-14
10
Table 1 Catalyst characteristics Catalyst no.
Heat treatment
Pt loading (%)
T (“C)
Amount of basic functional groups
Mass activity
MSA by CV
(mA/mg pt) at 0.9 V in Oz
(m’/g)
39 40 36 n.d. 41 57 33 35 46 44 n.d.
71 60 58 59 44 110 52 49 82 60 30
(meq/g) 1 2 3 4 5 6 7 8 9 10 11
800 850 850 900 950 900 900 900 900 900 900
16.4 17.6 16.1 17.1 15.4 13.7 13.8 15.2 14.5 16.1 19.3
The catalyst’s acid-base properties were evaluated by potentiometric titration in an aqueous suspension of Pt/C catalysts according to Parks and de Bruyn [9]. The experimental apparatus [lo] consisted of a Pyrex@ glass cell with an external jacket for circulation of thermostatic liquid, a glass electrode for pH measurement (Orion model 9 l-92) and a probe for temperature control (Orion ATC probe 917002). The two probes were connected to an Orion mode1 311 pH/mV meter. Titration of 500 ml of 0.1 M KNO, electrolyte solution containing the catalyst powder (2 g) was carried out under stirring. A 0.1 N HNO, solution was introduced (in increments of 5 ml to 20 ml) to the suspension in the electrolytic solution and the titrant (0.1 N KOH) was added via an automatic burette (Analytical Control Method). The XPS measurements were performed using a VSW Scientific Instruments (Manchester, England) apparatus; the X-ray source was Mg Ka at a power of 150 W. Spectra were obtained with pass energy of 90 eV for wide scans and 44 eV for individual elements. All of the spectra were obtained under identical conditions. The pressure of the spectrometer was 5 X 10-i’ mbar (Ti sublimation pump) and 5 X 10e9 mbar during the measurements. The XPS samples were prepared by making multiple coatings of a slurry consisting of catalyst with distilled water ( 1:5 by weight) on sample holders (stainless steel 304)) dried at 100 “C in a low-pressure (0.1 atm) oven for 1 h. These samples showed none of the Fe signals from stainless steel. The sample was introduced into the spectrometer using a separate differentially pumped fast entry chamber, then into a desorption chamber in ultra high vacuum for desorption of highly volatile species adsorbed on the carbon support, and finally into the analyzer chamber. The Pt 4f,,* and 4f,,* peaks of a Pt foil were taken each time after argon sputtering before sample observation for calibration of the spectrometer. The area analyzed was 0.51.O cm*. For data acquisition and processing, an IBM-compatible personal computer with RS232C was employed for data transport. The background correction by the Shirley
0.01 0.08 0.12 0.08 0.19 0.15 0.23 0.09 0.13 0.12 0.13
method [ 1 l] and curve fitting was performed by Gauss/ Lorentz routines. The quantitative evaluation of each peak was obtained by dividing the integrated peak area by atomic sensitivity factors, which were calculated from the ionization cross sections, the mean free electron escape depth and the measured transmission functions of the spectrometer expressed as the relative ratio with respect to carbon [ 1214].
3. Results and discussion In addition to chemical state information, XPS provides a considerable amount of information about the electronic properties of certain samples depending on the nature of the materials under examination. Electronic effects associated with the local and long-range charge transfer phenomena can be gleaned from the XPS spectra. In addition, differential charging and variations in electrical contact between the sample components with the spectrometer can cause asymmetric peak broadening and peak shifts from which more information can be drawn with careful analysis. The integral nature of the spectroscopy prevents differentiation between spectra1 changes due to chemical and structural effects. Therefore, additional microscopy data is often needed to better interpret XPS spectral changes. Thus the present XPS data are not intended to convey absolute values, but comparison of observations on similarly prepared Pt-C catalysts has demonstrated some significant differences worth elaborating. X-ray photoelectron spectra of the catalysts examined shows Pt 4f, 0 1s and C Is signals, and they are summarized in Table 2. The binding energies (BE) of all peaks are referenced to a C 1s value of 284.6 eV. Depending on the oxidation state of Pt, there are 2 or 3 pairs of platinum peaks. The most intense ones, located around 71 and 74 eV, are due to metallic platinum: Pt 4f,,, and Pt 4f,,,. The positions of these peaks are affected by the interactions between Pt and C. The second pair of Pt signals
V. Alderucci et al. /Materials Chemistry and Physics 41 (1995) 9-14
11
Table 2 XPS data of Pt/C catalysts Catalyst no.
