- Email: [email protected]
Electron energy loss spectroscopy of Pd and Pd–Au catalysts supported on multiwall carbon nanotubes
Journal of Electron Spectroscopy and Related Phenomena 185 (2012) 486–491
Contents lists available at SciVerse ScienceDirect
Journal of Electron Spectroscopy and Related Phenomena journal homepage: www.elsevier.com/locate/elspec
Electron energy loss spectroscopy of Pd and Pd–Au catalysts supported on multiwall carbon nanotubes B. Lesiak a,∗ , L. Stobinski a , L. Kövér b , J. Tóth b a b
Institute of Physical Chemistry, Polish Academy of Sciences, Kasprzaka 44/52, 01-224 Warsaw, Poland Institute of Nuclear Research, Hungarian Academy of Sciences (ATOMKI), P.O. Box 51, H-4001 Debrecen, Hungary
a r t i c l e
i n f o
Article history: Received 19 September 2012 Accepted 1 November 2012 Available online 9 November 2012 Keywords: X-ray PhotoElectron Energy Loss Spectroscopy XP-EELS Reflection Electron Energy Loss Spectroscopy REELS Pd/MWCNT Pd–Au/MWCNT
a b s t r a c t The X-ray PhotoElectron Energy Loss Spectroscopy (XP-EELS) and Reflection Electron Energy Loss Spectroscopy (REELS) were used for analysing surface layers of “as-received” and functionalised multiwall carbon nanotubes (MWCNT), and MWCNT decorated with Pd and Pd–Au particles after calcination/reduction. The decorated MWCNT were previously applied as catalysts in a reaction of formic acid electrooxidation. These spectroscopies, used as complementary methods of structural surface analysis, provide information on the energy position, intensity and full width at half maximum of the quasi-elastic peak and inelastic and + energy loss peaks. Analysing the + energy loss peak, the bulk and surface C sp2 /sp3 components can be separated. Functionalisation of MWCNT, catalyst reduction and Ar+ ion sputtering increase the C sp3 content in comparison to the “as-received” MWCNT and calcined catalysts. The intensity ratios of surface and bulk C sp3 and sp2 components evaluated from the REELS + energy loss peak indicate: (i) functionalisation leads to attachment of functional groups to the MWCNT surface, (ii) calcined catalysts show an amorphous carbon overlayer at the surface and (iii) reduction of calcined catalysts leads to increasing C sp3 hybridisations. © 2012 Elsevier B.V. All rights reserved.
1. Introduction The Pd and Pd–Au catalysts supported on the MWCNT have recently been by thermogravimetric analysis (TGA), X-ray diffraction (XRD), scanning electron microscopy (SEM), transmission electron microscopy (TEM) and electron spectroscopic methods, i.e. X-ray photoelectron spectroscopy (XPS) and X-ray excited Auger electron spectroscopy (XAES) [1–6]. Such Pd/MWCNT and Pd–Au/MWCNT samples were previously tested for their catalytic activity in the reaction of formic acid electrooxidation [4,5]. The reduced catalysts were active, the Pd–Au/MWCNT catalyst showed higher activity than Pd/MWCNT, whereas the calcined catalysts were inactive. This catalytic activity was attributed to a larger metallic surface area, a thinner palladium oxide and amorphous carbon layers on Pd metallic crystallites due to reduction causing increase of metallic palladium surface area [5–8]. The X-ray PhotoElectron Energy Loss Spectroscopy (XP-EELS) and the Reflection Electron Energy Loss Spectroscopy (REELS) spectra are also of importance in characterisation of surfaces and thin films [9]. The electron passing through a surface layer of a solid interacts with the free (quasi-free) and weakly bound electrons.
Quanta of collective excitations running as longitudinal charge density oscillations through a solid volume are referred to as volume plasmons, whereas along the surface region as surface plasmons. The expression “bulk excitation” denotes any energy loss phenomenon occurring in an infinite medium without any boundary, whereas “surface excitation” denotes excitation due to the presence of a solid–vacuum boundary. The surface modes of the inelastic excitation have a lowered resonance frequency and are orthogonal to the volume modes. The energy loss process involves the energy quanta of the oscillations (plasmons) and their excitation probability. The plasmon energy, Ep = ω, is related to the frequency, ω, of the oscillations, which depends on the density of weakly bound electrons for which the plasmon energy is large in comparison to their binding energy. The electron energy loss spectrum is characterised by the energy loss peaks, i.e. the loss involving only type valence electrons known as free-like electrons and + loss involving all the valence electrons. The energies and intensities (the latter ones measured by the peak areas) of the electron loss peaks (XP-EELS and REELS) characterise the near-surface valence electron density and the structural organisation of C sp2 /C sp3 bonds, which can be also quantified from the equation [10]: sp2 =
∗ Corresponding author. Tel.: +48 22 343 3432; fax: +48 22 632 5276. E-mail address: [email protected] (B. Lesiak). 0368-2048/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.elspec.2012.11.005
Area /Area+ (Area /Area+ )standard
(1)
with a normalisation to a standard, e.g. graphite. Therefore, fitting of the REELS spectra into the and + (surface and volume)
B. Lesiak et al. / Journal of Electron Spectroscopy and Related Phenomena 185 (2012) 486–491
loss peaks may be applied for characterisation of the surface and volume contributions [9]. Otherwise, the C sp2 /C sp3 content may be evaluated from the C 1s XPS spectra by fitting to C sp2 and C sp3 components and from the energy width of the first derivative of the C KLL XAES spectra (parameter D), using a linear interpolation of the parameter D for graphite (100% of C sp2 ) and diamond (100% of C sp3 ) [11]. The accuracy of this estimation depends on the accuracy of the assumed values of parameter D for diamond and graphite, which vary from 13.0 eV to 14.5 eV (diamond) and from 21.1 eV to 22.8 eV (graphite) [12 and references within]. The research using the REELS spectroscopy may have a broader impact for catalysis due to the possibility of distinguishing the electron signal from boundary from the respective signal from the outer layers since the catalytic reactions proceed at the surface. Such research can be especially useful for characterisation of catalyst performance and catalyst designing. Application of commonly used techniques of surface analysis such as XPS, XP-EELS and XAES provides the surface sensitivity from outer layers and is described by a few parameters, i.e. the inelastic mean free path (IMFP), the mean escape depth (MED), the information depth (ID), etc. where the IMFP is a material parameter dependent on electron kinetic energy (KE) and the other parameters depend on the IMFP and experimental conditions [13]. The above parameters are calculated from the emission depth distribution function (DDF), defining the probability that a photoelectron and/or Auger electron leaving a surface in a given direction originates from a specified depth measured inward along the surface normal. The MED is defined as an average depth normal to the surface from which the electron escapes. The ID is a maximum depth normal to the surface from which useful signal information is obtained, i.e. a thickness from which a specified percentage P (95%, 99%) of the detected signal originates. Neglecting elastic scattering, the following analytical expressions are valid: MED = cos ˛out and ID = cos ˛out ln(1/((1 − P)/100)), where is the IMFP and ˛out is the emission angle of the electron with respect to the surface normal. By analogy, in REELS the depths in the solid for a particular electron trajectory contribution are distributed according to probability density function referred to as a penetration depth distribution function (PDDF). This parameter is defined as a probability that an electron incident on the surface at an angle ˛in be elastically backscattered from a maximum depth, z, and emitted in the direction of the analyser at an angle ˛out and not inelastically backscattered, where ˛in and ˛out are measured with respect to the surface normal. Neglecting elastic scattering the mean penetration depth (MPD) and ID are expressed as: MPD = (cos ˛in cos ˛out /(cos ˛in + cos ˛out )) and ID = (cos ˛in cos ˛out /(cos ˛in + cos ˛out )) ln(1/((1 − P)/100)). Assuming an electron IMFP in graphite for respective electron KE of 272 eV, 1202 eV and 4000 eV as: 0.72 nm (C KLL), 2.10 nm (C 1s) and 5.60 nm (REELS) [14] the above parameters are: MED = 0.72 nm (C KLL), MED = 2.1 nm (C 1s) and MPD = 2.2 nm (REELS), with the ID values for P of 99%: 3.3 nm (C KLL), 9.7 nm (C 1s) and 10.1 nm (REELS). Therefore, the MED given by Auger electrons refers to about 2 carbon walls, whereas the MED and MPD from the C 1s XP-EELS and REELS spectra to about 6 carbon walls. For an overlayer/substrate system the ID may differ depending on the material atomic number, Z, of the overlayer and substrate and their electron elastic differential scattering cross-sections [15] and morphology [16]. The importance of characterisation using the REELS + energy loss spectra is based on the possibility of distinguishing the surface boundary and bulk components between the respective C sp2 /C sp3 contributions. The XP-EELS and REELS spectroscopies are applied for investigating the surface of “as-received” MWCNT, functionalised by oxidation MWCNT (ox-MWCNT), and MWCNT decorated with Pd and Pd–Au particles after different chemical and physical
487
treatments, i.e. reduction, calcination, and Ar+ ion surface cleaning of Pd–Au/MWCNT calcined sample. Structural changes in the XP-EELS and REELS spectra are discussed accounting for the energy positions and the intensity of the quasi-elastic peak, the inelastic and + loss peaks, and the + loss peak surface and bulk contributions.
2. Experimental 2.1. Samples and preparations Commercial MWCNT (Echo-nanobio. Tech. Co. Ltd., Japan) with an average diameter of 50 nm and length of 5–15 mm, were purified and functionalised (65% aq. HNO3 , 6 h, 118oC) [2,17]. Precursors of 17.8 wt% Pd/MWCNT and 8.8 wt% Pd–10.0 wt% Au/MWCNT catalysts were prepared by a polyol method described in details elsewhere [2].
2.2. Apparatus The XPS and REELS spectra were measured in the ultra-highvacuum ESA-31 spectrometer [18] equipped with a high energy resolution electron analyser, an electron gun, an X-ray excitation source and Ar+ ion source. The C 1s XPS, C KLL XAES and XP-EELS spectra were measured using Al Ka X-rays (h = 1486.67 eV), in a fixed retarding ratio (FRR) working mode, at photon incidence and electron emission angles of 70◦ and 0◦ , respectively, with respect to the surface normal. The REELS spectra were measured using the FRR working mode, a primary electron beam with a spot diameter of 1.5–2.0 mm, an electron current of about 5 nA and primary electron beam of 4000 eV KE, the incidence and emission angles of 50◦ and 0◦ , respectively, with respect to the surface normal. The applied measurement geometries resulted from optimisation of experimental signal intensity and apparatus set up. The ion sputtering was carried out using an Ar+ ion beam of 3 keV energy with a current of 30 A for 1 min.
3. Results 3.1. Samples – surface compositions and notations The investigated samples exhibit the following bulk compositions resulting from the chemical synthesis, i.e. 17.8 wt% Pd/MWCNT and 8.8 wt% Pd–10.0 wt% Au/MWCNT, i.e. below 1 at.% of metallic content. The XPS analysis indicated the presence of Au, Pd, C and O at the surface [4,5]. The surface content of Au and Pd resulting from XPS quantitative analysis [4,5], chemical treatments and assumed notations denoting the samples after chemical treatments are the following: (i) 10.6 wt% Pd/MWCNT calcined at 250 ◦ C in Ar: Pd-C250; (ii) 4.3 wt% Pd–4.6 wt%Au/MWCNT calcined at 250 ◦ C in Ar: Pd–AuC250; (iii) 9.1 wt% Pd/MWCNT reduced at 200 ◦ C in Ar and 5% H2 : Pd-R200; (iv) 4.4 wt% Pd–5.0 wt% Au/MWCNT reduced at 200 ◦ C in Ar and 5% H2 : Pd–Au-R200; (v) 5.5 wt% Pd–3.8 wt% Au/MWCNT calcined at 250 ◦ C in Ar and Ar+ ion sputtered in UHV: Pd–Au-C250–Ar+ . The O content for the Pd decorated samples after calcinations and reduction was 11.7 wt% and 7.7 wt%, respectively, whereas for the Pd–Au decorated samples – 6.5 wt% and 7.0 wt%, respectively [4,5].
488
B. Lesiak et al. / Journal of Electron Spectroscopy and Related Phenomena 185 (2012) 486–491
REELS KE=4000 eV 0.09 +
PdAu-C250-Ar
Intensity difference (cps)
0.08 0.07 0.06
PdAu-C250
0.05 0.04
PdAu-R200
0.03 0.02
Pd-C250
0.01 0.00
Pd-R200
-0.01
0
10
20 30 40 Energy loss (eV)
50
60
Fig. 2. Intensity difference between the REELS normalised intensity spectra recorded for Pd and PdAu decorated MWCNT and for the ox-MWCNT.
-3
8.0x10
-4
6.0x10
-4
4.0x10
-4
2.0x10
-4
Ma(d d )/
T
1.0x10
KE=4000 eV emission solid angle
C
Pd
0.0
Au 80
90
3.2. REELS and C 1s XP-EELS spectra The REELS, C 1s XP-EELS and C 1s spectra recorded from the investigated samples are fitted to asymmetric Gaussian–Lorentzian functions using the XPSPEAK4.11 software after linear (REELS) and Tougaard (C 1s XP-EELS) background2 subtraction. Comparison of C 1s XP-EELS and REELS spectra recorded from “as-received”, oxidised and Pd and Pd–Au decorated MWCNT is shown in Fig. 1. The REELS spectra recorded using the primary beam with an electron KE of 4000 eV were not corrected for effects of atomic recoil. No separated components resulting from the electron inelastic loss on metallic Pd and/or Au are shown distinctly in the spectra (Fig. 1a). This is confirmed by the intensity difference between the REELS spectra recorded for Pd and PdAu decorated MWCNT and for the ox-MWCNT (Fig. 2). In a single elastic electron scattering model the elastic backscattering intensity is proportional to a differential scattering cross section for elastic scattering, d/d˝, and inversely proportional to a total elastic scattering cross-section, T , where the elastic cross-sections and their angular dependence are
1
Kwok RWM. XPS Peak Fitting Program for WIN95/98 XPSPEAK Version 4.1, Department of Chemistry, The Chinese University of Hong Kong, [email protected]; http://www.uksaf.org/software.html, XPSPEAK4.1. 2 Tougaard S. Background analysis of XPS/AES. QUASES Simple Background, Ver. 2.2, Copyright 1994–2001; http://www.tougaard.com.
100 110 120 130 140 150 160 170 180 190
Scattering angle,
Fig. 1. Comparison of spectra recorded from “as-received”, oxidised and Pd and PdAu decorated MWCNT. (a) REELS. (b) C 1s XP-EELS.
(deg)
Fig. 3. Comparison of the angular dependence of ratio of differential elastic crosssections to total cross-sections accounting for atomic content of C, Pd and Au, Ma (number of atoms per unit volume) for primary electrons of KE = 4000 eV.