1
2 3
4
5
6 7
8
9 10 11
0 1s
c Is
Pt 4f7/2
Binding energy
Relative ratio
Binding energy
Relative ratio
(eV)
(%)
(eV)
(%)
5.20
287.4 289.0
76.8 23.2
533.3 535.0
78.3 21.7
0.4
2.96
0.4
4.43
286.9 288.5 286.8 288. I
86.9 13.1 81.4 18.6
533.4 534.7 532.2 534.7
31.7 68.3 69.9 30.1
0.5
4.62
286.8 287.6
86.6 13.4
534.6 536.9
86.5 13.5
0.6
6.45
85.8 14.2
2.58
1.1
6.87
533.3 535.1 533.7 535.4
38.2 61.8 65.6 34.4
0.4
4.53
66.6 23.4 10.0 86.9 13.4 79.9 12.7 7.4 88.1 11.9
533.3 534.7
0.1
286.9 288.5 291.0 286.9 288.6 287.5 289.2 292.1 287.0 288.5
533.2 534.9
91.2 8.8
0.2
2.00
0.5
2.43
0.4
5.45
287.3 289.1 286.7 288.1 286.9 288.3
86.7 13.3 87.9 12.1 81.5 18.5
533.3 535.2 533.0 534.7 533.2 534.8
66.5 33.5 75.4 24.6 39.9 60.1
Binding energy
Relative ratio
A& (vs. Pt ref.)
(eV)
(%I
(eW
73.5 75.1
68.3 21.5
0.8
77.0 73.3 75.2 73.6 75.1
10.2 83.2 16.8 72.5 17.2
77.0 73.4 75.0
10.3 73.0 15.6
77.0 73.5 76.0 78.3 73.0 74.4 74.5 76.9 78.8 73.7 75.8 78.2 73.1 74.7 73.4 75.5 73.7 75.0 77.0
11.4 58.1 30.7 11.2 81.2 18.8 50.2 26.2 23.6 70.2 19.6 10.2 86.2 13.8 84.9 15.1 72.8 15.6 11.6
appears around 1.5-2.8 eV higher than that for metallic Pt, and the third pair is observed at 3.4-5.3 eV higher than that for metallic Pt. The intensities of the third doublet are quite low, especially for the samples with two Pt 4f BE listed in the table, and an attempt to obtain deconvoluted data was made. The second set of doublets of PtO or Pt (OH) 2 was determined to be 2.6 eV higher than that of metallic Pt [ 1.51, and a few of our samples agreed with this value. Many of the samples, however, showed less than 2.0 eV difference, probably due to the chemical nature of the Pt in the dispersion because of the varying degree of electronic interaction of Pt with the carbon support [7,16,17]. Similar binding energy shifts due to Pt( II) were observed on graphite-supported Pt [ 18,191, Pt/CO,-treated Vulcan XC-72 [ 20-231, Pt single crystals [ 24,251, and on Pt group bimetallic alloys on carbon [26]. A small binding energy shift due to the presence of PtO,,, should not be ruled out [ 271. However, Dijkgraaf et al. [ 281 pointed out the possibility of oxygen diffusion into the Pt lattice without formation of Pt-0, based on the great differences between the heat of chemisorption of oxygen (71 kcal/mol) and the heat of formation of PtO, (32.2 kcal/mol). Because of the reasons
ptx (%)
mentioned above, the signal from the second set of doublets can be from either or both of the partially oxidized Pt and/or Pt-0 moieties. Signals observed 3.4-5.3 eV higher have been reported to be due to the presence of PtO, .xH,O or Pt( OH), [ 281. Two C 1s signals can be assigned as carbon (graphitic) for 284.6 eV and -C=O for 285.4-286.2 eV. The third peak, observed around 288.5 eV, seems due to either -COO or carbonate. These assignments agree very well with the extensive XPS study made on commercially available carbon used as supports by Albers et al. [ 291. The highly oxidized carbon signal (around 288 eV) was also observed by Goodenough et al. [ 231, but no precise identification of the species could be made; Bowker and Madix [ 301 have suggested that this highly oxidized species is M-O-CH-O-M after XPS and EELS investigation of the adsorption of formic acid on Cu. The oxygen signal observed around 530.9 eV is in between the values 530 eV for Pt-0 and 53 1.5 for Pt-OH,,, [ 241. It is very difficult to distinguish the Pt-0 signal in the presence of a large amount of oxygenated groups, because the samples were exposed to the air. Thus no quantitative ratio of Pt 4f,,, to 0 1s can be made for the same reason. A similar binding energy of 0 Is is obtained using electrochemically
12
V. Alderucci et al. /Materials
Chemistry and Physics 41 (1995) 9-14