provided by a NIST database [15]. The angular dependence of ratio of differential elastic cross-sections to total cross-sections accounting for atomic content of C, Pd and Au, Ma (number of atoms per unit volume), i.e. Ma (d/d˝)/ T , indicates significant contribution from C in contrary to Pd and Au (Fig. 3). Therefore contribution of Pd and Au to the REELS spectra can be neglected. The REELS spectra were fitted to the quasi-elastic peak, and + loss peaks, with the + loss peak separated into C sp2 and C sp3 surface and bulk components (Fig. 4a and Tables 1 and 2). A comparison of electron energy loss features in the REELS Table 1 Comparison of the electron energy loss features in the measured REELS spectra. Sample
“As-received” MWCNT ox-MWCNT Pd-C250 Pd-R200 Pd–Au-C250 Pd–Au-R200
REELS (primary beam KE = 4000 eV) Energy loss (eV)
Area peak/area total (%)
+
+
Quasi-el. peak
6.1 6.2 6.3 6.2 6.2 6.0
26.2 25.8 25.7 25.8 25.5 25.8
5.3 4.5 6.5 5.2 5.9 5.5
33.7 29.6 29.7 30.5 27.8 29.3
61.0 65.9 63.8 64.3 66.3 65.2
B. Lesiak et al. / Journal of Electron Spectroscopy and Related Phenomena 185 (2012) 486–491
489
Table 2 Comparison of the C sp2 and C sp3 (surface and bulk) contributions (REELS spectra) of the + electron energy loss peaks. Sample
“As-received” MWCNT ox-MWCNT Pd-C250 Pd-R200 Pd–Au-C250 Pd–Au-R200
+ energy loss peaks contribution (%) [energy loss (eV)] → *[13.5 ± 0.5]
sp2 S[18.4 ± 0.3]
sp2 B[27.4 ± 0.4]
sp3 S[23.5 ± 0.5]
sp3 B[33.5 ± 0.5]
→ *>38.0
1.4 1.0 5.3 8.5 4.1 6.9
11.5 9.5 14.1 8.5 12.8 12.0
37.0 37.1 27.1 41.0 25.8 31.0
28.6 26.9 30.1 16.4 29.5 23.3
18.2 21.2 20.3 22.7 23.9 22.9
3.3 4.4 3.1 2.9 3.9 3.9
spectra and these spectra + loss peak C sp2 and C sp3 contributions, → * interband transitions is given in Tables 1 and 2, respectively. Mean percentage deviations between the experimental values and those values obtained from fittings to experimental data range from 5.3% to 24.1%. The loss peak energy (at about 6.0 eV) and the , + loss peaks and quasi-elastic peak contributions resulting from the REELS spectra fitting (Fig. 4a) indicate the changes of the electronic structure in different samples (Fig. 1a and Table 1). Large scatter of the + loss peak energy values for the surface and bulk C sp2 /C sp3 contributions is observed in the spectra recorded from different samples (Table 2). These values for MWCNT and MWCNT decorated with Pd and Pd–Au differ from the values reported for graphite and diamond [9]. The → * interband transition at 13.5 eV may result from an overlap of the respective transitions occurring in graphite at 12.5 eV and in diamond
at 12.8 eV [9]. The surface C sp2 component is observed at 18.5 eV instead at 19.5 eV (graphite), whereas its bulk component is at 27.4 eV instead at 27.0 eV (graphite). However, the + loss peak surface and bulk C sp3 components for MWCNT are observed at similar (compared to that in the diamond) energy distance from the energy position of the elastic peak. The fitting results of the + loss peaks indicate also differences in the intensity of surface and bulk C sp2 /C sp3 contributions (Table 2) and differences in intensity ratios of surface and bulk contributions (Fig. 5). Oxidation of “as-received” MWCNT and reduction of calcined Pd-MWCNT and Pd–Au-MWCNT cause stronger increase of surface C sp3 (Fig. 5a). Also, after the respective oxidation and reduction the surface C sp3 content increases stronger than the overall C sp3 content (Fig. 5b). The C 1s XP-EELS were fitted to and + loss peaks [19] and different C chemical groups applying binding energy (BE) values according to Butenko et al. [20]. An exemplary fitted C 1s XP-EELS spectrum is shown in Fig. 4b, whereas comparison of C chemical group content in Table 3. Mean percentage deviations between the experimental values and those values obtained from fittings to experimental data of C 1s XP-EELS spectra range from 1.9% to 2.7%.
(a)
)y
2
3
)x/ I(
2
2
Pd-Au -C250 Pd-Au -R200
Pd-R200
Csp S/Csp B Pd-C250
0
"as-received" MWCNT oxMWCNT
I(
3
Csp S/Csp B
1
Samples
(b)
3 3
2
)y
2
)x/ I(
Csp S/Csp S
1
3
2
Pd-Au -C250 Pd-Au -R200
Pd-R200
Pd-C250
0
"as-received" MWCNT oxMWCNT
I(
Csp /Csp
Samples Fig. 4. The exemplary spectra and the fitting results to an asymmetric Gaussian–Lorentzian function (C C components) after linear background subtraction. (a) REELS recorded from “as-received” MWCNT. (b) C 1s XP-EELS recorded from PdAu-C250 sample.
Fig. 5. Comparison of the intensity ratios of the C sp2 /C sp3 surface and bulk components evaluated from the REELS spectra. (a) Surface to bulk. (b) Surface C sp3 /C sp2 and C sp3 /C sp2 .
490
B. Lesiak et al. / Journal of Electron Spectroscopy and Related Phenomena 185 (2012) 486–491
Table 3 Comparison of the C group content in the measured C 1s XP-EELS spectra recorded from “as-received”, oxidised and Pd and PdAu decorated MWCNT. C chemical groups content (%) [BE (eV)]
“As-received” MWCNT ox-MWCNT Pd-C250 Pd-R200 Pd–Au-C250 Pd–Au-R200 Pd–Au-C250–Ar+
sp2 [284.4]
sp3 [285.7]
C OH[286.3]
C O[287.7]
C OOH[288.5]
91.9 83.7 85.4 83.4 85.6 83.8 82.2
1.1 6.4 7.5 2.5 5.1 5.5 9.1
4.8 5.3 3.2 8.8 5.8 5.9 0
2.2 2.8 2.5 1.3 2.7 3.2 1.7
0 1.8 1.4 4.0 0.8 1.6 7.0
Similarly to the REELS spectra the C 1s XP-EELS spectra indicate changes of selected features, i.e. and + energy and intensity, depending on the sample preparation and treatment (Fig. 1b). The C group content resulting from the C 1s spectra fitting indicates: (i) increase of C sp3 , hydroxyl (C OH) and carbonyl (C = O) and carboxyl (C OOH) groups after oxidation of “as-received” MWCNT, (ii) increase of carboxyl (C OOH) and hydroxyl (C OH) groups and decrease of carbonyl (C O) after reduction of decorated MWCNT (Table 3). 3.3. C sp2 content The content of C sp2 was evaluated from: 1. the energy width of the C KLL Auger spectra assuming D values of 13.0 eV (diamond) and 22.6 eV (graphite) [11,12]; the mean percentage deviation resulting from a scatter of the assumed D values range from 3.4% to 10.4%; 2. the fitting of Gaussian–Lorentzian to the REELS and C 1s XP-EELS quasi-elastic, and + loss peaks and using Eq. (1), where – as a standard – the “as-received” MWCNT (C sp2 of 91.9% (Table 3)) sample was applied 3. the fitting of C 1s spectra according to BE suggested by Butenko et al. [20] to C chemical groups, i.e. C sp2 , C sp3 , C OH, C O and C OOH (Table 3). Comparison of the C sp2 content derived from the above procedures is shown in Fig. 6. The results indicate qualitative similarity, i.e. (i) increase of C sp3 after oxidation and after reduction, more significantly after oxidation and (ii) significant increase (up to about 50%) of C sp3 after Ar+ sputtering. A content of C sp2 is similar for “as-received” and calcined decorated MWCNT and larger than for oxidised and reduced samples, indicating traces of amorphous carbon resulting from synthesis. A quantitative agreement is observed between the results obtained from fitting of C 1s XP-EELS, REELS and evaluation of the CKLL C 1s fit REELS C 1s XP-EELS
100
PdAu-C250 + -Ar
PdAu-R200
PdAu-C250
Pd-R200
Pd-C250
40
ox-MWCNT
60
"as-received" MWCNT
2
C sp (%)
80
20
Sample Fig. 6. Comparison of the C sp2 percent resulting from evaluation of different spectra.
C KLL spectra width. The content of C sp2 resulting from the fitting of the C 1s spectra into C chemical groups is slightly smaller. Discrepancies seem not to result from differences in the ID, but rather from the systematic error in evaluating of C chemical group content from spectra fitting (Fig. 6).
4. Discussion Previous studies on the investigated Pd and Pd–Au MWCNT catalysts by XPS, TEM and XRD characterised the surface chemical features and metallic phase structural features [4–6]. The surface content of Pd and Au in Pd/MWCNT and Pd–Au/MWCNT after calcination, reduction and Ar+ sputtering was about a half of the content resulting from the chemical treatment [4,5]. The underestimation of the Pd content can be explained by an overlayer of Pd oxide and amorphous carbon on metallic Pd core, attenuating the measured intensity of photoelectrons emitted from Pd. Reduction decreases the content of oxygen [4,5]. For the calcined Pd/MWCNT sample reduction decreases Pd oxide and amorphous carbon overlayer thickness from 10.5 nm to 3 nm [4], whereas after Ar+ sputtering from 10.5 nm to 7.0 nm [6]. For the calcined Pd–Au/MWCNT reduction decreases an overlayer thickness from 4.3 nm to 2.2 nm [5]. The size of Pd and Pd–Au solid solution nanocrystallites is nearly similar, i.e. 4.2 nm and 4.8 nm [4,5]. The intensity of the X-ray photoelectron signal [21], quasielastic peak [22] and consequently energy loss peaks are influenced by the surface roughness. This depends on the angular spread of the local area distribution of slopes and the experimental geometry, i.e. the emission angle with respect to the surface normal. The XPS and quasi-elastic peak measurements and calculations showed that an error induced by a random surface corrugation is acceptably low (up to 10%) for emission angles to 35◦ [21] and 0◦ [22], respectively. The C 1s XP-EELS spectrum of graphite (sp2 ) exhibits the typical electron energy loss feature at 6.5 eV ( electrons) and at 19.5 eV and 27 eV ( + surface and bulk electrons), whereas diamond (sp3 ) shows loss peaks attributed to surface and bulk plasmons at 23 eV and 33.8 eV [9]. In sp2 graphene, the respective and + surface loss peaks for a free standing sheet are observed at lower energies than in graphite, i.e. 4.7 eV and 14.6 eV, approaching the graphite structure for about 10 sheets [23]. For the curved single layer C sp2 structure (onion-like fullerenes and nanotubes) the loss peak is observed at about 5 eV, whereas the surface and volume plasmons are observed at higher energies than in graphene, i.e. 16–19 eV and 23–27 eV, with values inversely proportional to the diameter [24]. The shift of energy loss peak values results from the coupling of tangential (in-plane) and radial (out-of-plane) modes on the internal and external surfaces [25]. The nanometallic particles in the CNT will modify the REELS spectra indicating the significant surface and bulk plasmon features resulting from the metallic component [26]. The following loss peaks may be observed: Pd – 6.5 eV, 16.0 eV, 25.1 eV and 31.9 eV, Au – 3.2 eV, 5.8 eV, 16.7 eV, 24.1 eV and 33.0 eV and Pd69 at.%Au31 at.% – 6.3 eV, 24.3 eV and 32.4 eV [27]. The REELS spectra of the
B. Lesiak et al. / Journal of Electron Spectroscopy and Related Phenomena 185 (2012) 486–491
“as-received”, oxidised and decorated with Pd and Pd–Au particles MWCNT show the and + energy loss values typical for carbon materials with no Pd and Au contribution due to small (below 1 at.%) content of metal [4,5] (Tables 1 and 2 and Figs. 1–4). Changes due to sample preparation and treatment result in variation of energy and intensity of and + loss peaks (Table 1 and Fig. 1) and + loss peak surface and bulk C sp2 and C sp3 contributions (Table 2 and Fig. 5). The surface and bulk C sp2 and C sp3 + loss peak energies do not coincide with the respective energies for graphite and diamond [9]. The larger energy difference between the C sp2 and C sp3 + loss peaks (Table 1), similarly to that between the BE of C 1s C sp2 and C sp3 [20] may result from the MWCNT curvature. Oxidation, reduction and Ar+ ion sputtering cause increase of C sp3 hybridisations (Figs. 5 and 6), at more significant rate for the surface (Fig. 5), indicating that these processes occur at the surface. A quantitative consistency of the results is obtained using different methods, i.e. evaluation of C sp2 from C 1s XP-EELS, REELS, C KLL (Fig. 6). No influence on ID is observed. 5. Conclusions The REELS were applied for the analysis of the C sp2 /sp3 bulk and surface ratio in “as-received”, functionalised and decorated with Pd and Pd–Au particles MWCNT. It is shown that MWCNT functionalisation and reduction of Pd and Pd–Au/MWCNT induce changes in the bulk different from that at the uppermost surface layer. Functionalisation – increasing significantly the surface C sp3 /C sp2 ratio – indicates that the oxidised carbon groups are created mainly at the outer walls of the MWCNT. The less significant increase of the surface C sp3 /C sp2 ratio after reduction confirms that nanometallic particles are placed on the MWCNT inhibiting reduction of oxygen groups at the surface. The proposed analysis provides information on the electronic structure of catalyst surface layers, what is of great importance for a better understanding of the catalytic processes. Acknowledgments This work was supported by the Polish Academy of Sciences and Hungarian Academy of Sciences, Projects of the Polish Council for Science: 15-0011-04/2008, KB/72/13447/IT1-B/U/08. The authors
491
L. Kövér and J. Tóth would like to thank for the support of the project OTKA-K67873. References [1] A. Borodzinski, A. Mikolajczuk, C.H. Chen, W.J. Liou, H.M. Lin, S.H. Wu, P. Kedzierzawski, L. Stobinski, K. Kurzydlowski, Proceedings of EFC2009, Rome, Italy, 2009. [2] C.H. Chen, W.J. Liou, H.M. Lin, S.H. Wu, A. Borodzinski, L. Stobinski, P. Kedzierzawski, Fuel Cells 10 (2010) 227. [3] C.H. Chen, W.J. Liou, H.M. Lin, S.H. Wu, A. Mikolajczuk, L. Stobinski, A. Borodzinski, P. Kedzierzawski, K. Kurzydlowski, Phys. Stat. Solidi A 207 (2010) 1160. [4] A. Mikolajczuk, A. Borodzinski, L. Stobinski, P. Kedzierzawski, B. Lesiak, L. Kövér, J. Tóth, H.M. Lin, Phys. Stat. Solidi B 247 (2010) 3063. [5] A. Mikolajczuk, A. Borodzinski, L. Stobinski, P. Kedzierzawski, B. Lesiak, L. Kövér, J. Tóth, H.M. Lin, Phys. Stat. Solidi B 247 (2010) 2717. [6] B. Lesiak, L. Stobinski, L. Kövér, J. Tóth, K.J. Kurzydłowski, Phys. Stat. Solidi A 208 (2011) 1791. [7] M. Mazurkiewicz, A. Malolepszy, A. Mikolajczuk, L. Stobinski, A. Borodzinski, B. Lesiak, J. Zemek, P. Jiricek, Phys. Stat. Solidi B 248 (2011) 2516. [8] A. Malolepszy, M. Mazurkiewicz, A. Mikolajczuk, L. Stobinski, A. Borodzinski, B. Mierzwa, B. Lesiak, J. Zemek, P. Jiricek, Phys. Stat. Solidi C 8 (2011) 3195. [9] L. Calliari, S. Fanchenko, M. Filippi, Carbon 45 (2007) 1410. [10] A.C. Ferrari, A. Libassi, B.K. Tanner, V. Rtolojan, J. Yuan, L.M. Brown, S.E. Rodil, B. Kleinsorge, J. Robertson, Phys. Rev. B 62 (2000) 11089. [11] J.C. Lascovich, S. Scaglione, Appl. Surf. Sci. 78 (1994) 17. ´ Anal. Sci. 26 (2010) 217. [12] B. Lesiak, J. Zemek, J. Houdkova, A. Kromka, A. Józwik, [13] A. Jablonski, C.J. Powell, J. Vac. Sci. Technol. A 27 (2009) 253. [14] S. Tanuma, C.J. Powell, D.R. Penn, Surf. Interface Anal. 43 (2011) 689. [15] F. Salvat, A. Jablonski, NIST Electron-Elastic-Scattering Cross-Section Database, ver. 3.1, NIST SRD 64m, Gaithersburg, USA, 2003. [16] L. Zommer, A. Jablonski, J. Electron Spectrosc. Relat. Phenom. 150 (2006) 56; L. Zommer, Surf. Sci. 600 (2006) 4735; L. Zommer, J. Electron Spectrosc. Relat. Phenom. 169 (2009) 57. [17] S. Kundu, Y. Wang, W. Xia, M. Muhler, J. Phys. Chem. C 112 (2008) 16869. ˝ Surf. Interface Anal. 19 (1992) 9. [18] L. Kövér, D. Varga, I. Cserny, J. Tóth, K. Tokési, [19] L. Stobinski, B. Lesiak, L. Kövér, J. Tóth, S. Biniak, S. Trykowski, J. Judek, J. Alloys Compd. 501 (2010) 77. [20] Yu.V. Butenko, S. Krishnamurthy, A.K. Chakraborty, V.L. Kuznetsov, V.R. Dhanak, ˇ Phys. Rev. B 71 (2005) 075420. M.R.C. Hunt, L. Siller, [21] J. Zemek, K. Olejnik, P. Kapetek, Surf. Sci. 602 (2008) 1440. ´ [22] A. Jablonski, K. Olejnik, J. Zemek, J. Electron Spectrosc. Relat. Phenom. 152 (2006) 100. [23] T. Eberlein, U. Bangert, R.R. Nair, R. Jones, M. Gass, A.L. Bleloch, K.S. Novoselov, A. Geim, P.R. Briddon, Phys. Rev. B 77 (2008) 233406. [24] L. Henrard, O. Stephan, C. Colliex, Synth. Met. 103 (1999) 2502. [25] O. Stéphan, D. Taverna, M. Kociak, K. Suenaga, L. Henrard, C. Colliex, Phys. Rev. B 66 (2002) 155422. [26] Z.J. Ding, H.M. Li, Q.R. Pu, Z.M. Zhang, R. Shimizu, Phys. Rev. B 66 (2002) 085411. [27] H.A.E. Hagelin-Weaver, J.F. Weaver, G.B. Hoflund, G.N. Salaita, J. Alloys Compd. 393 (2005) 93.