oxidized Pt [ 311. The peak at 532.5 eV is due to -t&O. Again, the presence of signals from Pt-HZ0 (533 eV) made it difficult to correlate the peak area to that from -C=O. The shift of the 0 1s signal from that of partially oxidized single crystal Pt to lower binding energy by heat treatment was observed by Peuckert [24] and confirms the presence of oxyhydrogen, which results from reduction of oxide, to give a lower BE. The binding energy shift (the difference between the BE of Pt 4f,,, of the sample and that of the platinum metal used as an external standard) is taken into account in the magnitude of the interaction between platinum and the carbon support. This represents the actual magnitude of the interaction on an absolute scale, because the BE of C 1s is influenced by the presence of Pt dispersion, as reported by Albers et al. [ 291. If the BE shift is measured against the BE of C 1s at 284.6 eV, the observed BE of Pt 4f will represent the difference from the already shifted C Is, depending on the nature of the charge transfer through the carbon support. This argument is acceptable as long as the energy calibration (work functions in a broad sense) of the instrument does not change during the measurements of the sample and platinum foil. It is believed to be so in this case, because the peaks of the external standard were obtained each time just before running each sample. Data derived from potentiometric titrations show that heat treatments varying between 800 and 950 “C induce marked changes in the basic properties of the catalysts samples, with the dissociation constants of surface functional groups in the range 3.4-3.8 and a concentration between 0.01 and 0.23 meq/g. Since the catalysts were prepared following the same experimental procedure (except for the final heat treatment temperature), the observed differences in the amount of oxygenated basic groups can be reconciled and explained in terms of different times of storage in air. It is well known, in fact, that upon exposure to air, oxygen fixation occurs on the active sites generated during heat treatment, giving rise to the formation of pyrone-type groups [ 61. By plotting the amounts of such basic functionalities (including both Crrand oxygenated pyrone-type groups) versus the relative fraction of oxidized Pt (derived from the deconvoluted Pt 4f spectra), it appears that higher Pt,, amounts are associated with the samples characterized by a high concentration of basic groups (Fig. 1 (a)). The scatter of the data is likely due to the fact that potentiometric titration does not allow distinction between CT and oxygenated groups. Such a differentiation can be made possible by analysis of the C 1s spectra, which show a noticeable asymmetry at high binding energies. This is caused by the oxidized carbon species at the surface, which have been separated, identified by curve fitting and attributed to the specific species by referring to Albers et al. [ 291. The less scattered trend of the data shown in Fig. 1 (b) seems to confirm the expectations, according to which increasing amounts of oxygen groups parallel the higher levels of surface Pt oxidation determined.
Fig. 2 shows a linear relationship between the Pt BE shift and the amount of Pt oxidized. This shows not only indirect evidence of the effect of oxidized carbon functionality in the catalyst, but also the ineffective screening of holes by oxidized Pt in the final state (after ejection of photoelectrons [ 321) . When all Pt atoms are oxidized and remain as Pt( II), the BE of Pt( II) will be about 2.6 eV higher than the values (4 : 0 q
0.05
0.1
Amount of baw
0.15
0.2
0.25
groups by ZPC, meqlg
(b)1
d 0
0
0
0’
5
10
15
20
I
/
1
J
25
30
35
40
Amount of oxidized C by XPS, %
Fig. 1. (a) Amount of oxidized Pt vs. concentration of basic groups detected by ZPC; (b) amount of oxidized Pt vs. oxygenated carbon functional groups detected by XPS. 10
0’ 0
I
’
’
0.1
0.2
0.3
0.4
’ 0.5
I
/
’
0.6
0.7
0.8
0.9
I
1
1
1.1
Pt binding energy shift, eV Fig. 2. Amount of Pt oxidized vs. Pt binding energy shift.