Contents lists available at SciVerse ScienceDirect
Journal of Electron Spectroscopy and Related Phenomena journal homepage: www.elsevier.com/locate/elspec
Electron energy loss spectroscopy of Pd and Pd–Au catalysts supported on multiwall carbon nanotubes B. Lesiak a,∗ , L. Stobinski a , L. Kövér b , J. Tóth b a b
Institute of Physical Chemistry, Polish Academy of Sciences, Kasprzaka 44/52, 01-224 Warsaw, Poland Institute of Nuclear Research, Hungarian Academy of Sciences (ATOMKI), P.O. Box 51, H-4001 Debrecen, Hungary
a r t i c l e
i n f o
Article history: Received 19 September 2012 Accepted 1 November 2012 Available online 9 November 2012 Keywords: X-ray PhotoElectron Energy Loss Spectroscopy XP-EELS Reflection Electron Energy Loss Spectroscopy REELS Pd/MWCNT Pd–Au/MWCNT
a b s t r a c t The X-ray PhotoElectron Energy Loss Spectroscopy (XP-EELS) and Reflection Electron Energy Loss Spectroscopy (REELS) were used for analysing surface layers of “as-received” and functionalised multiwall carbon nanotubes (MWCNT), and MWCNT decorated with Pd and Pd–Au particles after calcination/reduction. The decorated MWCNT were previously applied as catalysts in a reaction of formic acid electrooxidation. These spectroscopies, used as complementary methods of structural surface analysis, provide information on the energy position, intensity and full width at half maximum of the quasi-elastic peak and inelastic and + energy loss peaks. Analysing the + energy loss peak, the bulk and surface C sp2 /sp3 components can be separated. Functionalisation of MWCNT, catalyst reduction and Ar+ ion sputtering increase the C sp3 content in comparison to the “as-received” MWCNT and calcined catalysts. The intensity ratios of surface and bulk C sp3 and sp2 components evaluated from the REELS + energy loss peak indicate: (i) functionalisation leads to attachment of functional groups to the MWCNT surface, (ii) calcined catalysts show an amorphous carbon overlayer at the surface and (iii) reduction of calcined catalysts leads to increasing C sp3 hybridisations. © 2012 Elsevier B.V. All rights reserved.
1. Introduction The Pd and Pd–Au catalysts supported on the MWCNT have recently been by thermogravimetric analysis (TGA), X-ray diffraction (XRD), scanning electron microscopy (SEM), transmission electron microscopy (TEM) and electron spectroscopic methods, i.e. X-ray photoelectron spectroscopy (XPS) and X-ray excited Auger electron spectroscopy (XAES) [1–6]. Such Pd/MWCNT and Pd–Au/MWCNT samples were previously tested for their catalytic activity in the reaction of formic acid electrooxidation [4,5]. The reduced catalysts were active, the Pd–Au/MWCNT catalyst showed higher activity than Pd/MWCNT, whereas the calcined catalysts were inactive. This catalytic activity was attributed to a larger metallic surface area, a thinner palladium oxide and amorphous carbon layers on Pd metallic crystallites due to reduction causing increase of metallic palladium surface area [5–8]. The X-ray PhotoElectron Energy Loss Spectroscopy (XP-EELS) and the Reflection Electron Energy Loss Spectroscopy (REELS) spectra are also of importance in characterisation of surfaces and thin films [9]. The electron passing through a surface layer of a solid interacts with the free (quasi-free) and weakly bound electrons.
Quanta of collective excitations running as longitudinal charge density oscillations through a solid volume are referred to as volume plasmons, whereas along the surface region as surface plasmons. The expression “bulk excitation” denotes any energy loss phenomenon occurring in an infinite medium without any boundary, whereas “surface excitation” denotes excitation due to the presence of a solid–vacuum boundary. The surface modes of the inelastic excitation have a lowered resonance frequency and are orthogonal to the volume modes. The energy loss process involves the energy quanta of the oscillations (plasmons) and their excitation probability. The plasmon energy, Ep = ω, is related to the frequency, ω, of the oscillations, which depends on the density of weakly bound electrons for which the plasmon energy is large in comparison to their binding energy. The electron energy loss spectrum is characterised by the energy loss peaks, i.e. the loss involving only type valence electrons known as free-like electrons and + loss involving all the valence electrons. The energies and intensities (the latter ones measured by the peak areas) of the electron loss peaks (XP-EELS and REELS) characterise the near-surface valence electron density and the structural organisation of C sp2 /C sp3 bonds, which can be also quantified from the equation [10]: sp2 =
∗ Corresponding author. Tel.: +48 22 343 3432; fax: +48 22 632 5276. E-mail address: [email protected] (B. Lesiak). 0368-2048/$ – see front matter © 2012 Elsevier B.V. All rights reserved. http://dx.doi.org/10.1016/j.elspec.2012.11.005
Area /Area+ (Area /Area+ )standard
(1)
with a normalisation to a standard, e.g. graphite. Therefore, fitting of the REELS spectra into the and + (surface and volume)
B. Lesiak et al. / Journal of Electron Spectroscopy and Related Phenomena 185 (2012) 486–491
loss peaks may be applied for characterisation of the surface and volume contributions [9]. Otherwise, the C sp2 /C sp3 content may be evaluated from the C 1s XPS spectra by fitting to C sp2 and C sp3 components and from the energy width of the first derivative of the C KLL XAES spectra (parameter D), using a linear interpolation of the parameter D for graphite (100% of C sp2 ) and diamond (100% of C sp3 ) [11]. The accuracy of this estimation depends on the accuracy of the assumed values of parameter D for diamond and graphite, which vary from 13.0 eV to 14.5 eV (diamond) and from 21.1 eV to 22.8 eV (graphite) [12 and references within]. The research using the REELS spectroscopy may have a broader impact for catalysis due to the possibility of distinguishing the electron signal from boundary from the respective signal from the outer layers since the catalytic reactions proceed at the surface. Such research can be especially useful for characterisation of catalyst performance and catalyst designing. Application of commonly used techniques of surface analysis such as XPS, XP-EELS and XAES provides the surface sensitivity from outer layers and is described by a few parameters, i.e. the inelastic mean free path (IMFP), the mean escape depth (MED), the information depth (ID), etc. where the IMFP is a material parameter dependent on electron kinetic energy (KE) and the other parameters depend on the IMFP and experimental conditions [13]. The above parameters are calculated from the emission depth distribution function (DDF), defining the probability that a photoelectron and/or Auger electron leaving a surface in a given direction originates from a specified depth measured inward along the surface normal. The MED is defined as an average depth normal to the surface from which the electron escapes. The ID is a maximum depth normal to the surface from which useful signal information is obtained, i.e. a thickness from which a specified percentage P (95%, 99%) of the detected signal originates. Neglecting elastic scattering, the following analytical expressions are valid: MED = cos ˛out and ID = cos ˛out ln(1/((1 − P)/100)), where is the IMFP and ˛out is the emission angle of the electron with respect to the surface normal. By analogy, in REELS the depths in the solid for a particular electron trajectory contribution are distributed according to probability density function referred to as a penetration depth distribution function (PDDF). This parameter is defined as a probability that an electron incident on the surface at an angle ˛in be elastically backscattered from a maximum depth, z, and emitted in the direction of the analyser at an angle ˛out and not inelastically backscattered, where ˛in and ˛out are measured with respect to the surface normal. Neglecting elastic scattering the mean penetration depth (MPD) and ID are expressed as: MPD = (cos ˛in cos ˛out /(cos ˛in + cos ˛out )) and ID = (cos ˛in cos ˛out /(cos ˛in + cos ˛out )) ln(1/((1 − P)/100)). Assuming an electron IMFP in graphite for respective electron KE of 272 eV, 1202 eV and 4000 eV as: 0.72 nm (C KLL), 2.10 nm (C 1s) and 5.60 nm (REELS) [14] the above parameters are: MED = 0.72 nm (C KLL), MED = 2.1 nm (C 1s) and MPD = 2.2 nm (REELS), with the ID values for P of 99%: 3.3 nm (C KLL), 9.7 nm (C 1s) and 10.1 nm (REELS). Therefore, the MED given by Auger electrons refers to about 2 carbon walls, whereas the MED and MPD from the C 1s XP-EELS and REELS spectra to about 6 carbon walls. For an overlayer/substrate system the ID may differ depending on the material atomic number, Z, of the overlayer and substrate and their electron elastic differential scattering cross-sections [15] and morphology [16]. The importance of characterisation using the REELS + energy loss spectra is based on the possibility of distinguishing the surface boundary and bulk components between the respective C sp2 /C sp3 contributions. The XP-EELS and REELS spectroscopies are applied for investigating the surface of “as-received” MWCNT, functionalised by oxidation MWCNT (ox-MWCNT), and MWCNT decorated with Pd and Pd–Au particles after different chemical and physical
487
treatments, i.e. reduction, calcination, and Ar+ ion surface cleaning of Pd–Au/MWCNT calcined sample. Structural changes in the XP-EELS and REELS spectra are discussed accounting for the energy positions and the intensity of the quasi-elastic peak, the inelastic and + loss peaks, and the + loss peak surface and bulk contributions.
2. Experimental 2.1. Samples and preparations Commercial MWCNT (Echo-nanobio. Tech. Co. Ltd., Japan) with an average diameter of 50 nm and length of 5–15 mm, were purified and functionalised (65% aq. HNO3 , 6 h, 118oC) [2,17]. Precursors of 17.8 wt% Pd/MWCNT and 8.8 wt% Pd–10.0 wt% Au/MWCNT catalysts were prepared by a polyol method described in details elsewhere [2].
2.2. Apparatus The XPS and REELS spectra were measured in the ultra-highvacuum ESA-31 spectrometer [18] equipped with a high energy resolution electron analyser, an electron gun, an X-ray excitation source and Ar+ ion source. The C 1s XPS, C KLL XAES and XP-EELS spectra were measured using Al Ka X-rays (h = 1486.67 eV), in a fixed retarding ratio (FRR) working mode, at photon incidence and electron emission angles of 70◦ and 0◦ , respectively, with respect to the surface normal. The REELS spectra were measured using the FRR working mode, a primary electron beam with a spot diameter of 1.5–2.0 mm, an electron current of about 5 nA and primary electron beam of 4000 eV KE, the incidence and emission angles of 50◦ and 0◦ , respectively, with respect to the surface normal. The applied measurement geometries resulted from optimisation of experimental signal intensity and apparatus set up. The ion sputtering was carried out using an Ar+ ion beam of 3 keV energy with a current of 30 A for 1 min.
3. Results 3.1. Samples – surface compositions and notations The investigated samples exhibit the following bulk compositions resulting from the chemical synthesis, i.e. 17.8 wt% Pd/MWCNT and 8.8 wt% Pd–10.0 wt% Au/MWCNT, i.e. below 1 at.% of metallic content. The XPS analysis indicated the presence of Au, Pd, C and O at the surface [4,5]. The surface content of Au and Pd resulting from XPS quantitative analysis [4,5], chemical treatments and assumed notations denoting the samples after chemical treatments are the following: (i) 10.6 wt% Pd/MWCNT calcined at 250 ◦ C in Ar: Pd-C250; (ii) 4.3 wt% Pd–4.6 wt%Au/MWCNT calcined at 250 ◦ C in Ar: Pd–AuC250; (iii) 9.1 wt% Pd/MWCNT reduced at 200 ◦ C in Ar and 5% H2 : Pd-R200; (iv) 4.4 wt% Pd–5.0 wt% Au/MWCNT reduced at 200 ◦ C in Ar and 5% H2 : Pd–Au-R200; (v) 5.5 wt% Pd–3.8 wt% Au/MWCNT calcined at 250 ◦ C in Ar and Ar+ ion sputtered in UHV: Pd–Au-C250–Ar+ . The O content for the Pd decorated samples after calcinations and reduction was 11.7 wt% and 7.7 wt%, respectively, whereas for the Pd–Au decorated samples – 6.5 wt% and 7.0 wt%, respectively [4,5].
488
B. Lesiak et al. / Journal of Electron Spectroscopy and Related Phenomena 185 (2012) 486–491
REELS KE=4000 eV 0.09 +
PdAu-C250-Ar
Intensity difference (cps)
0.08 0.07 0.06
PdAu-C250
0.05 0.04
PdAu-R200
0.03 0.02
Pd-C250
0.01 0.00
Pd-R200
-0.01
0
10
20 30 40 Energy loss (eV)
50
60
Fig. 2. Intensity difference between the REELS normalised intensity spectra recorded for Pd and PdAu decorated MWCNT and for the ox-MWCNT.
-3
8.0x10
-4
6.0x10
-4
4.0x10
-4
2.0x10
-4
Ma(d d )/
T
1.0x10
KE=4000 eV emission solid angle
C
Pd
0.0
Au 80
90
3.2. REELS and C 1s XP-EELS spectra The REELS, C 1s XP-EELS and C 1s spectra recorded from the investigated samples are fitted to asymmetric Gaussian–Lorentzian functions using the XPSPEAK4.11 software after linear (REELS) and Tougaard (C 1s XP-EELS) background2 subtraction. Comparison of C 1s XP-EELS and REELS spectra recorded from “as-received”, oxidised and Pd and Pd–Au decorated MWCNT is shown in Fig. 1. The REELS spectra recorded using the primary beam with an electron KE of 4000 eV were not corrected for effects of atomic recoil. No separated components resulting from the electron inelastic loss on metallic Pd and/or Au are shown distinctly in the spectra (Fig. 1a). This is confirmed by the intensity difference between the REELS spectra recorded for Pd and PdAu decorated MWCNT and for the ox-MWCNT (Fig. 2). In a single elastic electron scattering model the elastic backscattering intensity is proportional to a differential scattering cross section for elastic scattering, d/d˝, and inversely proportional to a total elastic scattering cross-section, T , where the elastic cross-sections and their angular dependence are
1
Kwok RWM. XPS Peak Fitting Program for WIN95/98 XPSPEAK Version 4.1, Department of Chemistry, The Chinese University of Hong Kong, [email protected]; http://www.uksaf.org/software.html, XPSPEAK4.1. 2 Tougaard S. Background analysis of XPS/AES. QUASES Simple Background, Ver. 2.2, Copyright 1994–2001; http://www.tougaard.com.
100 110 120 130 140 150 160 170 180 190
Scattering angle,
Fig. 1. Comparison of spectra recorded from “as-received”, oxidised and Pd and PdAu decorated MWCNT. (a) REELS. (b) C 1s XP-EELS.
(deg)
Fig. 3. Comparison of the angular dependence of ratio of differential elastic crosssections to total cross-sections accounting for atomic content of C, Pd and Au, Ma (number of atoms per unit volume) for primary electrons of KE = 4000 eV.