’
1.2
V. Alderucci et al. /Materials
Chemistry and Physics 41 (1995) 9-14
13
the necessity of controlling the preparation and activation procedures in order to lower the deactivation effects resulting from a too high population of oxygen-containing surface groups.
4. Conclusions
OL
0
1
2
3
4
Amount
5
6
of oxidized
Fig. 3. Pt surface area by cyclic voltammetry
7
1
2
3
4 Amount
I 10
9
Pt, 96
vs. the amount of Pt oxidized.
1
0
6
5 of oxidized
6
7
1
1
6
9
10
Pt, %
Fig. 4. Mass activity in oxygen vs. the amount of Pt oxidized.
expected from metallic Pt [ 151. In the presence of a large amount of oxidizing functionalities, however, the electronic states of Pt will be disturbed in such a way that the electron density is less than that of a neutral metal atom. Also, oxidized Pt near Pt atoms (2-3 atomic neighbors) cannot screen effectively compared with the metal. These, in turn, make the BE of Pt metal species shift toward higher BE. We have previously reported [ 7] a decrease in Pt dispersion (measured by CV) with increasing number of oxygenated surface groups (either basic or mildly acidic); this has been explained by the higher electron withdrawing character of the carbon-oxygen complexes as compared to C edge and CT functionalities. The Pt MSA/Pt,, plot shown in Fig. 3 clearly reflects the effects exerted by such complexes on the stabilization of oxidized Pt states; it appears that highly electroactive Pt areas are found in samples having low amounts of Pt,, and low amounts of basic oxygenated groups. The high Pt-C bond strength and the resonance stabilization of the aromatic ring of the electron pair are able to anchor the Pt particles without any concurrent deactivation (i.e., oxidation) effect. This view seems to be confirmed by the mass activity in the oxygen reduction reaction/Pt,, relationship shown in Fig. 4. Accordingly, the trend of the data explains
Our results clearly show that: ( 1) the amount of Pt-0 increases with increasing amount of basic groups (Fig. 1 (a) ) ; (2) the oxidation of C is associated with higher Pt-0 amounts (Fig. 1 (b) ) ; (3) increasing amount of oxidized Pt produces more Pt BE shift (Fig. 2); (4) the amount of electroactive Pt area decreases with increasing amount of Pt-0 (Fig. 3) ; (5) the activity decreases as the amount of Pt-0 increases (Fig. 4). The high activation temperatures (800-950 “C) of Pt/C catalysts lead to a high concentration of basic carbon groups. Most of them are C edge and Cn- functionalities, but upon exposure to air, oxygen fixation occurs to different extents, generating pyrone-type groups. Both the Pt binding energy shift and the amount of Pt oxidized, which are measured by XPS, show that the differences arising in the electrochemical properties of Pt/C catalysts for the oxygen reduction reaction are related to the relative amount of such basic groups. Particularly, an optimal interaction between the supported Pt crystallites and the carbon surface appears to be associated with the edge sites and Z- sites of carbon, reflecting a positive effect on the electroactive surface area of the catalysts. On the other hand, the oxygenated groups having a lower electron donating character (in comparison with C edge and Crfunctionalities) interact less with zero-valent Pt, affecting the physical and electronic state of the metal in such a way as to produce negative effects on its electrochemical properties.
Acknowledgements One of the authors (HK) is grateful to the HAN Project for financial support through the New Energy Development Project (component and cell technology development for PAFC).
References [ 11 S. Mukerjee, J. Appl. Electrochem., 20 ( 1990) 537. [2] F. Coloma, C. Prado-Burguete and F. Rodriguez-Reinoso, Proc. ZOth ht. Congress on Catalysis, Budapest, 19-24 July 1992, p. 301. [3] P. Birke, S. Engels, K. Becker and H.D. Nenbaner, Chem. Tech. (Leipzig), 32 (1980) 245. [4] P. Ehrburger, Adv. Colloid Interface Sci., 21 ( 1984) 275. [5] C. Prado-Burguete, A. Linares-Solano, F. Rodriguez-Reinoso and C. Salinas-Martinez De Lecea, J. Catal., 125 ( 1989) 98.