provided by a NIST database [15]. The angular dependence of ratio of differential elastic cross-sections to total cross-sections accounting for atomic content of C, Pd and Au, Ma (number of atoms per unit volume), i.e. Ma (d/d˝)/ T , indicates significant contribution from C in contrary to Pd and Au (Fig. 3). Therefore contribution of Pd and Au to the REELS spectra can be neglected. The REELS spectra were fitted to the quasi-elastic peak, and + loss peaks, with the + loss peak separated into C sp2 and C sp3 surface and bulk components (Fig. 4a and Tables 1 and 2). A comparison of electron energy loss features in the REELS Table 1 Comparison of the electron energy loss features in the measured REELS spectra. Sample
“As-received” MWCNT ox-MWCNT Pd-C250 Pd-R200 Pd–Au-C250 Pd–Au-R200
REELS (primary beam KE = 4000 eV) Energy loss (eV)
Area peak/area total (%)
+
+
Quasi-el. peak
6.1 6.2 6.3 6.2 6.2 6.0
26.2 25.8 25.7 25.8 25.5 25.8
5.3 4.5 6.5 5.2 5.9 5.5
33.7 29.6 29.7 30.5 27.8 29.3
61.0 65.9 63.8 64.3 66.3 65.2
B. Lesiak et al. / Journal of Electron Spectroscopy and Related Phenomena 185 (2012) 486–491
489
Table 2 Comparison of the C sp2 and C sp3 (surface and bulk) contributions (REELS spectra) of the + electron energy loss peaks. Sample
“As-received” MWCNT ox-MWCNT Pd-C250 Pd-R200 Pd–Au-C250 Pd–Au-R200
+ energy loss peaks contribution (%) [energy loss (eV)] → *[13.5 ± 0.5]
sp2 S[18.4 ± 0.3]
sp2 B[27.4 ± 0.4]
sp3 S[23.5 ± 0.5]
sp3 B[33.5 ± 0.5]
→ *>38.0
1.4 1.0 5.3 8.5 4.1 6.9
11.5 9.5 14.1 8.5 12.8 12.0
37.0 37.1 27.1 41.0 25.8 31.0
28.6 26.9 30.1 16.4 29.5 23.3
18.2 21.2 20.3 22.7 23.9 22.9
3.3 4.4 3.1 2.9 3.9 3.9
spectra and these spectra + loss peak C sp2 and C sp3 contributions, → * interband transitions is given in Tables 1 and 2, respectively. Mean percentage deviations between the experimental values and those values obtained from fittings to experimental data range from 5.3% to 24.1%. The loss peak energy (at about 6.0 eV) and the , + loss peaks and quasi-elastic peak contributions resulting from the REELS spectra fitting (Fig. 4a) indicate the changes of the electronic structure in different samples (Fig. 1a and Table 1). Large scatter of the + loss peak energy values for the surface and bulk C sp2 /C sp3 contributions is observed in the spectra recorded from different samples (Table 2). These values for MWCNT and MWCNT decorated with Pd and Pd–Au differ from the values reported for graphite and diamond [9]. The → * interband transition at 13.5 eV may result from an overlap of the respective transitions occurring in graphite at 12.5 eV and in diamond
at 12.8 eV [9]. The surface C sp2 component is observed at 18.5 eV instead at 19.5 eV (graphite), whereas its bulk component is at 27.4 eV instead at 27.0 eV (graphite). However, the + loss peak surface and bulk C sp3 components for MWCNT are observed at similar (compared to that in the diamond) energy distance from the energy position of the elastic peak. The fitting results of the + loss peaks indicate also differences in the intensity of surface and bulk C sp2 /C sp3 contributions (Table 2) and differences in intensity ratios of surface and bulk contributions (Fig. 5). Oxidation of “as-received” MWCNT and reduction of calcined Pd-MWCNT and Pd–Au-MWCNT cause stronger increase of surface C sp3 (Fig. 5a). Also, after the respective oxidation and reduction the surface C sp3 content increases stronger than the overall C sp3 content (Fig. 5b). The C 1s XP-EELS were fitted to and + loss peaks [19] and different C chemical groups applying binding energy (BE) values according to Butenko et al. [20]. An exemplary fitted C 1s XP-EELS spectrum is shown in Fig. 4b, whereas comparison of C chemical group content in Table 3. Mean percentage deviations between the experimental values and those values obtained from fittings to experimental data of C 1s XP-EELS spectra range from 1.9% to 2.7%.
(a)
)y
2
3
)x/ I(
2
2
Pd-Au -C250 Pd-Au -R200
Pd-R200
Csp S/Csp B Pd-C250
0
"as-received" MWCNT oxMWCNT
I(
3
Csp S/Csp B
1
Samples
(b)
3 3
2
)y
2
)x/ I(
Csp S/Csp S
1
3
2
Pd-Au -C250 Pd-Au -R200
Pd-R200
Pd-C250
0
"as-received" MWCNT oxMWCNT
I(
Csp /Csp
Samples Fig. 4. The exemplary spectra and the fitting results to an asymmetric Gaussian–Lorentzian function (C C components) after linear background subtraction. (a) REELS recorded from “as-received” MWCNT. (b) C 1s XP-EELS recorded from PdAu-C250 sample.
Fig. 5. Comparison of the intensity ratios of the C sp2 /C sp3 surface and bulk components evaluated from the REELS spectra. (a) Surface to bulk. (b) Surface C sp3 /C sp2 and C sp3 /C sp2 .
490
B. Lesiak et al. / Journal of Electron Spectroscopy and Related Phenomena 185 (2012) 486–491
Table 3 Comparison of the C group content in the measured C 1s XP-EELS spectra recorded from “as-received”, oxidised and Pd and PdAu decorated MWCNT. C chemical groups content (%) [BE (eV)]
“As-received” MWCNT ox-MWCNT Pd-C250 Pd-R200 Pd–Au-C250 Pd–Au-R200 Pd–Au-C250–Ar+
sp2 [284.4]
sp3 [285.7]
C OH[286.3]
C O[287.7]
C OOH[288.5]
91.9 83.7 85.4 83.4 85.6 83.8 82.2
1.1 6.4 7.5 2.5 5.1 5.5 9.1
4.8 5.3 3.2 8.8 5.8 5.9 0
2.2 2.8 2.5 1.3 2.7 3.2 1.7
0 1.8 1.4 4.0 0.8 1.6 7.0
Similarly to the REELS spectra the C 1s XP-EELS spectra indicate changes of selected features, i.e. and + energy and intensity, depending on the sample preparation and treatment (Fig. 1b). The C group content resulting from the C 1s spectra fitting indicates: (i) increase of C sp3 , hydroxyl (C OH) and carbonyl (C = O) and carboxyl (C OOH) groups after oxidation of “as-received” MWCNT, (ii) increase of carboxyl (C OOH) and hydroxyl (C OH) groups and decrease of carbonyl (C O) after reduction of decorated MWCNT (Table 3). 3.3. C sp2 content The content of C sp2 was evaluated from: 1. the energy width of the C KLL Auger spectra assuming D values of 13.0 eV (diamond) and 22.6 eV (graphite) [11,12]; the mean percentage deviation resulting from a scatter of the assumed D values range from 3.4% to 10.4%; 2. the fitting of Gaussian–Lorentzian to the REELS and C 1s XP-EELS quasi-elastic, and + loss peaks and using Eq. (1), where – as a standard – the “as-received” MWCNT (C sp2 of 91.9% (Table 3)) sample was applied 3. the fitting of C 1s spectra according to BE suggested by Butenko et al. [20] to C chemical groups, i.e. C sp2 , C sp3 , C OH, C O and C OOH (Table 3). Comparison of the C sp2 content derived from the above procedures is shown in Fig. 6. The results indicate qualitative similarity, i.e. (i) increase of C sp3 after oxidation and after reduction, more significantly after oxidation and (ii) significant increase (up to about 50%) of C sp3 after Ar+ sputtering. A content of C sp2 is similar for “as-received” and calcined decorated MWCNT and larger than for oxidised and reduced samples, indicating traces of amorphous carbon resulting from synthesis. A quantitative agreement is observed between the results obtained from fitting of C 1s XP-EELS, REELS and evaluation of the CKLL C 1s fit REELS C 1s XP-EELS
100
PdAu-C250 + -Ar
PdAu-R200
PdAu-C250
Pd-R200
Pd-C250
40
ox-MWCNT
60
"as-received" MWCNT
2
C sp (%)
80
20
Sample Fig. 6. Comparison of the C sp2 percent resulting from evaluation of different spectra.
C KLL spectra width. The content of C sp2 resulting from the fitting of the C 1s spectra into C chemical groups is slightly smaller. Discrepancies seem not to result from differences in the ID, but rather from the systematic error in evaluating of C chemical group content from spectra fitting (Fig. 6).
4. Discussion Previous studies on the investigated Pd and Pd–Au MWCNT catalysts by XPS, TEM and XRD characterised the surface chemical features and metallic phase structural features [4–6]. The surface content of Pd and Au in Pd/MWCNT and Pd–Au/MWCNT after calcination, reduction and Ar+ sputtering was about a half of the content resulting from the chemical treatment [4,5]. The underestimation of the Pd content can be explained by an overlayer of Pd oxide and amorphous carbon on metallic Pd core, attenuating the measured intensity of photoelectrons emitted from Pd. Reduction decreases the content of oxygen [4,5]. For the calcined Pd/MWCNT sample reduction decreases Pd oxide and amorphous carbon overlayer thickness from 10.5 nm to 3 nm [4], whereas after Ar+ sputtering from 10.5 nm to 7.0 nm [6]. For the calcined Pd–Au/MWCNT reduction decreases an overlayer thickness from 4.3 nm to 2.2 nm [5]. The size of Pd and Pd–Au solid solution nanocrystallites is nearly similar, i.e. 4.2 nm and 4.8 nm [4,5]. The intensity of the X-ray photoelectron signal [21], quasielastic peak [22] and consequently energy loss peaks are influenced by the surface roughness. This depends on the angular spread of the local area distribution of slopes and the experimental geometry, i.e. the emission angle with respect to the surface normal. The XPS and quasi-elastic peak measurements and calculations showed that an error induced by a random surface corrugation is acceptably low (up to 10%) for emission angles to 35◦ [21] and 0◦ [22], respectively. The C 1s XP-EELS spectrum of graphite (sp2 ) exhibits the typical electron energy loss feature at 6.5 eV ( electrons) and at 19.5 eV and 27 eV ( + surface and bulk electrons), whereas diamond (sp3 ) shows loss peaks attributed to surface and bulk plasmons at 23 eV and 33.8 eV [9]. In sp2 graphene, the respective and + surface loss peaks for a free standing sheet are observed at lower energies than in graphite, i.e. 4.7 eV and 14.6 eV, approaching the graphite structure for about 10 sheets [23]. For the curved single layer C sp2 structure (onion-like fullerenes and nanotubes) the loss peak is observed at about 5 eV, whereas the surface and volume plasmons are observed at higher energies than in graphene, i.e. 16–19 eV and 23–27 eV, with values inversely proportional to the diameter [24]. The shift of energy loss peak values results from the coupling of tangential (in-plane) and radial (out-of-plane) modes on the internal and external surfaces [25]. The nanometallic particles in the CNT will modify the REELS spectra indicating the significant surface and bulk plasmon features resulting from the metallic component [26]. The following loss peaks may be observed: Pd – 6.5 eV, 16.0 eV, 25.1 eV and 31.9 eV, Au – 3.2 eV, 5.8 eV, 16.7 eV, 24.1 eV and 33.0 eV and Pd69 at.%Au31 at.% – 6.3 eV, 24.3 eV and 32.4 eV [27]. The REELS spectra of the
B. Lesiak et al. / Journal of Electron Spectroscopy and Related Phenomena 185 (2012) 486–491
“as-received”, oxidised and decorated with Pd and Pd–Au particles MWCNT show the and + energy loss values typical for carbon materials with no Pd and Au contribution due to small (below 1 at.%) content of metal [4,5] (Tables 1 and 2 and Figs. 1–4). Changes due to sample preparation and treatment result in variation of energy and intensity of and + loss peaks (Table 1 and Fig. 1) and + loss peak surface and bulk C sp2 and C sp3 contributions (Table 2 and Fig. 5). The surface and bulk C sp2 and C sp3 + loss peak energies do not coincide with the respective energies for graphite and diamond [9]. The larger energy difference between the C sp2 and C sp3 + loss peaks (Table 1), similarly to that between the BE of C 1s C sp2 and C sp3 [20] may result from the MWCNT curvature. Oxidation, reduction and Ar+ ion sputtering cause increase of C sp3 hybridisations (Figs. 5 and 6), at more significant rate for the surface (Fig. 5), indicating that these processes occur at the surface. A quantitative consistency of the results is obtained using different methods, i.e. evaluation of C sp2 from C 1s XP-EELS, REELS, C KLL (Fig. 6). No influence on ID is observed. 5. Conclusions The REELS were applied for the analysis of the C sp2 /sp3 bulk and surface ratio in “as-received”, functionalised and decorated with Pd and Pd–Au particles MWCNT. It is shown that MWCNT functionalisation and reduction of Pd and Pd–Au/MWCNT induce changes in the bulk different from that at the uppermost surface layer. Functionalisation – increasing significantly the surface C sp3 /C sp2 ratio – indicates that the oxidised carbon groups are created mainly at the outer walls of the MWCNT. The less significant increase of the surface C sp3 /C sp2 ratio after reduction confirms that nanometallic particles are placed on the MWCNT inhibiting reduction of oxygen groups at the surface. The proposed analysis provides information on the electronic structure of catalyst surface layers, what is of great importance for a better understanding of the catalytic processes. Acknowledgments This work was supported by the Polish Academy of Sciences and Hungarian Academy of Sciences, Projects of the Polish Council for Science: 15-0011-04/2008, KB/72/13447/IT1-B/U/08. The authors
491
L. Kövér and J. Tóth would like to thank for the support of the project OTKA-K67873. References [1] A. Borodzinski, A. Mikolajczuk, C.H. Chen, W.J. Liou, H.M. Lin, S.H. Wu, P. Kedzierzawski, L. Stobinski, K. Kurzydlowski, Proceedings of EFC2009, Rome, Italy, 2009. [2] C.H. Chen, W.J. Liou, H.M. Lin, S.H. Wu, A. Borodzinski, L. Stobinski, P. Kedzierzawski, Fuel Cells 10 (2010) 227. [3] C.H. Chen, W.J. Liou, H.M. Lin, S.H. Wu, A. Mikolajczuk, L. Stobinski, A. Borodzinski, P. Kedzierzawski, K. Kurzydlowski, Phys. Stat. Solidi A 207 (2010) 1160. [4] A. Mikolajczuk, A. Borodzinski, L. Stobinski, P. Kedzierzawski, B. Lesiak, L. Kövér, J. Tóth, H.M. Lin, Phys. Stat. Solidi B 247 (2010) 3063. [5] A. Mikolajczuk, A. Borodzinski, L. Stobinski, P. Kedzierzawski, B. Lesiak, L. Kövér, J. Tóth, H.M. Lin, Phys. Stat. Solidi B 247 (2010) 2717. [6] B. Lesiak, L. Stobinski, L. Kövér, J. Tóth, K.J. Kurzydłowski, Phys. Stat. Solidi A 208 (2011) 1791. [7] M. Mazurkiewicz, A. Malolepszy, A. Mikolajczuk, L. Stobinski, A. Borodzinski, B. Lesiak, J. Zemek, P. Jiricek, Phys. Stat. Solidi B 248 (2011) 2516. [8] A. Malolepszy, M. Mazurkiewicz, A. Mikolajczuk, L. Stobinski, A. Borodzinski, B. Mierzwa, B. Lesiak, J. Zemek, P. Jiricek, Phys. Stat. Solidi C 8 (2011) 3195. [9] L. Calliari, S. Fanchenko, M. Filippi, Carbon 45 (2007) 1410. [10] A.C. Ferrari, A. Libassi, B.K. Tanner, V. Rtolojan, J. Yuan, L.M. Brown, S.E. Rodil, B. Kleinsorge, J. Robertson, Phys. Rev. B 62 (2000) 11089. [11] J.C. Lascovich, S. Scaglione, Appl. Surf. Sci. 78 (1994) 17. ´ Anal. Sci. 26 (2010) 217. [12] B. Lesiak, J. Zemek, J. Houdkova, A. Kromka, A. Józwik, [13] A. Jablonski, C.J. Powell, J. Vac. Sci. Technol. A 27 (2009) 253. [14] S. Tanuma, C.J. Powell, D.R. Penn, Surf. Interface Anal. 43 (2011) 689. [15] F. Salvat, A. Jablonski, NIST Electron-Elastic-Scattering Cross-Section Database, ver. 3.1, NIST SRD 64m, Gaithersburg, USA, 2003. [16] L. Zommer, A. Jablonski, J. Electron Spectrosc. Relat. Phenom. 150 (2006) 56; L. Zommer, Surf. Sci. 600 (2006) 4735; L. Zommer, J. Electron Spectrosc. Relat. Phenom. 169 (2009) 57. [17] S. Kundu, Y. Wang, W. Xia, M. Muhler, J. Phys. Chem. C 112 (2008) 16869. ˝ Surf. Interface Anal. 19 (1992) 9. [18] L. Kövér, D. Varga, I. Cserny, J. Tóth, K. Tokési, [19] L. Stobinski, B. Lesiak, L. Kövér, J. Tóth, S. Biniak, S. Trykowski, J. Judek, J. Alloys Compd. 501 (2010) 77. [20] Yu.V. Butenko, S. Krishnamurthy, A.K. Chakraborty, V.L. Kuznetsov, V.R. Dhanak, ˇ Phys. Rev. B 71 (2005) 075420. M.R.C. Hunt, L. Siller, [21] J. Zemek, K. Olejnik, P. Kapetek, Surf. Sci. 602 (2008) 1440. ´ [22] A. Jablonski, K. Olejnik, J. Zemek, J. Electron Spectrosc. Relat. Phenom. 152 (2006) 100. [23] T. Eberlein, U. Bangert, R.R. Nair, R. Jones, M. Gass, A.L. Bleloch, K.S. Novoselov, A. Geim, P.R. Briddon, Phys. Rev. B 77 (2008) 233406. [24] L. Henrard, O. Stephan, C. Colliex, Synth. Met. 103 (1999) 2502. [25] O. Stéphan, D. Taverna, M. Kociak, K. Suenaga, L. Henrard, C. Colliex, Phys. Rev. B 66 (2002) 155422. [26] Z.J. Ding, H.M. Li, Q.R. Pu, Z.M. Zhang, R. Shimizu, Phys. Rev. B 66 (2002) 085411. [27] H.A.E. Hagelin-Weaver, J.F. Weaver, G.B. Hoflund, G.N. Salaita, J. Alloys Compd. 393 (2005) 93.