14
V. Alderucci et al. /Materials
Chemistry and Physics 41 (1995) 9-14
[6] AS. Aricb, V. Antonucci, L. Pino, P.L. Antonucci and N. Giordano, Carbon, 28 ( 1990) 599. [7] P.L. Antonucci, V. Alderucci, D.L. Cocke, H. Kim and N. Giordano, J. Appl. Electrochem.. 24 (1994) 58. [S] N. Giordano, E. Passalacqua, V. Alderucci, P. Staiti, L. Pino, H. Mirzaian, E.J. Taylor and G. Wilemski, Electrochim. Acra, 36 ( 1991) 1049. [ 91 G.A. Parks and P.L. de Bruyn, J. Phys. Chem., 66 ( 1962) 967. [lo] A.S. Aricb, V. Antonucci, M. Minutoli and N. Giordano, Carbon, 27 (1989) 337. [ 111 D.A. Shirley, Phys. Rev. B, 5 ( 1972) 4709. [ 121 C.D. Wagner, J. Electron Spectrosc. Relat. Phenom., 32 (1983) 99. [ 131 J.J. Yeh and I. Lindau, At. Data Nucl. Data Tables, 32 (1) (1985). [ 141 L.C. Feldman and J.W. Mayer, Fundamentals of Sur$ace and Thin Film Analysis, North-Holland, Amsterdam, 1986, pp. 226-228. [ 151 K.S. Kim, N. Winograd and R.E. Davis, J. Am. Gem. Sot., 93 (1974) 6296. [ 161 A.Q. Contractor and J. O’M. Bockris, Electrochim. Acta, 29 (1984) 427. [ 171 P.N. Ross, K. Kinoshita and P. Stonehart, J. Caral., 32 ( 1974) 163. [ 181 S. Wangand P.S. Fedkiw,J. Electrochem. Sot., 139 (1992) 3151. [ 191 A.K. Shulka, K.V. Ramesh, R. Manoharan, P.R. &rode and V. Vasudevan, Eler. Bunsenges. Phys. Chem., 89 (1985) 1261.
1201 B.J. Kennedy and A. Hamnett,J. 1211 J.B. Goodenough,
Electroanal. Chem., 283 ( 1990) 271.
A. Hamnett,
Electrochim. Acta, 32 (1987)
B.J. Kennedy
and S.A. Weeks,
1233.
1221 J.B. Goodenough, A. Hamnett, B.J. Kennedy, R. Manoharan Weeks, Electrochim. Acta, 35 ( 1990) 199.
and S.A.
[231 J.B. Goodenough, A. Hamnett, B.J. Kennedy, R. Manoharan Weeks, J. Electroanal. Chem., 240 (1988) 133.
and S.A.
[Xl M. Peuckert, Electrochim. Acta, 29 (1984) 1315. [251 M. Peuckert and H.P. Bonzel, Surf: Sci., 145 (1984) 239.
[261K.V. Ramesh,
P.R. Sarode,
S. Vasudevan
and A.K.
Shulka,
J.
Electronnal. Chem., 223 ( 1987) 91. 1271 C. Allen, P.M. Tucker,
A. Capon and R. Parsons,
J. Electroanal.
Chem., 50 (1974) 335. 1281 P.J.M. Dijkgraaf,
H.A.M.
Wiele, J. Cutal., 112 (1988)
Duisters,
B.F.M. Kuster and K. van der
337.
(291 P. Albers, K. Deller, B.M. Despeyroux,
A. Schafer and K. Seibold, J.
Catal., 133 (1992) 467. 1301 M. Bowker and R.J. Madix, Surf: Sci., 102 (1981) 542. [311 T. Dickinson, A.F. Prey and P.M.A. Sherwood, J. Chem. Sot., Faraday Trans. 1, 7i (1975) 298. 1321 B. Johansson and N. Martensson,
Phys. Rev. B. 21 ( 1980) 4427.